Antibiotic Resistance

Volume 105, Number 2
Introduction: Antibiotic Resistance
Christopher T. Walsh is the Hamilton Kuhn Professor of Biological

Chemistry and Molecular Pharmacology (BCMP) at Harvard Medical
School. For his biography, see p 426.
Gerry Wright is Professor and Chair of the Department of Biochemistry

and Biomedical Sciences at McMaster University in Hamilton, Ontario.
For his biography, see p 529.
One of the central themes of success in human

therapeutics in the 20th century was the discovery
and development of antibiotics and antibacterial
agents, for the treatment of bacterial infections. The
introduction of antibiotics helped drop the death rates
from infectious disease from 797 per hundred thousand in 1900 to 36 per hundred thousand in 1980, a
20-fold improvement. Two lines of chemical investigation proved fruitful: the isolation of natural products with antibiotic activity from microbial sources
and the purposeful synthesis of antibacterial agents
by medicinal chemists.
The first line of discovery, the isolation of microbial
metabolites from Nature, was initiated by Flemmings discovery of a penicillin-producing fungus and
was closely followed by systematic search of antibacterial producing microorganisms by pioneers such as
Dubos and Waksman. This strategy produced many
of the famous classes of antibiotics. These include
both the cephalosporin and penicillin branches of the
-lactams, the aromatic polyketides of the tetracycline class, the aminoglycosides represented by
streptomycin, the polyketide macrolactones exemplified by erythromycin, and the glycopeptides of the
vancomycin and teicoplanin family. The search by
medicinal chemists for antibacterial magic bullets by
synthetic efforts has produced the sulfa drugs, the
dihydrofolate reductase inhibitors, the fluoroquinolones, and most recently the oxazolidinones.
In the course of these discovery efforts, the active

natural products and synthetic antibacterials have
proven to be valuable probes for deciphering the
identity of targets in pathogenic bacteria. Historically, this has turned out to be a target poor therapeutic area with only four robust targets for widely
used groups of antibiotics: bacterial cell wall biosynthesis; bacterial protein biosynthesis; DNA replication and repair; and folate coenzyme biosynthesis.
The golden age of antibiotic discovery in the 20th
century was actually quite short (Table 1). The two
decades from 1940 to 1960 saw isolation of most of
the major classes of natural antibiotics. The sulfa
drugs were introduced in the 1930s and have been
in continuous use for 70 years. The first versions of
the quinolone synthetic drugs were introduced in
1962.
Much subsequent activity has involved refinement
of existing antibiotic structures to deal with the onset
of resistance. For example, the first generation
penicillin V was replaced by second generation versions, methicillin and ampicillin, which in turn were
superseded by such extended spectrum molecules as
piperacillin. A parallel effort at semisynthetic modification of cephalosporins has led from first generation (cephazolin) to second (cefoxatin) to third (ceftriaxone) and now fourth (cefipime) generation versions
where the bicyclic lactam core scaffold is maintained
and the periphery tailored to deal with succeeding
10.1021/cr030100y CCC: $53.50 2005 American Chemical Society

Published on Web 02/09/2005
392 Chemical Reviews, 2005, Vol. 105, No. 2
Editorial
Table 1. History of Antibiotic Classes in Clinical Use

year
1929 (activity)
1940 (purification)
1932
1944
1945
1947
1948
1950
1955
1955
1955
1955
1959
1962
1969
2000
2003
antibiotic
class
penicillin
-lactam
sufapyridine
streptomycin
cephalosporin
chloramphenicol
chlortetracycline
erythromycin
vancomycin
virginiamycin
amphotericin
lincomycin
rifamycin
nalidixic acid
fosfomycin
linezolid
daptomycin
sulfonamide
aminoglycoside
-lactam
phenypropanoid
tetracycline
macrolide
glycopeptide
streptogramin
polyene (antifungal)
lincosamide
ansamycin
quinolone
phosphonate
oxazolidinone
lipopeptide
waves of resistant bacteria. In the macrolide lineage,

the fully natural product erythromycin was tailored
to give the second generation semisynthetic azithromycin and clarithromycin and now the third generation ketolides such as telithromycin.
There was a 38-year interval between introduction
of the fluoroquinolone class of antibiotics in 1962 and
the next new structural class, the oxazolidinones, in
2000.
This innovation gap is directly relevant to the
current situation where the antibiotic cupboard is
rather bare for meeting the challenges of new outbreaks of resistant bacteria.
The pattern of introduction of successive generations of -lactam antibiotics and of macrolides, tetracyclines, and aminoglycosides is strong testimony to
the almost inescapable correlation that introduction
of a new antibiotic into widespread clinical use
induces the rise of resistant bacteria. Given the vast
numbers of bacteria, their short generation times,
and typical gene mutation frequencies of 1 in 107
bacteria, resistance is inevitable. Antibiotics select
for those very rare bacteria in a population that are
less susceptible and allow them to become dominant
in the populations as susceptible bacteria die off. For
all the major classes of antibiotics noted above, both
natural and synthetic, clinically significant antibiotic
resistance has ensued after introduction into human
therapeutic use. The time frame has been as short
as one year for penicillin V and as long as 30 years
for vancomycin. The resistance kinetics reflect a
range of intrinsic and acquired molecular mechanisms and the extent of dissemination of the antibiotic.
Three major mechanisms of antibiotic resistance
reveal a few common themes used by bacteria to fend
off antibiotics. One mechanism is destruction of the
antibiotic by bacterial enzymes, and this is the
quantitatively significant route for disabling -lactams by hydrolysis of the drug lactam warhead. A
second mechanism is bacterial reprogramming of the
antibiotic target to lowered susceptibility, and this
is the path vancomycin resistant enteroccci (VRE)
take to escape vancomycin action. The third major
route, especially prevalent in pseudomonads but
common to all bacteria, is to pump out antibiotics via
natural product
synthetic
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
transmembrane efflux pumps, keeping antibiotic

concentration within the bacterial cell below toxic
threshold concentrations.
Thus, the expectation in the first decade of the 21st
century is that as a new antibiotic is introduced to
combat pathogenic bacteria, resistant organisms will
be selected and will vitiate the utility of that antibiotic. New antibiotics will therefore be required periodically as waves of resistance follow. The anticipated
cyclical need for new antibiotics raises key questions
of where new antibiotics will come from and whether
monotherapy will continue to be an effective practice
for many life threatening infections.
To optimize the discovery and development time
for new antibiotics requires both sources of new
molecules and understanding of the molecular mechanisms of resistance. This thematic issue of Chemical
Reviews collects in one place reviews by experts both
on the mechanisms of action of the major classes of
antibacterial drugs and on the major resistance
mechanisms.
Articles 1-3 deal with cell wall biosynthetic targets, starting with the -lactam antibiotics and then
discussing the glycopeptides of the vancomcyin and
teicoplanin class. The glycopeptides are substrate
binders for un-cross-linked D-Ala-D-Ala termini of
peptidoglycan biosynthetic intermediates. The ramoplanin class of lipodepsipeptide antibiotics also
functions by substrate binding but targets the Lipid
II molecules that act as carriers for the nascent
peptidoglycan units.
Articles 4-7 deal with antibiotics that target
different facets of bacterial protein biosynthesis. The
first two articles in this cluster take up aminoglycosides, which bind at sites on 16S ribosomal RNA on
the small subunit of bacterial ribosomes, and macrolides, which bind to 23S ribosomal RNA near the
peptidyl exit tunnel. Article 6 evaluates the newer
streptogramins and oxazolidinones for their mechanism of ribosome inhibition. Article 7 reviews the
knowledge base for polyketide biosynthesis and combinatorial biosynthesis approaches to novel structural
and functional activities for third generation antibiotics in this polyketide class.
Articles 8-10 address antibiotics that act to interdict information flow from DNA to RNA in bacterial
Editorial
cells, discussing fluoroquinolones that target type II

DNA topoisomerases, antifolates that block provision
of monomers for DNA synthesis, and rifamycins that
block DNA-dependent RNA polymerases.
Articles 11 and 12 take up two variants of peptide
antibiotic classes, the ribosomally generated lantibiotics and the nonribosomal thiopeptide antibiotics.
Article 13 takes a broader look at the biosynthetic
logic for nonribosomal peptide assembly line machinery responsible not only for thiopeptides but also the
glycopeptides and the depsipeptide classes. The
antibiotic class most recently approved by the FDA
is the nonribosomal lipodepsipeptide daptomycin.
Article 14 examines the biosynthesis and mechanism of additional nonribosomal peptide and polyketide natural products, originally identified as antibiotics, but with special potency and utility as
antitumor agents. These include bleomycin, the enediynes, and mitomycin, all of which target DNA.
The last article addresses the pressing and continuing need for new antibiotics by reviewing how
Chemical Reviews, 2005, Vol. 105, No. 2 393
new targets for antibiotics are discovered in bacteria

by genetic and genomic approaches. It also summarizes contemporary approaches for screening of
leads in antimicrobial drug discovery.
This comprehensive collection of reviews on the
major classes of antibiotics in human clinical use
should be a central resource for scientists to grapple
with the questions of how antibiotics work, why they
stop working, and how the molecular insights into
molecules and their targets condition strategies for
much needed new antibiotics.
Christopher Walsh
Harvard Medical School
Gerard Wright
McMaster University
CR030100Y
Chem. Rev. 2005, 105, 395424
395
Bacterial Resistance to -Lactam Antibiotics: Compelling Opportunism,

Compelling Opportunity
Jed F. Fisher, Samy O. Meroueh, and Shahriar Mobashery*
Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556
Received June 3, 2004
Contents
1. Introduction
2. Penicillin-Binding Proteins
2.1. Enzymes of Cell Wall Biosynthesis
2.2. -Lactam-Sensing Proteins
3. -Lactamases
3.1. Overview and Classification
3.2. Class A -Lactamases
3.3. Class C -Lactamases
3.4. Class D -Lactamases
3.5. Class B Metallo--lactamases
4. Other Resistance Mechanisms
4.1. Porin Deletion
4.2. Transporter Expression
5. Envoi
6. Acknowledgments
7. Abbreviations
8. Note Added in Proof
9. References
395
398
398
402
404
404
406
411
412
413
416
416
417
417
419
419
419
419
1. Introduction
The simplistic image of the bacterium as an isolated, planktonic, self-cloning automaton is refuted.
We now recognize bacteria as microorganisms of
enormous diversity and adaptability. They can thrive
under conditions that we regard as extremesin the
absence of oxygen and at high temperatures, to
choose but two examplessand they can adjust with
surprising alacrity to their environment, and to their
circumstance, so as to improve their fitness for
survival.
The focus of this review is that of bacterial biochemical adaptation to a particular circumstance of
profound concern to the human species: that of
bacterial tolerance and resistance to the -lactam
antibiotics.1 The -lactam antibiotic family originally
was limited to the penicillin (sulfur-containing penams) and then the cephalosporin (sulfur-containing
This review is dedicated to our Ph.D. mentors: To Chris Walsh,
on the occasion of his 60th birthday and with appreciation for his
unique ability to inspire and guide self-discovery. To Bill Hase,
for his guidance and creating an environment that expected
perfection in every aspect of scientific investigation. To Michael
Johnston, for his creativity, passionate engagement, and singleminded ability to contribute to the many things that he has tackled
in life.
* To whom correspondence should be addressed. E-mail: mobashery
@nd.edu.
cephems) -lactams but now includes natural and

synthetic monocyclic -lactams, carbapenems, oxapenams, carbacephems, and oxacephems (Scheme 1).
The -lactams are one of the three largest antibiotic
classes (the others are the macrolides and fluoroquinolones).2,3 Mere words cannot properly emphasize
the role that these antibiotics have to the preservation of human health. Nor do words adequately
emphasize the disquieting reality that at the same
moment we profit from the use of antibiotics, that
there is cost and that this cost is inexorable bacterial
resistance.4-8
Bacterial resistance is not a new phenomenon. We
now recognize that resistance is the inevitable consequence of organisms competing for the same ecological niche. Yet it is only during the past 60 years
that resistance has been transformed by man (as the
driving evolutionary force)9 from what might be
reasonably described as stasissbacteria competing
against bacteriasto that of a disequilibrium of chemical warfare2,10-14 where bacteria additionally compete
directly with us. Assuredly, this is a competition with
uncertain outcome. While the phenomenon of bacterial resistance is evolutionarily ancient,15,16 the consequence of this (so very recent) warfare is that of
accelerated dispersion of the mechanisms for resistance across the bacterial kingdom, increasing selection for bacteria that have acquired these mechanisms, and devaluation of our antibiotic armamentarium.
Bacterial resistance mechanisms with respect to
the -lactam antibiotics are divided between those
that occur at the level of primary metabolism (altered
and acquired proteins and enzymes) and those that
occur at the level of secondary metabolism (the
biosynthesis of modified -lactams that are better
antagonists of the altered proteins). While the realization of the outstanding importance of secondary
metabolites as drug templates dates to the moment
of their very discovery by man,17,18 the recognition
that these secondary metabolites occupy a logical
place in the evolution of bacterial resistance is a more
recent consensus, as discussed in several lucid
reviews13,19-23 The parallel medicinal chemistry development of -lactam structure is presented by
Dalhoff and Thomson24 and is not further discussed
here. Our focus is the bacterial response to the
selection pressure exerted by the -lactam antibiotics.
Among the changes that are accomplished are alterations (mutations) in the molecular target of the
10.1021/cr030102i CCC: $53.50 2005 American Chemical Society

Jed F. Fisher is an alumnus of S.U.N.Y. Stony Brook (B.S., under Bill

Fowler) and M.I.T. (Ph.D., under Chris Walsh). Following a postdoctoral
stay at Harvard (with Jeremy Knowles), a faculty position at the University
of Minnesota, and nearly two decades in the pharmaceutical industry (at
Upjohn and Pharmacia), he joined the Mobashery group at the University
of Notre Dame in 2004. His scientific accomplishments include the
collaborative discovery of the utility of coenzyme mimetics to discern
flavoenzyme mechanism, the acylenzyme pathway for -lactamase
substrate hydrolysis and inhibitor inactivation, the autocatalytic pathway
for anthracycline and mitomycin reductive nucleophilic substitution, the
first rational design of an inhibitor of proteinprotein association, and the
virtue of glycosylation to alter peptidomimetic pharmacodynamics.
Samy Meroueh studied chemistry at Wayne State University, where he

received his Ph.D. degree in Physical Chemistry, under the supervision
of William L. Hase. He then joined the group of Shahriar Mobashery,
where he is currently a Walther Cancer Institute Fellow at the University
of Notre Dame in South Bend, IN. He is using a variety of tools, including
computational, calorimetry, and X-ray crystallography, toward understanding
the catalytic machinery of enzymes that are involved in antimicrobial
resistance and the design and discovery of molecules that would target
those enzymes as potential antimicrobial agents.
-lactams, the (ancient) transformation of these

enzymes into families of -lactam hydrolytic deactivating enzymes (the -lactamases), the expression of
protein inhibitors of the -lactamases, the deletion
of porin proteins in the membrane, the acquisition
and activation of efflux exporter proteins, and the
modification of the cell wall to minimize -lactam
antibiotic access to their targets. While the alteration
in the cell wall biosynthesizing targets and the
expression of -lactamases may be regarded as the
primary bacterial defensive measures, none of these
defensive measures is unimportant. Depending on
the bacterium and the particular circumstance of the
-lactam challenge, the bacterium that devises a
successful combination of these responses is the
bacterium that survives. This review covers the
Fisher et al.
Shahriar Mobashery received his undergraduate (1981) and doctoral (1985)

degrees from the University of Southern California and University of
Chicago, respectively. After postdoctoral research (19861988) at Rockefeller University, he joined the faculty at Wayne State University in 1989.
Mobashery and his research group relocated to the University of Notre
Dame in 2003, where he holds the position of Navari Family Professor of
Life Sciences in the Department of Chemistry and Biochemistry.
primary literature of the past 4 years with citation

to the preceding literature via the many outstanding
concurrent reviews on this topic.
Notwithstanding the likely familiarity of the reader
with the -lactam antibiotics as mechanism-based
enzyme inhibitors of cell wall biosynthesis (inhibiting
the -lactam binding protein enzymes) and enzyme
substrates (of the hydrolytic antibiotic-destroying
enzymes, the -lactamases),16 an overview of the
chemistry of the -lactams is essential context. The
point of reference is the eponymous -lactam (that
is, a 2-azetidinone) four-membered cyclic amide ring
of the penicillins and cephalosporins (Scheme 1). The
neighboring carboxylate (on the second ring of these
bicyclic structures) and the acylamino substituent
upon the -lactam immediately (from our contemporary perspective) define these structures as conformationally constrained tripeptides, having an exposed C-terminus and imbued with the capacity to
acylate (with opening of the -lactam) susceptible
nucleophiles. The antibiotic property implies that
these peptidomimetics mimic an essential peptide
motif possessed by the bacteria and engage this
mimicry to the purpose of confounding acylation of a
critical enzymatic target. Indeed, this hypothesis
coincides to the guiding -lactam presumptionsthe
Tipper-Strominger hypothesissthat dates from the
discovery by Tipper and Strominger (and simultaneously and independently by Wise and Park) that
a penicillin-derived entity is irreversibly incorporated
into the transpeptidase and carboxypeptidase enzymes of bacterial cell wall biosynthesis.25,26 Among
the cell wall biosynthetic enzymes an obvious candidate for such interference is the peptidoglycan
cross-linking transpeptidase wherein the peptidoglycan D-Ala-D-Ala terminus is cleaved in the serine
acylation half-reaction (with loss of the terminal
D-Ala) and the cross-link is formed in the deacylation
half-reaction (by acyl transfer to an amine substituent of the neighboring peptidoglycan strand). Should
this enzyme be presented with a substrate mimetic
wherein the initial acylation reaction remains enabled, but the capacity for deacylation is abolished,
Bacterial Resistance to -Lactam Antibiotics

Scheme 1
then the enzyme will fail to complete its catalytic

cycle.27,28 The loss of these enzymatic activities
disrupts the homeostasis of cell wall integrity, leading (through a poorly understood process that ultimately involves activation of cell wall degradative
enzymes, termed autolysins) to lethal cell wall
defects.1,29-31 A bacterium unable to maintain the
integrity of its cell wall will be unable to reproduce
(a bacteriostatic antibiotic) or survive (a bactericidal
antibiotic) wherein the impaired cell wall no longer
contains the osmotic pressure of the cytoplasm.32,33
Within this mechanistic perspective is found the
two important strategies for bacterial acquisition of
resistance to the -lactam antibiotics. If a cell wall
transpeptidase is deceived by a peptidomimetic in the
acylating half-reaction, then resistance can be achieved
by alteration of the transpeptidase such that acylation by the peptidomimetic does not happen. Second,
if a cell wall transpeptidase is unable to complete
deacylation of the peptidomimetic, then resistance
may be achieved by alteration of the transpeptidase

so as to enable this reaction. The first answer is that
of mutated (resistant) -lactam (penicillin) binding
proteins (PBPs). The second answer is that of the
transformation of the PBPs to new, fully capacitateds
often operational at the substrate diffusional rate
constantshydrolytic (wherein water is the acyl acceptor) enzymes, the -lactamases. As these alterations must be accomplished without loss of bacterial
fitness, it may be expected that specific circumstance
will make one the more effective strategy than the
other. This has proven to be so. Both contribute.
However, whereas the altered PBP is generally
regarded as a relatively recent development in Grampositive bacteria resistance, the advent of -lactamases is understood as an ancient evolutionary event
in bacteria resistance. With respect to the last 60
years, it is far less the evolutionary development of
these resistant enzymes as it is their broadened
distribution, which is of immediate concern.
We have posited a relationship between the mechanisms by which -lactams serve as antibiotics and
the primary mechanisms by which bacteria acquire
resistance to these mechanisms in terms of enzymatic
acylation and deacylation half-reactions. Further
development of these concepts requires, however, a
clearer description of these events as chemical reactions. Two reactions are pertinent. Under enzymatic
control each comprises an acylation event (wherein
an enzyme active site serine is involved for both) and
a deacylation event (wherein for the transpeptidase
the incoming nucleophile is usually a lysine, ornithine, or diaminopimelate amino group and wherein
for the -lactamase the acyl acceptor is water). For
the transpeptidase, the net reaction is the crosslinking of the peptidoglycan NAM-pentapeptide, the
major constituent of the cell wall. For the -lactamase, the net reaction is -lactam hydrolysis to the
biologically harmless penicilloate.
An objective of the two following sectionssthe
relationship of the PBPs and -lactamases to bacterial -lactam resistancesis to communicate at the
simplest possible level the operation of these acylation/deacylation reactions at the protein level. This
understanding remains a challenging task. The first
aspect of this task is the comparative transition-state
energies for ordinary (acyclic) amide (or peptide)
cleavage and -lactam opening. As Page discussed,34
the historical presumption that -lactam antibiotic
ability correlates to the release of significant strain
energy upon -lactam ring opening is overvalued.
Rather, the strain energy is modest and such greater
-lactam reactivity as does exist among the -lactam
family members (for example, the Sweet-Woodward
correlation)35 is unlikely to be expressed in the critical
(rate-limiting) enzymatic acylation step.36 Moreover,
the spatial requirements for general acid catalysis
of transient tetrahedral species collapse to the acylenzyme are rather different for a peptide compared
to a -lactam.34 The answers to the critical questions
as to why -lactams successfully inhibit the PBPs and
have become favorable substrates for the PBPderived -lactamases are found in the experiments
that address these questions: How is the -lactam
recognized as a peptidomimetic by both enzyme

classes? How does the -lactam exploit for acylation
the PBP catalytic machinery, for which it is not
intended? How does the resulting PBP acyl-enzyme
resist the catalytic machinery for deacylation (or
transacylation)? How has the -lactamase acylenzyme become fully competent for acyl transfer to
water? The extraction of answers to these questions
requires challenging experimental design. This design would be more straightforward if the mechanistic basis for enzyme catalysis was evident from
protein structure. It is not!34,37-39 Enzyme catalysis
is the subtle orchestration of a panoply of electrostatic
forces, often significantly influenced by distal (nonactive site) changes in protein structure. While we
can visualize atomssthe structures of several PBPs
and of numerous -lactamases are knownswe must
intuit forces and conjecture the transition states that
they stabilize. The relationship of such conjectures
(and transition states) to bacterial -lactam resistance is the objective of the remainder of this review.
Fisher et al.
Scheme 2
2. Penicillin-Binding Proteins
2.1. Enzymes of Cell Wall Biosynthesis
To have survived means to have been opportunistic. Among the survivors, across the eons of time, are
the single-cell microorganisms of the domain bacteria. The seminal observation that a crystal violet
stain is retained by some bacteria, but not by others,
is known now to signify two different exoskeleton
constructs. The positive staining bacteria have a
single multilayered polymericsthat of a cross-linked
peptidoglycansexoskeleton, while the nonstaining
bacteria have a thinner (two and in places three
layers) of a polymeric (and also a cross-linked peptidoglycan) exoskeleton, further surrounded by a
gellike periplasmic layer,32 itself enclosed by a complex (outer) membrane bilayer. Despite this substantial difference, there are at the functional level and
at the molecular level remarkable similarities between the two. The peptidoglycan exoskeleton (termed
the murein sacculus) is durable and elastic, strong
enough to contain the osmotic turgor of the living bacterium yet permitting nutrient access to the porins
and transporter proteins.40 The cell wall componentss
synthesized in the cytoplasm and transported across
the cytoplasmic membrane for polymerizationsof
both are remarkably similar.41 For Gram-negative
bacteria (and for many Gram-positive ones), in the
final cell wall assembly step the D-Ala terminus of
the pentapeptide-functionalized N-acetylglucosamine
(termed N-acetylmuramic acid or NAM, and assembled with N-acetylglucosamine or NAG, as a
NAG-NAM disaccharide repeat) is removed. The
resulting acyl species is then transferred (crosslinked) to an amino group of a neighboring chain,
thereby unifying the peptidoglycan sacculus as a
single polymeric macromolecule (Scheme 2).26 Given
the sophistication of this process (which is also
intimate to the resealing of the sacculus during cell
division), cooperative multienzyme catalysis (that
includes the transpeptidase just described) is implicated. The study of this enzyme ensemble (termed
the divisome when as part of cell division) to bacterial

shape, integrity, and function is of outstanding
scientific importance.26,32,42-45 With regard to antibiotics (not only the -lactams, although these are our
focus), this ensemble is the story of compelling
opportunity and compelling opportunism.
The events that lead to the cross-linking reaction
have been elucidated with the 1.2 resolution X-ray
structure of the acyl-enzyme formed from the Dalanyl-D-alanine carboxypeptidase/transpeptidase from
Streptomyces sp. strain R61 and a unique cephalosporin.33 This cephalosporin (compound 1, Scheme
3) was designed to incorporate components of the cell
Scheme 3
wall into its own structure. As anticipated, compound

1 modified the active site serine, binding between the
all R-helix and R/ domain. The portions of compound
1 that mimic strands 1 and 2 from the peptidoglycan
(see the green- and red-colored portions in Figure 1
and Scheme 3, for example) were found to be oriented
within the active site. The left portion of compound
1 mimics the acyl-D-Ala-D-Ala portion of the first
strand of the peptidoglycan (portion of 1 in green),
while the right portion mimics the approach of the
nucleophilesthe diaminopimelatesfrom the second
Figure 1. (A) Stereoview of the three-dimensional structures of two strands of peptidoglycan bound to the active
site of the D,D-transpeptidase/D,D-carboxypeptidase from
Streptomyces R61 PBP, constructed computationally from
the 1.2 resolution structure for the acyl-enzyme species
with compound 1 (the extension reaches the NAG-NAM
units on the peptidoglycan). The protein is shown in yellow
ribbon representation, while the bound computational
model representing the two strands of the peptidoglycan
are shown in green and red capped-sticks representation.
The blue van der Waals surface defines the active site. (B)
Schematic representing the peptidoglycan from A, showing
the various hydrogen-bonding interactions and color-coded
according to the three-dimensional model.
strand of the peptidoglycan (red portions in structure

2). The acylation is proposed to occur after activation
of the active site Ser62 (located at the N-terminus of
an R-helix, which is expected to modulate the serine
pKa as proposed by Moews et al.46) by the general
base Lys65swhich is in direct contact with Ser62 at
a distance of 3.0 . This mechanism is consistent
with previously proposed mechanisms of the acylation of PBPs, where the universally conserved lysine
acts as the general base, abstracting a proton from
serine, followed by the back-donation of the proton
(from lysine) to the peptide or -lactam nitrogen.47
To provide a more detailed picture of the cross-linking
event, a computational model was constructed from
the high-resolution structure by extending compound
2 to include the full pentapeptide and a NAG-NAM
extension. The resulting model was fully solvated and
energy minimized (shown in Figure 1A). The peptidoglycan strands were found to form a network of
electrostatic interactions (shown in Figure 1B). These
interactions should play important roles in properly
positioning the peptidoglycan strands and for other
important events such as deprotonation of diaminopimelate in the cross-linking event. It is of interest
to note that the three-dimensional model differed
from a previous model48 that used only the apo
PBP2x structure with respect to the location of the
saccharide-binding grooves.
The description of behavior as either moral or
immoral is (primarily) a human characteristic. For
other organisms this distinction is irrelevant, and the
focus of their behavior is survival to the point of
reproduction. Among the bacteria (and fungi, for
which bacteria are a food source) survivalsthat is,
preservation within an ecological nichesrequires
exploitation of vulnerability. In addition, the biosynthetic enzymes of bacterial cell wall biosynthesis are
vulnerable. The basis of their vulnerability (which
is one and the same with that of the bacterium) is
the combination of essentiality and exposure. These
enzymes are located underneath the very cell wall
that they assemble. For the Gram-negative bacteria
these enzymes are either within the periplasmic
space orsfor the most essential of these enzymess
with active sites exposed to the periplasmic space and
a transmembrane domain (with small cytoplasmic
anchor) within the cytoplasmic membrane. Hence,
bimolecular encounter with an inhibitor of these
enzymes requires only the successful passage of
the inhibitorsintermingling with solute nutrientss
through the lipopolysaccharide of the outer membrane into contact with the peptidoglycan surface of
the periplasmic space. While this simple requirement
cannot be underestimated (especially insofar as
antibacterial design and for resistance development)
for the penicillin and cephalosporin -lactams secreted by the biosynthesizing bacteria and fungi
within the niche, the passage and encounter with
these biosynthetic enzymes is straightforward. Astonishingly, each enzyme of the ensemble is capable
of inactivation (via the same mechanism of irreversible acylation) by an appropriately substituted -lactam. The inactivation is facile for susceptible Grampositive and -negative strains and less so for resistant
bacteria. This truly remarkable event is commemorated by historical nomenclature: these enzymes are
collectively the penicillin-binding proteins (or PBPs)
of the bacteria.
The chemically intriguing aspect of this event is
the recognition by each enzyme regardless of the
specific cell wall biosynthetic role. The inescapable
conclusion is a fundamental of homology of structure
and of alignment with the active site. However,
despite this homology, all -lactams do not inhibit
all PBPs, likely due to subtle differences in the active
site and distal regions in the protein. Regardless of
cell morphology (Gram positive or negative) and
regardless of individual specific synthetic function,
these enzymes must possess such similarity as to
implicate a mere handful of, if not a single, ancestral
progenitor(s). As the intact -lactam antibiotic was
recognized by this ancestral enzyme as a mimic of

the peptidic terminus of the peptidoglycan, so too
there is continued recognition of these antibiotics by
the offspring of this enzyme. Once the -lactam
antibiotic is recognized by the PBP, the recognition
culminates in the formation of a stable acyl-enzyme
species. Hydrolysis of the acyl-enzyme ester bond
is slow in PBPs, with half-lifes that substantially
exceed the doubling time of the organism. To appreciate the degree of inefficiency of this step in
PBPs, comparison of the deacylation rate constant
with that of -lactamasessresistance enzymes that
are believed to have descended from the PBPss
reveals that the rate of deacylation in PBPs is up to
6 orders of magnitude slower.49 The irreversibility of
the deacylation step in PBPs is at the root of the
antimicrobial action of -lactam antibiotics.
The basis for these differences must clearly derive
from the differences in the targetsPBP and -lactamasessstructures. The first X-ray diffraction structures of two low molecular weight PBPs were solved
nearly 25 years ago.50-52 Since then only a handful
of additional structures have been solved, despite the
very large number of PBPs that are now known.
Those with solved structures include the low Mr PBPs
from Streptomyces R6153 and Streptomyces K15 (PDB
code 1SKF),54,55 a zinc-dependent PBP from Streptomyces albus G (PDB code 1LBU),50 the PBP5 from
Escherichia coli (PDB code 1NZO),56 and the high Mr
(and -lactam resistant) PBP2x (PDB code 1PMD)48
from Streptococcus pneumoniae and PBP2a from
Staphylococcous aureus.57 A comparison of these
structures reveals similarities but also differences.
The fold of the transpeptidase/carboxypeptidase appears to be conserved among PBPssthe exception
would appear to be the abundant penicillin-binding
protein from Treponema pallidum, Tp47, which has
a unique multidomain fold.58 This unit consists of two
regions: one is an all R-helix, and the second is a
mixed R/ structure consisting of a -sheet that is
flanked on both sides with R-helices. An indication
of their similarities can be gleaned from a superimposition along the common backbone atoms of the E.
coli PBP5 and the Streptomyces K15 PBP that results
in a 1.2 rms deviation. Whereas the function of the
transpeptidase/carboxypeptidase domain among these
PBPs is known, the function of additional domains
that are found in the structures of both low- and highMr PBPs remain unknown. One example, in PBP5
from E. coli the additional unique domain is found
in a nearly perpendicular orientation relative to the
transpeptidase-like domain (that is shown in Figure
2A).59 This domain is composed of two- and threestranded antiparallel -sheets with noticeable hydrophobic properties.56,59 Other examples include two
high molecular weight class B PBPs: PBP2a from
S. aureus and PBP2x from S. pneumoniae. The threedimensional structure of PBP2a (Figure 2B) shows
a non-penicillin-binding domain and an N-terminal
domain whose functions remain unknown.
The similarity in the three-dimensional structure
of the carboxypeptidase/transpeptidase domains of
PBPs is also matched by a high degree of similarity
in the relative position of residues from three highly
Fisher et al.
Figure 2. (A) Stereoview of the active site of the PBP2x

from S. pneumoniae (cyan), the PBP2a from S. aureus
(magenta), the PBP from Streptomyces K15 (orange), the
D,D-transpeptidase/D,D-carboxypeptidase from Streptomyces
R61 (red), and the PBP5 from E. coli (yellow) superimposed
along the R-carbons of the conserved residues (shown in
capped-sticks representation). The cyan, yellow, green, and
white arrows point to the conserved lysine from the SXXK
motif, the serine from the SXXK motif, the lysine/histidine
from the KTS/KTG motif, and the serine/tyrosine from the
SDN/SGN/TXN motif, respectively. (B) Stereoview of the
three-dimensional structure of PBP2a from S. aureus
shown in ribbon representation. The non-penicillin-binding
domain is shown in yellow and green representation; the
yellow domain corresponds to the N-terminal domain. The
structure that is red corresponds to the D,D-transpeptidase
domain. (C) Stereoview of the three-dimensional structure
of PBP5 from E. coli shown in ribbon representation. The
arrow points to the flexible -like loop.
conserved motifs (Figure 2C). The first motif is the

strictly conserved (both PBPs and -lactamases)
SXXK tetrad. The serine corresponds to the amino
acid that is activated to undergo acylation in both
peptidase and -lactamase. It is located at the Nterminus of a helix, which likely modulates its pKa,
thus facilitating its activation as a nucleophile and

nucleofugacity as a leaving group.46 The lysine in this
motif is the general base that activates the serine for
acylation.60 The second conserved motif is the (S/Y)XN tripeptide sequence. This triad is SDN in Streptomyces R61, SGN in the Streptomyces K15 D,Dtranspeptidase, and YSN in the Streptomyces R61
D,D-peptidase. The S/Y residue in this motif is thought
to be required for back-donation of a proton to the
nitrogen atom of the -lactam ring after formation
of the tetrahedral intermediate. From the superimposition of the amino acids within the active sites of
these PBPs (as shown in Figure 2C) it appears that
the position of the hydroxyl (whether serine or
tyrosine) is conserved (comparing the SDN and YXN
motifs). The third conserved motif in PBPs is a KTS
or KTG motif (except in the case of Streptomyces R61
where it is an HT(S/G)). In -lactamases the role of
this lysine (or histidine) is important in modulating
the pKa of the universally conserved lysine that is
two residues downstream of the active site serine.
The overall similarity of the transpeptidase/carboxypeptidase region and the conserved relative
positions of highly conserved residues in PBPs do not
translate into similar catalytic (or functional) behavior. PBP5 of E. coli is a case in point. While most
PBPs have low deacylation rate constants, PBP5 has
an unusually high deacylation rate constant with a
half-life t1/2 < 10 min with penicillin G; this is to be
contrasted to the more typical deacylation rate
constant of 8.3 10-6 s-1 for the acyl-enzyme species
of PBP2a with the same substrate. What makes this
PBP so unusual? Earlier studies had shown that a
G105D mutation reduced the deacylation rate by 30fold.61 The X-ray structure of this mutant did not
reveal the basis for the reduced deacylation59 since
the mutation does not occur in the active site.
However, the X-ray structure of the wild-type enzyme56 revealed disorder at a loop located near the
active site serine. The position of this loop is reminiscent of the -loop (vide infra) in class A -lactamases, based on the superimposition of the TEM-1
-lactamase and PBP5.59 The different conformation
of the loop adopted by the mutant likely contributes
to the deacylation rate difference. This understanding
takes us to the curious event (and a theme of this
review). What are the molecular events that result
in failed recognition of -lactam antibiotics by the
PBPs? Alteration of the PBPs is a dominant mechanism of Gram-positive resistance. For this compelling reason the focus of Gram-positive PBP biochemistry has changed from the historical (the enzymes
of the nonpathogenic Bacillus subtilis)62 to those of
the resistant and pathogenic S. pneumoniae and S.
aureus. How are the PBPs of these resistant pathogens different? Two limiting possibilities exist. The
PBPs are the same but exist in increased copy
number or the PBPs are altered by selection of
mutant variants so as to diminish recognition of the
-lactam without compromise of the peptidoglycan
biosynthetic role. Both processes exist, but it is the
latter that has proven to be the most versatilesand
expandingsmechanism of Gram-positive bacterial
resistance.
The expansion of resistant PBPs is a medical

problem with microbiological (what is the ecological
circumstance where resistance is acquired from one
bacterium by another), molecular biological (what are
the mutations, and what is the genetic basis for their
transfer), biochemical (how do these mutations defeat
recognition of the -lactam as a mechanism-based
PBP inhibitor), and chemical (what will be the design
of new generation -lactam antibiotics effective against
resistant pathogens) manifestations. A decrease in
the spread of antimicrobial drug resistance will
require societal change and scientific discovery in
response to each of these manifestations. Of these
four we will briefly address the microbiological and
molecular biological, and focus on the biochemical.
The topic of the chemical is reviewed elsewhere.24
An excellent point of biochemical entry to the
reality of S. aureus -lactam resistance is Pucci and
Doughertys analysis, by saturating penicillin inactivation, of the PBP distribution and stoichiometry
in susceptible and resistant S. aureus.63 While it is
long known that substituent changes made to the
penicillin and cephalosporin periphery influence relative affinity (specificity) among the penicillin-binding
proteins, by judicious substituent choice and high
concentration it is nonetheless possible to saturate
(to titrate to the point of complete inactivation) the
entire ensemble and so obtain their relative abundance (copies per bacterium). Susceptible S. aureus
contains four PBPs, three of which are high molecular mass enzymes (70-80 kDa) and one of which is
a low molecular mass enzyme (46 kDa). The high Mr
enzymes are PBP1 (approximately 185 enzymes per
cell), PBP2 (460 enzymes), and PBP3 (150 enzymes).
The low Mr enzyme is PBP4 (285 enzymes per cell).
While this is a smaller number of enzymes than is
found in other bacteria species (B. subtilis has 12;
E. coli has 16) the critical biosynthetic enzymes (with
respect to -lactam lethality) for all are the (vide
infra) high Mr enzymes, which are usually bifunctional (one activity is the -lactam-sensitive transpeptidase activity, and the second is a non--lactamsensitive transglycosylase activity). For S. aureus the
bifunctional enzyme target of the -lactams (wherein
the transpeptidase but not the transglycosylase
activity is inhibited) is the PBP2 enzyme.
What is the PBP composition of the -lactamresistant S. aureus? A comparison of the PBP composition of a resistant strain, by saturating penicillin
inactivation, shows that the resistant bacterium
contains the same four PBPs as the susceptible
bacterium (and in the same quantity per bacterium
as the susceptible bacterium) and an additional fifth
enzyme (termed PBP2a).63,64 The PBP2a is present
in substantial quantity (approximately 800 copies per
cell with some variability) and is a low affinity
enzyme with respect to -lactam binding and inactivation. When resistant S. aureus is challenged by
a -lactam antibiotic, the transpeptidase activity of
its PBP2 enzyme is inactivated but the transpeptidase activity of its PBP2a enzyme is unaffected. The
PBP2a performs the transpeptidase function of the
now inactivated PBP2, and the bacterium survives.
Simply stated, the -lactam concentration attained
during chemotherapy is insufficient to inactivate the

transpeptidase activity of this new enzyme. (The
PBP2a contains a second domain that is presumed
to possess a catalytic function which is not transglycosylase activity. The role of this second PBP2a
domain remains unknown.) In the presence of -lactam antibiotics the functioning transglycosylase domain of the PBP2 (its transpeptidase having been
inactivated) works cooperatively with the active
transpeptidase of the PBP2a to maintain the cell wall
integrity of the resistant S. aureus.65,66 Therefore, the
salient issues to the understanding of Gram-positive
-lactam resistance are the circumstance of PBP2a
acquisition, the genetic origin of PBP2a, and the
molecular alteration(s) within the transpeptidase
PBP2a domain that result in a change from high to
low susceptibility to -lactam inactivation.
The appearance of methicillin resistance soon followed (within a year) the introduction of methicillin
in the clinic.67 The mechanism by which the mecA
gene, carried by mobile genetic elements known as
the staphylococcal cassette chromosome mec (SCCmec),68 was acquired by S. aureus remains unknown.
It is suggested that this gene was acquired from
Staphylococcus sciuri (a bacterium found in the gut
of animals) which possesses a close mecA gene
homologue.69 Upon activation of the mecA gene, the
PBP2a protein is expressed. It was shown in the mid1980s that the presence of PBP2a in S. aureus conferred resistance to the clinically used -lactams,70-72
as evidenced by a 500-fold increase in the MIC
(minimal inhibitory concentration) for penicillins.
These bacteria came to be known as methicillinresistant S. aureus (or MRSA). Kinetic characterization of the reaction of PBP2a with -lactams provided
valuable mechanistic information and revealed that
the resistance to -lactams was not merely due to a
large Kd value (a commonly held belief) but to the
acylation rate constant as well.73 The dissociation
constant of the PBP2a-benzylpenicillin complex was
13 mM,74 similar to what is found for other (susceptible) PBPs. A much clearer difference is the apparent
second-order rate constant for acylation of the active
site serine. For the reaction of PBP2a with benzylpenicillin this rate constant is 2-3 orders of magnitude smaller than those found for other high molecular weight PBPs (such as the S. aureus PBP2 and
S. pneumoniae PBP2x).74 The structural features of
PBP2a that are responsible for the poor acylation
were elucidated recently with X-ray structures for the
apo and acyl-enzyme complex of PBP2a with benzyl-penicillin, nitrocefin, and methicillin.57 Comparison of the apo-PBP and acyl-PBP structures revealed noticeable differences. In the acyl-enzyme
structure the CR, C, and O of Ser403sthe acylated
serinesare 1.1, 1.4, and 1.8 away from the same
atoms in the apo-structure, suggesting that the
R-helix that holds Ser403 must undergo a conformational change for acylation to occur.57
Structural modifications in PBPs that result in
increased resistance are not confined to S. aureus.
The three-dimensional structures of two additional
PBPs have been solved by X-ray crystallography. One
is PBP2x from S. pneumoniae. PBP2x is a class B
Fisher et al.
high molecular weight PBP, and its structure was

solved nearly a decade ago.48 The X-ray diffraction
structure of a mutant PBP2x provided the first
glimpse into the structural bases for resistance by
mutation of PBPs.75,76 Two drug-resistant PBP2x
mutants have been characterized. The first X-ray
structure describes the effects of mutations Thr338Ala
and Met339Phe, which along with other mutations
alter the acylation efficiency by 20-fold. Both of these
mutations occur close to the active site Ser-337, the
residue that is acylated by -lactam antibiotics. The
effect of these mutations is attributed to the disruption of hydrogen-bonding interactions between Thr388
and a conserved water molecule. Also, a conformational change that occurs for the 3 strand is attributed to the collective effects of the larger Phe339
side chain and smaller Ala338 chain. These conformational changes enable an alternative conformation
for Ser337 that might be less prone to activation. The
most likely candidate for the base that promotes
Ser337 acylation is Lys340, whose amine is in
hydrogen-bonding contact (3 ) with the serine
hydroxyl. The structure of another mutant of PBP2x
has been recently solved.75 It reveals similar conformational changes as the PBP2x above except that in
this case the effects are attributed to the change in
polarity introduced by Gln552Glu and to a narrower
active site.
PBP5, a transpeptidase from Enterococcus faeciums
a bacterium which incidentally does not produce
-lactamasesshas also been implicated in resistance
to -lactams through either modification or overexpression of the enzyme.77,78 The enterococci are less
virulent than S. aureus and S. pneumoniae but have
become prominent in the clinic due to their increased
levels of resistance to a variety of antibiotics.79
PBP5fm has low affinity for -lactam antibiotics. Two
reasons for this decreased affinity are suggested from
the structure of this protein.80 A glutamate (Glu622)
near the active site may present a steric barrier to
-lactam binding. This interpretation is consistent
with the reduced affinity for benzylpenicillin that
results when the equivalent site in PBP2x is changed
to glutamate.81 Second, an arginine (Arg464) may
interact with its neighbors in the conserved loop
spanning residues 461-465, resulting in a more rigid
cleft that would lead to the reduced affinity of
PBP5fm.80
2.2. -Lactam-Sensing Proteins

Resistance to antibiotics in Gram-positive and
Gram-negative bacteria has manifested itself through
various mechanisms, including the production of
-lactamases or PBPs that are insensitive to the
action of -lactams, such as the case of PBP2a from
S. aureus. In the mid-1980s Lampen and co-workers
observed that the synthesis of the -lactamase from
Bacillus licheniformis 749/I gradually peaks at 1-1.5
h after exposure to a -lactam and decreases slowly
in the following 1-2 h82 (induction in S. aureus is
far more rapid, complete within 11 min83). This
experiment confirmed that -lactamase production is
inducible and implied the presence of a transduction
mechanism. The identification of a membrane-spanning protein, BlaR, from B. licheniformis soon followed. That this enzyme contained a sensor/transduscer domain that is highly similar to the class D
-lactamases,84 a membrane domain consisting of a
four-helix bundle85 and an intracellular domain
containing a zinc ion,84 made it a strong candidate
to carry out the signal transmission events in a
transduction mechanism. The BlaR protein is the
product of the blaR1 gene, which is a member of a
triad of genes from the bla divergon; blaP and blaI
are the remaining genes that encode the effector
protein (-lactamase) and repressor protein BlaI.86
The BlaI proteinsa DNA-binding repressor protein
that is located immediately upstream of the genes
blaP and blaRsblocks expression of both structural
and regulatory genes, including itself.87
The regulation of the production of resistance
enzymes has also been recently studied in S. aureus,
an organism that, in addition to -lactamases, produces a low-affinity PBP, namely, PBP2a, which is
regulated by a similar mechanism that involves a
sensor-transducer (mecR), a DNA-binding repressor
(mecI), and a structural gene (mecA, Figure 3A). It
is worth noting that the mere presence of the mecA
gene is insufficient for expression of resistance in S.
aureus as yet other (and yet unknown) genetic
changes are also necessary.88,89 The bla or mec
regulatory genes regulate production of PBP2a and
-lactamase due to a high degree of homology between the two systems.90 However, the inability of
-lactams to induce PBP2a in S. aureus and the fact
that the blaI/blaR system interacts with the mecA
promoter indicate that this system could also be
responsible for the induction of mecA transcription.91
The sequence of events that lead to expression of the
blaZ gene (for -lactamase) in S. aureus is similar to
that of B. licheniformis: following the binding of a
-lactam to the sensor domain of BlaR, a signal is
transmitted across the membrane and leads to activation and autocatalytic cleavage of the intracellular
zinc-ion-dependent domain of blaR; the activated
metalloprotease either directly or with the aid of
cofactors cleaves the DNA-bound repressor protein
BlaI,92 which is left unable to dimerize and efficiently
bind to its operator for blockage of expression of the
structural genes.90 Autocleavage of BlaR leads to
incapacitation of the protein, which has to be regenerated continuously. The protein presumably expressed by the blaR2 gene is proposed to play a role
in the induction mechanism, but its exact role
remains unelucidated.
Transcription of the genes encoding -lactamases
is set in motion by the binding of a -lactam antibiotic
to the extracellular sensor domain of BlaR. The
kinetics of this process have been characterized for
B. licheniformis and S. aureus. Kinetics of BlaR from
S. aureus with various -lactams shows that a single
acylation event occurs over the lifetime of the organism, making BlaR like a PBP.93 The acylation step
is efficient (second-order rate constant k2/Ks of
104-106 M-1 s-1, while the first-order deacylation rate
constant has an exceedingly slow value of 10-5 s-1.93
The three-dimensional structure of BlaR from B.
Figure 3. (A) Schematic of the various proteins that are

involved in regulation of -lactamase and PBP2a. All
proteins are color-coded based on their secondary structures and shown in ribbon representation. The lipid bilayer
of the inner membrane is shown with green spheres
representing the phospholipid head, and the lines represent
the lipid tail. The sensor/transducer BlaR is anchored to
the surface through a helix bundle, shown in red ribbon
representation, which culminates in the C-terminal metalloproteinase domain, located in the cytoplasm. PBP2a is
shown anchored to the membrane, and a class A -lactamase (BlaZ) is shown unanchored in the periplasm. Shown
in the cytoplasm is the MecI dimer (ribbon) in complex with
its operator DNA (capped-sticks). (B) Close-up depiction
of the complex of MecI dimer with its operator DNA. The
MecI dimer is shown in ribbon representation with the
monomers colored in gray and yellow, while the DNA
oligonucleotide is shown in blue capped-sticks representation with a ribbon along the duplex backbone. The red
arrows point to the cleavage sites in the two monomers.
licheniformis has been recently solved by X-ray

crystallography94 and confirms the postulated high
degree of similarity between the C-terminal domain
of BlaR and the class D -lactamases,84 While the
three-dimensional structure of BlaR is nearly identical to that of the class D -lactamases, the X-ray
structure of this enzyme does not reveal N-carboxylation at the active site lysine,94 an event that is
widely believed to occur in the class D -lactamases.95-97 Previous kinetic studies of the BlaR
protein from S. aureus had found that BlaR from S.
aureus contains an active site lysine (Lys392) that
reacts with CO2 to form a carbamate.93 The authors
of the X-ray structure suggest that the presence of a
threonine residuestwo residues downstream of the
active serine residue where a valine residue would

be usually located in class D -lactamasesis likely
the reason behind the lack of carboxylation of that
lysine residue.94 However, the signature 13C NMR
signal for N-carboxylated lysine in BlaR of S. aureus
has been documented in solution.93
More recently, the structures of the acyl-enzyme
complex of BlaR from S. aureus with benzylpenicillin
and ceftazidime have been independently solved by
X-ray crystallography.98,99 These confirm the similarities in the structures of BlaR from S. aureus to
BlaR of B. licheniformis and the class D -lactamases.
Interestingly, both of these studies found that BlaR
was not carboxylated in the acyl-enzyme complex,
in contrast to the 13C NMR, fluorescence, and mutagenesis studies identifying the carboxylated lysine
in the resting BlaR protein.93 QM/MM calculations
carried out by Birck et al.99 reveal that upon protonation of the carbamate nitrogen in the class D
-lactamases a barrierless decarboxylation occurs.
The authors postulate that the same N-decarboxylation event occurs in BlaR, but unlike the class D
-lactamases decarboxylation results in an inactive
enzyme unable to recarboxylate due to a hydrogen
bond between Lys392 and Asn439. This conclusion
is consistent with the fact that BlaR undergoes a
single acylation event over the lifetime of the organism.
The nature of the signaling event across the
membrane that transpires as a result of -lactam
binding to the sensor domain of BlaR is not well
understood. Golemi-Kotra and co-workers have shown
that binding of the antibiotic to BlaR of S. aureus is
accompanied by significant conformational changes
that likely have a role in the signal transduction
mechanism.93 A recent study of the transduction
mechanism in B. licheniformis did not identify a
conformational change in the C-terminal domain.100
An alternative mechanism was noted based on an
interdomain conformational change in the membrane
protein consisting of the loss of interaction between
the C-terminal domain and an L2 loop of BlaR that
connects two R-helices of the four-helix bundle. It was
shown that in the absence of the antibiotic the sensor
domain and the L2 loop form an interaction. Recent
mutations of highly conserved residues in the L2 loop
appear to be lending credence to this mechanism, as
the organisms exhibited no -lactamase activity,
since the level of antibiotic remained at similar levels
to that of a -lactamase-negative control.98 Whereas
the mechanism that leads to the transduction of the
signal appears to be contentious, the event that
follows is accepted to consist of autocleavage of the
zinc-containing intracellular domain,87 which in the
case of S. aureus is followed by inactivation of the
repressor proteins MecI/BlaI, while in B. licheniformis this process is thought to occur through an
intermediary coactivator.101
The structures of BlaI102 and MecI103,104 have been
recently solved by NMR and X-ray crystallography,
respectively. The BlaI and MecI proteins from S.
aureus share 60% identity and 31 to 41% identity to
BlaI from B. licheniformis.102 The structure of MecI
consists of a dimer in the shape of a triangle (shown
Fisher et al.
in Figure 3B).104 Dimerization occurs at the C-terminal domain, while the DNA-binding domain is located
at the N-terminus (see Figure 3B for structure of
MecI-DNA complex). The topology of MecI follows a
winged-helix architecture103,104 with a helix-turnhelix DNA-binding motif; the second helix of this
motif binds to the major groove of DNA with up to
16 hydrogen bonds and salt bridges (see Figure
3B).104 The high level of conservation of the residues
that form contact between the repressor protein and
the operator DNA in S. aureus and B. licheniformis
suggests that this complexation is likely similar in
BlaI and MecI.104
3. -Lactamases
3.1. Overview and Classification
The -lactamases predate the antibiotic era. The
evolution of these enzymes is presumed to have taken
place in parallel to the biosynthetic steps leading to
-lactam antibiotics.105 Indeed, the first -lactamase
was identified in the early 1940s prior to the first
large-scale use of penicillins in Boston 2 years
subsequent.16 However, extensive clinical use of these
antibioticss-lactams comprise approximately 55%
of all antibiotics used currentlyshas accelerated the
selection process for the emergence of once rare genes
for these antibiotic-resistant enzymes. The rare
bacterium that harbored the gene for a -lactamase
would have had the opportunity to grow unencumbered once the susceptible organisms in a heterogeneous population of bacteria were eliminated in the
course of an antibiotic treatment regimen. In essence,
the less fit bacterium is eliminated by the -lactam
challenge and the resistant organism experiences
unlimited nutritional resources to propagate. The
once rare gene for the -lactamase is amplified.
Sharing of genetic materials among microbial
populations is relatively facile. Genes, often residing
on inherently mobile genetic elements such as plasmids and transposons, are shared not merely members of a given species of bacteria but also among
unrelated genera. The facility of genetic sharing is
underscored by the observation that some organisms,
such as the Streptoccoci, are able to acquire freestanding stretches of nucleic acids (containing entire
genes) directly from the environment. All these
processes have contributed by clinical selection to the
amplification of once rare antibiotic-resistant genes
and their liberal sharing among various bacterial
populations.
The account given above outlines the plausible
events that have given rise to the emergence of the
parental genes for -lactamases. As a consequence
of the inappropriate use of -lactam antibiotics for
the past half a century, especially in the community,
many variants of the parental enzymes have emerged.
This accelerated evolution of the antibiotic-resistant
genes has been abetted by the creative molecular
tinkering of medicinal chemists in the past decades,
the fruits of which are an ensemble of -lactam
antibiotic structures. The dynamics between the
discovery, the creation of new -lactam antibiotics,
and the clinical responses by microbial population to
these developments have been outlined by Bush and

Mobashery.106 In light of the different properties of
these various types of -lactam antibioticssfor example, antibiotics that target different sets of PBPs
or those that have been imparted with resistance to
the action of -lactamasessthe clinical response by
bacteria has been the selection of mutant variants
of -lactamases that often have broadened the catalytic ability of the enzymes. For example, the TEM-1
-lactamase, a plasmid-borne enzyme described first
in 1963, has given rise to 133 variants (as of February
2004).15 While this variety does reflect the successful
therapeutic use of -lactams, it also may be taken
as presaging the obsolescence of these versatile
antibiotics sometime in the future.
Prompted by a critical biochemical requirement,
nature deftly develops catalysts to meet the need.
Despite the outstanding stability of the peptide bond
(half-life of approximately 500 years when uncatalyzed for hydrolysis),107,108 multiple classes of proteases have evolved to hydrolyze this bond, in light
of the central importance of proteolysis to many
biological processes. The same has been true for
-lactamases in the face of the life or death options
to the organisms presented by the antibiotics. There
are four known classes of bona fide -lactamases,
each of which operates by a distinct reaction mechanism.15,105,109,110
While a handful of -lactamases were known in the
early 1970s, the number at the present exceeds 470
(communication by Dr. Karen Bush). Two general
types of -lactamases are known: those that require
the zinc ion for their function, and those that pursue
a transient serine acylation/deacylation strategy (if
the unique -lactamase activity of the T. pallidum
PBP is also found in other organisms, this will yet
be another type of -lactamase as it does not require
a zinc ion nor does it pursue the covalent catalytic
strategy of the serine enzymes).58,111 The widely
accepted molecular classification places -lactamases
into four classes: three serine-dependent enzyme
classes (classes A, C, and D) and one metal-dependent (class B). This classification is not to be confused
with that of Ghuysen for the PBPs in which the two
groups of low molecular weight and high molecular
weight PBPs are divided among classes A, B, and C
(for a total of six PBP classes).
Both PBPs and -lactamases are present in the
periplasmic space of Gram-negative bacteria. In
Gram-positive organisms (which lack the outer membrane) the PBPs are located on the outer surface of
the cytoplasmic membrane and the -lactamases are
either excreted or bound to the cytoplasmic membrane.112 All -lactamases are expected to have
divested completely their ability to bind the peptidoglycan substrate of their ancestral PBPs.14,113 If
not, the opportunity to function as a vanguard
against the incoming antibiotics would be lost (the
structural means to this end was revealed recently
for the class C -lactamases).113 Moreover, this
same conclusion may be intuited from the ability of
many of the class A and C -lactamases to act at
the diffusion-limited rate for their preferred substrates.114,115
A summary of the mechanistic expectations for

-lactamase catalysis is a useful prelude to the
discussion of the relationship between -lactamase
structure and the evolution of function (and mechanism) to confer -lactam resistance. Entry of the
-lactam substrate is guided by relatively long-range
electrostatic attractions between the cationic side
chain of an active site amino acid and the carboxylate
of the -lactam. Positioning of the -lactam then
occurs by shorter range attractive electrostatic interaction involving hydrogen bonding from the protein to the -lactam carbonyl (the oxyanion hole,
formed in the class A -lactamases by the hydrogen
bonding to the -lactam carbonyl by the backbone
amide nitrogens of Ser-70 and Ala-237).116 Appropriately functionalized -lactams (here with special
reference to the bicyclic structure of nearly all of the
antibacterial -lactams) sequester in the active site.
This complex formation results in localized (ground
state) destabilizing electrostatic interactions at the
locus that enable catalysis, compensated by stabilizing hydrophobic interactions elsewhere in the enzyme-substrate complex. The higher carbonyl IR
stretching frequency of the -lactam when it is in the
oxyanion hole, indicative of enhanced reactivity
toward nucleophile addition, exemplifies localized
ground-state destabilization.117 Moreover, within the
complex the total ensemble of the remaining amino
acids are now predisposed to initiate turnover. The
outcome of this positioning is a decrease in the
transition-state energy by the simultaneous operation
of Lewis (clearly exemplified by the zinc atom of the
metalloproteases but other through-space electrostatic interactions as well) and Bronsted catalysis.
The rate-limiting step of nonenzymatic -lactam
hydrolysis is oxyanion addition to the -lactam carbonyl, and there is little doubt that this event is also
rate limiting for hydrolysis by class A -lactamases.34,116
For the serine (classes A, C, and D) -lactamases
(the metallo--lactamases are discussed subsequently)
the critical result of Henri-Michaelis complex formation is to attain such stabilization of the incipient
tetrahedral species so as to enable general base
catalysis for serine addition to the -lactam. Upon
serine oxygen addition considerable basicity develops
on the nitrogen. At the point in the transition state
of (substantial) negative charge transfer (from the
general base, through the serine) to the tetrahedral
species, a concomitant increase in the -lactam
nitrogen basicity is attained as to accept proton
donation to this nitrogen. The source of this proton
is debated. Nonetheless, nitrogen protonation is an
obligatory event for productive tetrahedral collapse.
As noted by Page and Laws,34 oxyanion addition to
the less hindered re face of the -lactam and the
Bronsted general acid N-protonation both must occur
on this same face. This imposes a spatial (stereoelectronic) constraint on the positioning of these two
catalytic groups within the active site (if they are
allowed contact with each other they will simply selfannihilate by proton transfer). An obvious possibility
is the use of the protonated amino acid that was used
as the general base for oxyanion addition (as it is
already on this face and makes the self-annihilation

issue moot). Hydrolysis of the serine acyl-enzyme,
the deacylation step, is energetically less demanding
(a more reactive carbonyl and thus having a diminished need for general acid catalysis in tetrahedral
collapse).
Within the active sites of the serine -lactamases
are, therefore, two ensembles of amino acids. The
first is the catalytic ensemble comprised of a minimum of five amino acids: the serine, the general base
for the serine, the oxyanion hole (two amino acids),
and the cationic recognition site for the carboxylate.
These amino acids are expected to be invariant (or
very highly conserved). The second is the recognition
ensemble (in one form or another, all of the remaining amino acids) that complements the hydrophobic
and hydrophilic segments presented by the remaining structure of the -lactam. These recognition
amino acids will be variable and correspond to the
amino acids that will mutate under selection pressure. For a given -lactamase and given substrate
the complementarity to the rate-limiting transition
state and to overall recognition of the -lactam will
vary. This variability is, of course, quantified as the
unique kcat/Km for the substrate. While this may seem
obvious, there is a less appreciated corollary. The
relative individual role played by any member of
these ensembles during catalysis (measured, say, in
terms of an amino acid pKa) is therefore substrate
dependent. Deletion of an essential amino acidswhile
an essential tool of mechanistic enzymologysalters
the energetic landscape of the active site in such
fashion as to make all subsequent interpretation
cautionary. The criteria that determine which amino
acids of the recognition ensemble transform under
selection pressure are straightforward. The integrity
of the catalytic ensemble must be preserved, and the
integrity of the protein as the whole (for example, as
measured in terms of thermal stability or expression/
folding capability) can only be lightly varied.118-121
Likewise, mutations that require only a single nucleotide change and that preserve common codon usage
are more probable than those that do not.120 This
implies that certain -lactamase families are anticipated to be more plastic than others, and indeed,
the serine -lactamases provide such a contrast with
the great phenotypic diversity of the serine class A
TEM enzymes as compared to the class A SHV
enzymes, which have diversified but without substantial phenotypic evolution.15 Given these boundary
conditions and superlative kinetic and structural
data for these enzymes, one might presume that the
assignment of function within the catalytic ensemble
and within the recognition ensemble as these develop
under selection pressure would be straightforward.
One would be wrong.
3.2. Class A -Lactamases

This is the largest and best mechanistically characterized serine -lactamase class. Historically, these
-lactamases were described as penicillinases as
their ability to catalyze penicillin hydrolysis was
greater than that for cephalosporins. They have
become so efficient at their function that they are
Fisher et al.
diffusion controlled, where the apparent second-order

rate constant kcat/Km has reached the (upper) diffusion limit estimated from collision theory. As a result,
class A -lactamases may be described as having
reached catalytic perfection for their preferred substrates.114 Variants with much broader substrate
preferences are now known, including enzymes imparting clinical resistance to late generation -lactams.15,109,122-124 The class A -lactamases are closely
related in sequence to low molecular weight class
C PBPs such as PBP4 of E. coli, H. influenza, and
M. tuberculosis.60 As judged by the comparison of
crystal structures, the catalytic domain of the larger
E. coli PBP5 (low Mr PBP class A) shows high
similarity as well.56,125 In terms of bacterial resistance, three class A -lactamases subclasses dominate: the (historically Gram-negative plasmid penicillinase) TEM/SHV, the P. aeruginosa PER/OXA/
TOHO cephalosporinases, and the CTX-M (NMC-A)
carbapenemase subclasses.126 As of October 2004, 135
TEM and 57 SHV -lactamase variants are known
(http://www.lahey.org/studies/webt.htm). While the
sequence homology among the three is easily recognizable and the fundamental catalytic mechanism for
each is the same, the differences render broad structural and mechanistic generalizationssespecially as
they relate to resistance developmentsunwarranted.
Several class A variants resist inactivation by the
mechanism-based inhibitors (clavulanate and sulbactam) used in clinical formulations with otherwise
-lactamase-susceptible penicillins. Until very recently the occurrence of these inactivation-resistant
class A -lactamases was limited to the aforementioned TEM subclass, and hence, the term inhibitorresistant TEM (IRT) was coined. However, in light
of the discovery of inhibitor resistance in the SHVtype enzymes, this group is better referred to as
inhibitor-resistant -lactamases. Furthermore, new
class A -lactamases that are active against the more
recent cephalosporins (ceftazidime and cefotaxime
and the monobactam aztreonam) and others that are
active against the carbapenems are known collectively (also with other class C and D enzymes) as
expanded-spectrum -lactamases (ESBL).124
The crystal structure of several Gram-positive
(including enzymes from B. licheniformis and S.
aureus) and Gram-negative (from E. coli) class A
-lactamases were solved in the late 1980s and early
1990s.127-131 The TEM-1 -lactamase structure has
two domains: an R/ domain consisting of fivestranded -sheet and three R-helices and an R
domain consisting of eight R-helices.128,131 Together,
these two domains sandwich the core of the active
site. The class A -lactamases reveal striking similarities to the PBP structures132 (compare Figure 4A
and B, D, and F). Further similarities can be found
in the relative three-dimensional position of several
highly conserved residues, including Ser70, Lys73,
Lys234, and Ser130. While some of these residues
are not invariant, their substitutes contain similar
functional groups capable of carrying out the same
functions (for example, Ser130 is typically replaced
by tyrosine in the class C -lactamases, and Lys234
is sometimes replaced by a histidine as in the
Figure 4. Stereoviews of the three-dimensional structures of (A) a class A -lactamase (TEM-1; PDB code 1TEM), (C) a
class C -lactamase (AmpC; PDB code 1FCO), (E) and a class D -lactamase (OXA-10; PDB code 1K57). Close-up stereoviews
of the active sites of the acyl-enzyme complex are shown as (B) TEM-1 with 6R-hydroxymethylpenicillate, (D) AmpC with
moxalactam, and (F) OXA-10 with 6-(1-hydroxy-1-methylethyl)penicillanic acid. The enzymes are in yellow ribbon
representation with a van der Waals surface in blue for the active site. The important active site residues are depicted in
capped-sticks representation (color-coded according to atom type: S, yellow; O, red; N, blue; C, white). The hydrolytic
water molecule is shown as the red sphere. Hydrophobic residues in the active site and other residues that are close to
important residues but are not directly involved in the catalytic process are shown in orange capped-sticks representation.
structure of the Streptomyces R61 D,D-peptidase/

transpeptidase).
The remaining invariant amino acid is Glu166.
This glutamate is located on a loop (termed the
-loop) that sequesters, by hydrogen bonding, a
single water molecule as a bridge to the Ser70
adjacent to the re face of the -lactam carbonyl. This
glutamate has been proposed by some to have a role
in the rate-limiting acylation step.133-137 This mechanism envisions that this glutamate, acting through
the bridging water as a proton shuttle, activates
Ser70 for nucleophilic addition to the -lactam.136,138
The alternative possibility for the general base, the
invariant Lys73,129,131 must then be electrostatic
stabilization (such as to increase the Ser70 acidity
and hence nucleophilicity). The evidence in favor of
Glu166 as the serine-activating general base is
summarized. A compelling argument in favor of a
Lewis acid, rather than a Bronsted base, role for
Lys73 was the 15N NMR determination of its pKa as
greater than 10.139 This determination was followed
by Poisson-Boltzmann electrostatic calculations by
Lamotte-Brasseur et al. supporting the pKa > 10
assignment,140,141 although others (using the same
calculation method) estimated a value of 8.142 More
recently each of three methodssenzyme kinetics, 15N
NMR, and free-energy calculations using the ther-
modynamic integrationssupports a pKa value for

Lys73 of 8.0-8.5.143 Ultrahigh-resolution crystal
structures of the TEM-1133 and SHV-2136 enzymes
show a spatial arrangement of Glu166 and the
invariant water, in the presence of a bound transition
state mimic, to be consistent with serine activation
via the proton shuttle.133 The TEM-1 structure resolves a hydrogen atom on Glu166, while the SHV-2
structure shows the hydrogen of the Ser70 hydroxyl
pointing to the conserved water molecule. Following
serine addition the Ser130 hydroxyl is positioned
ideally to shuttle the proton on Lys73 to the -lactam
nitrogen of the tetrahedral species (Figure 4B),
driving ring opening to the acyl-enzyme species. It
is widely accepted that Glu166 is the general base
for hydrolysis of the acyl-enzyme ester. In fact, sitedirected mutagenesis of Glu166 (E166A) abolishes
deacylation while impairing (but not abolishing) the
acylation process by a factor of 103.127,137
Over the past three decades several strategies have
emerged, in the guise of new -lactams, to incapacitate the class A -lactamases. The first strategy is
exemplified by clavulanate and the penam sulfones
(sulbactam and tazobactam), which are poor PBP
inactivators but excellent -lactamase inactivators.144
The key mechanistic event for both is a quite similar
fragmentation reaction of the respective serine acyl-
enzyme intermediates that is competitive with hydrolytic deacylation and gives a new acyl-enzyme
intermediate improperly positioned for catalytic
deacylation.145-150 As noted previously, these -lactams are formulated with other -lactam PBP inactivators to target -lactamase-expressing pathogens.
The second strategy is exemplified by the carbapenems (such as imipenem) and the cephamycins (exemplified by cefoxitin), which resist -lactamase
hydrolysis by diminished ability at acylation and/or
(especially) deacylation events of -lactamase catalysis. Both of these -lactams have unusual -lactam
substituents that are believed to interfere with the
proper positioning of their -lactam segments when
in complex with the -lactamase. As these structural
features do not interfere with PBP affinity, these are
used therapeutically as single agents. The third
strategy is that of the third- (and fourth) generation
cephalosporins (exemplified by cefotaxime, ceftazidime, and cefepime), which are highly functionalized
cephalosporins that are poorly recognized by the class
A -lactamases. These too are used as single-agent
therapies, although this may change. With respect
to -lactam antibacterial design, the structural and
mechanistic basis for the evolution of -lactamases
that have overcome these barriers and now recognize
these -lactams as substrates is a topic of more than
idle curiosity.
In the event the acquisition of clavulanate and
penam sulfone inhibitor-resistant TEM -lactamases
is accomplished by single-point mutations at one of
several key amino acids,146-148,151-156 a compensatory
second point mutation, unrelated to resistance development but rather to restore enzyme stability,119,154 is also seen in some clinical isolates. The
relative ease of this transformation may be understood in terms of the required clavulanate (or penam
sulfone) acyl-enzyme fragmentation as an offpathway event, unrelated to normal catalysis, and
hence easily disposed.155 Moreover, it is evident from
the kinetic properties of the enzymes that incremental adjustment of the kinetic parameters suffice to
impart resistance to these inactivators. For example,
N276D mutation of TEM-1 is representative of common clavulanate-resistant IRT variants wherein the
clavulate Ki increases from 0.4 (TEM-1) to 17 M
(N276D TEM-1) and kcat increases from 0.02 to 0.16
s-1.151 The other kinetic parameters (kinact, krec) are
less altered. The critical fragmentation event in
clavulanate inactivation of the TEM -lactamase is
known to require a protonation event wherein the
proton is provided by a conserved structural water.146
Replacement of the neutral Asn276 with the charged
Asp276 results in substantial movement of the Asp
side chain so as to engage the Arg244 guanidinium
that is ordinarily involved in substrate carboxylate
recognition. The resulting electrostatic modulation
manifests in dissociation (and thus loss) of this
conserved water. Very similar kinetic changes are
seen with respect to clavulanate inactivation of the
M69L TEM-33 variant.152 This methionine, while
clearly not a conserved TEM residue, is nonetheless
located in a region of strong structural constraint (at
the beginning of the TEM H2 R-helix and in contact
Fisher et al.
with the B3 and B4 -strands and thus while

removed from the active site influences the active site
geometry).152,154 Replacement of the methionine with
leucine gives a -lactamase (the TEM-33 enzyme)
having an identical ability to hydrolyze penicillin G.
Clavulanate, however, binds more poorly (by 1.9 (
0.2 kcal mol-1 for the pre-acylation complex).152 An
explanation for this difference is not evident from the
protein structure but is suggested by computational
analyses that indicate less favorable van der Waals
and electrostatic energies for clavulanate binding to
the M69L mutant. Conversely, improved clavulanate
inactivation is seen for the (clinically not observed)
M69G TEM-1 mutant.154 Yet another mechanism by
which the TEM enzymes evade clavulanate has been
suggested to involve disruption of the active site
interactions of the Ser130 and whose oxygen engages
customarily in postfragmentation cross-linking to the
clavulanate (and penam sulfone) remnant acylenzyme.154 For both the TEM-32 (M69I/M182T) and
TEM-34 (M69V) variants the local environment of
this serine is perturbed such that the cross-linking,
leading to a long-lived acyl-enzyme, does not occur.154 An evaluation of Ser130 SHV mutations
identified only S130G as conferring clavulanate
resistance, again resulting from destabilization of the
clavulanate pre-acylation complex.149 It is evident
that the basis for the evasion is not loss of the ability
to cross-link157 but rather the simple result of an
overall diminished affinity for these variants to bind
these inactivators (even when this is accomplished
at the cost of loss of catalytic function, as measured
by kcat/Km, toward -lactam substrates).121,158 Last,
the ESBL -lactamases generally retain susceptibility toward clavulanate and penam sulfone inactivation (in contrast to the IRT enzymes), and thus, the
recent generation cephalosporins may eventually be
combined with these inactivators in clinical practice.159 This likelihood is also reflected in the continuing interest in other -lactam templates, such as the
6-methylidene penam sulfones and penems,160,161 that
have a broad-based ability to inactivate -lactamases
by acyl-enzyme fragmentations and nucleophile
additions.
Among the most common of the IRT TEM variants
are those with replacement of arginine-244 (alone
and in cooperation with other mutations).158 Arg244
is a conserved residue of the 4 strand of the parent
TEM -lactamase. Its replacement by serine gives the
TEM-30 () TEM-41)/IRT-2 enzyme, by cysteine the
TEM-31/IRT-1 enzyme, and by histidine the TEM51/IRT-15 enzyme. As this arginine is an active site
residue and as its replacement dramatically alters
the substrate specificity of the enzyme (exemplified
by a greater than 10-fold decrease in kcat/Km for
penicillin substrates), considerable effort has been
made to identify its role in catalysis. This effort is
further driven by the unique properties acquired
upon replacement of this arginine by these three
amino acids (the TEM-30, -31, and -51 -lactamases
are virtually identical).162 The most important of
these properties is resistance to the TEM inactivators
clavulanate, sulbactam, and tazobactam, accomplished by perturbation of the partitioning of the
acyl-enzyme between hydrolysis (where there is

little change in normal turnover) and the slower
fragmentation and cross-linking (where this inactivation event is even further suppressed).145,154,163 Two
possible roles for Arg244 in normal substrate turnover have been proposed; both are consistent with
the alteration in the steady-state kinetics (decreased
kcat/Km). The first possibility is that the arginine side
chain participates, with a highly ordered proximal
water, in -lactam substrate recognition and active
site orientation via electrostatic pairing with the
carboxylate substituent, which is found in all bicyclic
-lactam antibiotics.155,164 The likelihood of such an
interaction is well substantiated both by active site
simulation and by crystallography.118,119,154 The second proposal is that Arg244 facilitates turnover by
assisting, now via electrostatic repulsion, in product
dissociation.165 These two possibilities are not mutually exclusive. With respect to resistance, the paramount question is clearly that of the structural
consequence of arginine replacement and the correlation of that consequence to the mechanism of
clavulanate and penam sulfone TEM -lactamase
inactivation. The answer, it appears, is the pivotal
role of the proximal (to the arginine guanidinium)
water that is lost to the active site when this arginine
is replaced.146,154 This water molecule provides the
critical proton catalyst necessary to the fragmentation event of the clavulanate acyl-enzyme. Upon
arginine replacement this water molecule is lost and
these inactivators of the parent TEM become ordinary substrates for these IRT TEM variants. Other
IRT variants (such as occur at Met69) that retain this
arginine are nonetheless also able to resist these
inactivators by perturbing the second residue, Ser130,
critical to this fragmentation (and its sequelae).
With respect to resistance development, a particular objective is the understanding of the relationship
between the mutations securing the IRT class A
variants and the mutations (such as the R164S) that
secure to the class A -lactamases the ability to
hydrolyze late generation cephalosporins (the class
A ESBL).166-170 The consensussfor the momentsis
that these two phenotypes are mutually exclusive.
For example, the TEM -lactamase double mutant
(R164S, R244S) retains clavulanate resistance but is
no longer capable of ceftazidime hydrolysis.166,168 The
obvious possibility identified by this observation that
inactivator/late generation cephalosporin combination therapy might prove clinically advantageous is
now in the process of preliminary evaluation.159
The outstanding features of the carbapenem (e.g.,
imipenem) and cephamycin (e.g., cefoxitin) classes of
-lactamase inhibitors are the respective 6R-hydroxyethyl and 7R-methoxy substitution of these -lactams.
The potential for these substituents to interfere with
nucleophile approach to the -lactam carbonyl is
immediately evident, and this hypothesis is proven
especially with respect to catalytic deacylation.171 Of
the two, cefoxitin is the more straightforward. It has
a standard cephalosporin scaffold but with an unusual 7R-methoxy substituent in a position occupied
customarily by a hydrogen in the cephalosporins.
Thus, cefoxitin engages many of the standard recog-
nition features (albeit abnormally) for class A -lactamase substrates. The critical mechanistic event
occurs upon serine acylation. In the cefoxitin acylenzyme intermediate the 7R-methoxy not only displaces the catalytic water172,173 but also interferes
with the Asn132 side chain. This side chain is
compelled to move from its customary location (where
with normal substrates it is engaged in a hydrogen
bond with the substrate 7-amide).173 This movement
alters the cefoxitin 7-side chain in such a way as to
induce further active site distortion, especially of the
-loop. The cumulative effect is a remarkably stable
acyl-enzyme. Subtler (but no less complex nor less
intriguing) mechanisms operate for the class A carbapenemases. As the dominant mechanism for carbapenem resistance in pathogens involves the acquisition of class B or metallo--lactamases, however,
the enzymatic basis for resistance due to expression
of a modified class A -lactamase has been less well
studied. Yet there is no doubt that the new mechanism coincides, in part, with a stunning structural
transformation of the TEM peptide, the introduction
of a second disulfide bond (Cys69-Cys238). The
presence of this cystine is intimately related to the
acquisition of the carbapenemase activity.174-176 The
role(s) that this insertion has with respect to the
mechanism has only begun to be understood. To
begin with, for the E. cloacae NMC-A enzyme a
nearly 100-fold diminution in the ability to hydrolyze
penicillins but a 100-fold improvement in the ability
to hydrolyze imipenem is seen.177 Crystallographic
inspection of a stable acyl-enzyme species shows
that the positioning of the acyl-enzyme is very
similar to that seen for TEM acyl-enzymes (notably
normal oxyanion hole hydrogen bonding) but importantly a repositioning of Asn132 so as to open the
active site for efficient catalytic delivery of the
deacylation water. That even further structural and
mechanistic accommodation may be anticipated is
suggested by the structure of the related (70%
sequence identity to NMC-A) class A SME-1 carbapenemase (and which also shows impaired penicillin
hydrolysis). In the resting enzyme the presumptive
acylation/deacylation general base Glu166 (on the
-loop) hydrogen bonds directly to Ser70 (without a
water bridge).175 This suggests that the role of the
second cystine is to enable an alternative approach
(evading the steric barrier of the 6R-hydroxyethyl
substituent) of the serine in the acylation halfreaction.
The remaining aspect of class A -lactamase evolution as it relates to the acquisition of -lactam
resistance by bacterial pathogens is the ESBL enzymes.15 Following the clinical appearance of the
third-generation oxyimino-cephalosporins (ceftriaxone, cefotaxime) some two decades ago, resistant
bacteria appeared. The basis for the resistance was
the acquisition and dissemination of class A, C, and
D -lactamases capable (often with a narrow substrate spectrum) of hydrolysis of these oxyiminocephalosporins. Three related class A groups are
pertinent: the TEM and SHV ESBL variants initially
among the Enterobacteriaceae but increasingly among
the Pseudomonas;122 the VEB and PER variants
among the Pseudomonas; and the CTX-M variants.123

Of these three the CTX-M are of greatest concern.
These enzymes, formerly most commonly found in
nosocomial pathogens, are now found in community
strains (Vibrio, nontyphoid Salmonella, and Shigella). Moreover, the recent D240G CTX-M variants
show improved ability to hydrolyze ceftazidime.123
For the moment these enzymes remain capable of
inactivation by clavulanate, penam sulfones, cefoxitin, and imipenem, but there is no reason to believe
that this will not change. For this reason evaluation
of the structural basis for the acquisition of the
oxyiminocephalosporins as substrates by all of these
enzymes is a major current focus of -lactamase
enzymology.
The efforts concerning the simplest class A transformation into an ESBLsthat of the SHV (and TEM)
single G238S (SHV-2) point mutationsare instructive as to the difficulties inherent to such an understanding. This single change results in an MIC
increase (E. coli) from 2 to 8 g mL-1 for ceftazidime
and from 0.125 to 16 g mL-1 for cefotaxime (comparing to E. coli with the parent SHV-1 -lactamase,
for which neither oxyiminocephalosporin is a substrate).178 The G238S change gives a 4-fold improvement in kcat/Km for a typical cephalosporin (cephaloridine). Moreover, this mutation is exceptionally well
expressed, suggesting a basis (also with the optimal
kinetics) for the clinical selection of Ser238 as acquisition of (albeit weaker) ESBL activity occurs for
other G238 mutants (including G238A, G238N,
G238M, G238C, and G238I).120 An analogous transformation (appearance of cefotaxime and ceftazidime
as substrates) is seen for the G238S TEM mutant.179
The structural basis for this transformation (after
speculation, as summarized by Hujer et al.120) was
resolved by crystallography.136,180 Glycine 238 is
located on the b3 -strand, in close proximity to the
Asn170, Met69, and Ala237 residues of the active
site. Replacement by serine results in a significant
conformational alteration spanning the 238-242
positions but with overall preservation of R-carbon
positions elsewhere in the enzyme (especially those
of the -loop). The deformation that results from the
conformational alteration is borne fully by the b3
-strand, opening the distance between the lower
(active site) portion of this strand and the -loop by
nearly 3 .136,180 A hydrogen bond from the Ser238
hydroxyl to the main-chain carbonyl of the -loop
Asn170 is seen, which may be presumed (from its
orientation and distance and relatively poor solvent
accessibility) to be of sufficient stability as to be
unlikely to engage bound substrate. The most consistent explanation for the acquisition by this enzyme
of the oxyiminocephalosporins as substrates is that
the serine opens the active site just enough to
accommodate the larger mass of the oxyimino side
chain while preserving the orientations and roles of
all of the catalytic residues.
It is probable that a similar process (shape-selective expansion of the active site) is operative for the
other class A ESBL enzymes, although the process
(in terms of protein adjustment) differs for each. The
P. aeruginosa PER-1 enzyme, for example, is char-
Fisher et al.
acterized inter alia by an altered -loop,126 and the

P. vulgaris K1 enzyme lacks the Arg244 customarily
thought to be involved in substrate recognition and
has atypical residue replacements (notably a Ser237)
that may have specific roles in substrate recognition
or protein adjustment.181 Extensive evaluation of the
CTX-M Toho-1 structure by Shimamura et al.182
(acyl-enzymes with the E166A defective mutant)
and Ibuka et al.183 (apo-enzyme) demonstrates that
similar accommodations operate to embrace the oxyimino cephalosporins as substrates in this class A
enzyme.
These observations emphasize the remarkable
ability of bacteria to alter protein structure, under
selection pressure, to acquire new function. While the
perfect -lactamasesone that hydrolyzes all -lactam structures regardless of their substitutionshas
yet to be encountered, it is not necessary for such an
enzyme to be created in order to attain this high level
of resistance. As will become evident from the following sections, it is only necessary for the bacterium
to acquire a selection of complementary resistance
mechanisms. The CTX-M Toho-1 enzyme, for example, is quite susceptible to inactivation by cefoxitin. The bacterium does not need for this enzyme to
evolve to this function as the alternativesacquiring
a second -lactamase with this functionsis operationally more facile. Hence, the value of these exhaustive efforts to understand the relationship between structure and function of these -lactamases
is strategic: as the characteristics of the enzyme and
the design limits for its evolutionary adaptation are
better understood, the design of improved drug
structure (or the use of a new drug regimen) is made
possible.
A new and encouraging event (where there are not
many), with respect to our understanding of this
structure-function relationship, is the ability to
quickly assess the capability of -lactamases (not just
class A) to acquire new -lactams as substrates. In
the years following the classic experiment of Hall and
Knowles184 with the TEM -lactamases, which proved
that such a capability existed, new methods have
been developed to critically assess the structural
outcome on the -lactamase from the selection pressure exerted by a particular -lactam. Using DNA
shuffling (and E. coli hypermutator expression) Stemmer et al. isolated the triple-point mutation (E104K/
M182T/G238S) TEM-52 variant with ESBL cefotaxime activity.180,185 By the use of a highly error-prone
DNA polymerase, Camps et al.186 isolated three TEM
mutants (E104K/R164S, E104K/R164H/G267R, and
E104K/R164S/G267R) conferring a >50-fold resistance phenotype to aztreonam, a monobactam that
is a very poor substrate of the parent TEM-1 -lactamase. Two of these point mutations are known, but
the third (G267R) is new. Barlow and Hall developed
an in-vitro error-prone PCR method (see also Vakulenko et al.187 for the application of PCR to evaluate
the -lactamase response to clavulanate/sulbactam
with ampicillin) that is argued as predictive of
-lactamase evolution under clinical -lactam pressure. Using this method TEM -lactamase variants
resistant to the entire ensemble of third-generation
cephalosporins (cefotaxime, cefuroxime, ceftazidime,

and cefepime) as well as aztreonam were identified.188-190 Notwithstanding these examples, this
method also has identified examples where resistance
phentotypes have not emerged (as exemplified by
metallo--lactamases with imipenem, as discussed
later), strongly arguing against the presumption that
these enzymes have an unconstrained ability to adapt
to all variations in antibiotic structure.190 The value
of these methods to drug design and to the design of
clinical drug regimens is self-evident.189,190
3.3. Class C -Lactamases

Class C -lactamases share with the class A -lactamases a similar mechanismsactive site acylation
and hydrolytic deacylationsfor -lactam hydrolysis.
This ability was inherited from their respective
ancestral PBPs. Nonetheless, at the catalytic level
there is a significant difference for deacylation. As
first documented by Knox and colleagues,191 the two
classes use opposite faces of the acyl-enzyme species
for the approach of the hydrolytic water. In the class
C enzymes this water approaches from the -direction. This distinction refutes any possibility of a direct
evolutionary link between the two classes. Indeed,
class C -lactamases are evolutionarily closer to low
Mr class B PBPs.15,60 Furthermore, the residue responsible for activation of the hydrolytic water in the
deacylation has been proposed to be Tyr150,191-193 a
process that has been suggested to be assisted by the
amine of the acyl-enzyme species that was previously the -lactam nitrogen of the antibiotic.194,195
Therefore, the deacylation mechanism of the class C
enzymes is entirely distinct from that of class A
-lactamases.
The class C -lactamases originally were termed
cephalosporinases due to a substrate preference for
cephalosporins. They are found, with few exceptions,
in most Gram-negative bacteria and are chromosomally encoded in several organisms (including
Citrobacter freundii, Enterobacter aerogenes, and
Enterobacter cloacae).196 An increased incidence of
plasmid-encoded class C -lactamases15,196 was observed 15 years after their first discovery.197 Plasmidencoded class C enzymes have been found in E. coli,
K. pneumoniae, Salmonella spp., C. freundii, E.
aerogenes, and Proteus mirabilis.198-200 Most worrisome is that the rate of incidences of these enzymes
is highest in K. pneumoniae and E. coli, organisms
common to the hospital and community settings.196
Class C -lactamases have reached catalytic perfection for their preferred substrates, the cephalosporin-based -lactams.115 However, the efficiency
of turnover of the penicillins by the class C -lactamases also remains high. This is attributed to low
Km values as a result of deacylation as the ratelimiting step for the class C -lactamases (unlike
class A -lactamases), resulting in high kcat/Km values
(105-108 M-1 s-1).201 The structure of these enzymes
was first revealed for the class C -lactamase from
C. freundii (and E. cloacae strain P99).191,192 The C.
freundii enzyme has also been determined with
aztreonam bound, at 2.5 resolution.192 The structures reveal an overall similarity to the class A
-lactamases (see Figure 4A and C). Superimposition

of the class C -lactamase (from E. cloacae) and a
representative class A -lactamase reveals a handful
of active site residues that occupy similar positions.
In the class C -lactamase these are Ser64, Lys67,
Lys315, and Tyr150 that correspond, respectively, to
the class A residues Ser70, Lys73, Lys315, and
Ser130.191 On the basis of the structures released to
date, this correspondence is common among the
classes A, C, and D -lactamases and the PBPs.
The first crystal structure for a class C -lactamases prompted the proposal that the role of general
base, assisting Ser64 acylation, is carried out by the
conserved residue Tyr150.192 It is important to note
that the side chain functions of both Lys67 and
Tyr150 are in hydrogen-bonding contact with the
serine hydroxyl. After accepting a proton from the
serine in the formation of the tetrahedral intermediate, this tyrosine then donates the proton back to the
-lactam nitrogen to drive forward the collapse of the
tetrahedral intermediate.192 The same Tyr150 then
promotes a water molecule to achieve deacylation of
the acyl-enzyme species to complete the catalytic
cycle. This hypothesis is based in part on the structural superposition of the active sites of the class C
-lactamase with that of chymotrypsin (a serine
protease). In this superimposition the -lactamase
Tyr150 occupies a similar position to the histidine
general base in chymotrypsin.192 The viability of this
proposal is diminished by recent NMR and sitedirected mutagenesis studies. 13C NMR evaluation
shows that the chemical shifts of Tyr150 are invariant up to pH 11,202 implying a neutral Tyr150 in the
substrate-free enzyme. This result challenges earlier
calculations (using Poisson-Boltzmann methodology)
that predicted an unusually low pKa value (of 8.3)
for the Tyr150 phenol.203 The second line of evidence
arguing against Tyr150 as the general base is the
site-directed mutagenesis study by Dubus et al.204
The steady-state kinetics were not substantially
altered by replacement of the Tyr150 with a phenylalanine. This result is inconsistent with a role for
Tyr150 as a direct participant in the turnover events.
Tyr150 is suggested to contribute to the acylation
process indirectly, perhaps as a proton shuttle to the
-lactam ring nitrogen (with water serving this role
in the Phe mutant). It is of interest to note that the
effects on the kinetic parameters were akin to those
obtained with the D,D-transpeptidase/D,D-carboxypeptidase from Streptomyces species R61 at Tyr159, a
residue equivalent to Tyr150.
In light of the recent data by Kato-Toma et al.
indicating a normal pKa value for Tyr150,202 the
proposed mechanism for the second half of the
reaction of the class C -lactamases has to be
reevaluated as well. Oefner et al. indicated that the
deprotonated Tyr150 serves as the general base for
the second step of the reaction in activation of a water
molecule for the deacylation step, a proposal that has
been widely accepted.192 A protonated Tyr150 cannot
perform this function.
These observations invoke the involvement of
Lys67 (as a free-base) in activation of the active site
serine for the acylation event in class C enzymes or
that it may abstract a proton from Tyr150, which in

turn activates the serine. Furthermore, for the deacylation step, in the absence of a residue equivalent to
Glu166 in class A -lactamases, crystal structures
from E. cloacae P99191 and C. freundii192 argue for
the approach of the hydrolytic water from the -face
of the -lactam antibiotic. Bulychev et al. propose
that the nitrogen of the thiazoline (from -lactam
opening) is ideally positioned to promote hydrolytic
water addition to the acyl-enzyme to accomplish
deacylation.193 This would be an example of a substrate-assisted catalysis. Two non--lactam synthetic
molecules were used to test this concept. These
compounds were chemically predisposed to acylate
the active site serine in the E. cloacae class C
-lactamase, resulting in acyl-enzyme species. One
compound lacked the amine of the acyl-enzyme
species, whereas the other possessed it. The one with
the amine experienced turnover, whereas the one
without it served merely as an irreversible inhibitor
of the enzyme.194 A recent X-ray crystallographic
structure of an acyl-enzyme species of AmpC -lactamase and moxalactam lends further evidence to
this proposal. The authors note that a water molecule
is organized to take advantage of the ring nitrogen
in the hydrolytic step.195
Third-generation cephasloporins have been effectively used as a strategy against class C -lactamases for over a decade. However, class C -lactamases capable of hydrolyzing most third-generation
cephalosporins were first isolated in the 1980s205 and
yet more recently from a virulent strain of E. cloacae.206 These enzymes, termed extended-spectrum
-lactamases (ESBLs),124 mediate resistance to (extended-spectrum) third-generation cephalosporins
(exemplified by ceftazidime, cefotaxime, cefepime,
and ceftriaxone) and monobactams (aztreonam) but
do not affect cephamycins (cefoxitin and cefotetan)
or carbapenems (Meropenem or imipenem). The
structural basis for resistance mediated by these
ESBLs was revealed by the structure of the E. cloacae
GC1 enzyme, solved by Crichlow et al.207 These
authors suggest that conformational flexibility of the
expanded -loop facilitates hydrolysis of thirdgeneration cephalosporins by enabling greater mobility of the acyl moiety. Further evidence implicating
the -loop modification with resistance to these
third-generation cephalosporins came from the structures of the AmpC -lactamase in complex with
various third-generation cephalosporins.208 A more
recent structure of a phosphonate transition-state
mimetic bound to the E. cloacae GC1 and the wildtype C. freundii GN346 class C enzymes afforded
further insight to the basis for resistance. It was
found that the designer molecule adopted different
conformations in both enzymes, with the mutated
-loop of the GC1 enzyme able to accommodate the
cefotaxime side chain in a different conformation,
enabling it to allow the approach of the hydrolytic
water to the acyl-enzyme species.209
3.4. Class D -Lactamases

The class D -lactamases are increasingly encountered among the defensive -lactamase ensemble of
Fisher et al.
certain Gram-negative pathogens.15,109,124,210 These

-lactamases were first termed as oxacillinases (and
for this reason are still described as OXA -lactamase
variants) for their ability to hydrolyze the 5-methyl3-phenylisoxazole-4-carboxy side chain penicillin class,
exemplified by oxacillin and cloxacillin. Over 50 class
D OXA variants are now known,211 including variants
that have dispensed with their oxacillinase activity
in the process of acquiring ESBL activity against
carbapenem and third-generation cephalosporins.212
In the course of these transformations the class D
enzymes have expanded from their historical P.
aeruginosa niche 213,214 into other Gram-negative
pathogens including E. coli,215,216 P. mirabilis,217
Salmonella sp.,218 K. pneumoniae,219,220 and especially
Acinetobacter baumannii.221-228 While at present the
clinical impact of the OXA -lactamases is associated
with infections by P. aeruginosa and A. baumannii,
the widening Gram-negative distribution provides
powerful support for the concern that the clinical
value of carbapenems (and third-generation cephalosporins) may quickly diminish.211,220,222,225,229
These enzymes only recently have become the
subject of detailed structural and mechanistic studies. This is due in some measure to their very recent
appearance in clinically relevant pathogens (as distinct from the other three -lactamase classes) but
in greater measure to their heterogeneous properties
(not withstanding their interrelatedness) and the
historical difficulty of their in vitro assay. The
literature is replete with descriptions of poorly reproducible biphasic (burst-type) progress curves for
some variants, corresponding (in the apparent steadystate) to unusually low kcat/Km values (as exemplified
by the data of Danel et al.,230 Franceschini et al.,231
Pernot et al.,232 and Heritier et al.226). Among the
enzyme properties explored as possible explanations
for these difficulties were a monomer-dimer equilibrium with possible additional divalent metal dependency.230,231,233,234 No credible explanation was
found. For example, some OXA variants are monomeric,97 and the dimer Kd in any case (typically
micromolar) is relevant only to the in-vivo and not
in-vitro kinetics, while other variants show little
capacity for divalent metal binding.224 As a consequence of this dilemma, the possibility that other
mechanisms (in addition to the OXA -lactamase)
contribute to the -lactam resistance phenotype
has been discussed.221 Also, while the importance of
porin deletion211,216,219 and peptidoglycan remodeling
(in addition to the assembly of mixed class -lactamase ensembles or single -lactamase hyperexpression) to Gram-negative resistance cannot be underestimated, the identity of the likely critical and
confounding variable contributing to the in vitro
assay variabilitysCO2swas identified by Golemi et
al.95,96
The structural basis by which CO2 activates the
class D enzymes for -lactam hydrolysis was elucidated concurrent with several independent crystallographic studies of the class D -lactamases. The
first studies on a class D enzymes (the P. aeruginosa
OXA-10 -lactamase) showed that the class D domain
folding was similar to the other two serine -lacta-
mase classes.235,236 However, subsequent crystallographic analysis revealed that the active site lysine
was N-carboxylated as a result of addition of the
lysine side chain amine (as the free-base amine) to
CO2.96,237 The resulting carbamic acid ionizes to give
a carbamate functional group in hydrogen-bonding
contact with the active site serine. As the formation
of this carbamate is reversible, the earlier reports of
the absence of lysine carboxylation are explained. In
light of the high physiological concentration of CO2
(low millimolar) this lysine is expected to be fully
carboxylated in vivo.95 Kinetic analysis of a mutant
enzyme where this lysine is replaced shows the total
loss of catalytic activity, indicating a direct involvement in catalysis.95 The role for this unusual lysinederived carbamate is general base activation for both
acylation (activation of the serine) and deacylation
(activation of water) steps of catalysis.95,237 Moreover,
the assignment to this CO2-derived lysine carbamate
of this role as general base catalyst is consistent with
the absence of alternative possibilities. Not only does
the class D -loop not contain a counterpart for the
class A Glu166, but the tyrosine of the conserved
parental class D Y144-G145-N146 motif on this loop
is replaced by phenylalanine in ESBL (both carbapenem and cephalosporin) class D variants, obviating
direct participation of this tyrosine in catalysis (as
is more fully discussed elsewhere).97,222,226,238 The
OXA-1 crystal also shows the lysine carbamate97
whereas the OXA-13 enzyme232 does not (almost
certainly an artifact of crystallization). The relationship of the CO2 to the complicated kinetics extends
beyond the relative portion of the lysines that are
activated for catalysis. During catalysis the lysine
carbamate is prone to spontaneous decarboxylation
in the middle of the turnover process, thus arresting
catalysis at the acyl-enzyme stage. This must,
however, be regarded as an artifact of in vitro assay
since supplementation of the medium with bicarbonate (as a CO2 source) restores the kinetic profile.95 It
has been argued that the more complicated biphasic
turnover profile for these enzymes with some substrates is due to a branching mechanism. As the
enzyme experiences decarboxylation midcatalysis, it
becomes inactivated (the branching species), pending
the availability of a CO2 molecule to restore the lysine
to its active carboxylated form. The enzyme is then
able to complete the second step of catalysis.95,236
With a few substrates, however, it was shown that
supplementation of the medium with bicarbonate
does not simplify the turnover process. For these few
cases a branching mechanism (as might occur by a
conformational change at the acyl-enzyme state) has
been invoked.
[Dr. Roger Labia, a pioneer of studies of -lactamases, kindly communicated that his early investigations of the class D -lactamases in 1970s were
frustrating because of the complicated and erratic
kinetic behavior of the enzymes. He opted to abandon
studies of the class D enzymes. He now attributes
the erratic behavior of the enzymes to the seasonal
variations in the quality of the water, which has high
carbonation in the summers and low carbonation in
the winters.]
While there are numerous reports evaluating the

-lactam substrate profile for the class D OXA
enzymes, essentially all of these predate the discovery of requirement for in-vitro CO2 (but most unlikely
in vivo) activation. These earlier data may not be
reliable. Nonetheless, all these data suggest that the
substrate profile for these enzymes, individually and
as a class, is not broad.211 The value of the class D
enzymes to bacteria is the ability of this enzyme to
adapt, under selection pressure, to the specific -lactam. For example, while most of the ESBL OXA
variants hydrolyze ceftazidime better than cefepime,
the reverse is true for the OXA-30 variant (derived
from OXA-1).215 Full discussions of the orientation
(and contacts) within the class D active site for
(inhibitor) acyl-enzymes232,237 and possible orientations for substrates97 are presented elsewhere.
A final issue is the origin of the class D enzymes
in comparison to the two other serine -lactamase
classes. All three classes are now encountered as both
chromosomal- and plasmid-borne genes. However,
the evolutionary history of each class is distinct.15 In
contrast to the class C AmpC family, where the
mobilization to plasmids is a modern (antibiotic era)
event, plasmid mobilization of the class A TEM and
class D OXA are ancient events.239 The primary
distinction between the class A and C -lactamases
is the much more rapid and extensive diversifications
at this point in timesof the former within the Gramnegative bacteria. Also of particular interest is the
complete absence of this diversification within the
Gram-positive bacteria given the estimate by Hall
and Barlow that the horizontal transfer of the class
C gene from the Gram-negative to the Bacillus Grampositive bacteria is an ancient event of some 320-575
Myr (but is after the divergence of B. subtilis from
S. aureus).15 A homology between the hydrophilic
carboxy domain of the BlaR/MecR -lactam-signaling
receptors and the class D -lactamases was noted
previously.84,240 The recent discovery within B. subtilis of a gene encoding an enzyme with weak class
D -lactamase activity, yet also resembling a penicillin-binding protein,241 suggests that study of the class
D gene within those Gram-positive bacteria that
possess it may further refine our understanding of
the structural and functional relationships between
the PBP and -lactamase enzymes.
3.5. Class B Metallo--lactamases

The secondsand in many respects no less forebodingsvision of the -lactamase future is that of
the disseminated metallo--lactamases (MBLs).242
First observed in 1967 by Kawabata and Abraham
as chromosomal enzymes of the innocuous Grampositive B. cereus, these enzymes occupy a position
of concern (in terms of breadth of distribution and
breadth of -lactam catalytic activity) with respect
to the inexorable expansion of -lactam resistance.109,229,243,244 These metal-dependent (almost always divalent zinc) -lactamases have a broad -lactam substrate tolerance that encompasses many of
the newer generation cephalosporins, carbapenems,
and other -lactamase inhibitory (clavulanate and
penam sulfones) -lactams important to the treat-
ment of Gram-negative infection.210,229 As a different

chemical mechanism (compared to the serine -lactamases) is used by the metallo--lactamases for
-lactam hydrolysissnotably, a mechanism without
a covalent enzyme intermediatesan entirely different
strategy for their inhibition (or inactivation) will be
required should (or is it when) these enzymes expand
beyond their present niche (that of minor, and
opportunistic, Gram-negative pathogens). While several inhibition strategies have been identified, none
has yielded anything resembling that of a clinically
effective inhibitor. Also, although there exist as yet
chemical reasons (in terms of kcat/Km these are not
optimized enzymes) and biochemical reasons (zinc as
a limiting nutrient) that ultimately may limit the role
the metallo--lactamases will have as a resistance
mechanism (vide infra), the proposal discussed by
Fast et al.,245 Fabiane et al.,246 and Wommer et al.247
that these are young enzymes, only now in the
process of maturation under evolutionary pressure,
is both credible and worrisome.
The metallo--lactamases (also termed Ambler
class B and Bush-Jacoby-Medeiros Group 3 -lactamases)105,243 are small enzymes sharing a common
four-layer RR motif with a central -sandwich and
two R-helices on either side.246,248 This motif, arising
possibly by a gene duplication event,60,245 is found also
in other proteins (glyoxalase and certain flavoenzymes). The motif has an intrinsic metal-binding site
located at an edge of the -sandwich.249,250 For the
metallo--lactamases this site is occupied by a divalent zinc ion having a tetrahedral array of three
histidines and water. The importance of the zinc ion
to -lactam substrate binding is unquestionable.251,252
The role of the zinc in the hydrolytic mechanism,
beyond that of Lewis acid catalysis, is less certain.
Nonetheless, there is a consensus that the water
ligand of the zinc ion is the -lactam ring-opening
nucleophile (via a mechanism likely with parallel to
that of the zinc metalloproteases), but there are no
direct data establishing this role.
Beyond these structural commonalities the metallo--lactamases possess a surprising breadth of
primary structure that most notably includes the
creation (in some enzymes) of a second zinc binding
site.243 In these binuclear enzymes the two zincs are
proximal (approximately a 3.5 separation) with
both participating in the water coordination. The
ligand environment of the second zinc ion is very
different from that for the first both in array (trigonal
bipyramid) and amino acid ligands (variable among
the binuclear class). The purpose of the second zinc
ion is regarded as catalytic augmentationsthat is,
accomplishing an incremental increase in the -lactam substrate kcat/Kmsand not a role of catalytic
necessity.245,247,253-255 Nonetheless, the specific environment of the zinc ensemble determines the individual enzyme catalytic behavior toward substrates
and the detail of the rate-limiting (highest energy)
catalytic step.245,255-259 The relative affinity of the
enzyme for the two zinc ions is unequal, and in-vitro
enzyme kinetic analysis requires added Zn2+.247,255,260
Wommer et al.247 discussed in detail the role of the
-lactam substrate in the recruitment of zinc to, and
Fisher et al.
hence activation of, the -lactamase. This process is

argued as one of physiological necessity wherein the
-lactamase exists as an apoenzyme and the available zinc is reserved to other enzymes until the
-lactam antibiotic is encountered. The likelihood of
a fully zinc complemented metallo--lactamase (binuclear site fully occupied) in vivo is regarded as
small. It is therefore understandable why the bacterium possessing a metallo--lactamases is often found
with a serine -lactamase as well.261,262 The demonstrable fact that these enzymes are disseminating is
proof of evolutionary pressure for resistance development within human clinical practice.
While the details245,255,263 of the hydrolytic mechanism used by these enzymes is beyond the scope of
this review, the mechanistic fundamental is assuredly not. This fundamental is delivery of the zinccoordinated water, possibly as a hydroxide ion (pKa
) 4.9 to 5.6),245,260,264,265 to the -lactam carbonyl. (The
metalloprotease carboxypeptidase A has a similar
catalytic pKa that is also assigned to a zinc hydroxide.
However, the recent 67Zn NMR study by Lipton et
al. forcefully argues against this assumption.266 Accordingly, a wholesale mechanistic revision for metallo--lactamase catalysis may be necessary.) The
presumed involvement of a zinc hydroxide intermediate for the metallo--lactamases has stimulated
exquisite studies on the in-vitro mechanism of metalcatalyzed -lactam solvolysis.34,267-269 These studies
suggest the rate-limiting step to be either metalcoordinated hydroxide addition or (following the
hydroxide addition) the collapse of the tetrahedral
species. The mechanism is dependent on -lactam
structure in such fashion as to strongly implicate
zinc coordination of the -lactam in the transition
state.34,252,253,256,270-272 A similar mechanismsthat is,
hydrolysis without a covalent enzyme intermediates
is posited for the enzymatic reaction. Solvent kinetic
isotope effects for the enzymatic reaction implicate
additional transition-state stabilization by proton
flight.255,260 The identity of the enzymatic proton
donor/acceptor, however, remains a particular focus
of mechanistic study. A notable commonality of the
structure-kinetics analysis of the nonenzymatic solvolysis and the comparative enzymatic structurekinetics is the particular susceptibility of the carbapenems to hydrolysis.34,229,269,270
This realitysthe susceptibility toward metallo-lactamase hydrolysis of nearly all (vide infra) of the
serine -lactamase inhibitorssis a growing concern.
There are three reasons for this. The first is the
burgeoning presence of metallo--lactamases (first
the IMP and now the VIM and SMP class B1
variants) as mobile plasmid-encoded (and often also
as integron, or cassette) genes by the enterobacteria.273,274 Second, these enzyme variants have a
remarkable ability to alter their substrate capability
(as can be accomplished by simple-point mutation,
even remote to the active site), raising the possibility
of rapid adjustment to encompass new substrates
(and to thwart new inhibitors). Last, the metallo-lactamase plasmids often encode additional (multisubstrate) resistance mechanisms.
The class B1 metallo--lactamases are monomeric,

binuclear zinc enzymes constituting the largest metallo--lactamase subclass. Within this subclass the
dominant subtypes are the IMP and VIM enzymes.
While these subtype enzymes are defined by sequence, substantial diversity is found within each.
For example, the recently observed IMP-12 variant
has 89% sequence identity to its closest (IMP-8)
relative but includes 10 amino acid changes at
positions that are otherwise invariant among the 11
other subtypes.275 The IMP enzymes are originally
Asian (and the VIM enzymes originally European),
but it is quite evident that geographical characterization of both of these enzymes is irrelevant. First
observed in 1988, the IMP enzymes have a broad
substrate (primarily cephalosporins and carbapenems, less so penicillins) acceptance, although at the
level of specific -lactam structure the kcat/Km variation can be substantial. Replacement of glycine-196
(a noncatalytic residue adjacent to His197, a zinc
ligand) in the IMP-3 and IMP-6 variants with serine
(as is found in IMP-1) results in a significant kcat/Km
improvement toward certain cephalosporins and
toward imipenem (IMP-1 kcat/Km is 10-50-fold greater
compared to IMP-3).272,276 Nonetheless, a comprehensive in vitro substrate evaluation of IMP-6 and
microbiological evaluation of E. coli possessing the
TEM-1/IMP-6 plasmid indicate the IMP-6 to be the
better enzyme in terms of extended carbapenemase
activity.277 Whereas Meropenem and imipenem are
essentially equivalent (as measured by kcat/Km) IMP-1
substrates, the IMP-6 glycine replacement results in
a 6-fold improvement in the Meropenem kcat/Km (and
a 2-fold loss for imipenem). This improved kcat/Km
contributes to the significant difference in MIC (E.
coli with the IMP-1/TEM-1 plasmid, 64 g mL-1 for
Meropenem and 2-8 g mL-1 for imipenem; without,
0.25 g mL-1).277 Efforts to correlate IMP sequence
with altered substrate acceptance have been made
by Oelschlaeger et al. (comparing G196 IMP-6 to
S196 IMP-1)272 and Moali et al. (evaluating the role
of the distal 60-66 loop).271 In the former study a
favorable serine-196-lysine-33 interaction improves
packing and rigidifies histidine-197; this rigidity
propagates throughout the active site. A calculated
enzyme-substrate stability index was found to correlate well to the experimental kcat/Km. In the latter
study the loop was confirmed as nonessential for
catalysis but contributed (especially tryptophan-64)
to hydrophobic substrate binding. A mutagenesis
study by Materon and Palzkill of IMP-1 active site
proximal amino acids identified 52% of the IMP-1
amino acids as intolerant to substitution (by comparison, the TEM-1 value is 33%).278 Materon and
Palzkill278 suggest that the IMP metallo--lactamases
may have a relatively more limited ability to adjust
to extended spectrum -lactam structure compared
to the TEM enzymes (which have, of course, already
done so). A relatedsand no less pertinentsquestion
is whether the metallo--lactamases have the capability to develop their catalytic apparatus to function
at the substrate diffusion limit (that is, at full
catalytic competence), as is already the case for the
class A and C serine -lactamases.115 The extant IMP
kinetic data indicate that while certain cephalosporins have kcat/Km values that approach the diffusion
limit of (107-108 M-1 s-1), most substrates have lower
values (typical carbapenem kcat/Km values are approximately 106 M-1 s-1). Halls full mutagenesis invitro evaluation of the IMP-1 structure, which failed
to identify a mutant enzyme more capable of imipenem hydrolysis, is consistent with one (or both) of
these possibilities.279 As the diversity of known carbapenem structure is not nearly that of the penicillins
and cephalosporins, cautious optimism may be entertained that newer generation -lactams poorly
capable of metallo--lactamase hydrolysis may yet be
made.
Two additional aspects may temper this conclusion.
It is clearly not possible to determine, by evaluation
of enzyme sequence or enzyme kinetics, a direction
for metallo--lactamase variant evolution (which
variant arose from which variant). Hence, the apparent evolutionary limitation of IMP-1 with imipenem has no predictive value with respect to other
metallo--lactamase variants. As bluntly stated by
Hall, in order to understand the risks posed by
metallo--lactamases, it will be necessary to conduct
similar studies on representative members of each
of the three metallo--lactamase subfamilies and to
include all clinically relevant carbapenems.279 Second, it is evident that incremental changes in -lactam fitnesssin terms of PBP inactivation and competence as a substrate for -lactamase hydrolysiss
suffice to determine whether a bacterium is susceptible
or resistant. An effect need not be dramatic to be
important.
The VIM B1 subclass is newer (first observed in
1997) and biochemically less well studied.280 A VIM
sequence homology with IMP is recognizable (approximately 30-40%), and the overall pattern of
-lactam substrate recognition by the two enzymes
is similar.229 Seven VIM variants are extant.281 In less
than 7 years the VIM metallo--lactamases have
transitioned from chromosomal expression by nonfermenting Gram-negative bacteria (where it contributes to high-level antibiotic resistance in P.
aeruginosa pathogenic strains)282,283 to transferable
plasmid expression in Gram-negative enterobacteria.262,273 The presumptive circumstances leading to
this change (evolutionary pressure for plasmid dissemination is not necessarily carbapenem but rather
multidrug driven) and probable consequence of this
change (likelihood of carbapenem clinical failure)
underscore the caution expressed in the previous
paragraph.
A consistent observation from the in vitro evaluation of the metallo--lactamase substrate spectrum
is the inability of these enzymes to hydrolyze the
monocyclic N-sulfonyl -lactam antibiotic aztreonam.
Consequently, bacteria that have these metallo-lactamases can retain aztreonam susceptibility (although moderate to substantial increases in the
aztreonam MIC values, due to other plasmid-conferred resistance mechanisms or to the presence of
aztreonam-capable serine -lactamases, is common).
As the intrinsic reactivity toward solvolysis of the
aztreonam -lactam is identical to that of the penicil-
lins,34,116 the inability of the metallo--lactamases to

hydrolyze aztreonam must correspond to a failure
either to bind to the enzyme or to bind but in a
nonproductive orientation. Quite surprisingly, there
do not appear to be enzymatic kinetic data on this
matter. Diaz et al. predict, on the basis of computational study, the latter answer as correct.269 Accordingly, aztreonam may represent an unusual example
of what has emerged as a general strategy for
metallo--lactamase inhibition. Following the proven
basis for zinc protease inhibitor design (that of inhibitors providing sulfur coordination to the thiophilic
zinc), thiol-substituted carboxylic acids represent a
general metallo--lactamase inhibitor motif.248,259,284-286
The carboxylate of these inhibitors occupies the
identical site used in -lactam carboxylate recognition
and the thiol (as the thiolate) displaces the nucleophilic hydroxide from the zinc pair. When the juxtaposition is optimal (typically corresponding to Ki
values of 0.1-1 M) a reorganization of the active
site is seen that is believed to be similar to what
occurs in substrate binding.248,259,287,288 Given the
exemplary quality of these inhibitor-enzyme crystallographic structures and the increasing sophistication
with which proven -lactam synthetic methodology
is being applied to thiocarboxylate design, yet more
potent inhibitors (approaching what will likely be
required for clinical efficacy) are anticipated. Whether
these will also possess the appropriate pharmacodynamics to effectively synergize with a -lactam
antibiotic remains to be seen.
4. Other Resistance Mechanisms

4.1. Porin Deletion
The temporal response of bacteria to antibiotics is
both immediate (abrupt gene repression and gene
activation) and evolutionary (empirical gene mutation and gene acquisition). Not surprisingly the
increased sophistication with which these changes
may be assessed and the dramatic breadth of change
particularly in the immediate response289,290 have led
to justifiable optimism that understanding these
responses will identify new targets for antibiotic
design.291,292 At a simpler level the comparison of
bacterial protein expression before (susceptible) and
after (resistant) antibiotic exposure has been the
mainstay to the understanding of antibiotic resistance development, and it is these studies (as have
been just described) that support the generalization
that PBP alteration is a principle mechanism of
Gram-positive resistance and -lactamase expression
is a principle mechanism of Gram-negative resistance. Nonetheless, these same studies indicate that
other resistance mechanisms exist. In this section an
overview of two of these other mechanismssdecreased
antibiotic permeability and increased antibiotic effluxs
is given.
It is axiomatic that a successful antibiotic has
potency (the ability to incapacitate an essential
target) and access (to its target). Indeed, PBP alteration is a strategy to render impotence and -lactamase expression compromises access. A second method
for controlling access is by changes within the outer
Fisher et al.
membrane (in Gram-negative bacteria) and cell wall

(in Gram-positive bacteria). While the evaluation of
these changes is among the most difficult tasks to
accomplish at the microbiological level, the recent
observations concerning several resistant K. pneumoniae speciessthose of an important Gram-negative
pathogensare revealing. The report by Bradford et
al. is regrettably now typical.293 An examination of
12 highly -lactam-resistant K. pneumoniae strains
and 6 E. coli strains from a single hospital showed
that 17 possessed multiple -lactamases. The three
most resistant K. pneumoniae strains (as defined by
resistance to imipenem, the prototypical carbapenem)
achieved their resistance by the combined expression
of an AmpC extended-spectrum -lactamase and
deletion of a major 42 kDa outer membrane protein
(omp). A second report294 likewise describes this same
combination of the AmpC -lactamase and the omp
protein deletion giving an imipenem-resistant K.
pneumoniae. The surmise that this omp proteinsa
porin, or nonspecific solute poresis an important
point of ingress for the -lactam to the periplasmic
space is supported by the subsequent studies of
Nelson et al.295 and Domenech-Sanchez et al.296
Resistance derives from the synergistic combination
of reduced permeability and the -lactamase; the
degree of resistance of these two mechanisms in
concert exceeds that of each mechanism alone. Similar observations have been made recently (citing
representative examples) with respect to -lactamresistant E. coli,216,297 Salmonella enterica,298 Heliobacter pylori,299 Acinetobacter baumannii,300 E. aerogenes,301,302 K. pneumoniae,303 and P. aeruginosa.304,305
A simple conclusion as to the importance of porin
deletion (or modification) remains, however, elusive.
It most certainly contributes for certain bacteria
when selected for by certain antibiotics. The reasons
why a broader generalization is not possible are
straightforward. The variation in intrinsic antibiotic
permeability among the bacteria is substantial. For
example, Lakaye et al.306 estimate that E. coli is
approximately 20-1000-fold more permeable for a
given -lactam than E. cloacae. In addition, the
variation in relative permeability as a function of
-lactam structure is no less variable: as assessed
by Matsumura et al.,307 the relative E. coli permeability of imipenem (most permeable) is approximately 60-fold greater than ceftazidime (the least
permeable of six -lactams evaluated). An appreciation for the basis of this variability is provided by
Nikaidos superlative account45 of the utter complexity of the Gram-negative outer membrane dynamics: an extraordinarily asymmetric bilayer dominated
on the outside by the lipopolysacharide surface,
which itself exerts significant permeability selection
particularly against hydrophobic solutes,308,309 and
punctuated on both surfaces by an array of nonspecific protein pores (porins) and transporters. Relative
solute permeability is influenced by a nearly limitless
number of variables. Porin deletion in a -lactamresistant E. coli is accompanied by (uncharacterized)
changes in the outer membrane (and perhaps peptidoglycan) that could also contribute to resistance.297
The presumption that compensatory responses

such as porin deletion exact a fitness cost is almost
certainly correct.5,91,310 In this regard bacteria are
little different from other organisms: to the extent
that a choice is possible between death and discomfort, the latter is chosen. An example of accommodation between survival and vastly decreased solute
permeability is provided by the mycobacteria. The
mycobacteria are notable (and also opportunistic)
human pathogens that are suggested evolutionarily
to bridge the Gram-negative and -positive bacteria.
Their high intrinsic resistance to chemotherapys
including complete -lactam resistancesis believed
to result from a combination of restricted porin
ingress (either by limited abundance or by pore
character) and an impenetrable exoskeleton (consisting of a thick peptidoglycan and an outer membrane
having on its outer leaflet an additional barrier of
long mycolic fatty acids attached an arabinogalactan
chain).45,311 It may therefore be surmised that exoskeleton adjustment, so as to limit -lactam exposure,
is also a strategy exploited by the Gram-positive
bacteria (the more so because they are regarded as
more solute permeable than either the Gram-negative or mycobacteria). This surmise is correct. However, the astonishing complexity of these adjustments
is just now being appreciated. While it has been
known for some time that the cell wall composition
of both Gram-negative and -positive bacteria changes
in response to -lactam exposure,26,312,313 the presumption with respect to the Gram positives is that
these changes reflected the direct (for want of a better
term) resistance response: the peptidoglycan is
altered by compensatory overexpression of a PBP, or
as a result of differential inactivation by the -lactam
of a PBP from among the ensemble, or the differential
recognition of the selected low-affinity PBP for the
biosynthetic cell wall precursors results in the modified cell wall. That a more complex stratagem was
in play was discovered by Filipe and Tomasz314 from
the observation that inactivation of the murMN
operon, encoding the murM and murN cell wall crossbridge biosynthesizing enzymes, abolished -lactam
resistance in low-affinity PBP containing (and thus
previously) highly -lactam-resistant S. pneumoniae.315 The indisputable correlation of murM enzymatic activity (murN deletion has little effect) with
robust penicillin resistance is discussed by Fiser et
al.316 and Rohrer and Berger-Bachi.317 The facile
conclusionsthat the murM cross-link-enabling reaction, the acylation by serine of the lysine -amino of
Lipid II, is a necessary event in the construction of a
-lactam-impermeable peptidoglycansmay indeed
prove correct. Fiser et al. suggest a more intriguing
possibility based on homology modeling of the murM
sequence with FemA and myristoyl transferase: a
winged coiled-coil helical DNA-binding domain structure similar to the bacterial transcription factors
known to control multidrug exporter expression.
Their conjecture that murM additionally serves to
regulate bacterial gene expression following -lactam
exposure, such as by exporter expression, is consistent with the appearance of high-level -lactam
resistance. We know this conjecture to be credible
since the operation of a drug transporter is already

known to confer high-level -lactam resistance to
another insidious bacterial pathogen, P. aeruginosa.
4.2. Transporter Expression

The P. aruginosa species is a particular contributor
to highly drug-resistant biofilm infections in cystic
fibrosis patients for which carbapenem therapy is
often the only recourse. The breadth of its resistance
is believed to result from the combination of overall
membrane impermeability (especially porin deletion
and thickened peptidoglycan) and the action of active
transporter-catalyzed drug efflux (typically the RND
MexAB-OprM transporter).304,305,318,319 The bacterial
transporters (of which five families are now recognized, abbreviated as MFS, SMR, MATE, RND, and
ABC) have become a central issue to the understanding of overall bacterial vitality. Simply put, the
volume of regulated (such as by transporters) and
unregulated (such as by porins) molecular traffic in
and out of the bacterium is astonishing; it is now
believed that up to 20% of the E. coli genome encodes
transporters of one variety or the other.320 The
importance of this phenomenon, especially in relation
to drug resistance and virulence factor release,321,322
is attested to by the volume of exemplary reviews
that describe the rapidly changing status of this
field.320,323-328 What is particularly disconcerting is
the appearance of highly resistant bacteria wherein
the operation of these transporters, as part of an
ensemble of resistance mechanisms including porin
loss and -lactamase acquisition, suffice to compromise the carbapenems as an effective therapy. Moreover, bacteria that operate these transporters coincide to a multidrug-resistant phenotype. Also, while
imipenem is understood not to be an efficient substrate of the P. aeruginosa transporter, the increasing
appearance of imipenem-resistant P. aeruginosa
indicates that refinement of the panoply of P. aeruginosa resistance mechanisms to embrace imipenem
is in progress. Whether antibacterial design of new
-lactam structures that evade these structures is
possible remains to be seen. There is decreasing room
for optimism: imipenem is an exceptionally small
and permeable antibacterial. The strategy in medicinal chemistry design is invariably that of increased
functional-group complexity and concomitant increased molecular mass, entirely in the wrong direction for transporter evasion. Perhaps more promising
is the concept of synergism via co-administration of
transporter inhibitors, which has emerged as an
active area of drug design.329-332 Even more intense
interest in these approaches is likely as the role of
these transporters in Gram-negative333 and Grampositive316 drug resistance is clarified.
5. Envoi
That the thoughtless use of antibiotics is reckless
is an opinion that will fail to provoke dispute from
any reader of these words. Indeed, the issue is no
longer whether a clinical problem exists (the statistical data would be deemed indisputable even by
Samuel Clemens in his retelling of Disraelis quote
about statistics)6,334-338 nor largely what could (or

should) be done to forestall the inevitable crisis but
rather how to invest today in the scientific and
medical strategies that will provide preparedness
tomorrow. Jeffersons insightful connection between
vigilance and liberty is no less appropriate to parlous
geopolitical circumstances as it is the inexorable
progression of resistant human pathogens. Two questions address the relationship of scientific vigilance
to the future role of the -lactam antibacterials in
the treatment of infections. Do the -lactams remain
a viable template for drug discovery? Do their targetss
the enzymes of cell wall biosynthesissremain a
viable target for antibacterial design?
The former question is unquestionably the more
contentious, especially given the precipitous decline
in pharma investment in antibacterial discovery.2,339-343 The choice of template, always paramount
to ultimate success in drug discovery, is yet more so
when resources are limited. Notwithstanding the
historical dominance of the -lactams, which continues to this very day, the trend toward greater mass
and functional-group complexity in successive -lactam generations strongly suggests that design limits
are being approached. In the past these conceptual
limits have been breached by revelation from Natures
a vastly more imaginative engineer of chemical
structure than mansbut here as well pharma investment is in decline.344,345 It is evident that the era of
medicinal chemistry manipulation of the -lactam,
guided by the paradigm of iterative optimization of
MIC values, is coming to an end. This is not to say,
however, that future medicinal chemistry efforts
toward -lactam optimization are exercises in futility.
Rather, future antibacterial discovery (not just -lactams) will follow the much more complicated process
wherein chemical structure is evaluated in terms of
the interrelationships among bacterial genomics and
proteomics relating to several antibacterial targets.290-292,346,347 As has been noted on several occasions within this review, bacterial resistance as a
phenotype is the result of multiple compensatory
adjustments, many of which are incremental. An
example of the possible fragility of this accommodation (and as well the depth of our ignorance as to
these compensatory adjustments) is the synergistic
and compensatory relationship among the bacterial
genetic background, -lactamase expression, and the
maintenance and expression of plasmid-carried mecA
in -lactam-resistant S. aureus, as discussed by
Katayama et al.348 This observation would immediately suggest the beneficial combination of a -lactam
targeting the PBPs with a -lactam targeting the
-lactamase, which has been of course a mainstay of
clinical therapy for some two decades. The conclusion
that this combination therapy would prove successful
is arguably, in retrospect, facile. Whether future
antibacterial therapy of multidrug-resistant microorganisms will also require complex drug combinations (as is already the case for anti-HIV therapy) is
not nearly as obvious. This is, nonetheless, a likely
course of events. Fortunately, the experimental resources needed to accomplish this task are coming
into place. Single-cell microscopy, for example, may
Fisher et al.
allow one to identify the role of the target in cell wall

biosynthesis as the cell is seen to respond to drug
exposure.62,349-352 This response can be correlated to
proteomic analysis.289,291,353 The understanding of the
structural basis for -lactam induction of the bacterial SOS response354 and for creation of persister
bacteria populations290,355,356 are but two examples of
how future -lactam SAR may be guided.357 Screening to identify drug combinations that synergize with
-lactams (as is currently being done with efflux
pump inhibitors), with respect to target pairings, can
be done on a high-throughput scale. Old -lactams,
with proven safety and performance, can be given
continued clinical relevance.358 Nonetheless, the likelihood that future antibacterial chemotherapy will be
a multidrug regimen is real.359
The task of identifying and then optimizing multidrug safety and efficacy is daunting. Drug discovery
is already the zenith of the collaboration between
human scientific and engineering ingenuity, and the
emerging anticipation that this may now need to be
done on a multifactorial scale is part of the reason
that the economics for future antimicrobial drug
discovery are so dismal. While these economics may
change after the crisis (as only a crisis galvanizes
consensus of opinion), the prudent and sobering
reminder is that successful drug discovery is emphatically not instantaneous. Should our vigilance in
antibacterial drug discovery falter, the length and
depth of the crisis may be unlike anything modern
man has experienced.
There remain, of course, the companion questions
as to what should be done now to preserve -lactam
efficacy for future generations. There is no other antibacterial class that can substitute for -lactam antibiotics in the foreseeable future, and none are in the
pipeline.4,360 These are questions that involve the entire breadth of clinical practice, including minimizing
global environmental antibiotic exposure (the reduction of antibiotics in animal feeds is likely a step in
the right direction), re-appreciating the value of hospital quarantine, and re-emphasizing the incredible
importance of proper hygiene to minimizing infection.
At the level of drug therapy, additional possibilities
are emerging. We now appreciate that the gut is, to
use the delectable phrasing of Courvalin and Davies,4
a veritable microbial bordello with extensive capacity for genetic exchange.361,362 Exploratory therapies
that include -lactamases (to destroy nonabsorbed
-lactams, either in situ or post facto) have shown
promise.363-366 The interrelationship of communityand hospital-resistant microorganism reservoirs now
is recognized.367 Judicious use of early generation
-lactam therapy can mitigate resistance development against later generation -lactams.358,368-371
The central importance of the cell wall biosynthetic
enzymes as antibacterial targets is irrefutably validated by the -lactams themselves. Not only are
these enzymes accessible and essential, but also these
are enzymes with a demonstrated commonality for
inhibitor (and substrate) recognition at their active
sites. The proposal that the -lactams constitute the
only motif for inhibitor design is not merely unproven
but largely untested.372 Antibacterial screening is
only now transitioning from classical broth MIC

identification to discrete enzyme (target) screening
evaluation. As this is accomplished and as we more
clearly understand the relative importance of (say)
one specific PBP over another, opportunities for
structure-based -lactam design (a remarkably open
frontier!)14,28,373-376 and for new template identification will be created. The role and identity of each
enzyme contributing to the assembly of the cell wall
are only now emerging. As each falls into place, yet
another discrete target for antibacterial discovery is
acquired.
Nonetheless, the long-term future of the -lactams
is uncertain. While their widespread clinical use is
certain to continue for the foreseeable future, as with
all classes of anti-infective drugs continued efficacy
is the difference between chemical innovation and
clinical erosion by resistance development. The latter
is not merely certain but is both irreversible and
progressive. Also, while one should underestimate
neither human resilience nor human innovation, the
confluence of these attributes to accomplish drug
discovery has always required the context of need
and reward. While the imperfections of this system
have long been evident (look no further than the
obliviousness of major pharma to third-world disease), until now a demonstrable connection between
anti-infective need and reward has existed. This is
no longer true. Until society understands the difference between chemicals that are commodities and
chemicals that are creations, the investment of
human intellect in the -lactams may soon extinguish. Sheehans description of the -lactams as the
enchanted rings was a tribute to their intricacy,
safety, and efficacy. There is no reason to believe that
the burgeoning microbial resistance to their efficacy
is anything other than opportunity for further enchantment, unless we choose otherwise.
bacterial intervention. Two counterpoints on the

structure of the murein polymer are presented.383,384
Arbeloa et al. make the significant discovery that S.
aureus PBP2a can confer -lactam resistance to other
Gram-positive bacteria, with the synthesis of mosaic
peptidoglycan cross-bridges.385 Gardete et al. provide
further insight to the role of murE in S. aureus
resistance by control of PBP2 and PBP2a expression.386 The Class B (monofunctional) PBP2b transpeptidase from resistant S. pneumoniae has a T446A
mutation that reduces penicillin affinity by 99%.387
The crystal structures are disclosed of the R61 D,Dpeptidase (inactivated with a peptidoglycan-mimetic
penicillin),388 of a truncated S. pneumoniae Class A
cell division PBP1b enzyme (inactivated with nitrocefin and cefotaxime),389 and of the S. pneumoniae
PBP3 peptidoglycan synthesis regulatory factor.390
Labia reviews the structural evolution of the TEM
and SHV Class A -lactamases.391 Computational
modeling of the Class A -lactamase acylation supports Glu166 as the general base activating Ser70.392
An engineered cystine in the Toho-1 ESBL alters the
active site, reducing activity toward third-generation
cephalosporins.393 The sequence requirements of the
IMP-1 and FEZ-1 metallo--lactamases,394,395 evidence for direct -lactam-metal contact,396 and the
crystal structure of the CphA carbapenemase metallo--lactamase (complexed with biapenem)397 are
discussed. Two reports evaluate the structural basis
for the high affinity -lactamase-BLIP (-lactamase
inhibitory protein) protein-protein complex.398,399
Freiberg et al. discuss the impact of transcriptome
and proteome analysis on antibacterial drug discovery.400 The inactivation mechanism of broad-spectrum methylidene penem -lactamase inhibitors401 is
revealed by the structure of the inactivated -lactamase.402 A series of reviews update recent progress
in -lactam medicinal chemistry.403-409
6. Acknowledgments
9. References
The constructive comments of the referees are

appreciated. This work was supported by Grant
AI33170 from the National Institute of Allergy &
Infectious Disease and by Grant GM61629 from the
National Institute of General Medical Science.
7. Abbreviations
ESBL
IRT
MBL
MIC
MRSA
PBP
rms
extended spectrum -lactamase

inhibitor-resistant TEM (class A) -lactamase
metallo--lactamase
minimal inhibitory concentration of an antibacterial
methicillin-resistant S. aureus
penicillin-binding protein
root-mean square
8. Note Added in Proof

Levy and Marshall377 and Payne and Tomasz378
offer perspectives on the phenomenon of bacterial
resistance, whereas Poole379 provides a complimentary review of bacterial -lactam resistance. Mallorqui-Fernandez et al.,380 Walsh and Amyes,381 and
Gotz382 review aspects of the molecular basis for
MRSA/VRSA and possible new strategies for anti-
(1) Charpentier, E.; Tuomanen, E. Microbes Infect. 2000, 2, 1855.

(2) Walsh, C. T. Nat. Rev. Microbiol. 2003, 1, 65.
(3) Raja, A.; Lebbos, J.; Kirkpatrick, P. Nat. Rev. Drug Discovery
2004, 3, 733.
(4) Courvalin, P.; Davies, J. Curr. Opin. Microbiol. 2003, 6, 425.
(5) Andersson, D. I. Curr. Opin. Microbiol. 2003, 6, 452.
(6) Sutcliffe, J. A. Bioorg. Med. Chem. Lett. 2003, 13, 4159.
(7) Walsh, C. T. Nature 2000, 406, 775.
(8) Wise, R. J. Antimicrob. Chemother. 2004, 54, 306.
(9) Palumbi, S. Science 2001, 293, 1786.
(10) Czaran, T. L.; Hoekstra, R. F.; Pagie, L. Proc. Natl. Acad. Sci.
U.S.A. 2002, 99, 786.
(11) Enright, M. C. Curr. Opin. Pharmacol. 2003, 3, 474.
(12) Lenski, R. E.; Riley, M. A. Proc. Natl. Acad. Sci. U.S.A. 2002,
99, 556.
(13) Challis, G. L.; Hopwood, D. A. Proc. Natl. Acad. Sci. U.S.A. 2003,
100 (Suppl. 2), 14555.
(14) Pratt, R. F. J. Chem. Soc., Perkin Trans. 2 2002, 851.
(15) Hall, B. G.; Barlow, M. Drug Resist. Updates 2004, 7, 111.
(16) Abraham, E. BioEssays 1990, 12, 601.
(17) Bentley, R. Perspect. Biol. Med. 1997, 40, 197.
(18) Bentley, R. Perspect. Biol. Med. 1997, 40, 364.
(19) Williams, D. H.; Stone, M. J.; Hauck, P. R.; Rahman, S. K. J.
Nat. Prod. 1989, 52, 1189.
(20) Maplestone, R. A.; Stone, M. J.; Williams, D. H. Gene 1992, 115,
151.
(21) Stone, M. J.; Williams, D. H. Mol. Microbiol. 1992, 6, 29.
(22) Firn, R. D.; Jones, C. G. Mol. Microbiol. 2000, 37, 989.
(23) Firn, R. D.; Jones, C. G. Nat. Prod. Rep. 2003, 20, 382.
(24) Dalhoff, A.; Thomson, C. J. Chemotherapy 2003, 49, 105.
(25) Waxman, D. J.; Yocum, R. R.; Strominger, J. L. Philos. Trans.
R. Soc. London, Ser. B 1980, 289, 257.
(26) Koch, A. L. Clin. Microbiol. Rev. 2003, 16, 673.

(27) Silvaggi, N. R.; Kaur, K.; Adediran, S. A.; Pratt, R. F.; Kelly, J.
A. Biochemistry 2004, 43, 7046.
(28) Josephine, H. R.; Kumar, I.; Pratt, R. F. J. Am. Chem. Soc. 2004,
126, 8122.
(29) Giesbrecht, P.; Kerstein, T.; Maidhof, H.; Wecke, J. Microbiol.
Mol. Biol. Rev. 1998, 62, 1371.
(30) Sugai, M.; Yamada, S.; Nakashima, S.; Komatsuzawa, H.;
Matsumoto, A.; Oshida, T.; Suginaka, H. J. Bacteriol. 1997, 179,
2958.
(31) Severin, A.; Severina, E.; Tomasz, A. Antimicrob. Agents Chemother. 1997, 41, 504.
(32) Beveridge, T. J. J. Bacteriol. 1999, 181, 4725.
(33) Lee, W.; McDonough, M. A.; Kotra, L. P.; Li, Z. H.; Silvaggi, N.
R.; Takeda, Y.; Kelly, J. A.; Mobashery, S. Proc. Natl. Acad. Sci.
U.S.A. 2001, 98, 1427.
(34) Page, M. I.; Laws, A. P. Chem. Commun. 1998, 1611.
(35) Woodward, R. B. Philos. Trans. R. Soc. London, Ser. B 1980,
289, 239.
(36) Sykes, N. O.; Macdonald, S. J. F.; Page, M. I. J. Med. Chem.
2002, 45, 2850.
(37) Benkovic, S. J.; Hammes-Schiffer, S. Science 2003, 301, 1196.
(38) Kraut, D. A.; Carroll, K. S.; Herschlag, D. Annu. Rev. Biochem.
2003, 72, 517.
(39) Williams, D. H.; Stephens, E.; Zhou, M. Chem. Commun. 2003,
1973.
(40) Rogers, H. J.; Perkins, H. R.; Ward, J. B. Microbial cell walls
and membranes; Chapman and Hall: London, New York, 1980.
(41) Bugg, T. D. H. In Comprehensive Natural Products Chemistry;
Elservier: New York, 1999.
(42) Young, K. D. Mol. Microbiol. 2003, 49, 571.
(43) Popham, D. L.; Young, K. D. Curr. Opin. Microbiol. 2003, 6, 594.
(44) Nanninga, N. Microbiol. Mol. Biol. Rev. 1998, 62, 110.
(45) Nikaido, H. Microbiol. Mol. Biol. Rev. 2003, 67, 593.
(46) Moews, P. C.; Knox, J. R.; Dideberg, O.; Charlier, P.; Frere, J.
M. Proteins: Struct., Funct., Genet. 1990, 7, 156.
(47) Kuzin, A. P.; Liu, H.; Kelly, J. A.; Knox, J. R. Biochemistry 1995,
34, 9532.
(48) Pares, S.; Mouz, N.; Petillot, Y.; Hakenbeck, R.; Dideberg, O.
Nat. Struct. Biol. 1996, 3, 284.
(49) Knox, J. R.; Moews, P. C.; Frere, J. M. Chem. Biol. 1996, 3, 937.
(50) Dideberg, O.; Charlier, P.; Dupont, L.; Vermeire, M.; Frere, J.
M.; Ghuysen, J. M. FEBS Lett. 1980, 117, 212.
(51) Kelly, J. A.; Moews, P. C.; Knox, J. R.; Frere, J. M.; Ghuysen, J.
M. Science 1982, 218, 479.
(52) Knox, J. R.; DeLucia, M. L.; Murthy, N. S.; Kelly, J. A.; Moews,
P. C.; Frere, J. M.; Ghuysen, J. M. J. Mol. Biol. 1979, 127, 217.
(53) Kelly, J. A.; Kuzin, A. P. J. Mol. Biol. 1995, 254, 223.
(54) Rhazi, N.; Charlier, P.; Dehareng, D.; Engher, D.; Vermeire, M.;
Fre`re, J. M.; Nguyen-Diste`che, M.; Fonze, E. Biochemistry 2003,
42, 2895.
(55) Fonze, E.; Vermeire, M.; Nguyen-Disteche, M.; Brasseur, R.;
Charlier, P. J. Biol. Chem. 1999, 274, 21853.
(56) Nicholas, R. A.; Krings, S.; Tomberg, J.; Nicola, G.; Davies, C.
J. Biol. Chem. 2003, 278, 52826.
(57) Lim, D.; Strynadka, N. C. Nat. Struct. Biol. 2002, 9, 870.
(58) Deka, R. K.; Machius, M.; Norgard, M. V.; Tomchick, D. R. J.
Biol. Chem. 2002, 277, 41857.
(59) Davies, C.; White, S. W.; Nicholas, R. A. J. Biol. Chem. 2001,
276, 616.
(60) Massova, I.; Mobashery, S. Antimicrob. Agents Chemother. 1998,
42, 1.
(61) Amanuma, H.; Strominger, J. L. J. Biol. Chem. 1984, 259, 1294.
(62) Scheffers, D. J.; Jones, L. J. F.; Errington, J. Mol. Microbiol.
2004, 51, 749.
(63) Pucci, M. J.; Dougherty, T. J. J. Bacteriol. 2002, 184, 588.
(64) Hartman, B. J.; Tomasz, A. J. Bacteriol. 1984, 158, 513.
(65) Pinho, M. G.; Filipe, S. R.; de Lencastre, H.; Tomasz, A. J.
Bacteriol. 2001, 183, 6525.
(66) Couto, I.; Wu, S. W.; Tomasz, A.; de Lencastre, H. J. Bacteriol.
2003, 185, 645.
(67) Jevons, M. P.; Rolinson, G. N.; Knox, R. Br. Med. J. 1961, 1,
124.
(68) Ito, T.; Katayama, Y.; Asada, K.; Mori, N.; Tsutsumimoto, K.;
Tiensasitorn, C.; Hiramatsu, K. Antimicrob. Agents Chemother.
2001, 45, 1323.
(69) Wu, S. W.; de Lencastre, H.; Tomasz, A. J. Bacteriol. 2001, 183,
2417.
(70) Hartman, B. J.; Tomasz, A. Antimicrob. Agents Chemother. 1986,
29, 85.
(71) Reynolds, P. E.; Brown, D. F. FEBS Lett. 1985, 192, 28.
(72) Utsui, Y.; Yokota, T. Antimicrob. Agents Chemother. 1985, 28,
397.
(73) Fuda, C.; Suvorov, M.; Vakulenko, S. B.; Mobashery, S. J. Biol.
Chem. 2004, 279, 40802.
(74) Lu, W. P.; Kincaid, E.; Sun, Y.; Bauer, M. D. J. Biol. Chem. 2001,
276, 31494.
(75) Pernot, L.; Chesnel, L.; Legouellec, A.; Croize, J.; Vernet, T.;
Dideberg, O.; Dessen, A. J. Biol. Chem. 2004, 279, 16463.
Fisher et al.
(76) Chesnel, L.; Pernot, L.; Lemaire, D.; Champelovier, D.; Croize,
J.; Dideberg, O.; Vernet, T.; Zapun, A. J. Biol. Chem. 2003, 278,
44448.
(77) Rice, L. B.; Bellais, S.; Carias, L. L.; Hutton-Thomas, R.; Bonomo,
R. A.; Caspers, P.; Page, M. G.; Gutmann, L. Antimicrob. Agents
Chemother. 2004, 48, 3028.
(78) Sifaoui, F.; Arthur, M.; Rice, L.; Gutmann, L. Antimicrob. Agents
Chemother. 2001, 45, 2594.
(79) Jones, R. N.; Marshall, S. A.; Pfaller, M. A.; Wilke, W. W.; Hollis,
R. J.; Erwin, M. E.; Edmond, M. B.; Wenzel, R. P. Diagn.
Microbiol. Infect. Dis. 1997, 29, 95.
(80) Sauvage, E.; Kerff, F.; Fonze, E.; Herman, R.; Schoot, B.;
Marquette, J. P.; Taburet, Y.; Prevost, D.; Dumas, J.; Leonard,
G.; Stefanic, P.; Coyette, J.; Charlier, P. Cell Mol. Life Sci. 2002,
59, 1223.
(81) Mouz, N.; Di Guilmi, A. M.; Gordon, E.; Hakenbeck, R.; Dideberg,
O.; Vernet, T. J. Biol. Chem. 1999, 274, 19175.
(82) Salerno, A. J.; Lampen, J. O. J. Bacteriol. 1986, 166, 769.
(83) McKinney, T. K.; Sharma, V. K.; Craig, W. A.; Archer, G. L. J.
Bacteriol. 2001, 183, 6862.
(84) Zhu, Y. F.; Curran, I. H.; Joris, B.; Ghuysen, J. M.; Lampen, J.
O. J. Bacteriol. 1990, 172, 1137.
(85) Hardt, K.; Joris, B.; Lepage, S.; Brasseur, R.; Lampen, J. O.;
Frere, J. M.; Fink, A. L.; Ghuysen, J. M. Mol. Microbiol. 1997,
23, 935.
(86) Kobayashi, T.; Zhu, Y. F.; Nicholls, N. J.; Lampen, J. O. J.
Bacteriol. 1987, 169, 3873.
(87) Clarke, S. R.; Dyke, K. G. Microbiology 2001, 147, 803.
(88) Finan, J. E.; Archer, G. L.; Pucci, M. J.; Climo, M. W. Antimicrob.
Agents Chemother. 2001, 45, 3070.
(89) Niemeyer, D. M.; Pucci, M. J.; Thanassi, J. A.; Sharma, V. K.;
Archer, G. L. J. Bacteriol. 1996, 178, 5464.
(90) Zhang, H. Z.; Hackbarth, C. J.; Chansky, K. M.; Chambers, H.
F. Science 2001, 291, 1962.
(91) Ender, M.; McCallum, N.; Adhikari, R.; Berger-Bachi, B. Antimicrob. Agents Chemother. 2004, 48, 2295.
(92) Gregory, P. D.; Lewis, R. A.; Curnock, S. P.; Dyke, K. G. Mol.
Microbiol. 1997, 24, 1025.
(93) Golemi-Kotra, D.; Cha, J. Y.; Meroueh, S. O.; Vakulenko, S. B.;
Mobashery, S. J. Biol. Chem. 2003, 278, 18419.
(94) Kerff, F.; Charlier, P.; Colombo, M. L.; Sauvage, E.; Brans, A.;
Frere, J. M.; Joris, B.; Fonze, E. Biochemistry 2003, 42, 12835.
(95) Golemi, D.; Maveyraud, L.; Vakulenko, S.; Samama, J. P.;
Mobashery, S. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 14280.
(96) Maveyraud, L.; Golemi, D.; Kotra, L. P.; Tranier, S.; Vakulenko,
S.; Mobashery, S.; Samama, J. P. Struct. Fold. Des. 2000, 8,
1289.
(97) Sun, T.; Nukaga, M.; Mayama, K.; Braswell, E. H.; Knox, J. R.
Protein Sci. 2003, 12, 82.
(98) Wilke, M. S.; Hills, T. L.; Zhang, H. Z.; Chambers, H. F.;
Strynadka, N. C. J. Biol. Chem. 2004, 47278.
(99) Birck, C.; Cha, J. Y.; Cross, J.; Schulze-Briese, C.; Meroueh, S.
O.; Schlegel, H. B.; Mobashery, S.; Samama, J. P. J. Am. Chem.
Soc. 2004, 126, 13945.
(100) Hanique, S.; Colombo, M. L.; Goormaghtigh, E.; Soumillion, P.;
Frere, J. M.; Joris, B. J. Biol. Chem. 2004, 279, 14264.
(101) Filee, P.; Benlafya, K.; Delmarcelle, M.; Moutzourelis, G.; Frere,
J. M.; Brans, A.; Joris, B. Mol. Microbiol. 2002, 44, 685.
(102) Melckebeke, H. V.; Vreuls, C.; Gans, P.; Filee, P.; Llabres, G.;
Joris, B.; Simorre, J. P. J. Mol. Biol. 2003, 333, 711.
(103) Garcia-Castellanos, R.; Marrero, A.; Mallorqui-Fernandez, G.;
Potempa, J.; Coll, M.; Gomis-Ruth, F. X. J. Biol. Chem. 2003,
278, 39897.
(104) Garcia-Castellanos, R.; Mallorqui-Fernandez, G.; Marrero, A.;
Potempa, J.; Coll, M.; Gomis-Ruth, F. X. J. Biol. Chem. 2004,
279, 17888.
(105) Bush, K.; Jacoby, G. A.; Medeiros, A. A. Antimicrob. Agents
Chemother. 1995, 39, 1211.
(106) Bush, K.; Mobashery, S. In Resolving the Antibiotic Paradox;
Rosen, B. P., Mobashery, S., Eds.; Kluwer Academic/Plenum
Publishers: New York, 1998; Vol. 456.
(107) Bryant, R. A. R.; Hansen, D. E. J. Am. Chem. Soc. 1996, 118,
5498.
(108) Radzicka, A.; Wolfenden, R. J. Am. Chem. Soc. 1996, 118, 6105.
(109) Thomson, K. S.; Moland, E. S. Microbes Infect. 2000, 2, 1225.
(110) Kotra, L.; Samama, J. P.; Mobashery, S. In Bacterial Resistance
to Antimicrobials, Mechanisms, Genetics, Medical Practice and
Public Health; Lewis, K., Salyers, A. A., Haber, H. W., Wax, R.
G., Eds.; Marcel Dekkar, Inc.: New York, 2002.
(111) Cha, J. Y.; Ishiwata, A.; Mobashery, S. J. Biol. Chem. 2004, 279,
14917.
(112) Nielsen, B. K.; Lampen, J. O. J. Biol. Chem. 1982, 257, 490.
(113) Meroueh, S. O.; Minasov, G.; Lee, W.; Shoichet, B. K.; Mobashery, S. J. Am. Chem. Soc. 2003, 125, 9612.
(114) Hardy, L. W.; Kirsch, J. F. Biochemistry 1984, 23, 1275.
(115) Bulychev, A.; Mobashery, S. Antimicrob. Agents Chemother.
1999, 43, 1743.
(116) Page, M. I.; Laws, A. P. Tetrahedron 2000, 56, 5631.

(117) Hokenson, M. J.; Cope, G. A.; Lewis, E. R.; Oberg, K. A.; Fink,
A. L. Biochemistry 2000, 39, 6538.
(118) Wang, X.; Minasov, G.; Shoichet, B. K. Proteins: Struct., Funct.,
Genet. 2002, 47, 86.
(119) Wang, X.; Minasov, G.; Shoichet, B. K. J. Mol. Biol. 2002, 320,
85.
(120) Hujer, A. M.; Hujer, K. M.; Helfand, M. S.; Anderson, V. E.;
Bonomo, R. A. Antimicrob. Agents Chemother. 2002, 46, 3971.
(121) Helfand, M. S.; Bethel, C. R.; Hujer, A. M.; Hujer, K. M.;
Anderson, V. E.; Bonomo, R. A. J. Biol. Chem. 2003, 278, 52724.
(122) Weldhagen, G. F.; Poirel, L.; Nordmann, P. Antimicrob. Agents
Chemother. 2003, 47, 2385.
(123) Bonnet, R. Antimicrob. Agents Chemother. 2004, 48, 1.
(124) Bradford, P. A. Clin. Rev. Microbiol. 2001, 14, 933.
(125) Davies, C.; White, S. W.; Nicholas, R. A. J. Biol. Chem. 2000,
276, 616.
(126) Tranier, S.; Bouthors, A. T.; Maveyraud, L.; Guillet, V.; Sougakoff, W.; Samama, J. P. J. Biol. Chem. 2000, 275, 28075.
(127) Moult, J.; Sawyer, L.; Herzberg, O.; Jones, C. L.; Coulson, A. F.;
Green, D. W.; Harding, M. M.; Ambler, R. P. Biochem. J. 1985,
225, 167.
(128) Jelsch, C.; Lenfant, F.; Masson, J. M.; Samama, J. P. FEBS Lett.
1992, 299, 135.
(129) Jelsch, C.; Mourey, L.; Masson, J. M.; Samama, J. P. Proteins
1993, 16, 364.
(130) Knox, J. R.; Moews, P. C. J. Mol. Biol. 1991, 220, 435.
(131) Strynadka, N. C.; Adachi, H.; Jensen, S. E.; Johns, K.; Sielecki,
A.; Betzel, C.; Sutoh, K.; James, M. N. Nature 1992, 359, 700.
(132) Tipper, D. J.; Strominger, J. L. Proc. Natl. Acad. Sci. U.S.A.
1965, 54, 1133.
(133) Minasov, G.; Wang, X.; Shoichet, B. K. J. Am. Chem. Soc. 2002,
124, 5333.
(134) Fujii, Y.; Okimoto, N.; Hata, M.; Narumi, T.; Yasuoka, K.;
Susukita, R.; Suenaga, A.; Futatsugi, N.; Koishi, T.; Furusawa,
H.; Kawai, A.; Ebisuzaki, T.; Neya, S.; Hoshino, T. J. Phys.
Chem. B 2003, 107, 10274.
(135) Hermann, J. C.; Ridder, L.; Mulholland, A. J.; Holtje, H. D. J.
Am. Chem. Soc. 2003, 125, 9590.
(136) Nukaga, M.; Mayama, K.; Hujer, A. M.; Bonomo, R. A.; Knox,
J. R. J. Mol. Biol. 2003, 328, 289.
(137) Guillaume, G.; Vanhove, M.; Lamotte-Brasseur, J.; Ledent, P.;
Jamin, M.; Joris, B.; Frere, J. M. J. Biol. Chem. 1997, 272, 5438.
(138) Lamotte-Brasseur, J.; Dive, G.; Dideberg, O.; Charlier, P.; Frere,
J. M.; Ghuysen, J. M. Biochem. J. 1991, 279, 213.
(139) Damblon, C.; Raquet, X.; Lian, L. Y.; Lamotte-Brasseur, J.;
Fonze, E.; Charlier, P.; Roberts, G. C. K.; Frere, J. M. Proc. Natl.
Acad. Sci. U.S.A. 1996, 93, 1747.
(140) Lamotte-Brasseur, J.; Lounnas, V.; Raquet, X.; Wade, R. C.
Protein Sci. 1999, 8, 404.
(141) Raquet, X.; Lounnas, V.; Lamotte-Brasseur, J.; Frere, J. M.;
Wade, R. C. Biophys. J. 1997, 73, 2416.
(142) Swaren, P.; Maveyraud, L.; Guillet, V.; Masson, J. M.; Mourey,
L.; Samama, J. P. Structure 1995, 3, 603.
(143) Golemi-Kotra, D.; Meroueh, S. O.; Kim, C.; Vakulenko, S. B.;
Bulychev, A.; Stemmler, A. J.; Stemmler, T. L.; Mobashery, S.
J. Biol. Chem. 2004, 34665.
(144) Finlay, J.; Miller, L.; Poupard, J. A. J. Antimicrob. Chemother.
2003, 52, 18.
(145) Imtiaz, U.; Billings, E. M.; Knox, J. R.; Mobashery, S. Biochemistry 1994, 33, 5728.
(146) Imtiaz, U.; Billings, E.; Knox, J. R.; Manavathu, E. K.; Lerner,
S. A.; Mobashery, S. J. Am. Chem. Soc. 1993, 115, 4435.
(147) Yang, Y.; Janota, K.; Tabei, K.; Huang, N.; Siegel, M. M.; Lin,
Y. I.; Rasmussen, B. A.; Shlaes, D. M. J. Biol. Chem. 2000, 275,
26674.
(148) Kuzin, A. P.; Nukaga, M.; Nukaga, Y.; Hujer, A.; Bonomo, R.
A.; Knox, J. R. Biochemistry 2001, 40, 1861.
(149) Helfand, M. S.; Totir, M. A.; Carey, P.; Hujer, A. M.; Bonomo,
R. A.; Carey, P. R. Biochemistry 2003, 42, 13386.
(150) Padayatti, P. S.; Helfand, M. S.; Totir, M. A.; Carey, M. P.; Hujer,
A. M.; Carey, P. R.; Bonomo, R. A.; van den Akker, F. Biochemistry 2004, 43, 843.
(151) Swaren, P.; Golemi, D.; Cabantous, S.; Bulychev, A.; Maveyraud,
L.; Mobashery, S.; Samama, J. P. Biochemistry 1999, 38, 9570.
(152) Meroueh, S. O.; Roblin, P.; Golemi, D.; Maveyraud, L.; Vakulenko, S. B.; Zhang, Y.; Samama, J. P.; Mobashery, S. J. Am.
Chem. Soc. 2002, 124, 9422.
(153) Madec, S.; Blin, C.; Krishnamoorthy, R.; Picard, B.; Chaibi, E.
B.; Fouchereau-Peron, M.; Labia, R. FEMS Microbiol. Lett. 2002,
211, 13.
(154) Wang, X.; Minasov, G.; Shoichet, B. K. J. Biol. Chem. 2002, 277,
24744.
(155) Wang, X.; Minasov, G.; Blazquez, J.; Caselli, E.; Prati, F.;
Shoichet, B. K. Biochemistry 2003, 42, 8434.
(156) Doucet, N.; De Wals, P. Y.; Pelletier, J. N. J. Biol. Chem. 2004,
279, 46295.
(157) Pagan-Rodriguez, D.; Zhou, X.; Simmons, R.; Bethel, C. R.; Hujer,
A. M.; Helfand, M. S.; Jin, Z.; Guo, B.; Anderson, V. E.; Lily,
M.; Ng, L. M.; Bonomo, R. A. J. Biol. Chem. 2004, 279, 19494.

(158) Chaibi, E. B.; Sirot, D.; Paul, G.; Labia, R. J. Antimicrob.
Chemother. 1999, 43, 447.
(159) Lavigne, J. P.; Bonnet, R.; Michaux-Charachon, S.; Jourdan, J.;
Caillon, J.; Sotto, A. J. Antimicrob. Chemother. 2004, 53, 616.
(160) Nukaga, M.; Abe, T.; Venkatesan, A. M.; Mansour, T. S.; Bonomo,
R. A.; Knox, J. R. Biochemistry 2003, 42, 13152.
(161) Beharry, Z.; Chen, H.; Gadhachanda, V. R.; Buynak, J. D.;
Palzkill, T. Biochem. Biophys. Res. Commun. 2004, 313, 541.
(162) Bret, L.; Chaibi, E. B.; Chanal-Claris, C.; Sirot, D.; Labia, R.;
Sirot, J. Antimicrob. Agents Chemother. 1997, 41, 2547.
(163) Delaire, M.; Labia, R.; Samama, J. P.; Masson, J. M. J. Biol.
Chem. 1992, 267, 20600.
(164) Zafaralla, G.; Manavathu, E. K.; Lerner, S. A.; Mobashery, S.
Biochemistry 1992, 31, 3847.
(165) Delaire, M.; Labia, R.; Samama, J. P.; Masson, J. M. J. Biol.
Chem. 1992, 267, 20600.
(166) Schroeder, W. A.; Locke, T. R.; Jensen, S. E. Antimicrob. Agents
Chemother. 2002, 46, 3568.
(167) Randegger, C. C.; Hachler, H. J. Antimicrob. Chemother. 2001,
47, 547.
(168) Stapleton, P. D.; Shannon, K. P.; French, G. L. Antimicrob.
(169) Vakulenko, S. B.; Taibi-Tronche, P.; Toth, M.; Massova, I.;
Lerner, S. A.; Mobashery, S. J. Biol. Chem. 1999, 274, 23052.
(170) Vakulenko, S.; Golemi, D. Antimicrob. Agents Chemother. 2002,
46, 646.
(171) Maveyraud, L.; Mourey, L.; Kotra, L. P.; Pedelacq, J. D.; Guillet,
V.; Mobashery, S.; Samama, J. P. J. Am. Chem. Soc. 1998, 120,
9748.
(172) Vilanova, B.; Donoso, J.; Frau, J.; Munoz, F. Helv. Chim. Acta
1999, 82, 1274.
(173) Fonze, E.; Vanhove, M.; Dive, G.; Sauvage, E.; Fre`re, J. M.;
Charlier, P. Biochemistry 2002, 41, 1877.
(174) Sougakoff, W.; Naas, T.; Nordmann, P.; Collatz, E.; Jarlier, V.
Biochim. Biophys. Acta 1999, 1433, 153.
(175) Sougakoff, W.; LHermite, G.; Pernot, L.; Naas, T.; Guillet, V.;
Nordmann, P.; Jarlier, V.; Delettre, J. Acta Crystallogr., Sect.
D 2002, 58, 267.
(176) Majiduddin, F. K.; Palzkill, T. Antimicrob. Agents Chemother.
2003, 47, 1062.
(177) Mourey, L.; Miyashita, K.; Swaren, P.; Bulychev, A.; Samama,
J. P.; Mobashery, S. J. Am. Chem. Soc. 1998, 120, 9382.
(178) Hujer, A. M.; Hujer, K. M.; Bonomo, R. A. Biochim. Biophys.
Acta 2001, 1547, 37.
(179) Cantu, C., III; Palzkill, T. J. Biol. Chem. 1998, 273, 26603.
(180) Orencia, M. C.; Yoon, J. S.; Ness, J. E.; Stemmer, W. P. C.;
Stevens, R. C. Nat. Struct. Biol. 2001, 8, 238.
(181) Nukaga, K.; Mayama, K.; Crichlow, G. V.; Knox, J. R. J. Mol.
Biol. 2002, 317, 109.
(182) Shimamura, T.; Ibuka, A. S.; Fushinobu, S.; Wakagi, T.; Ishiguro,
M.; Ishii, Y.; Matsuzawa, H. J. Biol. Chem. 2002, 277, 46601.
(183) Ibuka, A. S.; Ishii, Y.; Galleni, M.; Ishiguro, M.; Yamaguchi, K.;
Fre`re, J. M.; Matsuzawa, H.; Sakai, H. Biochemistry 2003, 42,
10634.
(184) Hall, A.; Knowles, J. R. Nature 1976, 264, 803.
(185) Stemmer, W. P. C. J. Mol. Catal. B 2002, 19, 3.
(186) Camps, M.; Naukkarinen, J.; Johnson, B. P.; Loeb, L. A. Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 9727.
(187) Vakulenko, S. B.; Geryk, B.; Kotra, L. P.; Mobashery, S.; Lerner,
S. A. Antimicrob. Agents Chemother. 1998, 42, 1542.
(188) Barlow, M.; Hall, B. G. Genetics 2003, 163, 1237.
(189) Barlow, M.; Hall, B. G. Genetics 2003, 164, 23.
(190) Hall, B. G. Nat. Rev. Microbiol. 2004, 2, 430.
(191) Lobkovsky, E.; Moews, P. C.; Liu, H.; Zhao, H.; Frere, J. M.;
Knox, J. R. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 11257.
(192) Oefner, C.; DArcy, A.; Daly, J. J.; Gubernator, K.; Charnas, R.
L.; Heinze, I.; Hubschwerlen, C.; Winkler, F. K. Nature 1990,
343, 284.
(193) Gherman, B. F.; Goldberg, S. D.; Cornish, V. W.; Friesner, R. A.
J. Am. Chem. Soc. 2004, 126, 7652.
(194) Bulychev, A.; Massova, I.; Miyashita, K.; Mobashery, S. J. Am.
Chem. Soc. 1997, 119, 7619.
(195) Patera, A.; Blaszczak, L. C.; Shoichet, B. K. J. Am. Chem. Soc.
2000, 122, 10504.
(196) Rice, L. B.; Bonomo, R. A. Drug Resist. Update 2000, 3, 178.
(197) Bauernfeind, A.; Chong, Y.; Schweighart, S. Infection 1989, 17,
316.
(198) Bauernfeind, A.; Chong, Y.; Lee, K. Yonsei Med. J. 1998, 39,
520.
(199) Livermore, D. M. Clin. Microbiol. Rev. 1995, 8, 557.
(200) Philippon, A.; Arlet, G.; Lagrange, P. H. Eur. J. Clin. Microbiol.
Infect. Dis. 1994, 13 (Suppl 1), S17.
(201) Galleni, M.; Amicosante, G.; Frere, J. M. Biochem. J. 1988, 255,
123.
(202) Kato-Toma, Y.; Iwashita, T.; Masuda, K.; Oyama, Y.; Ishiguro,
M. Biochem. J. 2003, 371, 175.
(203) Lamotte-Brasseur, J.; Dubus, A.; Wade, R. C. Proteins 2000, 40,
23.

(204) Dubus, A.; Normark, S.; Kania, M.; Page, M. G. Biochemistry
1994, 33, 8577.
(205) Knothe, H.; Shah, P.; Krcmery, V.; Antal, M.; Mitsuhashi, S.
Infection 1983, 11, 315.
(206) Nukaga, M.; Haruta, S.; Tanimoto, K.; Kogure, K.; Taniguchi,
K.; Tamaki, M.; Sawai, T. J. Biol. Chem. 1995, 270, 5729.
(207) Crichlow, G. V.; Kuzin, A. P.; Nukaga, M.; Mayama, K.; Sawai,
T.; Knox, J. R. Biochemistry 1999, 38, 10256.
(208) Powers, R. A.; Caselli, E.; Focia, P. J.; Prati, F.; Shoichet, B. K.
(209) Nukaga, M.; Kumar, S.; Nukaga, K.; Pratt, R. F.; Knox, J. R. J.
Biol. Chem. 2004, 279, 9344.
(210) Nordmann, P.; Poirel, L. Clin. Microbiol. Infect. 2002, 8, 321.
(211) Heritier, C.; Poirel, L.; Nordmann, P. Antimicrob. Agents
Chemother. 2004, 48, 1670.
(212) Bou, G.; Martnez-Beltran, J. Antimicrob. Agents Chemother.
2000, 44, 428.
(213) Poirel, L.; Girlich, D.; Naas, T.; Nordmann, P. Antimicrob. Agents
Chemother. 2001, 45, 447.
(214) Toleman, M. A.; Rolston, K.; Jones, R. N.; Walsh, T. R. Antimicrob. Agents Chemother. 2003, 47, 2859.
(215) Dubois, V.; Arpin, C.; Quentin, C.; Texier-Maugein, J.; Poirel,
L.; Nordmann, P. Antimicrob. Agents Chemother. 2003, 47, 2380.
(216) Oliver, A.; Weigel, L. M.; Rasheed, J. K.; McGowan, J. E.; Raney,
P.; Tenover, F. C. Antimicrob. Agents Chemother. 2002, 46, 3829.
(217) Bonnet, R.; Marchandin, H.; Chanal, C.; Sirot, D.; Labia, R.; De
Champs, C.; Jumas-Bilak, E.; Sirot, D. Antimicrob. Agents
Chemother. 2002, 46, 2004.
(218) Orman, B. E.; Pineiro, S. A.; Arduino, S.; Galas, M.; Melano, R.;
Caffer, M. I.; Sordelli, D. O.; Centron, D. Antimicrob. Agents
Chemother. 2002, 46, 3963.
(219) Melano, R.; Corso, A.; Petroni, A.; Centron, D.; Orman, B.;
Pereyra, A.; Moreno, N.; Galas, M. J. Antimicrob. Chemother.
2003, 52, 36.
(220) Poirel, L.; Heritier, C.; Nordmann, P. Antimicrob. Agents
Chemother. 2004, 48, 348.
(221) Bou, G.; Cervero, G.; Dominguez, M. A.; Quereda, C.; MartinezBeltran, J. J. Clin. Microbiol. 2000, 38, 3299.
(222) Afzal-Shah, M.; Woodford, N.; Livemore, D. M. Antimicrob.
(223) Danes, C.; Navia, M. N.; Ruiz, J.; Marco, F.; Jurado, A.; Jimenez
de Anta, M. T.; Vila, J. J. Antimicrob. Chemother. 2002, 50, 261.
(224) Lopez-Otsoa, F.; Gallego, L.; Towner, K. J.; Tysall, L.; Woodford,
N.; Livermore, D. M. J. Clin. Microbiol. 2002, 40, 4741.
(225) Dalla-Costa, L. M.; Coelho, J. M.; Souza, H. A. P.; Castro, M. E.
S.; Stier, C. J. N.; Bragagnolo, K. L.; Rea-Neto, A.; PenteadoFilho, S. R.; Livermore, D. M.; Woodford, N. J. Clin. Microbiol.
2003, 41, 3403.
(226) Heritier, C.; Poirel, L.; Aubert, D.; Nordmann, P. Antimicrob.
(227) Mushtaq, S.; Ge, Y.; Livermore, D. Antimicrob. Agents Chemother. 2004, 48, 1313.
(228) Zarrilli, R.; Crispino, M.; Bagattini, M.; Barretta, E.; Di Popolo,
A.; Triassi, A. M.; Villari, P. J. Clin. Microbiol. 2004, 42, 946.
(229) Livermore, D. M.; Woodford, N. Curr. Opin. Microbiol. 2000, 3,
489.
(230) Danel, F.; Frere, J. M.; Livemore, D. M. Biochim. Biophys. Acta
2001, 1546, 132.
(231) Franceschini, N.; Segatore, B.; Perilli, M.; Vessilier, S.; Franchino, L.; Amicosante, G. J. Antimicrob. Chemother. 2002, 49, 395.
(232) Pernot, L.; Frenois, F.; Rybkine, T.; LHermite, G.; Petrella, S.;
Delettre, J.; Jarlier, V.; Collatz, E.; Sougakoff, W. J. Mol. Biol.
2001, 310, 859.
(233) Danel, F.; Paetzel, M.; Strynadka, N. C. J.; Page, M. G. P.
(234) Franceschini, N.; Boschi, L.; Pollini, S.; Herman, R.; Perilli, M.;
Galleni, M.; Frere, J. M.; Amicosante, G.; Rossolini, G. M.
Antimicrob. Agents Chemother. 2001, 45, 3509.
(235) Paetzel, M.; Danel, F.; de Castro, L.; Mosimann, S. C.; Page, M.
G. P.; Strynadka, N. C. J. Nat. Struct. Biol. 2000, 7, 918.
(236) Golemi, D.; Maveyraud, L.; Vakulenko, S.; Tranier, S.; Ishiwata,
A.; Kotra, L. P.; Samama, J. P.; Mobashery, S. J. Am. Chem.
Soc. 2000, 122, 6132.
(237) Maveyraud, L.; Golemi-Kotra, D.; Ishiwata, A.; Meroueh, O.;
Mobashery, S.; Samama, J. P. J. Am. Chem. Soc. 2002, 124,
2461.
(238) Donald, H. M.; Scaife, W.; Amyes, S. G. B.; Young, H. K.
(239) Barlow, M.; Hall, B. G. J. Mol. Evol. 2002, 55, 314.
(240) Zhu, Y.; Englebert, S.; Joris, B.; Ghuysen, J. M.; Kobayashi, T.;
Lampen, J. O. J. Bacteriol. 1992, 174, 6171.
(241) Colombo, M. L.; Hanique, S.; Baurin, S. L.; Bauvois, C.; De
Vriendt, K.; Van Beeumen, J. J.; Fre`re, J. M.; Joris, B. Antimicrob. Agents Chemother. 2004, 48, 484.
(242) Garau, G.; Garcia-Saez, I.; Bebrone, C.; Anne, C.; Mercuri, P.;
Galleni, M.; Fre`re, J. M.; Dideberg, O. Antimicrob. Agents
Chemother. 2004, 48, 2347.
Fisher et al.
(243) Galleni, M.; Lamotte-Brasseur, J.; Rossolini, G. M.; Spencer, J.;
Dideberg, O.; Fre`re, J. M. Antimicrob. Agents. Chemother. 2001,
45, 660.
(244) Gniadkowski, M. Clin. Microbiol. Infect. 2001, 7, 597.
(245) Fast, W.; Wang, Z.; Benkovic, S. J. Biochemistry 2001, 40, 1640.
(246) Fabiane, S. M.; Sohi, M. K.; Wan, T.; Payne, D. J.; Bateson, J.
H.; Mitchell, T.; Sutton, B. J. Biochemistry 1998, 37, 12404.
(247) Wommer, S.; Rival, S.; Hein, U.; Galleni, M.; Fre`re, J. M.;
Franceschini, N.; Amicosante, G.; Rasmussen, B.; Bauer, R.;
Adolph, H. W. J. Biol. Chem. 2002, 277, 24142.
(248) Concha, N. O.; Janson, C. A.; Rowling, P.; Pearson, S.; Cheever,
C. A.; Clarke, B. P.; Lewis, C.; Galleni, M.; Fre`re, J. M.; Payne,
D. J.; Bateson, J. H.; Abdel-Meguid, S. S. Biochemistry 2000,
39, 4288.
(249) Gomes, C. M.; Frazao, C.; Xavier, A. V.; Legall, J.; Teixeira, M.
Protein Sci. 2002, 11, 707.
(250) Schilling, O.; Wenzel, N.; Naylor, M.; Vogel, A.; Crowder, M. W.;
Makaroff, C.; Meyer-Klaucke, W. Biochemistry 2003, 42, 11777.
(251) Spencer, J.; Clarke, A. R.; Walsh, T. R. J. Biol. Chem. 2001, 276,
33638.
(252) Rasia, R. M.; Vila, A. J. J. Biol. Chem. 2004, 279, 26046.
(253) Ullah, J. H.; Walsh, T. R.; Taylor, I. A.; Emery, D. C.; Verma,
C. S.; Gamblin, S. J.; Spencer, J. J. Mol. Biol. 1998, 284, 125.
(254) Rasia, R. M.; Ceoln, M.; Vila, A. J. Protein Sci. 2003, 12, 1538.
(255) Garrity, J. D.; Carenbauer, A. L.; Herron, L. R.; Crowder, M.
W. J. Biol. Chem. 2004, 279, 920.
(256) Dal Peraro, M.; Vila, A. J.; Carloni, P. Proteins: Struct., Funct.,
Bioinform. 2004, 54, 412.
(257) Dal Peraro, M.; Vila, A. J.; Carloni, P. Inorg. Chem. 2003, 42,
4245.
(258) Garceia-Saez, I.; Mercuri, P. S.; Papamicael, C.; Kahn, R.; Fre`re,
J. M.; Galleni, M.; Rossolini, G. M.; Dideberg, O. J. Mol. Biol.
2003, 325, 651.
(259) Garceia-Saez, I.; Hopkins, J.; Papamicael, C.; Franceschini, N.;
Amicosante, G.; Rossolini, G. M.; Galleni, M.; Fre`re, J. M.;
Dideberg, O. J. Biol. Chem. 2003, 278, 23868.
(260) Bounaga, S.; Laws, A. P.; Galleni, M.; Page, M. I. Biochem. J.
1998, 331, 703.
(261) Bush, K. Clin. Infect. Dis. 2001, 32, 1085.
(262) Miriagou, V.; Tzelepi, E.; Gianneli, D.; Tzouvelekis, L. S.
(263) Yanchak, M. P.; Taylor, R. A.; Crowder, M. W. Biochemistry
2000, 39, 11330.
(264) Krauss, M.; Gilson, H. S. R.; Gresh, N. J. Phys. Chem. B 2001,
105, 8040.
(265) Rasia, R. M.; Vila, A. J. Biochemistry 2002, 41, 1853.
(266) Lipton, A. S.; Heck, R. W.; Ellis, P. D. J. Am. Chem. Soc. 2004,
126, 4735.
(267) Montoya-Pelaez, P. J.; Brown, R. S. Inorg. Chem. 2002, 41, 309.
(268) Montoya-Pelaez, P. J.; Gibson, G. T. T.; Neverov, A. A.; Brown,
R. S. Inorg. Chem. 2003, 42, 8624.
(269) Daz, N.; Sordo, T. L.; Suarez, D.; Mendez, R.; Martn-Villacorta,
J. New J. Chem. 2004, 28, 15.
(270) Diaz, N.; Suarez, D.; Merz, K. M., Jr. J. Am. Chem. Soc. 2001,
123, 9867.
(271) Moali, C.; Anne, C.; Lamotte-Brasseur, J.; Groslambert, S.;
Devreese, B.; Van Beeumen, J.; Galleni, M.; Fre`re, J. M. Chem.
Biol. 2003, 10, 319.
(272) Oelschlaeger, P.; Schmid, R. D.; Pleiss, J. Biochemistry 2003,
42, 8945.
(273) Luzzaro, F.; Docquier, J. D.; Colinon, C.; Endimiani, A.; Lombardi, G.; Amicosante, G.; Rossolini, G.; Toniolol, A. Antimicrob.
(274) Weldhagen, G. F. Int. J. Antimicrob. Agents 2004, 23, 556.
(275) Docquier, J. D.; Riccio, M. L.; Mugnaioli, C.; Luzzaro, F.;
Endimiani, A.; Toniolo, A.; Amicosante, G.; Rossolini, G. M.
(276) Iyobe, S.; Kusadokoro, H.; Ozaki, J.; Matsumura, N.; Minami,
S.; Haruata, S.; Sawai, T.; OHara, K. Antimicrob. Agents
Chemother. 2000, 44, 2023.
(277) Yano, H.; Kuga, A.; Okamoto, R.; Kitasato, H.; Kobayashi, T.;
Inoue, M. Antimicrob. Agents Chemother. 2001, 45, 1343.
(278) Materon, I. C.; Palzkill, T. Protein Sci. 2001, 10, 2556.
(279) Hall, B. G. Antimicrob. Agents Chemother. 2004, 48, 1032.
(280) Docquier, J. D.; Lamotte-Brasseur, J.; Galleni, M.; Amicosante,
G.; Frere, J. M.; Rossolini, G. M. J. Antimicrob. Chemother. 2003,
51, 257.
(281) Toleman, M. A.; Rolston, K.; Jones, R. N.; Walsh, T. R. Antimicrob. Agents Chemother. 2004, 48, 329.
(282) Cornaglia, G.; Mazzariol, A.; Lauretti, L.; Rossolini, M.; Fontana,
R. Clin. Infect. Dis. 2000, 31, 1119.
(283) Yatsuyanagi, J.; Saito, S.; Harata, S.; Suzuki, N.; Ito, Y.; Amano,
K.; Enomoto, K. Antimicrob. Agents Chemother. 2004, 48, 626.
(284) Siemann, S.; Clarke, A. J.; Viswanatha, T.; Dmitrienko, G. I.
(285) Tsang, W. Y.; Dhanda, A.; Schofield, C. J.; Fre`re, J. M.; Galleni,
M.; Page, M. I. Bioorg. Med. Chem. Lett. 2004, 2004, 1737.

(286) Buynak, J. D.; Chen, H.; Vogeti, L.; Gadhachanda, V. R.;
Buchanan, C. A.; Palzkill, T.; Shaw, R. W.; Spencer, J.; Walsh,
T. R. Bioorg. Med. Chem. Lett. 2004, 14, 1299.
(287) Heinz, U.; Bauer, R.; Wommer, S.; Meyer-Klaucke, W.; Papamichaels, C.; Bateson, J.; Adolph, H. W. J. Biol. Chem. 2003, 278,
20659.
(288) Mollard, C.; Moali, C.; Papamicael, C.; Damblon, C.; Vessilier,
S.; Amicosante, G.; Schofield, C. J.; Galleni, N.; Fre`re, J. M.;
Roberts, G. C. K. J. Biol. Chem. 2001, 276, 45015.
(289) Goh, E. B.; Yim, G.; Tsui, W.; McClure, J.; Surette, M. G.; Davies,
J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 17025.
(290) Kaldalu, N.; Mei, R.; Lewis, K. Antimicrob. Agents Chemother.
2004, 48, 890.
(291) Bandow, J. E.; Brotz, H.; Leichert, L. I. O.; Labischinski, H.;
Hecker, M. Antimicrob. Agents Chemother. 2003, 47, 948.
(292) Mills, S. D. J. Antimicrob. Chemother. 2003, 51, 749.
(293) Bradford, P. A.; Urban, C.; Mariano, N.; Projan, S. J.; Rahal, J.
J.; Bush, K. Antimicrob. Agents Chemother. 1997, 41, 563.
(294) Cao, V. T. B.; Arlet, G.; Ericsson, B. M.; Tammelin, A.; Courvalin,
P.; Lambert, T. J. Antimicrob. Chemother. 2000, 46, 895.
(295) Nelson, E. C.; Segal, H.; Elisha, B. G. J. Antimicrob. Chemother.
2003, 52, 899.
(296) Domenech-Sanchez, A.; Martnez-Martnez, L.; Hernandez-Alles,
S.; del Carmen Conejo, M.; Pascual, A
.; Tomas, J. M.; Albert,
S.; Bened, V. J. Antimicrob. Agents Chemother. 2003, 48, 3332.
(297) Martinez, M. B.; Flickinger, M.; Higgins, L.; Krick, T.; Nelsestuen, G. L. Biochemistry 2001, 40, 11965.
(298) Armand-Lefvre, L.; Leflon-Guibout, V.; Bredin, J.; Barguelli, F.;
Amor, A.; Pages, J. M.; Nicolas-Chanoine, M. H. Antimicrob.
(299) Kwon, D. H.; Dore, M. P.; Kim, J. J.; Kato, M.; Lee, M.; Wu, J.
Y.; Graham, D. Y. Antimicrob. Agents Chemother. 2003, 47, 2169.
(300) Fernandez-Cuencal, F.; Martnez-Martnez, L.; Conejo, M. C.;
Ayala, J. A.; Perea, E. J.; Pascual, A. J. Antimicrob. Chemother.
2003, 51, 565.
(301) Gayet, S.; Chollet, R.; Molle, G.; Page`s, J. M.; Chevalier, J.
(302) Bornet, C.; Saint, N.; Fetnaci, L.; Dupont, M.; Davin-Regli, A.;
Bollet, C.; Page`s, J. M. Antimicrob. Agents Chemother. 2004, 48,
2153.
(303) Jacoby, G. A.; Mills, D. M.; Chow, N. Antimicrob. Agents
Chemother. 2004, 48, 3203.
(304) Okamoto, K.; Gotoh, N.; Nishino, T. Antimicrob. Agents Chemother. 2001, 45, 1964.
(305) Pai, H.; Kim, J. W.; Kim, J.; Lee, J.; Choe, K. W.; Gotoh, N.
(306) Lakaye, B.; Dubus, A.; Joris, B.; Fre`re, J. M. Antimicrob. Agents
Chemother. 2002, 46, 2901.
(307) Matsumura, N.; Minami, S.; Watanabe, Y.; Iyobe, S.; Mitsuhashi,
S. Antimicrob. Agents Chemother. 1999, 43, 2084.
(308) Kotra, L. P.; Golemi, D.; Amro, N. A.; Liu, G. Y.; Mobashery, S.
J. Am. Chem. Soc. 1999, 121, 8707.
(309) Snyder, D. S.; McIntosh, T. J. Biochemistry 2000, 39, 11777.
(310) Gillespie, S. H.; McHugh, T. D. Trends Microbiol. 1997, 5, 337.
(311) Faller, M.; Niederweis, M.; Schulz, G. E. Science 2004, 303, 1189.
(312) Antignac, A.; Boneca, I. G.; Rousselle, J. C.; Namane, A.; Carlier,
J. P.; Vazquez, J. A.; Fox, A.; Alonso, J. M.; Taha, M. K. J. Biol.
Chem. 2003, 278, 31529.
(313) Antignac, A.; Rousselle, J. C.; Namane, A.; Labigne, A.; Taha,
M. K.; Boneca, I. G. J. Biol. Chem. 2003, 278, 31521.
(314) Filipe, S. R.; Tomasz, A. Proc. Natl. Acad. Sci. U.S.A. 2000, 99,
4891.
(315) Smith, A. M.; Klugman, K. P. Antimicrob. Agents Chemother.
2001, 45, 2393.
(316) Fiser, A.; Filipe, S. R.; Tomasz, A. Trends Microbiol. 2003, 11,
547.
(317) Rohrer, S.; Berger-Bachi, B. Antimicrob. Agents Chemother.
2003, 47, 837.
(318) Okamoto, K.; Gotoh, N.; Nishino, T. Antimicrob. Agents Chemother. 2002, 46, 2696.
(319) Middlemiss, J. K.; Poole, K. J. Bacteriol. 2004, 186, 1258.
(320) Van Bambekel, F.; Glupczynski, Y.; Plesiat, P.; Peche`re, J. C.;
Tulkens, P. M. J. Antimicrob. Chemother. 2003, 51, 1055.
(321) Burns, D. L. Curr. Opin. Microbiol. 2003, 6, 29.
(322) Cascales, E.; Christie, P. J. Nat. Rev. Microbiol. 2003, 1, 137.
(323) Borges-Walmsley, M. I.; McKeegan, K. S.; Walmsley, A. R.
Biochem. J. 2003, 376, 313.
(324) Grkovic, S.; Brown, M. H.; Skurray, R. A. Microbiol. Mol. Biol.
2002, 66, 671.
(325) McKeegan, K. S.; Borges-Walmsley, M. I.; Walmsley, A. R.
Trends Microbiol. 2003, 11, 21.
(326) Murakami, S.; Yamaguchi, A. Curr. Opin. Struct. Biol. 2003,
13, 443.
(327) Paulsen, I. T. Curr. Opin. Microbiol. 2003, 6, 446.
(328) Schmacher, M. A.; Brennan, R. G. Mol. Microbiol. 2002, 45, 885.
(329) Chevalier, J.; Bredin, J.; Mahamoud, A.; Mallea, M.; Barbe, J.;
Page`s, J. M. Antimicrob. Agents Chemother. 2004, 48, 1043.
(330) Nakayama, K.; Ishida, Y.; Ohtsuka, M.; Kawato, H.; Yoshida,
K.; Yokomizo, Y.; Hosono, S.; Ohta, T.; Hoshino, K.; Ishida, H.;
(331)
(332)
(333)
(334)
(335)
(336)
(337)
(338)
(339)
(340)
(341)
(342)
(343)
(344)
(345)
(346)
(347)
(348)
(349)
(350)
(351)
(352)
(353)
(354)
(355)
(356)
(357)
(358)
(359)
(360)
(361)
(362)
(363)
(364)
(365)
(366)
(367)
(368)
(369)
(370)
(371)
(372)
(373)
(374)
(375)
Renau, T. E.; Leger, R.; Zhang, J. Z.; Lee, V. J.; Watkins, W. J.

Bioorg. Med. Chem. Lett. 2003, 13, 4201.
Mallea, M.; Mahamoud, A.; Chevalier, J.; Alibert-Franco, S.;
Brouant, P.; Barbe, J.; Pages, J. M. Biochem. J. 2003, 376, 801.
Renau, T.; Leger, R.; Filonova, L.; Flamme, E.; Wang, M.; Yen,
R.; Madsen, D.; Griffith, D.; Chamberland, S.; Dudley, M.; Lee,
V.; Lomovskaya, O.; Watkins, W.; Ohta, T.; Nakayama, K.;
Ishida, Y. Bioorg. Med. Chem. Lett. 2003, 13, 2755.
Zgurskaya, H.; Nikaido, H. Mol. Microbiol. 2000, 37, 219.
McGeer, A.; Low, D. E. Nat. Med. 2003, 9, 390.
McCormick, A. W.; Whitney, C. G.; Farley, M. M.; Lynfield, R.;
Harrison, L. H.; Bennett, N. M.; Schaffner, W.; Reingold, A.;
Hadler, J.; Cieslak, P.; Samore, M. H.; Lipsitch, M. Nat. Med.
2003, 9, 424.
Service, R. F. Science 2004, 303, 1798.
Sheldon, T. Nat. Med. 2004, 10, 6.
Oliveira, D. C.; Tomasz, A.; de Lencastre, H. Lancet Infect. Dis.
2002, 2, 180.
Projan, S. J. Curr. Opin. Microbiol. 2003, 6, 427.
Projan, S. J. Curr. Opin. Pharmacol. 2003, 3, 457.
Shlaes, D. M. Curr. Opin. Pharmacol. 2003, 3, 470.
Projan, S. J. Curr. Opin. Pharmacol. 2002, 2, 513.
Livermore, D. Nat. Rev. Microbiol. 2004, 2, 73.
Eisner, T. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 14517.
Meinwald, J.; Eisner, T. Helv. Chim. Acta 2003, 86, 3633.
Coates, A.; Hu, Y.; Bax, R.; Page, C. Nat. Rev. Drug Discovery
2002, 1, 895.
Hutter, B.; Schaab, C.; Albrecht, S.; Borgmann, M.; Brunner,
N. A.; Freiberg, C.; Ziegelbauer, K.; Rock, C. O.; Ivanov, I.;
Loferer, H. Antimicrob. Agents Chemother. 2004, 48, 2838.
Katayama, Y.; Zhang, H. Z.; Hong, D.; Chambers, H. F. J.
Bacteriol. 2003, 185, 5465.
Weiss, D. S.; Chen, J. C.; Ghigo, J. M.; Boyd, D.; Beckwith, J. J.
Bacteriol. 1999, 181, 508.
Zhao, G.; Meier, T. I.; Kahl, S. D.; Gee, K. R.; Blaszczak, L. C.
Kyriacou, S. V.; Brownlow, W. J.; Xu, X. H. N. Biochemistry 2004,
43, 140.
Steel, C.; Wan, Q.; Xu, X. H. N. Biochemistry 2004, 43, 175.
Campbell, D. A.; Szardenings, A. K. Curr. Opin. Chem. Biol.
2003, 7, 296.
Miller, C.; Thomsen, L. E.; Gaggero, C.; Mosseri, R.; Ingmer,
H.; Cohen, S. N. Science 2004, 305, 1629.
Spoering, A. L.; Lewis, K. J. Bacteriol. 2001, 183, 6746.
Balaban, N. Q.; Merrin, J.; Chait, R.; Kowalik, L.; Leibler, S.
Science 2004, 305, 1622.
Hastings, P. J.; Rosenberg, S. M.; Slack, A. Trends Microbiol.
2004, 12, 401.
Rice, L. B. Curr. Opin. Pharmacol. 2003, 3, 459.
Rand, K. H.; Houck, H. J. Antimicrob. Agents Chemother. 2004,
48, 2871.
Barrett, C. T.; Barrett, J. F. Curr. Opin. Biotechnol. 2003, 14,
621.
Gustafsson, I.; Sjolund, M.; Torell, E.; Johannesson, M.; Engstrandl, L.; Cars, O.; Andersson, D. I. J. Antimicrob. Chemother.
2003, 52, 645.
Salyers, A. A.; Gupta, A.; Wang, Y. Trends Microbiol. 2004, 12,
412.
Kummerer, K.; Henniger, A. Clin. Microbiol. Infect. 2003, 9,
1203.
Harmoinen, J.; Mentula, S.; Heikkila, M.; van der Rest, M.;
Rajala-Schultz, P. J.; Donskey, C. J.; Frias, R.; Koski, P.;
Wickstrand, N.; Jousimies-Somer, H.; Westermarck, E.; Lindevall, K. Antimicrob. Agents Chemother. 2004, 48, 75.
Harmoinen, J.; Vaali, K.; Koski, P.; Syrjanen, K.; Laitinen, O.;
Lindevall, K.; Westermarck, E. J. Antimicrob. Chemother. 2003,
51, 361.
Stiefel, U.; Pultz, N. J.; Harmoinen, J.; Koski, P.; Lindevall, K.;
Helfand, M. S.; Donskey, C. J. J. Infect. Dis. 2003, 188, 1605.
Cooper, B. S.; Medley, G. F.; Stone, S. P.; Kibbler, C. C.; Cookson,
B. D.; Roberts, J. A.; Duckworth, G.; Lai, R.; Ebrahim, S. Proc.
Natl. Acad. Sci. U.S.A. 2004, 101, 10223.
Livermore, D. M.; Dudley, M. N. Curr. Opin. Microbiol. 2000,
3, 487.
Monroe, S.; Polk, R. Curr. Opin. Microbiol. 2000, 3, 496.
Bantar, C.; Vesco, E.; Heft, C.; Salamone, F.; Krayeski, M.;
Gomez, H.; Coassolo, M. A.; Fiorillo, A.; Franco, D.; Arango, C.;
Duret, F.; Oliva, M. E. Antimicrob. Agents Chemother. 2004, 48,
392.
Livemore, D. M.; Brown, D. F. J.; Quinn, J. P.; Carmeli, Y.;
Paterson, D. L.; Yu, V. L. Clin. Microbiol. Infect. 2004, 10, 84.
Chandrakala, B.; Shandil, R. K.; Mehra, U.; Ravishankar, S.;
Kaur, P.; Usha, V.; Joe, B.; deSousa, S. M. Antimicrob. Agents
Chemother. 2004, 48, 30.
Tondi, D.; Powers, R. A.; Caselli, E.; Negri, M. C.; Blazquez, J.;
Costi, M. P.; Shoichet, B. K. Chem. Biol. 2001, 8, 593.
Powers, R. A.; Shoichet, B. K. J. Med. Chem. 2002, 45, 3222.
Powers, R. A.; Morandi, F.; Shoichet, B. K. Structure 2002, 10,
1013.

(376) Trehan, I.; Morandi, F.; Blaszczak, L. C.; Shoichet, B. K. Chem.
Biol. 2002, 9, 971.
(377) Levy, S. B.; Marshall, B. Nat. Med. Suppl. 2004, 10, S122.
(378) Payne, D.; Tomasz, A. Curr. Opin. Microbiol. 2004, 7, 435.
(379) Poole, K. Cell. Mol. Life Sci. 2004, 61, 2200.
(380) Mallorqui-Fernandez, G.; Marrero, A.; Garcia-Pique, S.; GarciaCastellanos, R.; Gomis-Ruth, F. X. FEMS Microbiol. Lett. 2004,
235, 1.
(381) Walsh, F. M.; Amyes, S. G. B. Curr. Opin. Microbiol. 2004, 7,
439.
(382) Gotz, F. Curr. Opin. Microbiol. 2004, 7, 477.
(383) Dmitriev, B. A.; Toukach, F. V.; Holst, O.; Rietschel, E. T.; Ehlers,
S. J. Bacteriol. 2004, 186, 7141.
(384) Vollmer, W.; Holtje, J. V. J. Bacteriol. 2004, 186, 5978.
(385) Arbeloa, A.; Hugonnet, J.-E.; Sentilhes, A.-C.; Josseaume, N.;
Dubost, L.; Monsempes, C.; Blanot, D.; Brouard, J.-P.; Arthur,
M. J. Biol. Chem. 2004, 279, 41546.
(386) Gardete, S.; Ludovice, A. M.; Sobral, R. G.; Filipe, S. R.; De
Lencastre, H.; Tomasz, A. J. Bacteriol. 2004, 186, 1705.
(387) Pagliero, E.; Chesnel, L.; Hopkins, J.; Croize, J.; Dideberg, O.;
Vernet, T.; Di Guilmi, A. M. Antimicrob. Agents Chemother.
2004, 48, 1848.
(388) Silvaggi, N. R.; Josephine, H. R.; Kuzin, A. P.; Nagarajan, R.;
Pratt, R. F.; Kelly, J. A. J. Mol. Biol. 2005, 345, 251.
(389) Macheboeuf, P.; Di Guilmi, A. M.; Job, V.; Vernet, T.; Dideberg,
O.; Dessen, A. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 577.
(390) Morlot, C.; Pernot, L.; Le Gouellec, A.; Di Guilmi, A. M.; Vernet,
T.; Dideberg, O.; Dessen, A. J. Biol. Chem. 2005, in press.
(391) Labia, R. Curr. Med. Chem. Anti-Infect. Agents 2004, 3, 251.
(392) Ke, Y.-Y.; Lin, T.-H. Biophys. Chem. 2005, 114, 103.
(393) Shimizu-Ibuka, A.; Matsuzawa, H.; Sakai, H. Biochemistry 2004,
43, 15737.
(394) Materon, I. C.; Beharry, Z.; Huang, W.; Perez, C.; Palzkill, T. J.
Mol. Biol. 2004, 344, 653.
Fisher et al.
(395) Mercuri, P. S.; Garcia-Saez, I.; De Vriendt, K.; Thamm, I.;
Devreese, B.; Van Beeumen, J.; Dideberg, O.; Rossolini, G. M.;
Frere, J.-M.; Galleni, M. J. Biol. Chem. 2004, 279, 33630.
(396) Garrity, J. D.; Bennett, B.; Crowder, M. W. Biochemistry 2005,
44, 1078.
(397) Garau, G.; Bebrone, C.; Anne, C.; Galleni, M.; Frere, J.-M.;
Dideberg, O. J. Mol. Biol. 2005, 345, 785.
(398) Zhang, Z.; Palzkill, T. J. Biol. Chem. 2004, 279, 42860.
(399) Reichmann, D.; Rahat, O.; Albeck, S.; Meged, R.; Dym, O.;
Schreiber, G. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 57.
(400) Freiberg, C.; Brotz-Oesterhelt, H.; Labischinski, H. Curr. Opin.
Microbiol. 2004, 7, 451.
(401) Venkatesan, A. M.; Agarwal, A.; Abe, T.; Ushirogochi, H.;
Yamamura, I.; Kumagai, T.; Petersen, P. J.; Weiss, W. J.; Lenoy,
E.; Yang, Y.; Schlaes, D. M.; Ryan, J. L.; Mansour, T. S. Biorg.
Med. Chem. 2004, 12, 5807.
(402) Venkatesan, A. M.; Gu, Y.; Dos Santos, O.; Abe, T.; Agarwal,
A.; Yang, Y.; Petersen, P. J.; Weiss, W. J.; Mansour, T. S.;
Nukaga, M.; Hujer, A. M.; Bonomo, R. A.; Knox, J. R. J. Med.
Chem. 2004, 47, 6556.
(403) Bush, K.; Macielag, M.; Weidner-Wells, M. Curr. Opin. Microbiol.
2004, 7, 466.
(404) Buynak, J. D. Curr. Med. Chem. 2004, 11, 1951.
(405) Georgopapadakou, N., H. Exp. Opin. Invest. Drugs 2004, 13,
1307.
(406) Long, T. E. IDrugs 2003, 6, 351.
(407) Singh, G. S. Mini-Rev. Med. Chem. 2004, 4, 69.
(408) Singh, G. S. Mini-Rev. Med. Chem. 2004, 4, 93.
(409) Thomson, C. J.; Power, E.; Ruebsamen-Waigmann, H.; Labischinski, H. Curr. Opin. Microbiol. 2004, 7, 445.
CR030102I
Chem. Rev. 2005, 105, 425448
425
Glycopeptide and Lipoglycopeptide Antibiotics

Dan Kahne,*, Catherine Leimkuhler, Wei Lu, and Christopher Walsh*,
The Department of Chemistry, Princeton University, Princeton, New Jersey 08544, and the Department of Biological Chemistry and
Molecular Pharmacology, Harvard Medical School, Boston, Massachusetts 02115
Received May 12, 2004
Contents
1. Introduction
1. Introduction
2. Structures of Vancomycin and Teicoplanin
2.1. Variations in Vancomycin and Teicoplanin
Natural Analogues
3. Biosynthetic StrategiessEnzymatic Assembly
Lines and Tailoring Enzymes
4. Mechanism of Action of Glycopeptide Antibiotics
4.1. Cellular Targets of Antibiotics
4.2. Three Stages of Peptidoglycan Biosynthesis
4.3. Transglycosylase and Transpeptidase
Isoforms
4.4. Peptidoglycan Biosynthesis as an Antibiotic
Target
4.5. Early Studies on the Mechanism of Action of
Glycopeptide Antibiotics
5. Mechanisms of Resistance
5.1. Vancomycin Resistant Enterococci (VRE)
5.2. VRE Genotypes: VanB Resistance and How
To Overcome It
5.3. VRE Genotypes: VanA Resistance and the
Compounds That Overcome It
5.4. Models for How Vancomycin Analogues
Overcome Vancomycin Resistance
5.4.1. Dimerization and Membrane Anchoring
5.4.2. A Second Mechanism of ActionsDirect
Interaction with the Transglycosylase
5.5. Vancomycin-Resistant S. aureus (VRSA)
6. New Directions for Treating VRE and VRSA
6.1. Recently Approved Drugs for VRE
6.2. Second-Generation Semisynthetic
Lipoglycopeptides in Clinical Development
6.2.1. Oritavancin
6.2.2. Dalbavancin
6.2.3. TD-6424
6.3. In Discovery: Chemoenzymatic Routes To
Modify Sugars and Acyl Groups on
Heptapeptide Scaffolds
7. Acknowledgments
8. Note Added after ASAP Publication
9. References
425
426
428
428
431
431
432
433
435
436
437
437
437
439
441
441
441
442
442
443
443
443
444
444
444
446
446
446
* To whom correspondence should be addressed. C.W.: phone,

617-432-1715; fax, 617-432-0438; e-mail, Christopher_walsh@
hms.harvard.edu. D.K.: phone, 617-496-0208; fax, 617-496-0215;
e-mail, kahne@chemistry.harvard.edu.
Current address: The Department of Chemistry and Chemical
Biology, Harvard University, Cambridge, MA 02138.
Harvard Medical School.
Vancomycin and teicoplanin are the two glycopeptide antibiotics that are used clinically, and the
emergence of resistance to them poses a serious
threat to human health. Some microorganisms are
resistant to both vancomycin and teicoplanin, but
some resistant strains remain sensitive to teicoplanin when resistance to vancomycin develops.
Given the apparent structural and mechanistic similarities of these two drugs, one longstanding question is why they have different effects on some
microorganisms. This review will summarize what
is known about the structure and function of the
glycopeptide and lipoglycopeptide antibiotics, how
glycopeptide resistance develops, and how natural
and semisynthetic lipoglycopeptide derivatives are
being used to learn how to overcome glycopeptide
resistance and for development as novel therapeutic
agents.
Vancomycin and teicoplanin are used to treat
serious Gram-positive bacterial infections that are
resistant to other antibiotics, e.g. -lactams. The
frequency of resistance to the glycopeptide antibiotics
has increased significantly over the past decade and
now represents a serious threat to public health.1
Moreover, multiple genera, including Staphylococcus,
have developed resistance to these drugs.2 It is hoped
that research into the molecular basis for glycopeptide resistance will lead to the ability to design new
glycopeptide antibiotics with activity against both
sensitive and resistant bacterial strains. In this
review we will describe the structures of a set of
important natural glycopeptide antibiotics and outline their biosynthetic pathways. We will then discuss
what is known about the mode of action of the
glycopeptides and how structural differences influence biological activity. Next, we will analyze how
resistance to the glycopeptides develops and summarize the efforts to develop glycopeptides that can
overcome resistance. Finally, we will discuss how
access to unnatural glycopeptides has provided tools
that have been used to identify new targets of the
glycopeptide class and activity differences observed
for natural glycopeptides against different bacterial
pathogens.
Antibiotics can be classified on several axes. One
is by the nature of the targets in susceptible bacteria,
for example blockade of bacterial cell wall biosynthesis or bacterial protein synthesis, or DNA and
RNA replication.3 A second axis is whether the
antibiotics derive from natural product scaffolds or
10.1021/cr030103a CCC: $53.50 2005 American Chemical Society

Daniel Kahne recently moved to Harvard University from Princeton University, where he was on the faculty for 16 years. He holds appointments
in the departments of Chemistry and Chemical Biology (CCB) and
Biological Chemistry and Molecular Pharmacology (BCMP). He trained
as a synthetic organic chemist with Gilbert Stork and continued postdoctoral
training at Columbia with Clark Still. He has longstanding interests in the
chemistry and biology of natural products, and in recent years become
interested in how natural products can be used to probe cellular pathways.
Professor Kahnes research group is divided into students who develop
synthetic methods to make and modify complex natural products and
students who combine some chemistry with molecular and cellular biology
to address questions relating to how various natural products function. In
the last five years, the Kahne group has become interested in antibiotic
resistance and has made significant contributions to understanding the
mechanisms of action of glycopeptide antibiotics and derivatives that
overcome glycopeptide resistance. The Kahne group has also been using
glycopeptide derivatives and other antibiotics in conjunction with genetics
to probe pathways involving cell wall biosynthesis and outer membrane
biogenesis.
Catherine E. Leimkuhler attended Villanova University, where she received

her B.S. in Chemistry in 2001. Upon graduation, she attended Princeton
University, where she received her M.A. under the direction of Daniel
Kahne. She has worked on a chemoenyzmatic synthesis of novel
glycopeptide antibiotics. She is currently pursuing her Ph.D. with Daniel
Kahne at Harvard University.
are synthetic antibacterial drugs.4 The glycopeptide

antibiotics of the vancomycin (1) and teicoplanin (2)
(Figure 1) class block steps in the biosynthesis of the
peptidoglycan layer of bacterial cell walls. They are
natural products elaborated by actinomycete soil
bacteria, with vancomycin being isolated in the 1950s
from Amycolatopsis orientalis and teicoplanin around
1980 from Actinoplanes teichomyceticus.5 Each of
these antibiotics is effective against Gram-positive
bacteria but not Gram-negative ones, due to the
permeability barrier of the intact outer membrane
Kahne et al.
Wei Lu received his B.E. degree from East China University of Chemical
Technology in 1992. He did his Ph.D. in Philip Coles lab at Johns Hopkins
University. His Ph.D. research focused on the regulation of protein tyrosine
phosphatases. Since 2002, he has been pursuing his postdoctoral research
in Christopher Walshs group at Harvard Medical School. His current
research interest lies in functional analysis of glycosyltransferases involved
in natural product biosynthesis. He is currently a Ruth L. Kirschstein Nation
Institute of Health postdoctoral fellow.
Christopher T. Walsh is the Hamilton Kuhn Professor of Biological

Chemistry and Molecular Pharmacology (BCMP) at Harvard Medical
School. He has had extensive experience in academic administration,
including Chairmanship of the MIT Chemistry Dept (19821987) and the
HMS Biological Chemistry & Molecular Pharmacology Department (1987
1995) as well as serving as President and CEO of the Dana Farber Cancer
Institute (19921995). His research has focused on enzymes and enzyme
inhibitors, with recent specialization on antibiotics. He and his group have
authored over 590 research papers; a text, Enzymatic Reaction Mechanisms; and a book, Antibiotics: Origins, Actions, Resistance. He is a
member of the National Academy of Sciences, the Institute of Medicine,
and the American Philosophical Society.
in Gram-negative bacteria that keeps the glycopeptides from reaching their targets at the periplasmic
face of the cytoplasmic membrane.
Vancomycin has been approved for human use in
many countries, while teicoplanin, widely prescribed
in Europe, was never been brought successfully
through the FDA approval process in the US. The
drugs have been front line agents for treating endocarditis caused by enterococci, opportunistic pathogens, and have become mainstays in the treatment
of life-threatening infections due to methicillin resistant Staphylococcus aureus (MRSA).1
2. Structures of Vancomycin and Teicoplanin

In this review, the term lipopeptide is used for
acylated peptide natural products (such as daptomy-
Figure 1. Structures of vancomycin (1) and teicoplanin (2), glycopeptide and lipoglycopeptide antibiotics approved for
human therapy.
Figure 2. Oxidative cross-linking by hemeprotein enzymes of the nascent heptapeptides to yield the oxidized, crosslinked heptapeptide scaffolds: three cross-links at residues 2-4, 4-6, and 5-7 in the vancomycin scaffold and a fourth
cross-link at residues 1-3 in the teicoplanin scaffold.
cin); the term glycopeptide is used for molecules such

as vancomycin. Since teicoplanin has a peptide scaffold and both covalently attached glycosyl and long
chain acyl groups, the term lipoglycopoeptide is used
for such natural products.
Vancomycin and teicoplanin are closely related
structures, containing a homologous heptapeptide
scaffold that has undergone extensive oxidative crosslinking.6 Five of the seven residues in vancomycin
are aromatic, while all seven are in teicoplanin: Leu1
and Asn3 of vancomycin are replaced by hydroxyphenylglycine regioisomers in teicoplanin. There are five
such nonproteinogenic phenylglycines in teicoplanin
and three in vancomycin. The phenylglycines are
either 4-hydroxyphenylglycine (HPG) (residues 4 and
5 in 1 and 2 and also residue 1 in teicoplanin) or 3,5dihydroxyphenylglycine (DPG) (residue 7 in 1 and
residues 3 and 7 in 2). In vancomycin, the remaining
two residues are modified tyrosines, with chlorine at
the meta position of the aromatic ring and an OH
substituent at the benzylic carbon of the side chain.
In teicoplanin, only Tyr6 has been converted to the
-OH-Tyr.
The electron-rich side chains of these aromatic
amino acid residues facilitate the oxidative crosslinking (see below) that sets the rigid architecture of
the heptapeptide scaffolds of the two drugs. In
vancomycin, three cross-links are effected, joining the

aromatic rings of 2-4 and 4-6 in aryl ether linkages
and of 5-7 in a direct C-C coupling (Figure 2). In
teicoplanin, residues 1 and 3 are now hydroxyphenylglycines in the un-cross-linked precursor heptapeptide, and a fourth cross link, 1-3, is produced by
the teicoplanin tailoring enzymes (Figure 2). The four
cross-links of the teicoplanin scaffold involve all seven
side chains to produce the rigid, cup-shaped aglycon.
These are glycopeptide antibiotics, so subsequently,
the peptide framework is glycosylated, by a disaccharide chain on residue 4 in vancomycin or by three
monosaccharides at residues 4, 6, and 7 in teicoplanin. The disaccharide in vancomycin is a D-glucosyl2,1-D-vancosamine, linked through C1 of glucose to
the phenolic oxygen of OH-Phegly4 of the crosslinked peptide. The vancosamine sugar is a 2,3,6trideoxy-L-hexose, with methyl and amino substituents at C3. Such deoxyhexoses are typical in antibiotics
and mediate some of the recognition of target
molecules.7-11 In teicoplanin, there is a glucosamine
at the corresponding OH-Phegly4 residue, a GlcNAc
at the -hydroxyl substituent of -OH-Tyr6, and a
mannose on the cross-linked Dpg7. The N-decanoyl
moiety on the glucosamine installed at residue 4
blocks the elongation to the disaccharide chain
observed in the vancomycin family.
Kahne et al.
Teicoplanin, but not vancomycin, is a lipoglycopeptide with a C10 fatty acyl chain in an amide linkage
to the amino group of the glucosamine sugar moiety
(a series of fatty acyl variants are found in the
producer organism, but control of the feedstock
during fermentation yields the C10 acylated form of
teicoplanin). The hydrophobic acyl chain alters the
physical properties and presumably the partitioning
of lipoglycopeptide vs glycopeptide. This is likely a
determining factor not only in the differential activity
of 2 vs 1 against some forms of vancomycin-resistant
enterococci (VRE) noted below in section 5 but also
in the second-generation semisynthetic lipoglycopeptides oritavancin, dalbavancin, and TD-6424 discussed in section 6.
additional epivancosamine residue is found at the

benzylic oxygen of -OH-Tyr6. The biosynthetic gene
clusters for 3 and 412-14 have been reported and have
revealed much about the molecular logic for assembly
of these antibiotics.
In the teicoplanin subfamily, we note two congeners, A47934 (5), a sulfated aglycon form produced
by Streptomyces toyocaensis, and A40926 (6), from a
Nonomuraea species. In section 6, we shall note that
chloroeremomycin is the starting point for oritavancin and A40926 the starting point for dalbavancin,
the two most clinically advanced second-generation,
semisynthetic lipoglycopeptides. The DNA sequence
for the biosynthetic clusters of 5 and 6 have also been
reported.15-17
2.1. Variations in Vancomycin and Teicoplanin

Natural Analogues
3. Biosynthetic StrategiessEnzymatic Assembly

Lines and Tailoring Enzymes
After the clinical success of vancomycin and teicoplanin, dozens of additional glycopeptide congeners
have been isolated from strains of actinomycetes.6
Congeners with the vancomycin heptapeptide scaffold
include balhimycin (3) and chloroeremomycin (4)
(Figure 3, structures 3-6). Both have the identical
heptapeptide scaffold of vancomycin. Balhimycin
differs in having a vancosamine derivative, 4-oxovancosamine, at residue 6. Chloroeremomycin has
almost the same disaccharide appended to OHPhegly4. The difference is in the terminal 2,3,6trideoxy-3-amino-3-methyl hexose. In 4, the trideoxyhexose-4-OH is equatorial rather than the 4-axial
OH of vancosamine. This is epivancosamine, and an
Although the total synthesis of the vancomycin and

teicoplanin family of glycopeptides and lipoglycopeptides have been accomplished 6,18,19 with substantial
invention of new chemistry, the complexity of these
natural products makes fermentative routes the only
viable route to bulk production. Semisynthetic chemical tailoring of the scaffolds has been the predominant route to structure activity variation to produce
second-generation clinical development candidates,
as will be noted in section 6.20
Thus, understanding the biosynthetic logic for
construction of the heptapeptide core and its subsequent oxidations, glycosylations, and acylations is of
both basic and applied interest. It had been clear
Figure 3. Balhimycin (3) and chloroeremomycin (4) in the vancomycin family and A47934 (5) and A40926 (6) in the
teicoplanin family.
Figure 4. Core domains: C-A-T, in the elongation modules

of the NRPS assembly line for vancomcyin and teicoplanin
family members: C ) condensation domain, A ) adenylation domain, and T ) thiolation domain. Amino acids get
selected and activated by the A domain and installed as
thioester on the T domain. The C domain catalyzes peptide
bond formation.
from the prevalence of the nonproteinogenic phenylglycines in the peptide backbones that vancomycin and
teicoplanin family members must be assembled by
nonribosomal peptide synthetase (NRPS) enzymatic
machinery.5 The DNA sequence for the chloroeremomycin gene cluster12 revealed 30 adjacent genes
attributable to the pathway and validated the existence of seven NRPS modules, distributed over three
subunits, Cep ABC, in a 3/3/1 distribution.21 Each of
the seven modules selects and activates one amino
acid, and the order of the modules mandates the
heptapeptide sequence (Leu-Tyr-Asn-Hpg-Hpg-TyrDpg). In the teicoplanin subfamily, the catalytic
domains in modules one and three of the NRPS
assembly lines instead select and activate Hpg and
Dpg, respectively.16,22 In addition to a catalytic domain for amino acid selection and activation as the
aminoacyl-AMP, each module has a thiolation domain modified with phosphopantetheine23 to provide
a thiol for covalent aminoacyl-S-enzyme formation.
The third domain in each module is the condensation
domain, responsible for amide bond formation and
directional peptide chain growth and translocation
from N-terminal to C-terminal modules (Figure 4).
The source of the nonproteinogenic amino acids in
the microbial cell at the time antibiotic production
is switched on is relevant for yields and coordination.
Among the 30 clustered Cep biosynthetic genes are
four that encode enzymes that generate L-Hpg from
the common bacterial metabolic intermediate chorismate.21,24 There are another four encoded enzymes
in the cluster that channel four molecules of malonyl
CoA to the eight-carbon intermediate dihydroxyphenylacetyl CoA on the way to L-Dpg.25-27 Residues 2
and 6 in 1 and 4 differ from the proteinogenic amino
acid L-tyrosine by chlorine at the meta position of the
aromatic ring and the -OH that is the site of
glycosylation in 2-4. The Cep cluster contains a gene
encoding a flavoprotein halogenase12 as well as a pair
of enzymes28 that hydroxylate tyrosine while installed
as a tyrosyl-S-enzyme. Thus, 11 enzymes are coordinately induced to enable generation of residues 2,
4, 5, 6, and 7 required for the vancomycin-type NRPS
assembly lines and for generation of six of the seven
residues in teicoplanin NRPS multimodular enzymes.

One additional feature of these NRPS assembly
lines is worth note. Although the biosynthetic pathways generate the L-forms of modified tyrosines,
L-Hpg, and L-Dpg, the heptapeptide released by the
NRPS assembly line has the configuration D-D-L-LD-D-L. Inspection of the CepABC subunit domain
composition predicts epimerase (E) domains in module 2, 4, and 5 to convert L-aminoacyl-/peptidyl-Senzyme intermediates to the D-isomers. The first
module lacks an epimerase domain but can select and
activate D-Leu directly.29 All told there are 24 domains in the seven-module CepABC assembly line,
each with a well-defined function. Note that condensation domains downstream of E domains must be
chiral peptide synthase catalysts that recognize
upstream D-aminoacyl-/peptidyl donors. Depending
on the timing of the epimerization, the downstream
acceptors can be L- or D-amino acids.26,30
Following release from the last module of CepC in
the NRPS assembly line, the nascent heptapeptide
7 undergoes three kinds of postassembly tailoring
reactions to yield the glycopeptides 1 and 4 (Figure
5). The first reaction is N-methylation of the amino
terminal D-Leu1 residue carried out by orf 16 in the
Cep biosynthetic cluster.31
Then the oxidative cross-links are introduced. In
both the Cep and balhimycin gene clusters there are
three tandem heme proteins, termed OxyABC,32,33
that act ad seriatim to generate the fully cross-linked
aglycon 8. Genetic analysis in the balhimycin system
has defined the regiospecificity and timing of crosslinking, one bond each introduced by catalytic action
of the hemeproteins33 as determined by gene knockouts and structure determination of the accumulating
intermediates (OxyB > OxyA > OxyC) (Figure 6).
The coupling chemistry is probably via one-electron
pathways and regioselectivity and atropisomer selectivity may be imposed by the folding and orientation of acyclic and partially cross-linked substrates
in each hemeprotein active site. Although there are
X-ray structures of two of the three bahlimycin Oxy
proteins, OxyB and OxyC,34,35 in vitro activity of the
purified enzymes has not yet been reconstituted so
the possible promiscuity toward other substrates is
not yet known.
In teicoplanin heptapeptide scaffolds, a fourth
cross-link, connecting Hpg1 and Dpg3, needs to be
introduced. In the A47934, A40926, and teicoplanin
clusters,15,16,22 there is indeed a fourth tandem hemeprotein (OxyD) that is the obvious catalytic candidate. The timing of the 1-3 cross-link compared to
2-4, 4-6, and 5-7 is not yet known.
The last phase of tailoring involves the glycosyl
transferases (Gtfs) (Figure 7). There are three Gtfs,
GtfABC, embedded in the Cep cluster12 and corresponding two Gtfs, GtfDE, in the vancomycin biosynthetic gene cluster.36 GtfB ()GtfE) transfers
D-glucose from the nucleoside diphosphosugar dTDPD-glucose to the phenolic-OH at PheGly4 in the crosslinked aglycon peptide. This yields the monoglycosylated heptapeptide scaffold, 9, known historically as
desvancosaminyl vancomycin (DVV). The other two
Gtfs in the Cep cluster, GtfA and GtfC, transfer
Kahne et al.
Figure 5. Post-NRPS assembly line tailoring enzymatic modifications to convert the heptapeptides to vancomycin and
teicoplanin: N-methylation, oxidative cross-linking, and glycosylations.
Figure 6. Sequential action of OxyB, OxyA, OxyC to introduce the 2-4, 4-6, and 5-7 cross-links in the bahlimycin
scaffold.
L-epivancosamine to the -OH-Tyr6 and the C2 of the
glucosyl moiety of DVV, respectively37,38 accounting
for seven post-assembly line tailoring enzymes in the
Cep cluster.
The donor substrate for GtfA and GtfC is dTDPL--epivancosamine 10. It is not a standard primary
metabolite and is also made with just-in-time inventory control logic like the nonproteinogenic amino
acids. In particular, there are five genes, EvaA-E, in

the Cep cluster that channel the common NDP sugar
dTDP-D-glucose to dTDP-L-epivancosamine.39 These
enzymes effect remarkable changes of the hexose
attached to dTDP, deoxygenating C6 and then C2 and
reductively aminating a C3 keto intermediate and
then C-methylating it. This yields the dTDP-4oxovancosamine, the sugar donor to residue 6 in
Figure 7. Action of three glycosyltransferases, GtfABC, to add the three sugars in chloroeremoycin maturation and of
two Gtfs, GtfDE, in vancomycin maturation.
bahlimycin tailoring. The C4 ketone in dTDP-4oxovancosamine can be reduced with chirality control
to yield the 4-equatorial OH to finish the biosynthesis
of dTDP-L-epivancosamine for GtfC.39 Alternatively,
stereospecific enzymatic reduction to the other carbonyl face yields the 4-axial-OH and dTDP-L-vancosamine in vancomycin-producing actinomycetes,
presumably used by GtfD to build the D-glucosyl-2,1l-vancosamine disaccharide chain as the last step in
formation of 1.
Including the EvaA-E enzymes in the list of dedicated monomer generation and tailoring enzymes
yields a total of 24 proteins, over and above the three
NRPS assembly line subunits needed to make chloroeremomycin (one Gtf less to make vancomycin)
(Figure 8). The biosynthetic gene clusters for A40926
and for teicoplanin reveal the anticipated comparable
logic with a few variations, including a putative
mannosyl transferase for glycosylation of Dpg7 and
an acyl transferase that is the catalyst for the C10
fatty acyl amide formation to the glucosamine moiety
in 2 and 6.17
4. Mechanism of Action of Glycopeptide

Antibiotics
Figure 8. Twenty-four open reading frames (Orfs) in the

chloroeremomycin biosynthetic cluster.
The majority of antibacterial agents inhibit the

synthesis of DNA, RNA, proteins, or peptidoglycan.
These are all macromolecules containing different
monomer building blocks. Therefore, it is possible to
Kahne et al.
Figure 9. The peptidoglycan layer (PG) surrounding bacterial cells is a giant macromolecular meshwork with peptide
cross-bridges connecting glycan strands.
ascertain generally how an antibiotic functions by

examining whether it affects the incorporation of
radiolabeled monomer units into one of these four
different types of macromolecules. Vancomycin, the
first glycopeptide antibiotic discovered, was shown
to inhibit the incoporation of [14C]GlcNAc from UDP[14C]GlcNAc into bacterial peptidoglycan and was
therefore classified as a cell wall active antibiotic.40,41
An overview of the structure and biosynthesis of
bacterial peptidoglycan is provided in the following
section as a prelude to a detailed discussion of the
mechanism of action of vancomycin and other glycopeptide antibiotics.
4.2. Three Stages of Peptidoglycan Biosynthesis

The internal osmotic pressure inside a typical
bacterial cell fluctuates significantly, depending on
environmental conditions. For bacteria to survive,
their cell membranes must be able to withstand
osmotic pressures in excess of 5-15 atm without
rupturing. Bacterial cells are surrounded by layers
of peptidoglycan, a mesh-like carbohydrate polymer
that provides the mechanical support necessary to
prevent the cells from lysing as the osmotic pressure
fluctuates.42 Peptidoglycan is composed of linear
chains of -(1,4)-N-acetyl hexosamine units joined by
peptide cross-links (Figure 9).
The biosynthesis of peptidoglycan takes place in
three distinct stages (for a review on peptidoglycan
biosynthesis, see Bugg et al.43) (Figure 10). The first
stage takes place in the cytoplasm and involves the
conversion of UDP-GlcNAc to UDP-N-acetylmuramylpentapepide.4 The muramyl group is a D-2-N-acetyl3-O-lactylglucose, assembled from UDP-GlcNAc and
PEP by MurA and MurB catalysis. Mur C, D, E
sequentially add L-Ala-D--Gln and Lys (Gram positives) or meso diaminopimelate (DAP in Gram negatives), respectively, in ATP-dependent amide-forming
steps to create the UDP muramyl-tripeptide. MurF

adds preformed D-Ala-D-Ala in the fourth amideforming step to create the UDP-muramyl-L-Ala--DGln-L-Lys-D-Ala-D-Ala (pentapeptide) and complete
stage I.
Stage II involves the transfer of the muramylpentapeptide from UDP to a C55 isoprenol-P (bactoprenol) carrier. The first reaction (MraY) moves the
muramyl pentapeptide to the membrane interface
and creates lipid I. The second reaction, catalyzed by
MurG, adds GlcNAc in a -1,4-linkage to the muramyl moiety, generating the disaccharyl-pentapeptide (Figure 10) attached to bactoprenol-PP.44-47 This
is lipid II, which is the donor for the peptidoglycan
elongation reactions that occur on the external surface of the bacterial membrane. Once lipid II is
formed, the disaccharide-pentapeptide is somehow
flipped from the cytoplasmic phase of the membrane
to the external face (Figure 11). That flipping, still
mysterious in whether an enzyme accelerates the
facial equilibration, completes phase II of PG assembly.
The third stage of PG synthesis involves lipid II
molecules presenting disaccharyl-pentapeptide as
donor, on the outer face of membranes, to the 4-OH
of the GlcNAc termini of existing PG glycan strands
as acceptors. This is a net disaccharide transfer in
each elongation step, catalyzed by enzymes known
as transglycosylases (TGases) (Figure 12). These
newly elongated glycan strands are immature, or
nascent, in the sense that the pentapeptide units are
not cross-linked, so this portion of the PG layer will
be mechanically weak until transpeptidation between
adjacent peptide strands has occurred. This is the
task of a family of transpeptidases (TPases), using
the -NH2 of Lys3 on one strand to attack D-Ala4 on
an adjacent strand, liberating D-Ala5 as the free
amino acid (Figure 12). The Lys3-D-Ala4 interstrand
Figure 10. Cytoplasmic phases of peptidoglycan assembly. Phase I culminates in UDP murmamyl pentapeptide assembly
in the bacterial cytoplasm; phase II involves enzymatic transfer of muramyl pentapeptide to C55 bactoprnol-P to make
lipid I, followed by MurG-mediated GlcNAc transfer to make the C55-P-P-disaccharly pentapeptide, lipid II.
the fact that the active site serine side chains that
function as catalytic nucleophiles in the transpeptidases are covalently acylated by penicillins (and
cephalosporins).3
The first two stages of peptidoglycan biosynthesis,
leading to the production of lipid II, are wellunderstood at this point. Most of the enzymes involved in this part of the pathway have been studied
extensively, and crystal structures are available for
all of the essential enzymes in stage I and stage II,
except MraY.47-55 The final stage of peptidoglycan
biosynthesis, involving glycan polymerization and
cross-linking, is not nearly as well-understood. From
a chemical perspective, there are only two different
reactions involved in this final stage of peptidoglycan
synthesissa glycosidic bond-forming reaction and a
transpeptidation reaction; however, the resulting
product is a complex polymer, and there is presumed
to be considerable structural heterogeneity at the
molecular level in both the final product and the
various intermediates. Furthermore, there are several different transglycosylase domains and an even
larger number of transpeptidase domains involved
in the biosynthesis of this polymer, as noted in the
section below.
4.3. Transglycosylase and Transpeptidase

Isoforms
Figure 11. Translocation of lipid II. Lipid II is flipped
between the inner (cytoplasmic) and the external face of
the bacterial membranes via an unknown mechanism.
isopeptide bond thus generated is a mechanically

strengthening covalent cross-link. The crucial role of
transpeptidation in PG maturation is borne out by
There are multiple genes encoding TGases and

TPases in bacterial cells.42,56 These genes play different roles in cell growth and division.
Historically, transpeptidases were classified into
high and low molecular weight forms, collectively
known as penicillin-binding proteins (PBPs). Some
Kahne et al.
Figure 12. External phases of peptidoglycan assembly. Phase III involves lipid II as disaccharyl pentapeptide donor to
the 4-OH of GlcNAc at the ends of PG glycan chains undergoing elongation (transglycosylases) and then isopeptide bond
formation between Lys3 on one peptide strand and D-Ala4 on a neighboring peptide strand (transpeptidases).
Figure 13. Schematic of bifunctional TGase/Tpase domains in high molecular weight PBPs.
of the high molecular weight PBPs contain transglycosylase domains in addition to transpeptidase
domains. In Escherichia coli, for example, PBP1A,
-1B, and -1C are bifunctional TGase/TPase enzymes
(Figure 13).57-59 E. coli PBP1B has been proposed to
carry out 85% of PG elongation in that organism.60-62
The remaining 15% of peptidoglycan is made by some
combination of other transglycosylases and transpeptidases. Other high molecular weight PBPs, such as
PBP2 and PBP3 in E. coli, have domains that appear

to be vestigial transglycosylase domains.56 Genetics
experiments have shown that PBP2 is involved in cell
elongation/maintaining cell shape and PBP3 in cell
division.63-65 In addition to the high molecular weight
PBPs, there are several low molecular weight PBPs
containing only transpeptidase or carboxypeptidase
activities. These enzymes are thought to be important
in maintaining the shape of the cell.56,65
Table 1. List of Predicted Bifunctional TGase/TPases

in Gram-Positive Bacterial Pathogens
bifunctional
PBPs
bacteria
E. faecalis
E. faecium
S. aureus
S. pneumoniae
B. subtilis
enzyme
gene
unnamed
unnamed
unnamed
unnamed
unnamed
unnamed
PBP2
PBP1a
PBP2a
PBP1b
PBP1
PBP2c
PBP4
PBP2d
ponA
pbpF
pbpZ
ponA
pbpF
pbpZ
pbpB
pbp1a
pbp2a
pbp1b
ponA
pbpF
pbpd
ywhE
Several high molecular weight TGase/TPase enzyme forms have been identified in other bacterial
strains, including Gram-positive organisms such as
S. aureus, Streptococcus pneumoniae, and Enterococcus faecalis (Table 1).59,66 The putative functions
of most of these enzymes are based on what has been
learned about the corresponding enzymes in E. coli
from a combination of genetic and biochemical studies.
Although genetics has provided considerable insight into the general roles of different PBPs in cell
growth and division, only limited biochemical work
on either the transglycosylase or transpeptidase
domains of the high molecular weight PBPs has been
carried out. In large part, this is because assays to
monitor the activity of the Tgase and Tpase domains
of high molecular weight PBPs were not available
until recently. The lack of good assays for the high
molecular weight PBPs was related to difficulties in
obtaining adequate quantities of lipid II for study.
Both chemical and enzymatic methods to produce
quantities of lipid II have recently enabled the study
of high molecular weight PBPs, and kinetic characterizations of E. coli PBP1b and S. pneumoniae
PBP2a have been reported.10,67-71,136
The historical lack of methods to study and deconvolute the roles of individual transglycosylase and
transpeptidase domains in vitro and in bacterial cells
has made it difficult to understand the detailed
mechanisms of antibiotics that inhibit the final steps
of peptidoglycan biosynthesis. Nevertheless, it is clear
from studying the glycopeptides that some of the
differences in activity between antibiotics with ostensibly similar mechanisms of action are related to
differential inhibition of related targets (i.e., PBPs).
Now that better biochemical and genetic tools to
probe the function of different PBPs have become
available, it should be possible to learn more about
why different glycopeptides have different effects on
cells. This knowledge, in turn, may lead to the
development of better antibiotics.
4.4. Peptidoglycan Biosynthesis as an Antibiotic

Target
Peptidoglycan biosynthesis is a good pathway to
target for antimicrobial chemotherapy, because it is
essential for survival and highly conserved even in

disparate microorganisms. In fact, many antibiotics
that inhibit the biosynthesis of bacterial peptidoglycan are derived from natural products produced by
microorganisms themselves.3 For reasons that are
not clear, a disproportionate number of natural
product-based, cell wall-active antibiotics inhibit the
final steps of peptidoglycan biosynthesis of peptidoglycan biosynthesis. It is possible that these steps
are preferentially targeted because the enzymes
involved are extracellular and are thus accessible to
compounds that would not penetrate the cell membranes. It has also been argued, however, that it is
advantageous to inhibit metabolic processes that
involve multiple, related enzymes because spontaneous mutations in single enzymes do not result in
resistance.72
Antibiotics that inhibit the final stages of peptidoglycan biosynthesis fall into three major classes
with respect to mechanism of action (Figure 14).73
The first class, which includes the -lactams, carbapenems, and cephalosporins, comprises those antibiotics that prevent glycan cross-linking by inhibiting the active sites of enzymes catalyzing transpeptidation. The second class includes those antibiotics that inhibit glycan polymerization by binding to
bacterial transglycosylases. Moenomycin, which is
used commercially in animal feed, is the prototypical
member of this class. It is notable for its outstanding
potency, but is not used in humans because it has
poor physical properties. The third class includes
antibiotics that can inhibit glycan polymerization
and/or cross-linking by binding to the substrates of
transglycosylases and transpeptidases. The glycopeptides are the best known of these substratebinding antibiotics, but there are many others,
including ramoplanin and members of the lantibiotic
family of antibiotics, both of which are reviewed in
this issue in the reviews by Walker and Boger and
by Van der Donk et al. The glycopeptides present
special challenges for mechanistic analysis, because
they are potentially able to inhibit multiple enzymes
involved in two distinct processessi.e., glycan polymerization and cross-linkingsat the same time.
Vancomycin is the prototypical member of the
glycopeptide family of antibiotics and has served as
the model system for many mechanistic investigations of glycopeptides. It is commonly assumed that
teicoplanin and other glycopeptides kill bacterial cells
by the same mechanism of action as vancomycin,
because they recognize the same dipeptide motif on
peptidoglycan precursors (see below). However, there
is increasing evidence that different glycopeptides
preferentially target different enzymatic steps. A
major goal of this review is to draw attention to the
differences between vancomycin, teicoplanin, and
other glycopeptide antibiotics and to address possible
mechanistic explanations for these differences. In the
following sections, we provide an overview of the
general mechanism of action and the molecular basis
for resistance of the glycopeptide class of antibiotics
as a prelude to discussing the differences between
compounds.
Kahne et al.
Figure 14. Three classes of antibiotics inhibiting stage III of peptidoglycan assembly: -lactams, moenomycin, and
glycopeptides.
4.5. Early Studies on the Mechanism of Action of

Glycopeptide Antibiotics
Vancomycin was discovered in 1956 and its general
mechanism of action was elucidated in the 1960s.41,74
In 1965, Strominger and co-workers reported the first
studies on the mechanism of action of vancomycin
and the related glycopeptide ristocetin.40 These authors carried out a series of studies using crude
bacterial membranes containing peptidoglycan-synthesizing enzymes supplemented with radiolabeled
UDP-MurNAc pentapeptide and UDP-GlcNAc in
which they showed that the glycopeptide antibiotics
block peptidoglycan biosynthesis at either transglyoslyation or transpeptidation. This work represented
a tour de force at the time, because key steps in the
biosynthetic pathway to peptidoglycan were not
understood. Subsequent work by Perkins showed
that vancomycin actually binds to peptidoglycan
precursors, specifically to the D-Ala-D-Ala terminus,
leading to the hypothesis that enzyme inhibition is
related to this phenomenon.75,76 Stepwise degradation
of UDP-MurNAc-pentapeptide indicated that vancomycin interacts with the D-Ala-D-Ala terminus of this
precursor. In addition to UDP-MurNAc-pentapeptide,
other peptidoglycan intermediates that contain the
D-Ala-D-Ala dipeptide include (inward facing) lipid I,
(inward and outward facing) lipid II, and nascent (uncross-linked) peptidoglycan (Figures 10 and 12). Since
experiments with radioactive vancomycin derivatives
showed that vancomycin does not enter cells, it was
concluded that vancomycin and other glycopeptides
affect the extracellular enzymes that utilize intermediates downstream of lipid I, such as lipid II.77
Many of the structural features of vancomycin were
determined in 1978, when Sheldrick and Williams
determined the structure of a degradation product,
CDP-1, through X-ray analysis.78 However, the structure of vancomycin was damaged during the degradation process and the correct structure was finally
Figure 15. Complexation of vancomycin with N-acyl-DAla4-D-Ala5 termini: five hydrogen bonds between the
underside of the glycopeptide and the acyl-D,D-dipeptide
moiety.
put forth in 1982, based on important contributions

from the Harris and Williams labs.79,80 Shortly after
the structure of vancomycin was solved, the Williams
lab used NMR to show the binding interaction
between vancomycin and the D-Ala-D-Ala dipeptide.81
Binding was shown to occur through a set of backbone contacts between the D-Ala-D-Ala dipeptide and
the amides that line a cleft formed by the cross-linked
heptapeptide of the glycopeptide. (Figure 15) All
glycopeptides have a similar binding pocket and are
believed to make the same contacts.82 Through binding to D-Ala-D-Ala, the bound glycopeptide acts as a
steric impediment that prevents lipid II and/or the
nascent glycan chain from being processed further.83

The net result is inhibition of the transglycosylation
and/or transpeptidation steps of peptidoglycan synthesis, which overall weakens the peptidoglycan
layers and leaves the cell susceptible to lysis due to
changes in osmotic pressure.
5. Mechanisms of Resistance
Scientists have been intrigued by the unusual
mechanism of action and behavior of vancomycin
since its discovery. Researchers noticed early on that
it is difficult to induce resistance to vancomycin.
Ziegler et al., for example, compared penicillin and
vancomycin, both of which inhibit stage III of peptidoglycan synthesis, and found that the MICs of
penicillin against a range of S. aureus strains increased by more than 100 000-fold after 25 serial
passages in antibiotic-containing media.84 In contrast,
the MICs of vancomycin increased by only 8-fold.
Furthermore, whereas resistance to the -lactams
appeared almost immediately upon the introduction
of penicillin into clinical use, glycopeptide resistance
was not observed for a very long time. Resistance to
antibiotics typically develops when either an antibiotic itself or its target is altered in some way.11 Unlike
the -lactams, the glycopeptides did not appear to be
susceptible to modifications that rendered them
inactive. Furthermore, it was widely believed that
microorganisms could not readily alter the target of
the glycopeptides, the D-Ala-D-Ala peptide terminus,
because that would entail simultaneous, coordinated
changes to multiple enzymes in the pathway to
peptidoglycan synthesis. In 1986, Cooper and Given
noted that during the three decades in which vancomycin has been in clinical use, there has been no
trend toward resistance among organisms usually
susceptible, and speculated that the mode of action
of glycopeptide antibiotics made the development of
high-level resistance virtually impossible.85 One year
later, vancomycin-resistant enterococcal strains began to appear in hospitals, and 15 years after that
the incidence of VRE in hospitalized patients with
enterococcal infections in the US had spread to 30%.1
In retrospect, the appearance of significant glycopeptide resistance is not surprising, because the
widespread use of an antibiotic virtually guarantees
the emergence of resistance.86 However, high-level
resistance to the glycopeptides in enterococci is not
the result of spontaneous mutations in clinically
relevant microorganisms. Instead, the genes conferring glycopeptide resistance in the organisms producing glycopeptide antibiotics appear to have been
transferred to pathogenic microorganisms. The mechanism of glycopeptide resistance in enterococci was
elucidated by Courvalin, Walsh, and their co-workers
in the 1990s. Subsequent work on glycopeptide
resistance in producer organisms has revealed that
they contain the same sets of resistance genes as the
resistant enterococcal strains. The mechanism of
glycopeptide resistance in enterococci is described
below.
Figure 16. Schematic of the Five Gene Van RSHAX

cassette in VanA and VanB phenotypes of VRE.
including additional variants of these classes89). The

VanA and VanB phenotypes were initially distinguished by differential susceptibility to vancomycin
vs teicoplanin. The VanA phenotype shows 1000-fold
increased resistance to both drugs, while VanB VRE
isolates have equivalent resistance to vancomycin but
remain susceptible to teicoplanin.4 Additionally, a
VanC VRE phenotype has been observed: it confers
about 1 log increase in resistance to vancomycin and
has not been a widespread problem in humans.88
VanA and VanB isolates of VRE contain five van
genes, VanRSHAX, necessary and sufficient to cause
high-level resistance (Figure 16).
The encoded five proteins sort into two categories.
The VanR and VanS pair up to function as a twocomponent regulatory system.90 VanS is a transmembrane receptor histidine kinase. The extramembrane domain, directly or indirectly, senses vancomycin on the outside and transmits that information
to the cytoplasmic domain, which autophosphorylates
on the His side chain. The phospho-VanS then
transfers the PO3 group to an aspartyl side chain in
the N-terminal domain of VanR in the enterococcal
cytoplasm. The C-terminal domain of phospho-VanR
now acts as a transcriptional regulator91,92 to activate
the transcription of the VanHAX genes.
The VanHAX proteins comprise the second category. All three are enzymes, coordinately acting to
reprogram the peptidoglycan termini from N-acyl-DAla-D-Ala, a high-affinity target of vancomycin and
teicoplanin, as noted above in section 4, to N-acyl-DAla-D-lactate. VanH reduces the common metabolite
pyruvate to D-lactate. VanA is a D,D-depsipeptide
ligase, making D-Ala-D-lactate. VanX is a D-Ala-D-Ala
dipeptidase, removing the normal D-D-dipeptide intermediate hydrolytically while sparing D-Ala-Dlactate. The D-Ala-D-Lac accumulates and gets added
by the MurF ligase to UDP muramyltripeptide to
generate the UDP muramyl-L-Ala-D--Glu-L-Lys-DAla-D-Lac that can serve as cross-linking substrate
for transpeptidase action (Figure 17).
The 1000-fold resistance in VanA and -B VRE
phenotypes results from the reprogramming of the
peptidoglycan termini from D-Ala-D-Ala to D-Ala-Dlactate.90,93 The absence of the amide NH bond and
the ground-state repulsion of the oxygen lone pair
cause a 1000-fold loss of binding affinity for vancomycin and teicoplanin to N-acyl-D-Ala-D-lactate, correlating genotype with phenotype (Figure 18).93,94
5.1. Vancomycin Reistant Enterococci (VRE)
5.2. VRE Genotypes: VanB Resistance and How

To Overcome It
VRE has been categorized into distinct clinical

phenotypes, originally VanA and VanB87,88 (but now
One of the interesting aspects of glycopeptide

resistance is that the VanB genotype remains sus-
Figure 17. Action of VanHAX to make D-Ala-D-Lac and

destroy competing D-Ala-D-Ala as a substrate for MurF.
Figure 18. Loss of the one H-bond in the vancomycin

complex with D-Ala-D-Lac vs D-Ala-D-Ala and ground-state
repulsion by the ester oxygen accounts for a 1000-fold drop
in affinity.
ceptible to teicoplanin. VanB strains of VRE have

vanRSHAX genes and can reprogram PG termini to
produce D-Ala-D-Lac. However, only vancomycin induces these strains to make altered peptidoglycan
precursors.87 The difference in behavior between
vancomycin and teicoplanin with respect to VanB
strains is puzzling, because vancomycin and teicoplanin have remarkably similar structures (Figure
1). There are many hypotheses that attempt to
explain these observations.
Williams and co-workers found that lipidated glycopeptides are anchored to the bacterial cell membrane, whereas nonlipidated glycopeptides are distributed more broadly in the peptidoglycan layers.95,96
They reasoned that teicoplanin inserts itself into the
membrane, where the peptidoglycan precursors terminating in D-Ala-D-Lac are located. Thus, the two
binding partners are located in close proximity to
each other and the interaction between teicoplanin
and D-Ala-D-Lac becomes intramolecular, which is
more favorable than an intermolecular interac-
Kahne et al.
tion.97,98 Williams and co-workers concluded that it

is the restriction of motion of intermediates and the
intramolecular bond formation that allow teicoplanin
to overcome VanB resistance.31
In 1996, Courvalin and co-workers found that
VanB VRE remain susceptible to teicoplanin, because
this antibiotic does not induce the cell wall reprogramming machinery. They showed that vancomycin,
but not teicoplanin, activates the VanSb sensor kinase.99 Furthermore, they found that mutations in
the VanSb sensor kinase render cells resistant to
teicoplanin.100 From these findings, they reasoned
that activation of VanSb required direct interaction
with the glycopeptide.100,101
To determine which structural features of teicoplanin and vancomycin correlate with induction of
resistance, Dong et al. undertook a direct comparison
of pairs of teicoplanin and vancomycin analogues.
The compounds examined included the teicoplanin
and vancomycin aglycons, the monoglucosylated scaffolds bearing the same disaccharyl chain, and the two
scaffolds bearing a C10-acyl glucosamine group (Figure 19).102
The teicoplanin and vancomycin aglycons and the
monoglucosylated scaffolds induced resistance, but
the lipoglycopeptides did not (as judged by susceptibility of VanB VRE strains to the test compounds).
Thus, adding a simple C10 acyl chain to vancomycin
abrogated its ability to activate the resistance genes.
More recently, Boger and co-workers have investigated the activity of methyl ester derivatives of
vancomycin and teicoplanin, replacing the carbohydrate portion of the glycopeptide with a methyl
group.103,104 They found that these compounds were
still active against VanB VRE strains and attributed
their activity to the hydorophobic nature of these
compounds. Thus, they reasoned that it might not
be necessary to add a lipid chain to the glycopeptide
to overcome VanB resistance; rather, a simple methyl
group will prove effective.
Two hypotheses to explain the differences between
lipoglycopeptides and the corresponding glycopeptides vis-a` -vis the induction of resistance have been
put forth.102 Both hypotheses attribute the differences
between lipidated/hydrophobic and nonlipidated glycopeptides to differential partitioning among available targets on bacterial cell surfaces. Williams and
co-workers showed that the lipid chain of teicoplanin
localizes the molecule to the membrane. Therefore,
it has been suggested that membrane localization
renders teicoplanin and other lipidated glycopeptides
inaccessible to the sensor kinase, so that induction
does not occur. This hypothesis presumes that induction of resistance involves a direct interaction between the sensor kinase and the antibiotic.100,101
Alternatively, it has been suggested that the sensor
kinase interacts not with the glycopeptide itself but
with peptidoglycan intermediates or degradation
products produced by the metabolic blockade.102
Teicoplanin does not induce resistance according to
this hypothesis, because it blocks a different enzymatic step than vancomycin, which means that
different intermediates and degradation products
buildup.
Figure 19. Teicoplanin and vancomycin analogues. A set of matched pairs of vancomycin and teicoplanin derivatives
(a-f) were used to probe what triggers the sensor kinase. It was found that C and D did not induce resistance, whereas
all other compounds did.
There is no direct evidence about how the sensor

kinase is activated. However, using an E. coli model
system, Kahne and co-workers have shown that
lipidated glycopeptides preferentially inhibit the
transglycosylation step of peptidoglycan biosynthesis,
whereas nonlipidated glycopeptides inhibit the
transpeptidation step.8,105 Membrane-anchored glycopeptides evidently inhibit transglycosylation, because they bind preferentially to lipid II, the membrane-anchored substrate for the transglycosylases,
rather than to nascent peptidoglycan (Figure 20).
Thus, relatively small structural changes in glycopeptides can lead to differential partitioning among
available binding sites and thus, in turn, can have
functional consequences in terms of the metabolic
steps that are affected. While it is still not known

how the VanSB sensor kinases are activated, mechanistic investigations aimed at understanding the
molecular basis for glycopeptide resistance have
already led to a clear prescription for how to avoid
reprogramming the bacterial cell wall, namely, use
lipoglycopeptides.74,102
5.3. VRE Genotypes: VanA Resistance and the

Compounds That Overcome It
VanA-resistant enterococcal strains contain similar
sets of genes for sensing glycopeptide activity and for
reprogramming peptidoglycan precursors as VanB
strains. Unlike the latter, however, VanA strains are
Kahne et al.
Figure 20. Two types of TPase substrates containing -D-Ala-D-Ala termini (in red) that can be complexed with vancomycin.
lipid II are molecules embedded in the bacterial membrane via the C55 anchor. Immature PG chains that have been elongated
by TGase action and are now ready for TPase action.
resistant to all natural glycopeptides that have been

examined, including teicoplanin. Thus, irrespective
of differences in how they are distributed among the
possible binding sites or which steps they preferentially block, natural glycopeptides induce VanA strains
to make altered cell wall precursors. Nevertheless,
some semisynthetic glycopeptide derivatives overcome VanA resistance, and studies on these compounds are beginning to provide clues as to how this
form of resistance can be overcome.
Researchers at Eli Lilly reported the first glycopeptide derivatives capable of overcoming both VanA
and VanB resistance. These compounds, which were
vancomycin derivatives containing lipid substituents
attached to the nitrogen of the vancosamino sugar
(Figure 21), showed good activity against a broad
panel of vancomycin-resistant strains as well as
outstanding activity against sensitive strains.106,107
The vancosamino nitrogen of vancomycin can be
readily modified without extensive protecting group
manipulations and was, therefore, a logical position
to explore. The motivation for attaching hydrophobic
substituents was not made clear in the reports on
these compounds, but it is possible that teicoplanins
activity against VanB strains provided the inspiration. It would have been reasonable to speculate that
activity against VanB strains could be recovered by
attaching a hydrophobic substituent to the vanco-
Figure 21. Lipoglycopeptide derivatives found at Eli Lilly

to reverse VanA phenotypic resistance.
mycin carbohydrate, because the teicoplanin carbohydrate, which is attached to the same amino acid,
contains a hydrophobic substituent. Remarkably,
some of the glycolipid derivatives of vancomycin

proved to be active not only against VanB strains but
also against VanA strains, and the Lilly group
ultimately took a lipid derivative of chloroeremomycin, now known as oritavancin, into the clinic (vide
infra). Understanding the mechanism by which VanA
resistance is overcome is crucial to developing better
second-generation glycolipid derivatives. Two different models for how compounds such as oritavancin
overcome VanA resistance have been proposed and
are discussed below.

Table 2. MICs for a Series of Glycopeptides
Derivatives against Sensitive and Resistant Strainsa
5.4. Models for How Vancomycin Analogues

Overcome Vancomycin Resistance
5.4.1. Dimerization and Membrane Anchoring
Williams and co-workers proposed the first model
which holds that vancomycin analogues kill resistant
bacterial strains by the same mechanism that they
kill sensitive strains, i.e., these analogues still bind
to peptide termini of peptidoglycan precursors such
as lipid II, but they have acquired the ability to bind
to D-Ala-D-Lac better than vancomycin itself can (for
a review, see Williams et al.).31 Williams has proposed that the lipid substituent on vancomycin
facilitates dimerization and membrane localization.
It is these characteristics that promote better binding
to D-Ala-D-Lac peptides that are presented in multiple
copies in close proximity on the cell surface.108 Williams argues that the second binding event is entropically favored to the extent that it compensates
for the loss of the hydrogen bond between the
depsipeptide ligand and the glycopeptide. There is
considerable evidence to support the proposal that
dimerization can enhance binding avidity to ligands
that are presented in multiple copies on a surface of
vesicles.109 Moreover, it has been shown that covalent
dimers and covalent trimers of vancomycin bind with
high avidity to polyvalent peptide ligands.110,111 However, there is no evidence that improved biological
activity is related to the ability of glycopeptide
analogues to dimerize. Furthermore, there is no
evidence that these glycopeptides kill resistant bacteria by binding to D-Ala-D-Lac.
Membrane anchoring has also been suggested as
a factor contributing to enhanced activity against
resistant strains. It has been proposed that localizing
the glycopeptide near the membrane partially compensates for the weak binding to D-Ala-D-Lac. Again,
there is both theoretical and experimental support
for the hypothesis that anchoring to membranes
should increase avidity of glycopeptide analogues for
membrane-anchored precursors presenting D-Ala-DAla or D-Ala-D-Lac.95,108 Williams argues that it is a
combination of membrane anchoring and dimerization that allows lipidated derivatives of vancomycin
to bind to D-Ala-D-lactate. To show the effectiveness
of the dimerization/membrane localization combination, Allen and co-workers damaged the lipidated
derivative of vancomycin so that it could no longer
bind D-Ala-D-Ala.112 It has been shown that if the
N-terminal amino acid is removed from vancomycin,
it can no longer bind to peptidoglycan precursors.113
Allen and co-workers showed that the damaged
MIC (g/mL)
glycopeptide
vancomycin
chlorobiphenyl vancomycin
damaged vancomycin
damaged chlorobiphenyl
vancomycin
sensitive
E. faecium
resistant
E. faecium
1
0.03
no activity
10
2048
16
no activity
40
a Compounds lacking the N-terminal methylleucine amino

acid were used to evaluate which component of the activity
derives from a peptide-binding-independent mechanism.
compounds still formed dimers and that these dimers

were able to bind D-Ala-D-Ala, albeit very weakly.112
Allen and co-workers attributed the biological activity
of these damaged compounds to their ability to
dimerize, which allows them to weakly bind to
precursors. However, it is interesting to note that
Ellman and co-workers have published data showing
that covalent dimers of damaged vancomycin derivatives, nevertheless, have biological activity.114,115 The
following section describes a second mechanism of
action that explains how these lipidated compounds
kill resistant bacteria, despite being unable to bind
peptidoglycan precursors.
5.4.2. A Second Mechanism of ActionsDirect Interaction

with the Transglycosylase
Biophysical studies of binding using model systems
are at best suggestive with respect to mechanism of
action. Kahne and co-workers, therefore, designed
experiments to test the hypothesis that activity
against resistant bacterial strains depends on binding
to D-Ala-D-Lac. To address this issue, vancomycin
analogues in which the peptide binding pocket was
damaged were prepared (Table 2). These analogues
lack the N-methylleucine moiety of the aglycon,
which makes both hydrogen bonding and electrostatic
contacts to D-Ala-D-Ala. It has been established that
these damaged vancomycin analogues are unable to
bind N-acyl-D-Ala-D-Ala.113 The damaged vancomycin
analogues were tested for activity against both sensitive and vancomycin-resistant strains. The chlorobiphenyl derivative lost considerable activity against
sensitive strains, as expected; however, its activity
against resistant strains did not change significantly.
While it is conceivable that membrane-anchoring
or dimerization could compensate for the loss of a
single hydrogen bond, it would not be reasonable to
Figure 22. Two mechanisms of action for glycopeptides:

(a) inhibition of the transpeptidase step by binding to the
D-Ala-D-Ala terminus and (b) direct inhibition of the
transglycosylases.
conclude that these phenomena compensate for the

loss of two hydrogen bonds and an electrostatic
contact. Kahne and co-workers therefore concluded
that depsipeptide binding does not play a role in the
mechanism of action by which chlorobiphenyl vancomycin analogues kill VanA resistant bacteria. They
proposed, instead, that there must be a second mechanism of action that explains the activity against resistant strains. Studies aimed at probing the site of
inhibition of the damaged chlorobiphenyl vancomycin
analogues established that the damaged compound,
like the intact parent, blocks peptidoglycan biosynthesis8 at the transglycosylation step (Figure 22).
On the basis of these experiments, they suggested
that chlorobiphenyl vancomycin analogues might
interact directly with a key component of the transglycosylation machinery.
Goldman and co-workers provided additional evidence for a second mechanism of action for chlorobiphenyl derivatives of vancomycin when they showed
that both chlorobiphenyl vancomycin and damaged
chlorobiphenyl vancomycin are able to inhibit formation of nascent peptidoglycan when UDP-MurNactetrapeptide (L-Ala--D-Gln-L-Lys-D-Ala) is provided
to bacterial cells as a substrate. Vancomycin, in
contrast, does not inhibit the incorporation of this
substrate into peptidoglycan.116 The following year,
Chapman and co-workers showed through affinity
chromatography of bacterial cell extracts that chlorobiphenyl derivatives of vancomycin retained multiple PBPs,117 including E. coli PBP1b, whereas
vancomycin was unable to retain PBPs. These experiments provided support for the hypothesis that
chlorobiphenyl derivatives of vancomycin interact
directly with PBPs involved in peptidoglycan biosynthesis. Chen et al. subsequently developed an assay
to monitor the activity of E. coli PBP1b and found
that chlorobiphenyl vancomycin analogues can inhibit this enzyme without binding substrate.10 Inhibition by vancomycin, in contrast, depends on peptide
binding. Taken together, these studies provide strong
evidence for a second mechanism of action. At this
point, however, it is essential to develop methods to
study Gram-positive transglycosylases found in relevant microorganisms in order to determine whether
chlorobiphenyl vancomycin analogues directly inhibit
these enzymes.137 If, in fact, vancomycin analogues
with activity against resistant microorganisms func-
Kahne et al.
Figure 23. Proposal for recombination in a patient

infected with VRE and MRSA allowing TN1546 to move
to MRSA and create VRSA.
tion because the derivatized carbohydrate moiety

facilitates a direct and inhibitory interaction with
transglycosylases, then it should be possible to design
improved antibiotics by attaching to the vancomycin
aglycon structural elements that have better activity
against relevant lipid II-utilizing transglycosylases.8,20,46,118 The design of better antibiotics is important not only for overcoming VanA resistance but also
for tackling the problem of the newly emerging
vancomycin-resistant staphylococcus.
5.5. Vancomycin-Resistant S. aureus (VRSA)

For some years the levels of vancomcyin resistance
among clinical isolates of Staphylococci were low. S.
aureus with such diminished susceptibility to vancomycin were termed VISA (vancomycin intermediate S. aureus).119,120 The genotypes of these isolates
were not of the VanHAX type noted above for VRE.
Rather, they typically generated multilayered, thickened cell walls as though more sites for stoichiometric
binding of drug were the cause of lessened susceptibility. But in 2003, VRSA were isolated from a
dialysis patient who also had a chronic infection with
VRE.121 Genotyping of the VRSA showed the same
five VanRSHAX genes found also on transposons in
the VRE (Figure 23). Every indication is that the
transposon hopped from the enterococcal host to the
S. aureus.122 It remains to be seen what the rate of
spread of the VanRSHAX genes will be into MRSA
strains, but there is no doubt that this only hastens
the need for second- and third-generation forms of
glycopeptides and lipoglycopeptides that can combat
various phenotypes of VRE and VRSA.
6. New Directions for Treating VRE and VRSA

The long time frame and progressively broader
clinical use of the first-generation natural product
glycopeptides vancomycin and teicoplanin as front
line agents has eventually led to widespread incidence of VRE (up to 30% of enterocccal infections in
US hospitals in 2002) and the recent emergence of
VRSA. Over the past decade there has been keen
interest in the discovery and development of new
treatment regimes for these life-threatening bacterial
Figure 24. Three nonglycopeptide classes of antibiotics recently approved for treatment of VRE infections: Synercid (11)
combination of Quinupristin and Dalfopristin blocks protein synthesis; the oxazolidinone Linezolid (12) blocks the first
peptide-bond-forming step at bacterial ribosomes; the lipopeptide daptomycin (13) damages membranes and causes ion
leaks in bacteria.
Figure 25. Lipoglycopeptide antibiotics in clinical development: oritavancin (14) and TD-624 (15) are N-aryl and N-alkyl
hydrophobic derivatives modified on the vancomycin-type heptapeptide scaffold; dalbavancin (16) is a lipoglycopeptide
modified in the amide linkage on a teicoplanin-type scaffold.
pathogens. Two parallel lines of discovery and development have ensued. One has been the search for
a replacement to vancomycin and teicoplanin and has

yielded newly approved drugs. The other line began
Kahne et al.
Figure 26. (a) Reconstitution of chloroermomycin from the aglycon scaffold by consecutive action of Gtfs B, A, and C. (b)
Decoration of the teicoplanin aglycon with GlcNAc at residue 6, with glucosamine at residue 4 by tGtfA and tGtfB, and
N-acylation of the glucosamine residue by acyl transferase action.
with semisynthetic improvements to the vancomcyin

and teicoplanin subfamily scaffolds.
6.1. Recently Approved Drugs for VRE

Over the past 5 years three new classes of antibiotics have been approved in the US with efficacy
against enterococcal infections (Figure 24).3 None of
the drugs are glycopeptides, so they do not induce
the VanRSHAX enzymes and therefore are active
against all VRE strains. Synercid (11), approved in
1998, is a mixture of two natural product cyclic
lactones, pristinamycins I and II. They block bacterial
protein synthesis in the elongation phase.4 Linezolid
(12), a synthetic antibacterial agent with an oxazolidinone backbone approved in 2000, blocks the first
peptide-bond-forming step at the peptidyl transferase
center of bacterial ribosomes. The third molecule,
daptomycin (13), approved in 2003 under the trade
name Cubicin, is also a natural product. It is a
lipopeptidolactone, made nonribosomally,4 that dis-
rupts bacterial membrane integrity, causing ions to

leak out and thus bacterial death. All three of these
drugs then act by mechanisms independent of those
used by vancomycin and teicoplanin and will not be
subject to cross-resistance with the glycopeptides.
6.2. Second-Generation Semisynthetic

Lipoglycopeptides in Clinical Development
Given the structural complexity of the glycopeptide
antibiotic class, in particular, the oxidatively crosslinked heptapeptide scaffolds that are variant in
vancomcyin and teicoplanin family members, efforts
have been undertaken to create semisynthetic drugs
by chemical modification of the natural products.
One successful route has involved the single-step
reductive alkylation of the amino group of the vancosamine sugar in the vancomycin/chloroeremomycin
scaffold and led to oritavancin (14). Use of an N-alkyl
rather than N-aryl group has generated TD-624 (15)
(Figure 25). A comparable single-step chemical modi-
Figure 27. Enzymatic synthesis of a variant glycopeptide, starting from the teicoplanin aglycon, UDP-4-deoxyglucose
and GtfB, and TDP-l-epivancosamine and GtfC.
fication by amidation of the carboxy terminus of the

heptapeptide scaffold of the teicoplanin family member A40926 has generated dalbavancin (16).
6.2.1. Oritavancin
Screening of the natural and semisynthetic glycopeptide collection at Lilly led to the choice of the
chloroeremomycin scaffold and a series of N-alkyl or
N-acyl derivatives of the terminal aminosugar on the
glucosyl-epivancosamine chain.123 The oritavancin
molecule, originally known as LY333328, is a simple
N-aryl derivative of the natural product chloroeremomycin 4. The epivancosamine moiety in the
disaccharide chain could be selectively derivatized at
the amino group by imine formation with aryl aldehydes, followed by reduction to the stable secondary
amine. The chlorobiphenyl group was one of several
lipophilic groups that conferred activity against VRE,
allowing a regain of about 2 of the 3 logs of activity
lost in VanA type VRE.123 Arylamines have been used
elsewhere to decorate natural product scaffolds,
notably in the antifungal echinocandins as replacements for fatty acyl amide groups,124 and they may
mimic the N-acyl substituents found in the teicoplanin series of natural lipoglycopeptides. Oritavancin is active against both VanA and VanB phenotypes
of VRE and also MRSA and VRSA at MIC values of
<1 g/mL.123 Once-a-day administration is proposed.
Oritavancin shows strong bacteriocidal properties
under conditions where vancomycin is bacteriostatic.
Oritavancin has shown clinical effectiveness in complicated skin and soft tissue infections caused by
Gram-positive bacteria, including MRSA.125
6.2.2. Dalbavancin
Dalbavancin is a semisynthetic variant of the
teicoplanin family member A40926.22,126,137 Like
A40926 and teicoplanin, it has a long fatty acyl
moiety, in this case a C12 terminally branched chain,
in amide linkage to the glucosamine. The mannose
at residue 7 is present but not the GlcNAc at residue
6. The synthetic modification is amidation of the
C-terminal carboxylic acid of the cross-linked heptapeptide scaffold by a N,N-dimethyl propylamide

group.
This lipoglycopeptide is active against VanB type
VRE at 0.1 g/mL but is ineffective against VanA
phenotypes. It has enhanced potency over vancomycin and teicoplanin against susceptible enterococci
and against MRSA.126 Dalbavancin also has bacteriocidal activity. A once weekly administration is
being planned, given the long lifetime.127 Dalbavancin
has also shown efficacy in complicated skin and skin
structure infections.128
6.2.3. TD-6424
This is the third of the second-generation, semisynthetic lipoglycopeptide variants to enter clinical
development. In some analogy to oritavnacin, TD6424 utilizes a vancomycin-type cross-linked heptapeptide scaffold. There is one modification in that
scaffold at residue 7. The Dpg7 has been replaced
with a 4-aminoethyl side chain, in turn bearing a
CH2PO32- substituent, to enhance solubility of the
scaffold. The other modification maintains the lipoglycopeptide character of these second-generation
compounds. The vancosamine sugar is alkylated by
a long alkyl chain, with an NH at the two positions.
The reductive alkylation is again a presumptive
mimic of the normal fatty acyl chain attached to the
single sugar of teicoplanin. TD-6424 is rapdly bacteriocidal129 and more potent than teicoplanin or
vancomycin against MSSA and MRSA.
6.3. In Discovery: Chemoenzymatic Routes To

Modify Sugars and Acyl Groups on Heptapeptide
Scaffolds
The second-generation lipoglycopeptides in development noted above represent different one-step
chemical manipulations of chloroeremomycin or
A40926, and multiple-step chemical manipulation of
both scaffold and aminosugar N-alkylation in TD-624
on the natural products derived from fermentation.
The current knowledge of the biosynthetic gene
clusters for both vancomycin and teicoplanin family
members noted in section 3 and the importance of
the sugars and acyl groups decorating the heptapeptide scaffold noted in section 5 have focused attention
on using the tailoring enzymes to embellish the
scaffolds in new combinations.
For example, three glycosyltransferases, GtfA, -B,
and -C, convert the vancomycin aglycon to chloroeremomycin, a pathway that has been reconstituted in
vitro with the purified enzymes (Figure 26a)38 Analogously, the two homologous Gtfs from the teicoplanin
cluster, tGtfA and tGtfB, have been overproduced
and shown to transfer GlcNAc to residues 6 and 4 of
the teicoplanin aglycon.16,17 Further, tGtfB will transfer glucosamine, which can be N-acylated by an acyl
transferase also encoded in the biosynthetic cluster
(Figure 26b).17 Different sugars and scaffolds can be
used by some of the Gtfs such that novel glycopeptides can be assembled. Thus, GtfB will take several
UDP-glucose derivatives,130,131 some of which can be
elongated by GtfC or by GtfD from the vancomycin
biosynthetic cluster. It is possible, for example, to
Kahne et al.
build a variant disaccharide, e.g. 4-deoxy-D-glucose2,1-L-epivancosamine, on an altered scaffold, the

teicoplanin aglycon, introducing three variations via
the action of GtfB and -D (Figure 27).132 X-ray
structures of GtfA, -B, and -D133-135 have been
determined to allow for molecular insights into Gtf
specificity for NDP sugar donor and scaffold acceptor
and aid in structure-based Gtf evolution.
Finally, it is possible to N-alkylate and N-arylate
variant aminosugars introduced by Gtfs on the heptapeptide scaffolds to effect chemoenzymatic manipulations that enable multistep variations of structure
in glycopeptide and lipoglycopeptide antibiotics.102
This approach should allow construction of focused
libraries of novel lipoglycopeptide structures to optimize properties desired in a third-generation glycopeptide antibiotic, including broad spectrum, rapid
cidality, oral activity, high potency, good pharmacokinetic, and therapeutic ratio.
7. Acknowledgments
The authors are indebted to many colleagues in the
Kahne and Walsh research groups for contributions
to glycopeptide and lipoglycopeptide projects, some
of which are acknowledged in the references cited.
Research in the Walsh group was supported in part
by NIH Grant GM49338 and in the Kahne group by
NIH Grant 66174. C.W. is a member of the board of
directors of Vicuron Pharmaceuticals.
8. Note Added after ASAP Publication

This review was posted ASAP on January 22, 2005.
The spelling of transglycosylase was corrected in
Figures 12 and 13, and refs 136 and 137 were added.
The review was reposted January 27, 2005.
9. References
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
Murray, B. E. N. Engl. J. Med. 2000, 342, 710.

Ferber, D. Science 2003, 302, 1488.
Walsh, C. T. Nat. Rev. Microbiol. 2003, 1, 65.
Walsh, C. T. Antibiotics: Actions, Origins, Resistance; ASM
Press: 2003.
Williams, D. H. Nat. Prod. Rep. 1996, 13, 469.
Nicolaou, K. C.; Boddy, C. N.; Brase, S.; Winssinger, N. Angew.
Chem., Int. Ed. 1999, 38, 2096.
Thorson, J. S., Hosted, T. J., Jiang, J. Q., Biggins, J. B., Ahlert,
J. Curr. Org. Chem. 2001, 5, 139.
Ge, M.; Chen, Z.; Onishi, H. R.; Kohler, J.; Silver, L. L.; Kerns,
R.; Fukuzawa, S.; Thompson, C.; Kahne, D. Science 1999, 284,
507.
Kahne, D.; Silva, D.; Walker, S. In Bioorganic Chemistry:
Carbohydrates; Hecht, S. M., Ed.; Oxford University Press: New
York, 1999.
Chen, L.; Walker, D.; Sun, B.; Hu, Y.; Walker, S.; Kahne, D.
Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 5658.
Walsh, C.; Freel Meyers, C. L.; Losey, H. C. J. Med. Chem. 2003,
46, 3425.
van Wageningen, A. M.; Kirkpatrick, P. N.; Williams, D. H.;
Harris, B. R.; Kershaw, J. K.; Lennard, N. J.; Jones, M.; Jones,
S. J.; Solenberg, P. J. Chem. Biol. 1998, 5, 155.
Pelzer, S.; Sussmuth, R.; Heckmann, D.; Recktenwald, J.; Huber,
P.; Jung, G.; Wohlleben, W. Antimicrob. Agents Chemother. 1999,
43, 1565.
Recktenwald, J.; Shawky, R.; Puk, O.; Pfennig, F.; Keller, U.;
Wohlleben, W.; Pelzer, S. Microbiology 2002, 148, 1105.
Pootoolal, J.; Thomas, M. G.; Marshall, C. G.; Neu, J. M.;
Hubbard, B. K.; Walsh, C. T.; Wright, G. D. Proc. Natl. Acad.
Sci. U.S.A. 2002, 99, 8962.
Sosio, M.; Kloosterman, H.; Bianchi, A.; de Vreugd, P.; Dijkhuizen, L.; Donadio, S. Microbiology 2004, 150, 95.

(17) Li, T. L.; Huang, F.; Haydock, S. F.; Mironenko, T.; Leadlay, P.
F.; Spencer, J. B. Chem. Biol. 2004, 11, 107.
(18) Evans, D.; Wood, M.; Trotter, B.; Richardson, T.; Barrow, J.;
Katz, J. Angew. Chem., Int. Ed. 1998, 37, 2700.
(19) Boger, D. L. Med. Res. Rev. 2001, 21, 356.
(20) Thompson, C.; Ge, M.; Kahne, D. J. Am. Chem. Soc. 1999, 121,
1237.
(21) Hubbard, B. K.; Walsh, C. T. Angew. Chem., Int. Ed. 2003, 42,
730.
(22) Sosio, M.; Stinchi, S.; Beltrametti, F.; Lazzarini, A.; Donadio,
S. Chem. Biol. 2003, 10, 541.
(23) Lambalot, R. H.; Gehring, A. M.; Flugel, R. S.; Zuber, P.; LaCelle,
M.; Marahiel, M. A.; Reid, R.; Khosla, C.; Walsh, C. T. Chem.
Biol. 1996, 3, 923.
(24) Hubbard, B. K.; Thomas, M. G.; Walsh, C. T. Chem. Biol. 2000,
7, 931.
(25) Pfeifer, V.; Nicholson, G. J.; Ries, J.; Recktenwald, J.; Schefer,
A. B.; Shawky, R. M.; Schroder, J.; Wohlleben, W.; Pelzer, S. J.
Biol. Chem. 2001, 276, 38370.
(26) Luo, L.; Kohli, R. M.; Onishi, M.; Linne, U.; Marahiel, M. A.;
Walsh, C. T. Biochemistry 2002, 41, 9184.
(27) Chen, H.; Tseng, C. C.; Hubbard, B. K.; Walsh, C. T. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 14901.
(28) Chen, H.; Walsh, C. T. Chem. Biol. 2001, 8, 301.
(29) Trauger, J. W.; Walsh, C. T. Proc. Natl. Acad. Sci. U.S.A. 2000,
97, 3112.
(30) Luo, L.; Walsh, C. T. Biochemistry 2001, 40, 5329.
(31) Williams, D. H., Bardsley, B. Angew. Chem., Int. Ed. 1999, 38,
1172.
(32) Bischoff, D.; Pelzer, S.; Bister, B.; Nicholson, G. J.; Stockert, S.;
Schirle, M.; Wohlleben, W.; Jung, G.; Sussmuth, R. D. Angew.
Chem., Int. Ed. 2001, 40, 4688.
(33) Bischoff, D.; Pelzer, S.; Holtzel, A.; Nicholson, G. J.; Stockert,
S.; Wohlleben, W.; Jung, G.; Sussmuth, R. D. Angew. Chem.,
Int. Ed. 2001, 40, 1693.
(34) Zerbe, K.; Pylypenko, O.; Vitali, F.; Zhang, W.; Rouset, S.; Heck,
M.; Vrijbloed, J. W.; Bischoff, D.; Bister, B.; Sussmuth, R. D.;
Pelzer, S.; Wohlleben, W.; Robinson, J. A.; Schlichting, I. J. Biol.
Chem. 2002, 277, 47476.
(35) Pylypenko, O.; Vitali, F.; Zerbe, K.; Robinson, J. A.; Schlichting,
I. J. Biol. Chem. 2003, 278, 46727.
(36) Solenberg, P. J.; Matsushima, P.; Stack, D. R.; Wilkie, S. C.;
Thompson, R. C.; Baltz, R. H. Chem. Biol. 1997, 4, 195.
(37) Losey, H. C.; Peczuh, M. W.; Chen, Z.; Eggert, U. S.; Dong, S.
D.; Pelczer, I.; Kahne, D.; Walsh, C. T. Biochemistry 2001, 40,
4745.
(38) Lu, W.; Oberthur, M.; Leimkuhler, C.; Tao, J.; Kahne, D.; Walsh,
C. T. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 4390.
(39) Chen, H.; Thomas, M. G.; Hubbard, B. K.; Losey, H. C.; Walsh,
C. T.; Burkart, M. D. Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
11942.
(40) Anderson, J. S.; Matsuhashi, M.; Haskin, M. A.; Strominger, J.
L. Proc. Natl. Acad. Sci. U.S.A. 1965, 53, 881.
(41) Gale, E. F.; Cundliffe, E.; Reynold, P. E.; Richmond, M. H.;
Waring, M. J. In Molecular Basis of Antibiotic Actions; Wiley
and Sons Ltd.: London, 1981.
(42) Holtje, J. V. Microbiol. Mol. Biol. Rev. 1998, 62, 181.
(43) Bugg, T. D. In Comprehensive Natural Products Chemistry;
Pinto, B. M., Ed.; Elsevier: Oxford, 1999; Vol. 3.
(44) Men, H.; Park, P.; Ge, M.; Walker, S. J. Am. Chem. Soc. 1998,
120, 2484.
(45) Bupp, K.; van Heijenoort, J. J. Bacteriol. 1993, 175, 1841.
(46) Chen, Z.; Eggert, U. S.; Dong, S. D.; Shaw, S. J.; Sun, B.; LaTour,
J. V.; Kahne, D. Tetrahedron 2002, 58, 6585.
(47) Ha, S., Walker, D., Shi, Y., Walker, S. Protein Sci. 2000, 9, 1045.
(48) Yan, Y.; Munshi, S.; Leiting, B.; Anderson, M. S.; Chrzas, J.;
Chen, Z. J. Mol. Biol. 2000, 304, 435.
(49) Schonbrunn, E.; Sack, S.; Eschenburg, S.; Perrakis, A.; Krekel,
F.; Amrhein, N.; Mandelkow, E. Structure 1996, 4, 1065.
(50) Benson, T. E.; Harris, M. S.; Choi, G. H.; Cialdella, J. I.; Herberg,
J. T.; Martin, J. P., Jr.; Baldwin, E. T. Biochemistry 2001, 40,
2340.
(51) Mol, C. D.; Brooun, A.; Dougan, D. R.; Hilgers, M. T.; Tari, L.
W.; Wijnands, R. A.; Knuth, M. W.; McRee, D. E.; Swanson, R.
V. J. Bacteriol. 2003, 185, 4152.
(52) Bertrand, J.; Auger, G.; Fanchon, E.; Marin, L.; Blanot, D.; van
Heijenoort, J.; Dideberg, O. EMBO J. 2000, 16, 3416.
(53) Gordon, E.; Flouret, B.; Chantalat, L.; van Heijenoort, J.;
Mengin-Lecreulx, D.; Dideberg, O. J. Biol. Chem. 2001, 276,
10999.
(54) Fan, C.; Park, I. S.; Walsh, C. T.; Knox, J. R. Biochemistry 1997,
36, 6, 2531.
(55) Yamaguchi, H.; Kato, H.; Hata, Y.; Nishioka, T.; Kimura, A.;
Oda, J.; Katsube, Y. J. Mol. Biol. 1993, 229, 1083.
(56) Goffin, C.; Ghuysen, J. Microbiol. Mol. Biol. Rev. 1998, 62, 1079.
(57) Schiffer, G.; Holtje, J.-V. J. Biol. Chem. 1999, 274, 32031.
(58) Broome-Smith, J. K.; Edelman, A.; Yousif, S.; Spratt, B. G. Eur.
J. Biochem. 1985, 147, 437.

(59) DiGuilmi, A. M.; Dessen, A.; Dideberg, O.; Vernet, T. Curr.
Pharam. Biotechnol. 2002, 3, 63.
(60) Suzuki, H.; Nishimura, Y.; Hirota, Y. Proc. Natl. Acad. Sci.
U.S.A. 1978, 75, 664.
(61) Tamaki, S.; Nakajima, S.; Matsuhashi, M. Proc. Natl. Acad. Sci.
U.S.A. 1977, 74, 5472.
(62) Nakagawa, J.; Tamaki, S.; Tomioka, S.; Matsuhashi, M. J. Biol.
Chem. 1984, 259, 13937.
(63) Nanninga, N. Microbiol. Mol. Biol. Rev. 1998, 62, 110.
(64) Buddelmeijer, N.; Beckwith, J. Curr. Opin. Microbiol. 2002, 5,
553.
(66) Arbeloa, A.; Segal, H.; Hugonnet, J.-E.; Josseaume, N.; Dubost,
L.; Brouard, J.-P.; Gutmann, L.; Mengin-Lecreulx, D.; Arthur,
M. J. Bacteriol. 2004, 186, 1221.
(67) DiGuilmi, A. M.; Dessen, A.; Dideberg, O.; Vernet, T. J. Bacteriol.
2003, 185, 4418.
(68) Ye, X. Y.; Lo, M. C.; Brunner, L.; Walker, D.; Kahne, D.; Walker,
S. J. Am. Chem. Soc. 2001, 123, 3155.
(69) VanNieuwenhze, M. S.; Mauldin, S. C.; Zia-Ebrahimi, M.;
Winger, B. E.; Hornback, W. J.; Saha, S. L.; Aikins, J. A.;
Blaszczak, L. C. J. Am. Chem. Soc. 2002, 124, 3656.
(70) Schwartz, B.; Markwalder, J. A.; Wang, Y. J. Am. Chem. Soc.
2001, 123, 11638.
(71) Schwartz, B.; Markwalder, J. A.; Seitz, S. P.; Wang, Y.; Stein,
R. L. Biochemistry 2002, 41, 12552.
(72) Silver, L. L.; Bostian, K. A. Antimicrob. Agents Chemother. 1993,
37, 377.
(73) Silver, L. L. Curr. Opin. Microbiol. 2003, 6, 431.
(74) McCormick, M. H.; McGuire, J. M.; Pittenger, G. E.; Pittenger,
R. C.; Stark, W. M. Antibiot. Annu. 1956, 3, 606.
(75) Perkins, H. R. Biochem. J. 1969, 111, 195.
(76) Chatterjee, A. N.; Perkins, H. R. Biochem. Biophys. Res. Commun. 1966, 24, 489.
(77) Bordet, C.; Perkins, H. R. Biochem. J. 1970, 119, 877.
(78) Sheldrick, G. M.; Jones, P. G.; Kennard, O.; Williams, D. H.;
Smith, G. A. Nature 1978, 271, 223.
(79) Williamson, M. P.; Williams, D. H. J. Am. Chem. Soc. 1981, 103,
6580.
(80) Harris, C. M.; Harris, T. M. J. Am. Chem. Soc. 1982, 104, 4293.
(81) Williams, D. H.; Williamson, M. P.; Butcher, D. W.; Hammond,
S. J. J. Am. Chem. Soc. 1983, 105, 2.
(82) Barna, J. C.; Williams, D. H. Annu. Rev. Microbiol. 1984, 38,
339.
(83) Reynolds, P. Eur. J. Clin. Microbiol. Infect. Dis. 1989, 943.
(84) Ziegler, D. W.; Wolfe, R. N.; McGuire, J. M. Antibiot. Annu. 1956,
612.
(85) Cooper, G. L.; Given, D. B. Vancomycin, A Comprehensive Review
of 30 Years of Clinical Experience; Park Row Publishers: 1986.
(86) Levy, S. B. Sci. Am. 1998, 278, 46.
(87) Arthur, M.; Courvalin, P. Antimicrob. Agents Chemother. 1993,
37, 1563.
(88) Evers, S.; Quintiliani, R., Jr.; Courvalin, P. Microb. Drug Resist.
1996, 2, 219.
(89) Cetinkaya, Y.; Falk, P.; Mayhall, C. G. Clin. Microbiol. Rev. 2000,
13, 686.
(90) Walsh, C. T.; Fisher, S. L.; Park, I. S.; Prahalad, M.; Wu, Z.
Chem. Biol. 1996, 3, 21.
(91) Wright, G. D.; Holman, T. R.; Walsh, C. T. Biochemistry 1993,
32, 5057.
(92) Holman, T. R.; Wu, Z.; Wanner, B. L.; Walsh, C. T. Biochemistry
1994, 33, 4625.
(93) Bugg, T. D.; Wright, G. D.; Dutka-Malen, S.; Arthur, M.;
Courvalin, P.; Walsh, C. T. Biochemistry 1991, 30, 10408.
(94) McComas, C. C.; Crowley, B. M.; Boger, D. L. J. Am. Chem. Soc.
2003, 125, 9314.
(95) Beauregard, D. A.; Williams, D. H.; Gwynn, M. N.; Knowles, D.
J. Antimicrob. Agents Chemother. 1995, 39, 781.
(96) Westwell, M. S.; Gerhard, U.; Williams, D. H. J. Antibiot. 1995,
48, 1292.
(97) Cooper, M. A.; Williams, D. H. Chem. Biol. 1999, 6, 891.
(98) Westwell, M. S.; Bardsley, B.; Dancer, R. J.; Try, A. C.; Williams,
D. H. Chem. Commun. 1996, 6, 589.
(99) Evers, S.; Courvalin, P. J. Bacteriol. 1996, 178, 1302.
(100) Baptista, M.; Rodrigues, P.; Depardieu, F.; Courvalin, P.; Arthur,
M. Mol. Microbiol. 1999, 32, 17.
(101) Arthur, M.; Depardieu, F.; Courvalin, P. Microbiology 1999, 145
(Pt 8), 1849.
(102) Dong, S. D.; Oberthur, M.; Losey, H. C.; Anderson, J. W.; Eggert,
U. S.; Peczuh, M. W.; Walsh, C. T.; Kahne, D. J. Am. Chem.
Soc. 2002, 124, 9064.
(103) McAtee, J. J.; Castle, S. L.; Jin, Q.; Boger, D. L. Bioorg. Med.
Chem. Lett. 2002, 12, 1319.
(104) McComas, C. C.; Crowley, B. M.; Hwang, I.; Boger, D. L. Bioorg.
Med. Chem. Lett. 2003, 13, 2933.
(105) Kerns, R.; Dong, S. D.; Fukuzawa, S.; Carbeck, J.; Kohler, J.;
Silver, L. L.; Kahne, D. J. Am. Chem. Soc. 2000, 122, 12608.
(106) Nagarajan, R.; Schabel, A. A.; Occolowitz, J. L.; Counter, F. T.;
Ott, J. L.; Felty-Duckworth, A. M. J. Antibiot. 1989, 42, 63.

(107) Nagarajan, R. J. Antibiot. 1993, 46, 118.
(108) Sharman, G. J.; Try, A. C.; Dancer, R. J.; Cho, Y. R.; Staroske,
T.; Bardsley, B.; Maguire, A. J.; Coper, M. A.; OBrien, D. P.;
Williams, D. H. J. Am. Chem. Soc. 1997, 119, 12041.
(109) OBrien, D. P.; Entress, R. M. H.; Cooper, M. A.; OBrien, S. W.;
Hopkinson, A.; Williams, D. H. J. Am. Chem. Soc. 1999, 121,
5259.
(110) Rao, J.; Lahiri, J.; Isaacs, L.; Weis, R. M.; Whitesides, G. M.
Science 1998, 280, 708.
(111) Rao, J.; Yan, L.; Lahiri, J.; Whitesides, G. M.; Weis, R. M.;
Warren, H. S. Chem. Biol. 1999, 6, 353.
(112) Allen, N. E.; LeTourneau, D. L.; Hobbs, J. N., Jr.; Thompson,
R. C. Antimicrob. Agents Chemother. 2002, 46, 2344.
(113) Booth, P. M.; Williams, D. H. J. Chem. Soc., Perkin Trans 1 1989,
2335.
(114) Ahrendt, K. A.; Olsen, J. A.; Wakao, M.; Trias, J.; Ellman, J. A.
Bioorg. Med. Chem. Lett. 2003, 13, 1683.
(115) Jain, R. K.; Trias, J.; Ellman, J. A. J. Am. Chem. Soc. 2003,
125, 8740.
(116) Goldman, R. C.; Baizman, E. R.; Longley, C. B.; Branstrom, A.
A. FEMS Microbiol. Lett. 2000, 183, 209.
(117) Roy, R. S.; Yang, P.; Kodali, S.; Xiong, Y.; Kim, R. M.; Griffin,
P. R.; Onishi, R.; Kohler, J.; Silver, L. L.; Chapman, K. Chem.
Biol. 2001, 8, 1095.
(118) Sun, B.; Chen, Z.; Eggert, U. S.; Shaw, S. J.; LaTour, J. V.;
Kahne, D. J. Am. Chem. Soc. 2001, 123, 12722.
(119) Smith, T. L.; Pearson, M. L.; Wilcox, K. R.; Cruz, C.; Lancaster,
M. V.; Robinson-Dunn, B.; Tenover, F. C.; Zervos, M. J.; Band,
J. D.; White, E.; Jarvis, W. R. N. Engl. J. Med. 1999, 340, 493.
(120) Sieradzki, K.; Tomasz, A. J. Bacteriol. 1999, 181, 7566.
(121) Chang, S.; Sievert, D. M.; Hageman, J. C.; Boulton, M. L.;
Tenover, F. C.; Downes, F. P.; Shah, S.; Rudrik, J. T.; Pupp, G.
R.; Brown, W. J.; Cardo, D.; Fridkin, S. K. N. Engl. J. Med. 2003,
348, 1342.
(122) Weigel, L. M.; Clewell, D. B.; Gill, S. R.; Clark, N. C.; McDougal,
L. K.; Flannagan, S. E.; Kolonay, J. F.; Shetty, J.; Killgore, G.
E.; Tenover, F. C. Science 2003, 302, 1569.
(123) Allen, N. E.; Nicas, T. I. FEMS Microbiol. Rev. 2003, 26, 511.
Kahne et al.
(124) Pfaller, M. A.; Marco, F.; Messer, S. A.; Jones, R. N. Diagn.
(125) Wasilewski, M. 41st ICAAC, late breaking poster, 2001.
(126) Candiani, G.; Abbondi, M.; Borgonovi, M.; Romano, G.; Parenti,
F. J. Antimicrob. Chemother. 1999, 44, 179.
(127) Steiert, S. Cuur. Opin. Investig. Drugs 2002, 3, 229.
(128) Seltzer, E.; Dorr, M. B.; Goldstein, B. P.; Perry, M.; Dowell, J.
A.; Henkel, T. Clin. Infect. Dis. 2003, 37, 1298.
(129) Pace, J. L.; Krause, K.; Johnston, D.; Debabov, D.; Wu, T.;
Farrington, L.; Lane, C.; Higgins, D. L.; Christensen, B.; Judice,
J. K.; Kaniga, K. Antimicrob. Agents Chemother. 2003, 47, 3602.
(130) Losey, H. C.; Jiang, J.; Biggins, J. B.; Oberthur, M.; Ye, X. Y.;
Dong, S. D.; Kahne, D.; Thorson, J. S.; Walsh, C. T. Chem. Biol.
2002, 9, 1305.
(131) Fu, X.; Albermann, C.; Jiang, J.; Liao, J.; Zhang, C.; Thorson, J.
S. Nat. Biotechnol. 2003, 21, 1467.
(132) Losey, H. C.; Jiang, J.; Biggins, J. B.; Oberthur, M.; Ye, X. Y.;
Dong, S. D.; Kahne, D.; Thorson, J. S.; Walsh, C. T. Chem. Biol.
2002, 9, 1305.
(133) Mulichak, A. M.; Losey, H. C.; Walsh, C. T.; Garavito, R. M.
Structure (Camb) 2001, 9, 547.
(134) Mulichak, A. M.; Losey, H. C.; Lu, W.; Wawrzak, Z.; Walsh, C.
T.; Garavito, R. M. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 9238.
(135) Mulichak, A. M.; Lu, W.; Losey, H. C.; Walsh, C. T.; Garavito,
R. M. Biochemistry 2004, 43, 5170.
(136) S. aureus PBP2, the major transglycosylase in Staph, was
recently overexpressed, purified, and characterized. Barrett, D.;
Leimkuhler, C.; Chen, L.; Walker, D.; Kahne, D.; Walker, S. J.
Bacteriol., in press.
(137) An assay to monitor inhibition of the major transglycosylase in
S. aureus has been developed. Various glycopeptides, including
damaged chlorobiphenyl vancomycin and damaged dalbavancin,
were shown to inhibit S. aureus PBP2. Leimkuhler, C.; Chen,
L.; Barrett, D.; Panzone, G.; Sun, B.; Falcone, B.; Oberthur, M.;
Donadio, S.; Walker, S.; Kahne, D. J. Am. Chem. Soc., in press.
CR030103A
Chem. Rev. 2005, 105, 449475
449
Chemistry and Biology of Ramoplanin: A Lipoglycodepsipeptide with Potent

Antibiotic Activity
Suzanne Walker,*, Lan Chen, Yanan Hu, Yosup Rew, Dongwoo Shin, and Dale L. Boger*,
The Department of Microbiology & Molecular Genetics, Harvard Medical School, Boston, Massachusetts 02115, The Department of Chemistry and
The Skaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, California 92037, and The Department of Chemistry,
Princeton University, Princeton, New Jersey 08544
Received July 12, 2004
Contents
1. Introduction
2. Ramoplanin Basics
2.1. Structural Overview
2.2. Antimicrobial Activity
2.3. Clinical Status
3. Mechanism of Action of RamoplaninsEarly Work
3.2. Peptidoglycan Structure and Biosynthesis
3.3. MurG Is Proposed as the Target of
Ramoplanin
3.4. Ramoplanin Is Shown To Bind to an
Intermediate in Peptidoglycan Biosynthesis
3.5. Ramoplanin Is Proposed To Block the
Transglycosylation Step of Peptidoglycan
Biosynthesis
4. Mechanism of Action of RamoplaninsRecent
Work
4.1. Technical Advances in the Study of
Peptidoglycan-Synthesizing Enzymes
4.2. Expected Inhibition Kinetics for Substrate
Binders
4.3. Inhibition Kinetics of Ramoplanin
4.3.1. Transglycosylase Inhibition
4.3.2. MurG Inhibition
4.4. Evaluating the Proposed Cellular Targets of
Ramoplanin
5. Molecular Recognition by Ramoplanin
5.1. Problem of Fibril Formation
5.2. Comparison of Substrate-Binding Affinities
5.3. Structural Studies on Ramoplanin and
Ramoplanin Complexes
6. Total Synthesis of Ramoplanin and Key
Analogues
6.1. Preparation of Key Amino Acids
6.1.1. aThr (allo-Threonines)
6.1.2 HAsn (L-threo--Hydroxyasparagine)
6.2. Total Synthesis of the Ramoplanin A2 and
Ramoplanose Aglycon
6.3. Total Synthesis and Structure of the
Ramoplanin A1 and A3 Aglycons
449
452
452
453
454
454
454
454
455
456
456
457
457
457
458
458
459
459
460
460
461
462
463
463
463
463
464
467
* To whom correspondence should be addressed. E-mail:

suzanne_walker@hms.harvard.edu and boger@scripps.edu.
Harvard Medical School.
Princeton University.
The Scripps Research Institute.
6.4. [N-Acetyl-Asn1]ramoplanin Aglycon

6.5. Total Synthesis of Ramoplanin Amide
Analogues
6.6. Solid-Phase Synthesis of a Simplified
Analogue
7. Degradation and Semisynthetic Studies
7.1. Ramoplanin Aglycons
7.2. Depsipeptide Hydrolysis
7.3. Lipid Side-Chain Reduction
7.4. Orn4 and Orn10 Derivatization
7.5. Lipid Side-Chain Replacement
7.6. Summary
8. StructureActivity Studies on Ramoplanin and Its
Synthetic and Semisynthetic Analogues
8.1. Antimicrobial Activity of Key Derivatives and
Analogues
8.2. Mechanistic Analysis of Ramoplanin
AnaloguessThe Path Forward
9. Conclusion
10. Abbreviations
11. References
467
468
469
470
470
470
470
470
471
471
471
471
472
473
473
474
1. Introduction
Before the introduction of antibiotics in the 1940s
and 1950s, patients with bacteremiasbacteria in the
blood streamshad almost no chance of survival.1,2
Antibiotics (Figure 1) were hailed as miracle drugs
because they rapidly cured infections that would
otherwise have proven fatal. In the belief that
antibiotics, vaccination, and modern sanitation methods had defeated infectious disease, the Surgeon
General declared in 1970 that the United States was
ready to close the book on infectious disease as a
major health threat. Twenty-five years later, hospitalacquired (nosocomial) infections cost several billion
dollars and contribute to 100 000 deaths annually.3
According to one estimate, 20% of patients admitted
to hospitals have or will develop an infection and 70%
of the bacteria that give rise to these infections are
resistant to at least one of the main antimicrobial
agents used to fight infection.4
More than one-third of nosocomial infections are
caused by three Gram-positive pathogenssStaphylococcus aureus, coagulase-negative staphylococci, and
enterococci. Acquired antibiotic resistance in these
organisms is a major concern. Clinical isolates of
10.1021/cr030106n CCC: $53.50 2005 American Chemical Society

Suzanne Walker (above right) received her Ph.D. degree from Princeton
University in 1992. She joined the Department of Chemistry at Princeton
as a faculty member in 1995, where she initiated a program to study
peptidoglycan biosynthesis and its inhibition. In 2004 she moved to the
Department of Microbiology & Molecular Genetics at Harvard Medical
School. She is interested in bacterial cell growth and division as well as
the mechanism of action of unusual antibiotics.
Lan Chen (above left), born in Wuhan, China, obtained her M.S. degree
in Chemistry from Wuhan University in 1993. She received her Ph.D.
degree in 2000 from Rutgers University, where she studied mechanistic
enzymology under the supervision of Professor W. Phillip Huskey. In 2000
she joined the laboratory of Professor Suzanne Walker as a postdoctoral
associate, where she studied the kinetic mechanism of E. coli MurG,
developed methods to study bacterial transglycosylases, and helped
characterize the mechanism of action of ramoplanin. She joined Cubist
(Lexington, MA) as a Research Scientist in 2004.
Yanan Hu was born in Dalian, China, and obtained her M.S. degree in
Chemistry from Tsinghua University in Beijing in 2000. She joined the
laboratory of Professor Suzanne Walker as a graduate research associate
at Princeton University in 2000. She solved the X-ray structure of a cocomplex of E. coli MurG with UDP-GlcNAc bound, developed a highthroughput screen for MurG inhibitors, and helped characterize the
mechanism of action of ramoplanin. After receiving her Ph.D. degree in
2004, she joined Colgate-Palmolive (Piscataway, NJ) as a Research
Scientist.
S. aureus, which infects burns, skin, and surgical

wounds, are typically resistant to a range of antibiotics, including methicillin.5 The glycopeptide antibiotic
vancomycin is used to treat infections caused by drugresistant S. aureus strains and other Gram-positive
pathogens.6 Unfortunately, vancomycin resistance
has become increasingly common in enterococci and
is now beginning to spread to other organisms.7,8 For
example, a methicillin-resistant S. aureus strain
displaying high-level resistance to vancomycin was
isolated recently from a dialysis patient in Michigan.9,10 This isolate had apparently acquired the
Walker et al.
Yosup Rew, born in 1968 in Andong, South Korea, received his B.Sc.
and M.Sc. degrees in Chemistry from the Seoul National University, South
Korea, in 1990 and 1992, respectively. After working as a research scientist
of the new fungicide discovery team in the agrochemical division of LG
Chemical Ltd., Daejeon, South Korea. for 5 years, he continued his studies
in chemistry at the University of California, San Diego. where he received
his Ph.D. degree in 2002 under the supervision of the late Professor
Murray Goodman in the field of peptide chemistry. Following this he was
a postdoctoral fellow at The Scripps Research Institute working with
Professor Boger and involved in the total synthesis of key analogues of
ramoplanin. He joined Amgen (San Francisco) in 2004.
Dongwoo Shin (above left), born in 1971 in Seoul, South Korea, obtained
his B.Sc. degree in Chemistry from Hanyang University, Seoul, South
Korea in 1996. He received his Ph.D. degree (2002) at The University of
Illinois at Chicago, where he worked on the design and synthesis of
protease inhibitors under supervision of Professor Arun K. Ghosh. In 2002
he joined the laboratory of Professor Dale L. Boger as a postdoctoral
associate. His current interests cover the total synthesis and structure
activity studies of bioactive molecules.
Dale L. Boger (above right) received his Ph.D. degree from Harvard
University in 1980. He joined the Department of Medicinal Chemistry of
the University of Kansas, moved to Purdue University in 1986, and joined
the newly founded Department of Chemistry at The Scripps Research
Institute in 1990. His research interests include the total synthesis and
structural exploration of biologically active natural products.
genes that confer vancomycin resistance from a

coexisting E. faecalis strain and was now harboring
these genes on a large multiresistance conjugative
plasmid. Additional genes on the plasmid encoded
resistance to trimethoprim, aminoglycosides, -lactams, and disinfectants.10
As the foregoing discussion emphasizes, there is a
clear need for new antibiotics to treat drug-resistant
bacterial infections. Antibiotics that operate by new
mechanisms or belong to new structural classes are
particularly attractive because they are less likely to
show cross-resistance with existing antibiotics. Un-
Chemistry and Biology of Ramoplanin
Figure 1. Structures of selected natural products.
fortunately, between 1962 and 2000 only one such

antibiotic, linezolid, was introduced clinically, although others have since been introduced. All of the
other agents that entered the market between 1962
and 2000 were analogues of existing drugs.4,11 Collaborations between industrial and academic scientists may be required to meet the challenges involved
in discovering, understanding, and developing new
antimicrobial agents. Here we review an antibiotic

that belongs to a new structural class of compounds
and has not shown any cross-resistance to other
antibacterial agents. This antibiotic, ramoplanin
(Figure 2), is currently in Phase III clinical trials for
the prevention of vancomycin-resistant enterococcal
infections in hospitalized patients, and it possesses
some potentially important advantages over many
Walker et al.
either in vitro or in vivo. However, ramoplanin has

not yet been developed for systemic use, and its
potential applications, albeit important, are currently
limited.14 A better understanding of the mechanism
of action of ramoplanin, combined with the ability to
prepare analogues, could lead to new derivatives with
broader utility.
This review will present our understanding of the
mechanism of action of ramoplanin. An introduction
to the primary structure of ramoplanin, its spectrum
of activity, and current clinical indications and
progress will be followed by an overview of the early
mechanistic investigations of this antibiotic. Technical limitations restricted the studies that were initially conducted on ramoplanin, and its mechanism
of action continues to be refined and revised since
the early investigations. Recent chemical and biochemical advances that have made it possible to
probe ramoplanins mechanism of action in greater
detail will be described, and the current mechanistic
understanding will be presented. This will be followed by a description of the synthetic studies that
have been conducted on ramoplanin. The ability to
synthesize ramoplanin and analogues is critical to
addressing the relationship between structure and
activity, which in turn is central to developing better
antibiotics. The review will conclude with an assessment of where we are with respect to understanding
ramoplanin and where we need to go.
2. Ramoplanin Basics
2.1. Structural Overview
Figure 2. (a) Structures of the ramoplanins. (b) Structures

of ramoplanin A2 and synthetic analogues described in the
text.
other antibiotics.12,13 Chief among them is that

resistance to ramoplanin does not develop readily
The discovery, structure elucidation, and biosynthesis of ramoplanin have been reviewed recently by
McCafferty et al., and only a few salient features will
be repeated here.15 Ramoplanin factors A1 (1), A2 (2),
and A3 (3) were discovered in 1984 from a fermentation broth of Actinoplanes and identified as cell wall
active antibiotics by Biosearch Italia (then Gruppo
LePetit).16,17 Ramoplanins A1-A3 consist of a 49membered macrocyclic depsipeptide containing 17
amino acids joined by a lactone linkage between a
beta hydroxy group on amino acid 2 (HAsn2) and the
carboxyl terminus of amino acid 17 (Chp17). Ramoplanin is produced by nonribosomal peptide synthesis
and, like many such natural products, contains a
mixture of L and D amino acids (nine L, seven D) as
well as several nonproteinogenic side chains, including ornithine (Orn), hydroxyphenylglycine (Hpg), and
chlorohydroxyphenylglycine (Chp) in addition to
-hydroxyasparagine (-OH-Asn, HAsn). Ten of
the 16 residues in the ramoplanin macrocycle are
-branched: L--HAsn2, D-Hpg3, D-allo-Thr5, L-Hpg6,
D-Hpg7, L-allo-Thr8, L-Hpg11, D-allo-Thr12, L-Hpg13,
and L-Chp17. The first amino acid in the depsipeptide
is acylated at the amino terminus with a lipid
substituent. Ramoplanins A1-A3 differ in minor
ways in the length and structure of this appended
lipid substituent; however, all three have R,,,unsaturated chains. The lipid chains of ramoplanins
A1-A3 were originally assigned the (2Z,4Z) stereochemistry around these double bonds,18 but all three
have been reassigned as (2Z,4E) based on work by
Kurz and Guba (A2)19 and Boger and co-workers20

(see section 6.3). Ramoplanin A2 is the compound
now in clinical trials because it can be produced in
much larger quantities than factors A1 and A3;
however, the antimicrobial activities of the lipid
variants are virtually indistinguishable. The ramoplanins also contain a saccharide moiety attached by
an R-glycosidic linkage to the phenol of amino acid
11. In ramoplanins A1-A3 the saccharide moiety is
an R-1,2-dimannosyl group; however, ramoplanin
variants in which the terminal mannose has been
removed or the saccharide consists of a branched
trimannosyl group (ramoplanose, 4) have been reported.21 The activities of the saccharide variants
appear to be similar to those of ramoplanins A1-A3.
nonpolar surfaces such as the plastic plates that are

usually used for MIC measurements. BSA reduces
this nonspecific binding, thereby increasing the
amount of ramoplanin in solution.
Ramoplanin has been compared to vancomycin in
a number of different in vitro studies and consistently
has been found to be a more potent antibiotic on a
molar basis regardless of the method of testing
used.15,30 Ramoplanin is also bactericidal at concentrations close to its MIC, which may provide it some
advantages over vancomycin, which is bacteriostatic
at concentrations near its MIC. Finally, ramoplanin
retains full activity against vancomycin-resistant
enterococcal strains and methicillin-resistant staphylococcal strains,27,28 and significant ramoplanin
resistance has not been observed to date, despite
efforts to elicit it in the laboratory. The mechanism
of action of ramoplanin makes the spontaneous
development of high-level resistance improbable.
However, as with other natural product antibiotics,
resistance in clinically relevant pathogens could
develop to ramoplanin by the horizontal transfer of
genetic information from the producing organism to
other microorganisms. The likelihood of this occurring rapidly depends on usage patterns of ramoplanin
and related antibiotics, the biology of the producing
microorganisms, and the mechanism of resistance in
the producer organisms. It should be noted that the
mechanism by which the ramoplanin producer resists
the effects of ramoplanin has not yet been elucidated.
It should also be noted that only one confirmed
structurally related compound has been reported,
enduracidin (Figure 3), and this compound is not
used clinically or commercially.31 Another potentially
related antibiotic, janiemycin, has also been reported
but not yet structurally characterized.32
2.2. Antimicrobial Activity

Ramoplanin is active against a wide range of
Gram-positive organisms, including many different
species of Staphylococcus, Enterococcus, and Bacillus
(Table 1). Like vancomycin and many other antibiotics, ramoplanin displays no activity against Gramnegative bacteria,22,23 presumably because it cannot penetrate the outer membrane. In general, ramoplanin has MICs (MIC ) minimum inhibitory
concentration, defined as the lowest concentration at
which bacterial growth ceases) below 1 g/mL against
most Gram-positive strains, although the measured
MICs can vary widely based upon the method used
to test growth inhibition.24-28 It has been reported
in two separate studies that the addition of BSA to
the antibiotic stock solution or to plate wells containing bacteria that are used to carry out the measurements results in values up to 30-fold lower (more
active) than without BSA.29 This is consistent with
the observation that ramoplanin tends to stick to
Table 1. Minimum Inhibitory Concentrations (MICs) of Ramoplanin and Its Analogues Against Different
Bacterial Strains (g/mL; compound structures are shown in Figure 2b)
compound
S. aureusa
S. aureusb
2
5b
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
0.5-1.56
0.25-0.78
>128
>128
0.39
6.25-15
>35-50
>100
0.125
S. aureusc
E. faeciume
E. faecalisf
B. subtilisd
0.1
0.1
0.03
0.03
64
0.3
30
25
0.1
15
15
15
20
0.8
50
0.8
50
2
12.5-80
>128
4
8
4
0.125
0.25
0.25
1
0.06
0.06
a
S. aureus ATCC25923, refs 109 and 110. b S. aureus Smith SA819, ref 111. c S. aureus Tour, ref 112.
ref 30. e E. faecium L19624, refs 93 and 113. f E. faecalis Z9212, refs 36 and 113.
0.25
4
16
d
B. subtilis ATCC 8037,
Figure 3. Structures of the enduracidins.
2.3. Clinical Status

Ramoplanin has not yet been developed for parenteral use due, inter alia, to problems associated with
its stability in the blood stream.33 However, it is
currently in phase III clinical trials for the prevention
of systemic nosocomial infections caused by vancomycin-resistant enterococci (VRE). Certain groups of
patients, including those undergoing abdominal surgery and those receiving chemotherapy, are at high
risk of developing systemic VRE infections in the
hospital. It is believed that the GI tracts of many of
these patients are colonized with VRE (possibly
acquired in the hospital from other patients or
contaminated instruments, etc.) and that damage to
the intestinal mucosa caused by surgery or certain
drugs can lead to systemic VRE infections. Therefore,
eradicating VRE from the GI tracts of these patients
prior to treatment may reduce the incidence of
systemic VRE infections. Early clinical results are
promising, and because the incidence of nosocomial
VRE infections is increasing, the FDA has assigned
ramoplanin Fast Track status for this indication.14
In addition to its use for the prevention of blood
stream VRE infections, ramoplanin is also in clinical
trials for the treatment of Clostridium difficileassociated diarrhea (CDAD).3 C. difficile-associated
diarrhea affects over 400 000 patients a year, and
strains resistant to many commonly used therapeutic
agents have appeared. As a result, the FDA recently
designated this application of ramoplanin for Fast
Track status as well.
3. Mechanism of Action of RamoplaninsEarly

Work
The majority of antibiotics act by inhibiting DNA
synthesis, RNA synthesis, protein synthesis, or peptidoglycan synthesis.34 DNA, RNA, and proteins are
found in eukaryotic cells as well as prokaryotes, but
peptidoglycan is unique to bacteria.35 Peptidoglycan
is a polymeric mesh that forms layers around bacterial cell membranes. One of its primary functions is
Walker et al.
to stabilize bacterial membranes so that they do not

rupture as the osmotic pressure within the cells
fluctuates.36 Both the chemical structure of peptidoglycan and the process by which it is made are highly
conserved in bacteria, even across genera that occupy
very different evolutionary niches.35 Microorganisms
produce a range of natural products that kill other
microorganisms by blocking peptidoglycan biosynthesis.34 Because there are no analogous metabolic
pathways in mammalian cells, many of these natural
products have proven to be safe as well as efficacious
antibiotics. Because peptidoglycan biosynthesis is
regarded as an attractive pathway with which to
interfere, many research groups in academia and
industry have established programs to identify inhibitors of this pathway.37,38 Ramoplanin was discovered in one such program. An overview of peptidoglycan structure and biosynthesis is provided in the
following section to provide a context for discussing
the current understanding of the mechanism of
action of ramoplanin.
3.2. Peptidoglycan Structure and Biosynthesis

Peptidoglycan is comprised of linear chains of a
repeating disaccharide held together by cross-links
between peptides appended to one of the sugars of
the disaccharide.39-42 The repeating disaccharide unit
consists of N-acetyl glucosamine attached via a
-glycosidic linkage to the C4 alcohol of an N-acetyl
muramic acid derivative. Muramic acid, a sugar
found only in bacteria, is identical to glucose except
that the C3 hydroxyl is replaced by a lactic acid unit.
The peptide chain that is involved in cross-linking
the glycan strands is attached via its N-terminus to
the lactate moiety. Prior to cross-linking, the peptide
typically consists of a pentapeptide terminating in
D-Ala-D-Ala. There are strain-dependent variations
in the composition of the peptides, but all contain a
nucleophilic group that attacks the D-Ala-D-Ala peptide bond to form cross-links.39-42
Peptidoglycan is synthesized in three stages (Figure 4), the first of which takes place in the cytoplasm
and involves the conversion of UDP-N-acetyl glucosamine to UDP-N-acetyl muramic acid pentapeptide by a series of seven enzymes.43-54 The first two,
MurA and MurB, convert UDP-GlcNAc to UDPMurNAc; the rest are ligases involved in the assembly of the peptide chain. The second stage of
peptidoglycan biosynthesis, which takes place on the
cytoplasmic surface of the bacterial membrane, commences when MraY, an enzyme containing several
membrane-spanning regions, catalyzes a pyrophosphate exchange reaction in which membrane-bound
undecaprenyl phosphate attacks the R-phosphate of
UDP-MurNAc pentapeptide to form a lipid-anchored
MurNAc pentapeptide known as Lipid I with release
of UMP.43,55,56 A membrane-associated glycosyltransferase, MurG, then catalyzes the transfer GlcNAc
from UDP to the C4 hydroxyl of the MurNAc sugar
of Lipid I to form Lipid II.40,57-59 Once Lipid II is
made, the disaccharide is somehow transported from
the cytoplasmic surface of the bacterial membrane
to the external surface where it is polymerized and
cross-linked in the final stage of peptidoglycan bio-
Figure 4. Three stages of peptidoglycan biosynthesis.
synthesis.39,60 The enzymes responsible for polymerizing the GlcNAc-MurNAc disaccharide to form the
glycan chains of peptidoglycan are known as transglycosylases; the enzymes involved in cross-linking
the glycan chains are known as transpeptidases.39
Bacteria contain several transglycosylases and transpeptidases that play different roles in cell growth and
division.61 Most, but not all, of the transglycosylases
are found as domains in bifunctional enzymes that
also contain transpeptidase domains. It is believed
that the bulk of peptidoglycan biosynthesis is carried
out by a subset of the transglycosylases.62,63 In E. coli,
for example, the major transglycosylases are contained in the bifunctional enzymes PBP1a and PBP1b,
with PBP1b believed to be responsible for 85% of
glycan strand synthesis.64,65 Homologous enzymes in
other bacterial strains can be readily identified based
on sequence similarities and are assumed to play
comparable roles.62
Several antibiotics in clinical use block key steps
in peptidoglycan biosynthesis.34 For example, fosfomycin, used to treat urinary track infections, inhibits
MurA, which catalyzes the first committed step in
the biosynthetic pathway.66 Vancomycin and other
glycopeptide antibiotics block glycan polymerization
and cross-linking by binding to the D-Ala-D-Ala
dipeptide terminus of Lipid II and nascent peptidoglycan.6 Penicillin and other -lactams covalently
modify the active site of the enzymes involved in
transpeptidation.67,68
3.3. MurG Is Proposed as the Target of

Ramoplanin
Ramoplanin was discovered by scientists at Biosearch Italia in the course of screening fermentation
products for compounds that killed a methicillinresistant S. aureus strain without affecting a variant
lacking a cell wall and grown under hypotonic conditions.16,17 This screen was designed to identify com-
pounds that block peptidoglycan biosynthesis. Experiments probing inhibition of macromolecular

synthesis showed that ramoplanin blocks the incorporation of radiolabeled precursors into peptidoglycan
without affecting the incorporation of radiolabeled
building blocks into DNA, RNA, or protein, thereby
confirming that ramoplanin inhibits cell wall synthesis. The challenge then became identifying the
step that ramoplanin affects. At that time, identifying
the site of inhibition of a compound that blocks
peptidoglycan synthesis was a daunting challenge.
While the structures of all the intermediates in the
pathway were known, most of the enzymes involved
in the individual chemical transformations had not
been isolated or characterized, and, with few exceptions, direct assays for the enzymes did not exist.
In 1990, Somner and Reynolds reported detailed
investigations of the mechanism of action of ramoplanin.69,70 They first showed that bacteria treated with
ramoplanin accumulate UDP-MurNAc-pentapeptide,
implying that ramoplanin blocks one of the membraneassociated steps that takes place after the formation
of the final cytoplasmic intermediate. They then tried
to determine which membrane-associated step ramoplanin affects. In 1990, two related assays existed
to probe the membrane-associated steps of peptidoglycan synthesis. One assay involved monitoring the
conversion of UDP-MurNAc pentapeptide to nascent
peptidoglycan using washed, particulate bacterial
membranes as a source of the required membranebound enzymes, including MraY, MurG, and the
transglycosylases/transpeptidases.71,72 The second
assay involved monitoring the conversion of UDPMurNAc pentapeptide to mature peptidoglycan in
bacterial cells rendered membrane permeable with
an organic solvent such as toluene.73 Methods to
isolate UDP-MurNAc pentapeptide from bacterial
cells, combined with techniques to separate Lipid I/II
from peptidoglycan and to distinguish cross-linked
Figure 5. Schematic of the membrane assay used to

determine the site of inhibition of ramoplanin.
from un-cross-linked peptidoglycan, had enabled the

development of these assays.71,72 By using isotopically
labeled UDP-MurNAc pentapeptide and/or UDPGlcNAc, Somner and Reynolds were able to follow
the formation of various intermediates along the
pathway and monitor changes in product formation
upon addition of ramoplanin and other antibiotics.
Somner and Reynolds noted that addition of ramoplanin to permeabilized bacterial cells supplemented
with radiolabeled UDP-MurNAc-pentapeptide prevented incorporation of radiolabel into both un-crosslinked and cross-linked peptidoglycan. Un-crosslinked peptidoglycan was formed in the presence of
an antibiotic known to block only the cross-linking
step of peptidoglycan biosynthesis. Therefore, Somner
and Reynolds concluded that ramoplanin acts before
the transpeptidation step. They then showed that
ramoplanin blocks the formation of Lipid II but does
not prevent the formation of Lipid I. Thus, Somner
and Reynolds proposed that ramoplanin acts by
blocking peptidoglycan biosynthesis at the step catalyzed by the glycosyltransferase, MurG (Figure 5).
On the basis of preliminary experiments that were
mentioned but not described in the paper, Somner
and Reynolds also proposed that ramoplanin inhibits
its target by binding to and sequestering Lipid I.
At the time of their experiments Somner and
Reynolds were missing an important piece of information about MurG, namely, that it is located on the
intracellular surface of the cytoplasmic membrane. 74
Therefore, to reach MurG and/or its substrate, ramoplanin would have to pass through the cell membrane. Somner and Reynolds observed in 1990 that
ramoplanin is unlikely to penetrate the cytoplasmic
membrane because of its size and polar nature, but
they did not consider alternative mechanisms of
action because the cellular location of MurG had not
been established. Moreover, the biochemical results
were clear.
The hypothesis that ramoplanin inhibits peptidoglycan biosynthesis at the MurG step became widely
accepted over the next decade. Although Bupp et al.
reported in 1993 that MurG is located on the intracellular surface of the cytoplasmic membrane,74 this
finding did not initially prompt a reexamination of
ramoplanins mechanism. In fact, tools to test the
hypothesis that ramoplanin inhibits MurG by binding
to Lipid I were still not available. Lipid I, the MurG
substrate that ramoplanin was proposed to bind, is
present in miniscule quantities in bacterial cells,
making isolation from natural sources all but impos-
Walker et al.
sible.75,76 Because Lipid I could not be obtained, there

was no direct assay to monitor MurG activity and
no straightforward way to assess whether ramoplanin binds to Lipid I. Although Lipid II is found in
somewhat larger amounts than Lipid I, it also
presented major challenges for isolation. Furthermore, natural Lipid I and Lipid II both contain a 55carbon undecaprenyl chain that renders them insoluble in water and thus difficult to manipulate.
Until these issues of substrate availability and physical properties were addressed, it was not feasible to
carry out detailed investigations of how ramoplanin
inhibits individual membrane-associated enzymes of
peptidoglycan synthesis.
3.4. Ramoplanin Is Shown To Bind to an

Intermediate in Peptidoglycan Biosynthesis
The first direct evidence that ramoplanin is a
substrate binder was reported by Brotz et al. in
1998.77,78 Brotz, Sahl, and co-workers were studying
several lantibiotics, including nisin and epidermin
(Figure 1). Lantibiotics are ribosomally synthesized
peptides that are posttranslationally dehydrated at
serine and threonine side chains.79,80 They then
undergo a series of enzyme-catalyzed cyclizations in
which cysteine residues attack the dehydroamino
acids to form thioether-linked macrocycles. Lantibiotics display a range of mechanisms of action, but
many appear to target Lipid II and other molecules
presented on bacterial cell surfaces (the lantibiotics
are reviewed by van der Donk in this issue). Brotz
et al. showed that ramoplanin as well as the lantibiotics nisin and epidermin comigrate with Lipid II
upon thin-layer chromatography, indicating that all
three molecules interact with this peptidoglycan
intermediate.78 They also reported that nisin and
epidermin block the formation of Lipid II in a
particulate membrane assay similar to that employed
by Somner and Reynolds in their study of ramoplanin. Thus, like ramoplanin, both nisin and epidermin
can inhibit the MurG step of peptidoglycan biosynthesis. Brotz et al. did not conclude, however, that
nisin and epidermin function in cells by inhibiting
MurG because other information pointed to a different mechanism of action for these lantibiotics. By
showing that nisin and epidermin behave similarly
to ramoplanin with respect to MurG inhibition in a
particulate membrane assay and yet are presumed
not to inhibit MurG in cells, Brotz et al. thus provided
the basis for questioning the interpretation that
ramoplanin inhibits the MurG in a cellular context.
3.5. Ramoplanin Is Proposed To Block the

Transglycosylation Step of Peptidoglycan
Biosynthesis
The possibility that ramoplanin functions by inhibiting a different cellular target than MurG was
addressed in 2000 by Lo et al., who noted that the
original mechanistic investigations on the antibiotic
were not conclusive.81 Assays involving the use of a
precursor substrate to monitor flux through several
steps in a pathway can provide inaccurate information about cellular targets if more than one step in
the pathway is inhibited by a particular compound.

Under these circumstances, such vectorial assays
report only the first step to be inhibited, which may
not be the relevant step in a cellular context. In the
case of a proposed substrate binder, there is special
cause for concern because the substrates along a
particular pathway often bear a close resemblance
to one another. Furthermore, in the case of peptidoglycan biosynthesis, it is necessary to carry out these
vectorial assays under conditions in which the polar
precursor substrates can access enzymes that are
normally found on the internal surface of the bacterial membrane. Thus, the integrity of the membrane
must be disrupted by organic solvents, detergents,
or mechanical means for the assays to proceed, and
so a major constraint on the mechanism of actions
the membrane barriersis removed. Lo et al. speculated that in the presence of an intact membrane
barrier, ramoplanin might block the polymerization
of Lipid II, which is catalyzed by bacterial transglycosylases located on the external surface of the
bacterial membrane, instead of the synthesis of Lipid
II, which is catalyzed by an enzyme located on the
internal surface of the bacterial membrane. Inhibition of the transglycosylation step would not have
been observed under assay conditions in which the
formation of Lipid II is blocked.
The hypothesis that ramoplanin can block transglycosylation was tested by Lo et al. using a modified
version of a particulate membrane assay in which
UDP-MurNAc pentapeptide and radiolabeled UDPGlcNAc were added as precursor substrates.81,82 In
this modified assay, Triton X-100 was added to the
bacterial membranes, causing inhibition of the bacterial transglycosylases and allowing radiolabeled Lipid
II to accumulate. The addition of a Triton X-100
scavenger led to reactivation of the transglycosylases
and concomitant formation of peptidoglycan. This
modified assay permits one to add inhibitors after
Lipid II is formed and thus evaluate steps after the
formation of this intermediate.82 Ramoplanin was
found to block transglycosylation in this assay,
implicating a second target for this antibiotic. Thus,
depending on the assay conditions, ramoplanin is
capable of inhibiting either MurG or the bacterial
transglycosylases. Lo et al. suggested that the bacterial transglycosylases were likely to be the primary
targets of ramoplanin because they are extracellular
and would be encountered first by ramoplanin. The
validity of this argument depends on issues such as
whether transglycosylase inhibition is comparable to
or better than MurG inhibition and whether ramoplanin is able to penetrate cell membranes. The
former issue is addressed in the sections below.
4. Mechanism of Action of RamoplaninsRecent

Work
4.1. Technical Advances in the Study of
Peptidoglycan-Synthesizing Enzymes
The development of better methods to obtain
natural Lipid I and II substrates and analogues has
made it possible to assay MurG and the bacterial
transglycosylases directly, enabling in turn detailed
studies of ramoplanin inhibition. For example, in

1997 Auger et al. reported a method to convert UDPMurNAc pentapeptide, which can be isolated in large
quantities from natural sources, to a heptaprenyl
Lipid I analogue, potentially enabling the study of
MurG.83 Shortly thereafter Men et al. reported the
chemical synthesis of a Lipid I analogue (29, Table
2) and its use in a biotin capture assay to monitor
MurG activity.84 Two total syntheses of natural Lipid
I containing the 55-carbon undecaprenyl chain were
reported in 2001.85,86 E. coli MurG was purified and
characterized using synthetic Lipid I analogues,87,88
and several different fluorescence-based and radiometric assays to monitor enzymatic activity using
synthetic substrate analogues have been reported.84,87-94 Moreover, access to purified MurG has
made possible the enzymatic conversion of synthetic
Lipid I to Lipid II. The first synthesis of a Lipid II
analogue containing a 10-carbon lipid chain (30,
Table 2) was reported in 2000.81 This accomplishment
was followed in 2001 by the conversion of synthetic
Lipid I containing the intact 55-carbon chain to Lipid
II using MurG to form the glycosidic linkage.85 This
synthesis of Lipid II was followed by two alternative
routes to the compound in which the glycosidic bond
was formed by chemical methods.95,96 Finally, in 2003
an efficient enzymatic strategy to produce quantities
of Lipid II and analogues from UDP-MurNAc pentapeptide using bacterial membranes was reported.97
Methods to study purified transglycosylases have
followed developments in Lipid II synthesis, enabling
the kinetic analysis of a variety of antibiotics proposed to block transglycosylation.95,98,99
4.2. Expected Inhibition Kinetics for Substrate

Binders
The availability of substrates and assays to monitor
MurG and the bacterial transglycosylases has made
it possible to examine how ramoplanin inhibits these
enzymes. Before detailing the results of recent studies on ramoplanin, it is worth describing the inhibition curves that would be expected for a substrate
binder.100 Figure 6a shows a calculated curve for
inhibition of enzymatic activity by a substrate binder
that forms a 1:1 complex and has a dissociation
constant that is low relative to the Km of the substrate. The reaction rate is negligible at low substrate
concentrations because the substrate is fully bound
and unable to react. When the substrate concentration increases to the level where free substrate
becomes available, the reaction rate jumps rapidly.
At high substrate concentrations inhibition is overcome. Thus, the plot of reaction rate as a function of
substrate concentration has a pronounced sigmoidal
shape. Compounds that bind the substrate less
tightly still yield sigmoidal inhibition curves, but the
behavior is not as pronounced. Figure 6b shows
inhibition curves for three hypothetical substrate
binders having Kds ranging from 0.01 to 1 M. Hence,
the shape of the inhibition curves can provide information both on the mechanism of inhibition and on
the affinity of substrate binding.
Walker et al.
Table 2. Substrate Recognition of Ramoplanina
R1 ) L-Ala--D-Glu-L-Lys-D-Ala-D-Ala. R2 ) L-Ala--D-Glu-L-Dap-D-Ala-D-Ala. R3 ) L-Ala--D-Glu-CONHCH3. R4 ) L-Ala--D-Gln.
4.3. Inhibition Kinetics of Ramoplanin

4.3.1. Transglycosylase Inhibition
Plots of reaction rate versus substrate concentration for inhibition of E. coli PBP1b, a prototypical
bacterial transglycosylase, as well as E. coli MurG
are shown in Figure 7a and b.93,101 The plots for
inhibition of the transglycosylase are sigmoidal, and
inhibition is overcome at high substrate concentrations, consistent with a substrate-binding mechanism
for inhibition of this enzyme. The absence of enzymatic activity in the first part of the curve implies
that ramoplanin binds to Lipid II with a dissociation
constant that is low relative to the substrate concen-
trations used. In addition, the substrate concentration at which the reaction rate increases suggests
that ramoplanin binds to Lipid II with a stoichiometry of 2:1. For example, when the ramoplanin
concentration is 6 M, the reaction rate jumps when
the substrate concentration exceeds 3 M; when the
ramoplanin concentration is 8 M, the reaction rate
jumps when the substrate concentration exceeds 4
M. Thus, the inhibition kinetics reveal that ramoplanin binds with submicromolar affinity and a 2:1
ramoplanin:Lipid II stoichiometry. Assuming that
ramoplanin acts by a pure equilibrium binding mechanism, a dissociation constant of 50 nM can be
calculated from the kinetic data.101
Figure 6. Calculated inhibition curves for hypothetical substrate binders. (a) Curves generated at four different inhibitor
concentrations (0, 5, 10, and 15 mM) for a substrate binder that forms a 1:1 complex and has a Kd of 10 nM. (b) Curves
generated at a single inhibitor concentration for three substrate binders that form 1:1 complexes and have Kds of 1, 0.1,
and 0.01 mM.
Figure 7. Inhibition kinetics of ramoplanin against transglycosylase PBP1b and MurG.
4.3.2. MurG Inhibition

The curves for MurG inhibition are different from
those for transglycosylase inhibition in two respects
(Figure 7).93,101 First, inhibition cannot be overcome
by increasing the substrate concentration. Second,
the shapes of the curves are different. Whereas the
curves for transglycosylase inhibition are strongly
sigmoidal, those for MurG inhibition are not. In fact,
only at the highest ramoplanin concentration is there
any kinetic evidence that substrate binding may
occur. This observation, combined with the fact that
inhibition cannot be overcome, reveals that ramoplanin does not inhibit MurG simply by sequestering
Lipid I. The kinetic data do not, however, rule out
the possibility that ramoplanin forms a complex with
Lipid I and that this complex inhibits MurG in a
noncompetitive manner.
To determine whether Lipid I binding is required
for inhibition of MurG by ramoplanin, alanine was
attached to the Orn4 and Orn10 amines to produce
derivatives 24 and 25 (Figure 16). The Orn10 derivative was found to be incapable of binding to Lipid I,
while the Orn4 derivative retained the ability to bind
Lipid I.93 The compounds were tested for MurG
inhibition, and both had IC50s comparable to the
parent compound, indicating that Lipid I binding is
not required for MurG inhibition. On the basis of

these results, it was proposed that ramoplanin inhibits E. coli MurG not by binding to Lipid I, as
originally suggested by Somner and Reynolds,69 but
by binding directly to the enzyme. This hypothesis
is supported by studies showing that fluorescent
ramoplanin derivatives bind directly to E. coli MurG
at concentrations comparable to the concentrations
required to inhibit enzymatic activity.93
4.4. Evaluating the Proposed Cellular Targets of

Ramoplanin
The aforementioned studies have revealed that
ramoplanin inhibits MurG and the bacterial transglycosylase, PBP1b, by different mechanisms. These
differences in the mode of inhibition are unexpected
and, therefore, interesting; however, the results do
not by themselves provide insight into which target
is preferred in a cellular context. To draw conclusions
as to the cellular target from the biochemical experiments, it is necessary to consider other issues as well.
One issue that has already been raised is that of
accessibility. Lipid II and the bacterial transglycosylases are located on the external surface of the
bacterial membrane, where they are accessible to
ramoplanin. MurG, however, is located on the cyto-
plasmic surface of the bacterial membrane, and

ramoplanin would have to penetrate the membrane
to reach this enzyme. Ramoplanin has a molecular
weight of more than 2500 and an estimated water
solubility of greater than 100 mg/mL. In the absence
of a transport system, a molecule having these
properties is unlikely to penetrate bacterial membranes efficiently. No experiments addressing whether
ramoplanin accumulates inside bacterial cells have
yet been reported, and the existence of a transport
mechanism cannot be discounted.
Another issue that bears on the mechanism of
action is that of the correspondence between the
concentrations of antibiotic needed for substrate
binding or enzyme inhibition relative to those needed
to inhibit bacterial growth. As noted earlier, ramoplanin is a potent antibiotic that inhibits a range of
Gram-positive bacterial strains at submicromolar
concentrations. Ramoplanin binds to and inhibits E.
coli MurG at concentrations at least 10-fold higher
than these MICs. Unless ramoplanin inhibits MurG
homologues from clinically relevant Gram-positive
microorganisms at significantly different concentrations (or somehow becomes concentrated at the site
of action), these low MICs cannot be explained by
inhibition of MurG. In contrast, the kinetics for
transglycosylase inhibition indicate that ramoplanin
binds to Lipid II very tightly, with an estimated
dissociation constant below 10-7 M.101 Moreover, a
qualitative comparison of ramoplanin binding to
Lipid I and Lipid II analogues shows that the
molecule binds to Lipid II better than to Lipid I (see
below). Thus, a mechanism of action involving inhibition of enzymes that utilize Lipid II would be
consistent with the experimental evidence. On the
basis of both target accessibility and the correspondence between the results of biochemical and cellbased assays, a strong argument can now be made
that ramoplanin inhibits bacterial transglycosylases
rather than MurG in bacterial cells.
Recently, another set of experiments addressing
the cellular target of ramoplanin was reported.
Scientists at Genome Therapeutics constructed a S.
aureus strain in which the expression of MurG is
increased. Compounds that inhibit MurG directly
should show an increase in MICs against this strain
relative to typical S. aureus strains. The MIC of
ramoplanin against this strain was found to be
comparable to its MICs against other strains, implying that the MurG enzyme itself is not the cellular
target of the antibiotic.102
Walker et al.
challenging to study because ramoplanin does not

form a discrete complex with Lipid II in vitro. Rather,
it associates to form ordered fibrils that do not yield
readily to structural analysis.81
5.1. Problem of Fibril Formation

The first attempts to characterize the interaction
between ramoplanin and Lipid II were reported by
Lo et al., who prepared a water-soluble analogue of
Lipid II containing a citronellyl chain in order to
probe the interaction of this molecule with ramoplanin.81 Natural Lipid II contains a 55-carbon lipid
chain, which causes it to aggregate extensively in
water, and it was expected that the water-soluble
analogue would facilitate structural analysis of the
complex. To address this possibility, Lo et al. carried
out an NMR titration of ramoplanin with the Lipid
II analogue (30) (Figure 8).81 Unexpectedly, the NMR
resonances of ramoplanin disappeared upon addition
of approximately 0.5 equiv of Lipid II analogue. A
CD titration of ramoplanin with the Lipid II analogue
revealed significant changes in the peptide backbone
region of the antibiotic along with the appearance of
a new band in the far UV region of the spectrum,
implying a nonrandom orientation of the aromatic
side chains in the complex (Figure 9). The presence
of isosbestic points in spectra acquired at different
Lipid II concentrations indicated a two state transi-
Figure 8. NMR titrations of ramoplanin with Lipid I and

Lipid II analogues. Spectra show downfield region of
ramoplanin in D2O upon addition of increasing amounts
of Lipid I/II analogues.
5. Molecular Recognition by Ramoplanin

Brotz et al. provided the first direct evidence for
an interaction between ramoplanin and a substrate
involved in peptidoglycan biosynthesis when they
showed that ramoplanin comigrates with Lipid II.77
Subsequent work from members of the Walker lab
has shown that ramoplanin inhibits the transglycosylation step of peptidoglycan biosynthesis by binding
to Lipid II.81,101 Thus, understanding how ramoplanin
recognizes Lipid II is essential for understanding its
mechanism of action. As described below, the recognition of Lipid II by ramoplanin has proven extremely
Figure 9. CD spectra of ramoplanin upon titration with

Lipid II analogue.
lactate in 33 disfavors binding. In any event, the

peptide chain does not appear to be essential for
binding of lipid intermediates to ramoplanin. There
is evidence, however, that the intact peptide chain
plays a role in the affinity of UDP-MurNAc-pentapeptide for ramoplanin.104
5.2. Comparison of Substrate-Binding Affinities
Figure 10. Transmission electron micrograph of ramoplanin with Lipid II analogue.
tion between free ramoplanin and the complexed

species. These data, combined with the NMR results,
suggested that ramoplanin forms an ordered assembly in the presence of the Lipid II analogue.
Electron-microscopic analysis of the mixture of ramoplanin and the substrate analogue revealed, in
fact, that the ramoplanin complexes assemble to form
fibrils (Figure 10).81,103 The formation of these fibrils
explained the disappearing NMR resonances but
raised questions about whether the ability of ramoplanin to self-assemble in the presence of substrate
is in some way biologically relevant. This is a question that has not yet been answered.
The propensity of ramoplanin complexes to form
fibrils in the presence of Lipid II greatly complicates
structural analysis and makes it difficult to evaluate
differences in the binding affinities of different
substrates. Therefore, the relative contributions of
the different parts of Lipid II to binding are not well
understood. Lo and Cudic et al. evaluated the behavior of ramoplanin in the presence of different
Lipid I analogues or fragments thereof using CD
changes and fibril formation as a measure of binding.
From these studies it is known that the diphosphate
portion of the molecule is essential because the
MurNAc monophosphate derivatives 31 and 39 do
not interact with ramoplanin nor does the lipid
monophosphate derivative 32 (Table 2).30,81,103,104
However, lipid diphosphates such as farnesyl pyrophosphate 34 do interact, promoting the formation
of ramoplanin fibrils.103 It is also known that the
GlcNAc sugar plays a role in binding of Lipid II
because NMR titrations with Lipid I and Lipid II
analogues show that ramoplanin binds Lipid II more
tightly than it binds Lipid I (Figure 8). A comparison
of the inhibition kinetics of transglycosylase and
MurG also suggests that Lipid II is a better ligand
for ramoplanin than Lipid I.93,101 Finally, Cudic et al.
also suggested that the peptide chain on the MurNAc
sugar is essential for recognition based on studies
showing that compound 33 (Table 2) does not interact
with ramoplanin.30 Lo reported, however, that compounds 34, 35, and 36 (Table 2) do interact with
ramoplanin, with the binding of 35 and 36 qualitatively indistinguishable from 29, which contains the
intact peptide chain.103 It is possible that the free
A quantitative binding assay is required to understand the structural features that are important for
substrate recognition by ramoplanin. In an effort to
develop such an assay, Cudic et al. explored the
behavior of a variety of ramoplanin complexes (i.e.,
ramoplanin with UDP-MurNAc pentapeptide and
several Lipid I analogues) under different solution
conditions and found that some complexes remain in
solution even at relatively high concentrations (10-4
M) provided that 20% DMSO is added.104 Therefore,
it is possible to measure dissociation constants for
the compounds that form these soluble complexes
using NMR methods.104 Cudic et al. measured dissociation constants for UDP-MurNAc-pentapeptide
37 and the corresponding dipeptide analogue 38 and
found a 10-fold difference in affinity, indicating that
the terminal L-Dap-D-Ala-D-Ala tripeptide plays a role
in the binding of these substrate analogues to ramoplanin. Whether this is also true for Lipid II is
not clear because there may be significant differences
in how the UDP analogues and Lipid II bind to
ramoplanin (see below).
To characterize the binding interface of the DMSOsolubilized ramoplanin:UDP-MurNAc pentapeptide
complex, Cudic et al. analyzed NMR chemical shift
changes and intermolecular NOEs. The data showed
that ramoplanin binds to UDP-MurNAc-pentapeptide
in a 1:1 ratio and that the majority of the contacts
from ramoplanin to UDP-MurNAc-pentapeptide involve residues 2-8. Cudic et al. suggested that both
ornithine 4 and ornithine 10 make stabilizing electrostatic contacts to the ligand, helping to orient it.
They proposed that a contact from Orn4 to the
carboxylate of the terminal D-Ala-D-Ala dipeptide
plays a role in discriminating the pentapeptide from
the dipeptide. This NMR study was carried out based
on the assumption that structural information on
weakly binding peptidoglycan intermediates, which
are not as susceptible to fibril formation as Lipid II,
would be relevant to understanding how ramoplanin
recognizes Lipid II, which is presumed to be the
biologically relevant substrate. The validity of this
assumption has been called into question by the
kinetic results of Hu et al., showing that ramoplanin
binds to Lipid II with a stoichiometry of 2:1.101 A Job
titration has confirmed ramoplanin binds to Lipid II
in a 2:1 mode rather than the 1:1 mode reported for
the UDP-MurNAc-pentapeptide complex (Figure 11).
Therefore, UDP-MurNAc-pentapeptide may not be a
good model system for Lipid II. There is clearly still
a need for direct structural information on ramoplanin bound to Lipid II or a suitable analogue. It
may be necessary to produce isotopically labeled
ramoplanin and Lipid II analogues in order to obtain
the required structural information.
Walker et al.
Figure 11. (a) Titration of a fluorescently labeled ramoplanin analogue (Orn4-fluorescein) with Lipid II. (b) Job titration
of Orn4-fluorescein ramoplanin with Lipid II.
5.3. Structural Studies on Ramoplanin and

Ramoplanin Complexes
In the absence of direct structural information on
the ramoplanin complex, models for how ramoplanin
interacts with Lipid II must be pieced together based
on the solution structure of free ramoplanin combined
with available data on substrate recognition. The first
solution structure of a ramoplanin factor was determined more than a decade ago by Williams and coworkers, who were studying an analogue of ramoplanin A2 called ramoplanose.21 Ramoplanose is identical
to ramoplanin A2 except that it contains a branched
trisaccharide rather than a disaccharide moiety at
position Hpg11. In water, ramoplanose was found to
have an antiparallel -sheet structure with a reverse
turn at L-Thr8/L-Phe9 and another turn in the vicinity
of Gly14-D-Ala16. The chirality of the amino acids
flanking L-Thr8/L-Phe9 alternates in a D,L,D pattern,
with the result that the R groups on both sides of
this turn are located on the same surface of the
-sheet. Williams and co-workers noted that the
-sheet curves upward slightly in order to alleviate
the steric congestion between side chains on the same
side of the molecule.
The first solution structure for ramoplanin A2 itself
was reported by Kurz and Guba in 1996.19 The
structure of ramoplanin A2 was found to be similar
in many respects to that reported for ramoplanose,
as expected based on the structural similarities
between the molecules. For example, like Williams
and co-workers, Kurz and Guba observed that ramoplanin adopts an antiparallel -stranded structure
with turns in the vicinity of residues 8-9 and 1416.19,21 However, there were also significant differences between the two NMR structures, and the
overall topologies were different. For example, whereas
ramoplanose was reported to have a slight curvature,
ramoplanin A2 was found be distinctly U-shaped,
with amino acid side chains located at the two ends
of the antiparallel sheet structure coming into contact. There were minor differences in the solution
conditions under which the NMR data were acquired.
Ramoplanose was studied in water, and ramoplanin
was studied in a 90:10 water:DMSO mixture. In
addition, the carbohydrate moieties on the two compounds were slightly different. However, the differences in the structures are probably related to the
Figure 12. NMR structure of ramoplanin dimer in methanol.
fact that Kurz and Guba used many more NOE

constraints in their calculations, including constraints between amino acids 9 and 17 at opposite
ends of the molecule. Williams and co-workers noted
that they were unable to achieve convergence due to
limitations in the modeling packages unless the
structural calculations were performed with a reduced set of distance constraints and atoms.21 Subsequent NMR studies on ramoplanin A2 by Lo et al.
and by Cudic et al. support the U-shaped topology
observed by Kurz and Guba.104,105
The third NMR study of ramoplanin in the absence
of substrate was reported by Lo et al., who observed
that ramoplanin A2 exists as a mixture of monomer
and dimer forms in slow exchange in methanol.105
The structure of the dimer was solved in the hope
that it might shed light on how ramoplanin binds to
Lipid II or on how it self-associates or both. The
ramoplanin dimer in methanol was found to consist
of a C2-symmetric antiparallel structure in which the
two monomer units are held together by interactions
between amino acids 10-14 (Figure 12). The structures of the monomer units themselves are very
similar to the monomer structure reported by Kurz
and Guba. The chirality of the residues in the
-strands spanning residues 10-14 alternates in a
D,L,D,L,Gly pattern. Cyclic peptides consisting of
alternating D,L residues are known to self-associate
via a combination of hydrogen bonds and interactions
between side chains aligned on the same side of the
peptide backbone, and the interface between ramoplanin monomers is reminiscent of the interface
between these self-associating peptides.106-108 The

ramoplanin dimer contains a cleft that provides a
possible binding site for Lipid II. An Orn10 residue
from each monomer flanks this cleft. The Orn4
residues are presented on opposite side of the dimer.
Derivatives of ramoplanin containing alanine at
either Orn4 or Orn10 have been prepared and tested
for biological activity and the ability to bind peptidoglycan intermediates.93 The Orn4Ala derivative
retained biological activity and the ability to bind
Lipid I substrate analogues, but the Orn10Ala derivative was inactive and unable to bind Lipid I analogues. These results suggest that Orn10 makes
critical contacts to the substrate but Orn4 does not.
The results are consistent with a model in which
Lipid II binds in a dimer cleft flanked by the two
Orn10 residues (Figure 12). However, confidence in
the model will require additional experiments, particularly since results that might not be consistent
with this model have also been reported (see section
8). Work on various synthetic and semi-synthetic
derivatives of ramoplanin that shed more light on the
Lipid II recognition event and on the structural
requirements for biological activity is discussed in the
following section.
In summary, three different solution structures of
ramoplanin factors have now been reported.19,21,105
Two of the structures are of monomers that differ in
overall topology,19,21 and one is of a dimer comprised
of two molecules that closely resemble one of the
reported monomers.105 The two monomer structures
were determined in largely aqueous media, while the
dimer structure was determined in methanol. Ramoplanin is believed to act at a membrane interface, and
Lo et al. suggested that methanol may mimic the
conditions under which ramoplanin acts better than
water.105 Whether this is so remains to be established. However, it is worth noting that the dimer
structure is consistent with other evidence showing
that ramoplanin complexes Lipid II with a stoichiometry of 2:1. Structural studies of ramoplanin in the
presence of mono- or bilayers could provide more
insight into the bioactive conformation of ramoplanin.
At this point, however, it is evident that ramoplanin
can exist as a monomer, dimer, and, in the presence
of Lipid II, higher order assembly with a 2:1 stoichiometry.101 The changeable nature of ramoplanin is
remarkable.
6. Total Synthesis of Ramoplanin and Key

Analogues
6.1. Preparation of Key Amino Acids
Ramoplanin consists of 17 amino acids of which
seven possess the nonproteinogenic D-configuration
and 12 possess nonstandard side chains. Of these,
the -hydroxy R-amino acids (D/L-allo-threonines and
L-threo--hydroxyasparagine) are the most abundant components of the ramoplanins whose preparations constitute significant elements of any projected
total synthesis. Since -hydroxy R-amino acids are
important components of many complex natural
products, a number of approaches have been reported
for their synthesis. These can be classified as (a)
aldol-type additions of anions or enolates possessing

chiral auxiliaries including bis-lactams,114 oxazolidinones,115,116 oxazinones,117 and imidazolidinones,118
(b) the use of glycine Schiff bases,119 (c) asymmetric
epoxidation,120,121 dihydroxylation,122,123 and aminohydroxylation of olefins,124,125 (d) enantioselective
hydrogenation,126-128 (e) preparation from naturally
occurring amino acids,129-133 and (f) enzymatic
synthesis.134-137 Summarized below are the approaches implemented in efforts leading to the total
synthesis of ramoplanin.
6.1.1. aThr (allo-Threonines)

Despite their biological significance, the availability
of allo-threonines, nonproteinogenic amino acids, is
limited. Consequently, various methods for their
preparation have been developed. Of these, derivatization of a readily available natural amino acid is
the simplest. For efforts directed at ramoplanin, the
approach described by Beaulieu where D- or L-serine
is converted to L- or D-,,-trichloro-allo-threonine,
followed by catalytic hydrogenolysis to provide diasteromerically pure L- or D-allo-threonine, was employed (Scheme 1).132
Scheme 1
6.1.2. HAsn (L-threo--Hydroxyasparagine)

In the ramoplanins the hydroxyl group of L-threo-HAsn2 forms a labile ester bond with the Cterminus of 3-chloro-4-hydroxyphenylglycine (Chp17),
and the preparation of this unnatural amino acid,
L-threo--HAsn, constitutes one of the key elements
in the total synthesis of the ramoplanin aglycons. The
isomers of -HAsn were originally isolated from
human urine138 and had been synthesized as a
racemic mixture that was separated by resolution.139,140 Although there are reports of the enantioselective synthesis of the isomers of L--hydroxyaspartic acid (L--HAsp)141-144 and L-erythro--HAsn,145
an asymmetric synthesis of L-threo--HAsn had not
been disclosed. In efforts leading to the total synthesis of the ramoplanin aglycons, Boger et al.
disclosed the preparation of L-threo--HAsn enlisting
the Sharpless asymmetric aminohydroxylation (AA)
of methyl 4-methoxycinnamate to provide the corresponding amino alcohol in 71% yield and >99% ee
(Scheme 2).109,146 Sequential protection of the alcohol
(TBDMSOTf), one-pot N-Cbz/Boc exchange, and direct aminolysis of the methyl ester afforded the
corresponding amide in 94% overall yield with no
Walker et al.
Scheme 2
evidence of epimerization. The aryl ring was oxidatively cleaved with RuO4 to release the carboxylic
acid, which was subsequently protected to provide the
benzyl ester in 65% and 84% yields, respectively.
Treatment of Boc-L-threo-HAsn(OTBS)-OBn with 4
N HCl-EtOAc led to the single-step removal of the
Boc and TBS protecting groups, and the resulting
amine was Fmoc protected (92%, 2 steps). A final
treatment with trityl alcohol and acetic anhydride
under acidic conditions provided Fmoc-L-threo-HAsn(Trt)-OBn in 71% yield and suitably protected
for direct incorporation into the following synthetic
efforts.
Figure 13. Key disconnections for synthesis of the ramoplanins.
6.2. Total Synthesis of the Ramoplanin A2 and

Ramoplanose Aglycon
The first, and to date only, total synthesis of the
ramoplanin A2 and ramoplanose aglycon was disclosed in 2001.109,147,148 Three key subunits composed
of residues 3-9 (heptapeptide 53), the pentadepsipeptide 64 (residues 1, 2, and 15-17), and
pentapeptide 72 (residues 10-14) were sequentially
coupled and cyclized in a solution-phase, convergent
approach to the 49-membered depsipeptide core of
1-3 (Figure 13).
The indicated coupling sites were chosen to maximize the convergency of the synthesis including that
of the three subunits, to minimize the use of protecting groups, to prevent late-stage opportunities for
racemization of carboxylate-activated phenylglycinederived residues, and to enlist -sheet preorganization of an acyclic macrocyclization substrate for ring
closure. Macrocyclization at the Phe9-D-Orn10 site
proved unusually successful and represents a site
found at the corner of a -turn at the end of the
H-bonded antiparallel -strands of ramoplanin (Figure 14). Consequently, closure at this site may benefit
from both -sheet preorganization of the substrate149
as well as closure at a D-amine terminus.150,151 A
second alternative cyclization site at Gly14-Leu15 site
was also successfully examined and lies within a
small flexible loop at the other end of the H-bonded
antiparallel -strands. In addition to potentially
benefiting from -sheet preorganization of the macrocyclization substrate, it represents closure at a nonhindered site incapable of racemization. Also key to
the implementation of the approach was the judicious
choice of the Orn4/Orn10 SES protection and the Asn1
Figure 14. Macrolactamization sites.
Fmoc protection providing orthogonal protecting

groups stable to Boc, Cbz, and benzyl ester deprotections yet capable of sequential and selective removal
in the presence of the labile depsipeptide ester.
Scheme 3
Scheme 4
Heptapeptide 53, the first of the three key subunits

which contains all but Orn10 of the putative Hpg3Orn10 recognition sequence,30 was assembled as shown
in Scheme 3 from the tripeptide 45 and tetrapeptide
51 followed by Boc deprotection. This coupling cleanly
provided 52 without deliberate protection of the free
alcohols and with no evidence of competitive -elimination or trace racemization. Tripeptide 45 was
prepared by benzyl ester deprotection of 44 obtained
by coupling D-aThr-OBn132,152 with the dipeptide 43.
In turn, 43 was obtained by benzyl ester deprotection
of 42 derived from coupling Boc-D-Hpg-OH and
D-Orn(SES)-OBn.153,154 Tetrapeptide 51 was prepared
from the dipeptides 46 and 48 secured by the
coupling of Boc-L-Hpg-OH with D-Hpg-OBn and BocL-aThr-OH132,152 with L-Phe-OBn, respectively. Notably, benzyl ester hydrogenolysis of 46 followed by
coupling with 49 activated by DEPBT155,156 was
accomplished with no detectable racemization of the
sensitive D-Hpg residue (>99% de) or -elimination
of the hindered L-aThr, whereas alternative coupling
reagents gave lower conversions accompanied by
substantial epimerization.
Synthesis of the second key subunit, depsipentapeptide 64 which contains the labile backbone ester,
is summarized in Scheme 4. Fmoc deprotection of
Fmoc-L-threo--HAsn(Trt)-OBn (54, synthesis described in section 6.1.2) and coupling of 55 with
Fmoc-L-Asn(Trt)-OH provided 56. The key esterification of 56 with 57157 required activation with EDCI
conducted in the presence of DMAP158,159 catalyst. A

wide range of alternative esterification protocols were
examined, but these and EDCI- or DCC-promoted
couplings in the absence of DMAP provided no or
lower conversions and lower diastereomeric ratios or
suffered competitive -elimination. Even the EDCIDMAP, or a less satisfactory DCC-DMAP,158,159
reaction required carefully controlled reaction conditions with higher reaction temperatures (25 vs 0 C),
longer reaction times, or more DMAP (0.5-2 equiv
vs 0.15-0.2 equiv) offering less satisfactory conversions. Boc removal,160 coupling with 61, buffered TBS
deprotection,161,162 and benzyl ester hydrogenolysis
provided 64. Efforts to shorten this sequence by first
coupling 61 with L-Chp(OTBS)-OBn followed by
benzyl ester deprotection and esterification with 57
were not successful, and an unbuffered Bu4NF deprotection of 62 (no HOAc) led to competitive epimerization of the Chp R-aminoester center.
The preparation of the final subunit 72 is detailed
in Scheme 5. D-aThr-OBn132,152 was coupled with BocL-Hpg-OH to give dipeptide 65 devoid of diastereomeric contaminants. Boc deprotection and coupling
of 66 with Boc-D-Orn(SES)-OH provided tripeptide
67. Benzyl ester deprotection and coupling of 68 with
70 gave pentapeptide 71. In turn, the dipeptide 70
was obtained in two steps by coupling Boc-L-Hpg-OH
and Gly-OBn hydrochloride to give dipeptide 69,
followed by Boc deprotection (HCl-EtOAc). Notably,
C-terminus protection of D-aThr as its benzyl ester
in 67 and elsewhere throughout the synthesis permitted hydrogenolysis deprotection, avoiding basecatalyzed -elimination. Similarly, coupling each
sensitive Hpg-OH residue in 72, like that throughout the synthesis, conducted upon activation with
DEPBT156 minimized epimerization especially when
coupled with a hindered and sensitive aThr amine
terminus.
Assembly of three key fragments and macrocyclization at the Phe9-Orn10 site are detailed in
Walker et al.
Scheme 5
Scheme 6
Scheme 6. The coupling of subunits 53 and 64 proved

to be the most challenging step of the synthesis.
Carboxylate activation of 64 typically resulted in
preferential -elimination of the acyloxy substituent.
This can be attributed to the combination of a superb
leaving group, the hindered nature of the activated
carboxylate resulting from its R,,-trisubstitution
and the large protecting groups (trityl and Fmoc), and
the enhanced R-carbon acidity of the activated carboxylate derived from 64 resulting from the use of a
N-acyl versus carbamate derivative. Only DEPBT156
promoted the coupling to provide 74 in superb yields
with no competitive -elimination, whereas all other
alternative coupling reagents and conditions surveyed provided predominantly -elimination products. Boc removal was accomplished under mild
conditions160 that preserved the trityl protecting
groups. Coupling of the crude free amine with 72
provided the key acyclic depsipeptide 75. Successive Boc removal, benzyl ester hydrogenolysis, and
macrocyclization afforded the cyclic depsipeptide core
76. Presumably, -sheet preorganization of the cyclization substrate149 and closure at a D-amine terminus150,151 at the corner of a -turn contributes to
the superb conversions for closure of the 49-membered ring (89% when conducted with purified amino
acid).
The alternative Gly14-Leu15 site was also examined
for the closure of the linear peptide (Scheme 7). This
approach would be beneficial for the preparation of
analogues of the sensitive depsipeptide region of the
natural products enlisting an alternative coupling
order of first assembling Hpg3-Gly14 as a common
precursor, followed by coupling with modified depsipentapeptide subunits, and a final key macrocyclization at the Gly14-Leu15 site. It constitutes closure at
the other end of the H-bonded antiparallel -strands

at a site within a small flexible loop and should
benefit from -sheet preorganization of the cyclization
substrate like closure at the Phe9-Orn10 site. Unlike
the Phe9-Orn10 site it does not benefit from closure
at a D-amine terminus, although the carboxylate
terminus now lacks a L-amino acid side chain that
may decelerate such ring closures. It also precludes
competitive racemization of the activated carboxylate.
As such, and based on these criteria alone, the two
closures might be expected to perform comparably.
The distinction between the two sites rests with Phe9Orn10 lying at the corner of a conformationally
defined -turn adjacent to the H-bonded antiparallel
-strands whereas the Gly14-Leu15 site is embedded
in what appears to be a conformationally more
flexible region of the molecule. Benzyl ester deprotection of 74 followed by coupling with 73 provided
77. Boc deprotection, benzyl ester deprotection, followed by macrocyclization provided 76 in good con-

Scheme 7
Scheme 8
versions (40-50%). This preliminary observation,

which was not subjected to optimization efforts,
suggests that closure at the Gly14-Leu15 site, while
perfectly acceptable, may not be as facile as closure
at the Phe9-Orn10 site.
Fmoc removal of 76 under specially developed
conditions (8 equiv of Bu4NF, 10 equiv of i-PrOH,
DMF, 25 C, 1 h) that do not promote competitive
Asn2 -elimination followed by side-chain acylation
of crude free amine 78 with anhydride 79 provided
80 in excellent overall conversions, Scheme 8. A
single-step HF deprotection of the trityl and SES
groups163,164 or sequential trityl and SES deprotection
provided the ramoplanin A2 and ramoplanose aglycon 5b. Notably, the Fmoc deprotection required the
development of a new method of Fmoc cleavage that
preserves the sensitive depsipeptide ester and the
Orn4 and Orn10 SES deprotections upon treatment
with HF enlisted conditions first introduced and
developed to avoid a competitive depsipeptide cleavage under typical strongly basic conditions.163,164
6.3. Total Synthesis and Structure of the

Ramoplanin A1 and A3 Aglycons
The structure of ramoplanins (1-3) differs only
in the acyl group attached to the Asn1 N-terminus.15,17,18,165 Initially the stereochemistry of the two
double bonds in the three different acyl groups was
assigned as cis-cis.18 Three years after its structural
elucidation, Williams disclosed the structure of ramoplanose (4) whose composition was identical to
ramoplanin A2 with the exception of the branched
mannose trisaccharide attached at Hpg11 and the
stereochemistry of the lipid side chain (cis,trans- vs
cis,cis-7-methyloctadi-2,4-enoic acid).21 Soon thereafter, the stereochemistry of the 7-methyloctadi-2,4enoic acid side chain of ramoplanin A2 was also
revised to cis-trans by Kurz et al. in studies that
also served to provide a solution-phase conformation
of the natural product.19 The synthesis of the ramoplanin A2 aglycon detailed above confirmed this
revised structure of ramoplanin A2.109,147,148 Key to
the strategic planning of the approach was the
introduction of the lipid side chain onto the fully
functionalized cyclic depsipeptide core, thereby potentially providing direct access to all natural aglycons from a common, late-stage intermediate. In
recent studies this has been implemented for the total
synthesis of the ramoplanin A1 and A3 aglycons,
which confirmed an expected analogous structural
revision of the lipid side-chain stereochemistry of
ramoplanins A1 and A3.20
Removal of the Fmoc protecting group from 76
(Bu4NF, i-PrOH) afforded the free amine, the advanced synthetic intermediate prepared en route to
the ramoplanin A2 aglycon (Scheme 9).20,109,148 Subsequent treatment of the resulting free amine with
the anhydrides 81/83 provided the protected aglycons
82/84. Global deprotection using HF furnished the
ramoplanin A1 and A3 aglycons (5a/5c).
For comparison, the authentic ramoplanin A1 and
A3 aglycons were obtained from the natural ramoplanin complex by deglycosidation (HF) followed by
HPLC purification.166 The 1H NMR spectroscopic data
of the synthetic ramoplanin A1 and A3 aglycons (5a/
5c) were in complete agreement with the authentic
compounds. Although the H and H proton signals
of the lipid side chains are not clearly resolved in the
1H NMR spectra due to partial coincidence with other
resonances, JR and J can be measured directly from
the remaining two olefin proton signals, HR and H,
in the spectra. Not only are their values very similar
in ramoplanin A1-A3 (Figure 15), but the coupling
constants of 11.3-11.8 Hz and 14.9-15.4 Hz for JR
and J, respectively, define a cis stereochemistry for
the CR-C double bonds and a revised trans stereochemistry for C-C double bonds.
6.4. [N-Acetyl-Asn1]ramoplanin Aglycon

[N-Acetyl-Asn1]ramoplanin aglycon (13) was prepared and assessed in which the variable Asn1 lipid
N-acyl group was replaced with a minimal N-acetyl
group. Semisynthetic modifications of the natural
products and their aglycons have established that
removal of the side-chain unsaturation (hydro-
Walker et al.
Scheme 9
Figure 15. Diagnostic coupling constants.

Scheme 10
6.5. Total Synthesis of Ramoplanin Amide

Analogues
genation) appears to have a minimal effect,167 but

the impact of the lipid side chain presence and its
potential role was not known. This was addressed
with the total synthesis of 13 as shown in Scheme
10.110 Removal of the Fmoc protecting group from
76109,110,148 (Bu4NF, i-PrOH), the advanced synthetic
intermediate we prepared en route to the ramoplanin
aglycon, followed by treatment with Ac2O provided
the protected N-acetyl derivative 85. Global deprotection using HF furnished 13.
Recently, Boger et al. described the total synthesis

of two key analogues of ramoplanin which entailed
the replacement of HAsn2 with L-2,3-diaminopropionic acid (Dap) and L-2,4-diaminobutyric acid (Dab).110
These deep-seated modifications provide a new generation of stable ramoplanin analogues incorporating
a hydrolytically stable amide bond (vs ester) between
residues 17 and 2.
Not only is the HAsn2 residue the most lengthy to
secure,146 but its incorporation into and maintenance
throughout the late stages of synthesis constitutes
the most difficult challenge limiting the ease of a total
synthesis of ramoplanin.109,147,148 More significantly,
the depsipeptide ester is the most fragile linkage in
the molecule and is central to the instability of the
natural products. Williams described the mild hydrolysis of ramoplanose providing the inactive linear
peptide.149 McCafferty and co-workers reported the
time-dependent depsipeptide cleavage of ramoplanin
A2 aglycon in acidic solutions,104 and Boger and coworkers described depsipeptide cleavage of ramoplanin A2 or its aglycon that occurs rapidly even under
mild basic conditions (1% Et3N-H2O, 25 C, 40%
hydrolysis in 2 min).109,147,148 Since it lies at the end
of the antiparallel -sheet with the ester adopting a
transoid carbonyl-eclipsed conformation analogous to
that of an amide and forms the link to the short
flexible loop (residue 15-17, see Figure 14),19 ester
replacement with an amide with incorporation of
L-Dap or L-Dab to provide 8 and 10 was expected to
have a minimal impact on the ramoplanin conformation but would preclude hydrolysis (8 and 10) and
significantly reduce (8) or preclude (10) -eliminationderived cleavage. Both would simplify the synthetic
challenges associated with assembling a library of
ramoplanin analogues required to probe each residue
and would be expected to improve in vivo stability.
The modified pentapeptide subunits 92a and 92b,
incorporating the Dap2 and Dab2 residues in place

Scheme 11
of HAsn2, were prepared from tripeptide 86 (Scheme

11), which in turn was obtained by coupling Boc-LeuD-Ala-OH109,148 with Chp-OBn109,148 followed by deprotection of the benzyl ester. Fmoc-Dap-OBn (89a) and
Fmoc-Dab-OBn (89b), obtained from Fmoc-Dap(Boc)OH/Fmoc-Dab(Boc)-OH, were coupled with 87 using
DEPBT to give a single diastereomer of each tetrapeptide 90a and 90b. Fmoc removal upon treatment
with the Bu4NF and i-PrOH,109,148 coupling with
Fmoc-Asn(Trt)-OH, and benzyl ester hydrogenolysis
provided 92a and 92b.
The convergent assembly of the modified pentapeptide subunits into the linear precursors for macrocyclization highlights a strategic advantage of the
solution-phase modular synthesis developed for the
ramoplanin synthesis. DEPBT coupling of 92a and
92b with heptapeptide 53109,148 provided 93a/93b. Boc
removal was accomplished under conditions that
preserve the trityl protecting group, and the crude
amines 94a and 94b were coupled with pentapeptide
72109,148 to yield the macrocyclization substrates 95a
and 95b. Successive Boc removal, benzyl ester hydrogenolysis, and macrocyclization afforded the cyclic
peptide cores 96a and 96b. Fmoc removal (Bu4NF
and i-PrOH),109,148 Asn1 acyl side-chain introduction,
and global deprotection of the trityl and SES groups
upon HF treatment provided the key amide analogues 8 and 10.
The amide analogues lacking the Asn1 side chain,
9 and 11, were also prepared from 96a and 96b by
successive Fmoc removal and HF deprotection of the
trityl and SES groups.
6.6. Solid-Phase Synthesis of a Simplified

Analogue
Although a total synthesis of ramoplanin using
solid-phase chemistry has not yet been reported, a
simplified analogue has been assembled. McCafferty
and co-workers30 synthesized a cyclic peptide, derived
from the putative ramoplanin peptidoglycan recognition sequence, NR-octanoyl-Asn1-Asn2-c[Cys3-D-Orn4D-aThr5-Hpg6-D-Hpg7-aThr8-Cys9]-D-Orn10-NH2, by
solid-phase peptide synthesis of the linear peptide,
Scheme 12
Walker et al.
Figure 16. Degradation and derivatization of the ramoplanins.
Asp1 N-acylation, resin cleavage, and oxidative intramolecular cyclization (Scheme 12).
7. Degradation and Semisynthetic Studies

In efforts to examine the role of structural features
and functional groups in ramoplanin, accessible
moieties have been chemically deleted, masked, or
modified in the naturally occurring structures by
several groups. Figure 16 summarizes the semisynthetic analogues examined to date and conditions
used for their synthesis.
7.1. Ramoplanin Aglycons

In initial studies treatment of the ramoplanin
complex with either (1) TMSI or TMSCl/NaI followed
by hydrolysis or (2) HCl in anhydrous BuOH provided
the deglycosylated derivatives of ramoplanin in 2030% yield.167 A recent improved procedure relied on
reaction of the ramoplanin complex with anhydrous
HF.166 Reverse-phase HPLC purification of the crude
mixture which serves to separate the A1-A3 aglycons gave pure A1 (3%, 5a), A2 (46%, 5b), and A3
(3%, 5c).
7.2. Depsipeptide Hydrolysis

The most fragile linkage in the ramoplanins, the
depsipeptide ester formed between HAsn2 and Chp17,
can be hydrolyzed under mildly basic conditions to
afford acyclic ramoplanin derivatives.109,149 Williams
first reported this hydrolysis with ramoplanose (1%
Et3N-H2O, 25 C, 1 h), and more recently the ease
of hydrolysis was examined by Boger and co-workers
in the course of synthetic studies. Treatment of either
ramoplanin A2 or the ramoplanin A2 aglycon under
these conditions resulted in rapid hydrolysis of the
depsipeptide ester to cleanly provide the linear acyclic
derivatives 6 and 7 (>90% isolated yield, 20 min; 80%
conversion, 5 min; 30% conversion, 1 min) with little
or no difference in the rates of reaction. Significantly,
Williams and co-workers observed that the corresponding acyclic ramoplanose partially retained the
secondary -sheet structure of the parent molecule
(1H NMR NOEs). On the basis of this presumed
preorganized -sheet conformation of a linear peptide
derivative, the Phe9-D-Orn10 macrocyclization site
found at the corner of the -turn and the end of the
H-bonded antiparallel -sheet proved unusually successful in the total synthesis of the ramoplanin
aglycons.
7.3. Lipid Side-Chain Reduction

Tetrahydroramoplanin (e.g., tetrahydroramoplanin
A2, 12) and ramoplanin aglycon derivatives167,112 can
be easily accessed by catalytic hydrogenation (5% Pd/
C) of the natural products or their aglycons.
7.4. Orn4 and Orn10 Derivatization

In efforts to probe the biological importance of
the Orn4 and Orn10 -amino groups, several groups
have described methods for their derivatization.
McCafferty and co-workers converted the terminal
amines of both Orn4 and Orn10 to the diguanidylated,
diisovaleryl, and diacetyl derivatives.30 Thus, treatment of ramoplanin A2 with either 1H-pyrazole-1carboxamidine or acetic anhydride in the presence
of pyridine provided the [Orn4, Orn10]-diguanidylated
(26) or -diacetylated (28) products. Reductive amination of ramoplanin A2 with isovaleraldehyde and
NaBH3CN afforded [Orn4, Orn10]-diisovaleryl ramoplanin A2 (27). In elegant studies designed to address
the relative role of each Orn residue, Walker and coworkers disclosed the reaction of ramoplanin A2 with
Boc-Ala-OSu followed by removal of Boc groups with
TFA to provide the [Orn4-Ala] (24) and [Orn10-Ala]
(25) derivatives.93 Both retain the charged (protonated) amine of the parent molecule but displace its
positioning by three atoms. Finally, treatment of the
ramoplanin A2 aglycon with SESCl (Et3N, DMF, -20
C) afforded a mixture of the [Orn4,Orn10]-diSES
derivative 22 and the [Orn10]-monoSES derivative 23

which were separated by reverse-phase HPLC.109
7.5. Lipid Side-Chain Replacement

Complementary to the totally synthetic preparation
of the [N-acetyl-Asn1]ramoplanin aglycon, Ciabatti
recently detailed the preparation of an extensive
series of semisynthetic side-chain-modified ramoplanin analogues.111 The Orn4 and Orn10 -amino groups
of ramoplanin A2 were first masked with Fmoc or
Cbz protecting groups. Ozonolysis of the ramoplanin
derivatives 98 and subsequent reductive amination
of the resulting N-acyl glyoxal with benzylamine
afforded a precursor suitable for Edman degradation
with phenylisothiocyanate in pyridine-H2O to provide the ramoplanin-free amine. The resulting amine
was then acylated with an extensive series of lipid
side-chain replacements and the Orn protecting
groups removed to give over 130 analogues of the
natural products (Scheme 13).
Scheme 13
7.6. Summary
Notably, degradative and semisynthetic modifications of the ramoplanins have been limited to date
and probed only small regions of the natural product.
A renewed effort in such studies could shed insight
into additional roles of undefined peripheral functional groups complementary to more deep-seated
structural modifications and amino acid side-chain

substitutions that are now accessible by total synthesis.
8. StructureActivity Studies on Ramoplanin and

Its Synthetic and Semisynthetic Analogues
8.1. Antimicrobial Activity of Key Derivatives and
Analogues
Ramoplanin A2 is the most abundant component
of the ramoplanin complex. Consequently, synthetic
and semisynthetic analogues of ramoplanin A2 have
been the subject of most investigations seeking to
define the key structural requirements for activity.
The antimicrobial activities of such key derivatives
are summarized in Table 1.
The in vitro antimicrobial activity of ramoplanin
A2 (2) and its aglycon (5b) are not distinguishable109,110,167 with the latter typically displaying slightly
greater in vitro potency.109,110 Consequently, glycosidation does not contribute to the intrinsic in vitro
antimicrobial activity and the role of the natural
product di- or trisaccharides remains to be established. Acycloramoplanin A2 (6) and its aglycon (7),
in which the macrocyclic lactone was hydrolyzed,
have been reported to be inactive (MIC >128 g/mL)
and >250- to 500-fold less potent than the natural
product109 or roughly 2000-fold less active than
ramoplanin A2 when tested against a more sensitive
organism.30 Although it is difficult to establish whether
such residual activity (e0.05%) can be attributed to
contaminant natural product in such samples, it is
clear that the intact macrocyclic ring of the ramoplanin is key to their properties.
Two depsipeptide amide analogues were recently
synthesized, one of which was found to be slightly
more potent than the natural aglycon in antimicrobial assays providing a new lead structure with an
improved profile and a more stable and accessible
macrocyclic template on which to conduct structurefunction studies.110 The depsipeptide amine analogue
8 of the ramoplanin A2 aglycon, in which the backbone ester was replaced with an amide, retained or
exhibited a slightly increased antimicrobial potency
(MIC ) 0.39 g/mL) relative to the ramoplanin A2
aglycon (MIC ) 0.78 g/mL). Not only did this
demonstrate that the depsipeptide ester may be
replaced with a more stable amide without impacting
the in vitro antimicrobial activity, but also that the
HAsn2 carboxamide found in ramoplanin (but is
absent in 8) does not appear to contribute directly to
its properties. Notably, the ramoplanin ester adopts
a typical trans (carbonyl eclipsed) conformation
within the flexible loop capping one end of the
antiparallel -sheet, and it might not be surprising
that its replacement with an amide was well tolerated. Just as importantly, 8 proved to be completely
stable to mildly basic conditions (1% Et3N-H2O, 25
C, 24 h, 0% disappearance)110 that rapidly consume
the ramoplanin aglycon (80% within 5 min at 25
C).109 In sharp contrast, the depsipeptide amide
analogue 10 containing the single additional methylene in the macrocycle was inactive (MIC g 35-50
g/mL), exhibiting a >100-fold loss in activity relative
to the ramoplanin A2 aglycon and 8.110 Thus, even a

simple methylene insertion with incorporation of the
Dab2 residue into the cyclic peptide abolished activity
completely, suggesting not only that the cyclic structure is critical to ramoplanins properties, but that
subtle perturbations to even the flexible loop can
have a pronounced impact. The impact of this modification was subsequently explored in greater detail
as described in the next section.
Recent studies have also defined the impact of the
lipid side chain. The tetrahydroramoplanin A1-A3,
which are under investigation for use as topical
antibiotics for treatment of wound infections, exhibited slightly reduced activities (e.g., 12, Table 1).112
The active ramoplanin amide analogue (9), lacking
the Asn1 lipid side chain, and [N-acetyl-Asn1]ramoplanin aglycon (13) both were approximately 16-110
or 100-fold113 less potent than the corresponding
compounds containing the natural A2 side chain.
These results indicate that the lipid side chain of
ramoplanin A2 contributes significantly to its properties but is not essential.110 These observations are in
accord with those of Ciabatti and co-workers, who
prepared an extensive series of over 130 semisynthetic ramoplanin analogues with modifications to
the lipid side chain. When the length and composition
of the aliphatic side chain is similar to that of the
naturally occurring lipid side chains, the antimicrobial activities of the corresponding analogues are
retained, whereas the activities are significantly
reduced with longer and shorter lengths (e.g., 1419 in Table 1 and Scheme 13). Interestingly, some
analogues which possess aromatic side chains including naphthyl or biphenyl were found to possess the
same or more potent antimicrobial activity than
ramoplanin (e.g., 20 and 21).111 Further insight into
the role of the lipid side chain was obtained from
studies detailed in the following section.
The importance of the free amines of the two
ornithine residues (Orn4 and Orn10) has been studied
extensively. McCafferty and co-workers found that
[Orn4,Orn10-diacetyl]ramoplanin A2 (28), in which
both amines are acetylated, was 500-fold less potent
than ramoplanin A2, highlighting the importance of
these two basic amines which are conserved among
all members of this class of natural products. They
also found that preservation of cationic charge by
guanidylation (26) or transformation to a secondary
amine (27) by reductive amination resulted in only
small to modest alterations in the antimicrobial
activity.30 Similarly, the [Orn4,Orn10-diSES]ramoplanin A2 aglycon (22), in which both amines were
converted to sulfonamide derivatives, was found to
be inactive (MIC > 128 g/mL, >500-fold loss in
activity).109 Much more interestingly, [Orn10-SES]ramoplanin A2 aglycon (23), in which only the Orn10
-amino group is protected, was 16-fold less potent
than the free aglycon but >32-fold more potent than
the diSES derivative.109 Clearly, both the Orn4 and
Orn10 amines contribute to the antimicrobial activity,
and the latter result suggested that the Orn4 free
amine might be more important than the Orn10
amine. However, more recent work of Walkers,93
enlisting alanine derivatives of the Orn4 and Orn10
Walker et al.
-amines, which maintain but move or extend the

position of the free amines, found that [Orn4-Ala]ramoplanin A2 (24) was an active antimicrobial agent
(MIC ) 0.8 g/mL) whereas [Orn10-Ala]ramoplanin
A2 (25) was nearly inactive (MIC ) 50 g/mL),
implicating Orn10 versus Orn4 as the more important
of the two residues. Notably, the positively charged
amine of Orn4 flanks the one side of the ramoplanin
solution structure, whereas that of Orn10 flanks the
other, and they lie at opposite ends of the conserved
putative peptidoglycan-binding domain Hpg3-Hpg.10
Further studies on these and additional synthetic and
semisynthetic derivatives should serve to clarify the
role of Orn4 and Orn10 amines and the putative Hpg3Orn10 recognition domain and offer insight into the
mechanism of action of ramoplanin.
8.2. Mechanistic Analysis of Ramoplanin

AnaloguessThe Path Forward
Most recently, several of the key ramoplanin
analogues have been examined for transglycosylation
inhibition enlisting the assay introduced by Walker
and co-workers113 to quantitate the direct conversion
of Lipid II to peptidoglycan. Combined with the
antimicrobial activity of the analogues, the results
provide insight into the roles of the key structural
features required for Lipid II binding, inhibition of
the transglycosylation reaction, and biological activity. Just as importantly, the work defines a protocol
of assays from which key insights can be garnered
about the functional role of ramoplanin structural
features that can be used to design new, related, and
improved analogues or derivatives. Thus, the active
amide analogue 8 of the ramoplanin A2 aglycon, its
free amine derivative 9 lacking the lipid side chain,
[N-acetyl-Asn1]ramoplanin aglycon (13), and the
inactive ring-expanded amide analogue 10 were
examined in the kinetic assay for inhibition of the
transglycosylase PBP1b. Notably, the kinetics of
inhibition in this assay is a sensitive measure of
substrate binding and provides insight into both the
affinity and stoichiometry of Lipid II binding. As
shown in Figure 17 each analogue bound to Lipid II
with comparable affinity and the same 2:1 stoichiometry as ramoplanin itself. Thus, replacing the Asn2
ester with an amide does not affect Lipid II binding,
even when an additional methylene unit is inserted.
Furthermore, truncating or removing the lipid side
chain does not affect the inhibition kinetics, indicating that it does not play a direct role in substrate
binding or transglycosylase inhibition.
Nonetheless, and with the exception of 8, which
was equipotent with 2, each of these analogues was
at least 2 orders of magnitude less potent than
ramoplanin A2 in antimicrobial assays. Thus, although the lipid side chain does not play a role in
substrate binding or in vitro inhibition of the transglycosylase reaction, it plays a key role in the
biological activity. Most likely this arises by membrane targeting and localization by the lipid side
chain, positioning the antibiotic near its target, Lipid
II, which is located on the outer surface of the
bacterial membrane.
Similarly, 10, which contains an extra methylene
in the macrocyclic ring, is also 100-fold less potent
Figure 17. Inhibition of E. coli PBP1b by ramoplanin and

its key analogues: 2 (O), 8 (0), 10 (4), and 13 (2) at 6 M
concentration and no inhibitor (b).
than 2 or 8. NMR analysis revealed that 10, unlike

ramoplanin and 8, aggregates extensively in aqueous
buffer. Williams and co-workers previously reported
that the ring-opened form of ramoplanin maintains
the same general conformation as the cyclized parent
compound but shows a pronounced tendency to
aggregate, presumably because the increased flexibility permits the -strands in the molecule to
associate in an intermolecular fashion.149 The extra
methylene in 10 may increase macrocycle flexibility,
promoting self-association and perhaps enabling
undesirable interactions with other molecules as well.
Unfavorable partitioning of the more flexible ramoplanin analogue 10 in cell-based assays would explain its higher MICs.
Thus, substituting an amide linkage lacking the
HAsn2 carboxamide side chain for the more labile and
more complex ester linkage does not affect substrate
binding, in vitro activity, or in vivo antimicrobial
activity provided that the ring size is maintained and
the lipid chain is not removed. The amide-linked
macrocycle is considerably more stable than the ester
and may have significant advantages as a therapeutic agent. Increasing the ring size does not appear to
affect substrate binding or inhibition of the transglycosylation reaction but greatly increases the tendency of the molecule to associate, which likely leads
to the decrease in biological activity. Any modifications that increase the flexibility of the molecule may,
therefore, have a generally deleterious effect on
biological activity. Most importantly, the lipid chain
plays a key role in biological activity without directly
influencing binding to Lipid II or in vitro inhibition
of transglycosylation. It is proposed that the lipid
helps target ramoplanin to bacterial membranes. If
so, substitution of the lipid chain with other groups
that also facilitate localization may lead to analogues
with improved activity.
9. Conclusion
Ramoplanin is an antibiotic with a novel structure
and distinct mechanism of action. As such, it repre-
sents an excellent target for clinical development to

treat bacteria resistant to the existing panel of
clinically used antibiotics. Ramoplanin is currently
in Phase III trials for the treatment of VRE; however,
further development of this promising antibiotic
requires a deeper understanding of its mechanism
and the development of structural analogues that
may address its present limitations (e.g., stability).
Though it was originally believed to inhibit MurG, a
series of experiments has shown that the transglycosylation step of peptidoglycan biosynthesis is the
likely target for this molecule. The results of the
mechanistic work also show that ramoplanain may
bind to its target, Lipid II, with a 2:1 stoichiometry.
The reevaluation of the enzymatic target and mechanism of action for ramoplanin has important implications for the design and synthesis of analogues with
potentially improved potency, stability, and solubility.
Concurrent with the mechanistic work, a series of
elegant synthetic studies resulted in the first total
synthesis of the ramoplanin aglycon. The total synthesis enabled the construction of ramoplanin analogues that probed the contribution of specific aspects
of the ramoplanin structure to biological activity. In
particular, the role of the lipid tail, the size of the
macrocycle, and the contribution of the ornithine
residues to activity have been examined to date. Most
significantly, a synthetic stable amide analogue of the
ramoplanin aglycon has been identified (8, [L-Dap2]ramoplanin A2 aglycon) that can now serve as an
accessible macrocyclic template upon which systematic modifications to the structure can be conducted.
Using this stable and accessible template, the key
structural elements of the putative recognition domain and the dimerization domain can be probed
with subsequent synthetic analogues to provide
insight into the mechanism of action and potentially
provide compounds that address issues limiting the
clinical use of ramoplanin itself (e.g., stability). Thus,
development of synthetic methodology and a synthesis of ramoplanin analogues have provided a means
to further probe the mechanism of action.
The combination of synthetic and mechanistic work
has furthered our understanding of ramoplanin.
However, a number of questions remain about the
drug, in particular, the role of ligand-induced polymerization in antibiotic activity and the structure of
the ramoplanin:Lipid II complex. The ability to
synthesize ramoplanin analogues now makes it possible to prepare specific derivatives for such structural and mechanistic analysis.
10. Abbreviations
BCB
Chp
Dab
Dap
DCC
DEPBT
DMAP
EDCI
HOAt
B-bromocatecholborane
L-chlorohydroxyphenylglycine
L-2,3-diaminobutyric acid
L-2,3-diaminopropionic acid
1,3-dicyclohexylcarbodiimide
3-(diethoxyphosphoryloxy)-1,2,3-benzotriazin4(3H)-one
4-(dimethylamino)pyridine
1-(3-dimethylaminopropyl)-3-ethylcarbodiimide
hydrochloride
1-hydroxy-7-azabenzotriazole

HOBt
Hpg
SES
1-hydroxybenzotriazole
L-hydroxyphenylglycine
2-trimethylsilylethanesulfonyl
11. References
(1) Austrian, R.; Gold, J. Ann. Intern. Med. 1964, 60, 759.
(2) Dineen, P.; Homan, W. P.; Grafe, W. R. Ann. Surg. 1976, 184,
717.
(3) U.S. Congress, Office of Technology Assessment; Impacts of
Antibiotic-Resistant Bacteria; OTA-H-629; U.S. Government
Printing Office: Washington, DC, Sept 1995.
(4) Coates, A.; Hu, Y.; Bax, R.; Page, C. Nat. Rev. Drug Discov. 2002,
1, 895.
(5) Sievert, D. M.; Boulton, M. L.; Stoltman, G.; Johnson, D.;
Stobierski, M. G.; Downes, F. P.; Somsel, P. A.; Rudrik, J. T.;
Brown, W.; Hafeez, W.; Lundstrom, T.; Flanagan, E.; Johnson,
R.; Mitchell, J.; Chang, S. CDC Morbid. Mortal. Weekly Rep.
2002, 51, 565.
(6) Williams, D. H.; Bardsley, B. Angew. Chem., Int. Ed. Engl. 1999,
38, 1172.
(7) Derlot, E.; Courvalin, P. Am. J. Med. 1991, 91, S82.
(8) Walsh, F. M.; Amyes, S. G. Curr. Opin. Microbiol. 2004, 7, 439.
(9) Smith, T. L.; Pearson, M. L.; Wilcox, K. R.; Cruz, C.; Lancaster,
M. V.; Robinson-Dunn, B.; Tenover, F. C.; Zervos, M. J.; Band,
J. D.; White, E.; Jarvis, W. R. N. Engl. J. Med. 1999, 340, 493.
(11) Walsh, C. Nat. Rev. Microbiol. 2003, 1, 65.
(12) Woodford, N. Expert Opin. Inv. Drug 2003, 12, 117.
(13) Patel, R. J. Antimicrob. Chemother. 2003, 51, 13.
(14) Montecalvo, M. A. J. Antimicrob. Chemother. 2003, 51, 31.
(15) McCafferty, D. G.; Cudic, P.; Frankel, B. A.; Barkallah, S.;
Kruger, R. G.; Li, W. Biopolymers 2002, 66, 261.
(16) Cavalleri, B.; Pagani, H.; Volpe, G.; Selva, E.; Parenti, F. J.
Antibiot. 1984, 37, 309.
(17) Parenti, F.; Ciabatti, R.; Cavalleri, B.; Kettenring, J. Drugs Exp.
Clin. Res. 1990, 16, 451.
(18) Ciabatti, R.; Kettenring, J. K.; Winters, G.; Tuan, G.; Zerilli, L.;
Cavalleri, B. J. Antibiot. 1989, 42, 254.
(19) Kurz, M.; Guba, W. Biochemistry 1996, 35, 12570.
(20) Shin, D.; Rew, Y.; Boger, D. L. Proc. Natl. Acad. Sci. U.S.A. 2004,
101, 11977.
(21) Skelton, N. J.; Harding, M. M.; Mortishire-Smith, R. J.; Rahman,
S. K.; Williams, D. H.; Rance, M. J.; Ruddock, J. C. J. Am. Chem.
Soc. 1991, 113, 7522.
(22) Jones, R. N.; Barry, A. L. Diagn. Microbiol. Infect. Dis. 1989,
12, 279.
(23) Maple, P. A. C.; Hamiltonmiller, J. M. T.; Brumfitt, W. J.
Antimicrob. Chemother. 1989, 23, 517.
(24) Francis, J.; Webster, H.; Newsom, S. W. B. Drugs Exp. Clin. Res.
1990, 16, 457.
(25) Ohare, M. D.; Felmingham, D.; Gruneberg, R. N. Drug. Exp.
Clin. Res. 1988, 14, 617.
(26) Ohare, M. D.; Ghosh, G.; Felmingham, D.; Gruneberg, R. N. J.
(27) Ristow, T. A.; Noskin, G. A.; Warren, J. R.; Peterson, L. R.
Microb. Drug Resist. 1995, 1, 335.
(28) Mobarakai, N.; Quale, J. M.; Landman, D. Antimicrob. Agents
Chemother. 1994, 38, 385.
(29) Scotti, R.; Dulworth, J. K.; Kenny, M. T.; Goldstein, B. P. Diagn.
(30) Cudic, P.; Behenna, D. C.; Kranz, J. K.; Kruger, R. G.; Wand,
A. J.; Veklich, Y. I.; Weisel, J. W.; McCafferty, D. G. Chem. Biol.
2002, 9, 897.
(31) Higashide, E.; Hatano, K.; Shibata, M.; Nakazawa, K. J. Antibiot.
1968, 21, 126.
(32) Meyers, E.; Weisenborn, F. L.; Pansy, F. E.; Slusarchyk, D. S.;
Von Saltza, M. H.; Rathnum, M. L.; Parker, W. L. J. Antibiot.
1970, 23, 502.
(33) Espersen, F. Curr. Opin. Anti-Infect. Invest. Drugs 1999, 1, 78.
(34) Walsh, C. T. Antibiotics: Actions, Origins, Resistance; ASM
Press: Washington, D.C., 2003.
(35) Gale, E. F.; Cundliffe, E.; Reynolds, P. E.; Richmond, M. H.;
Waring, M. J. The Molecular Basis of Antibiotic Action; WileyInterscience: New York, 1981.
(36) Rogers, H. J.; Perkins, H. R.; Ward, J. B. Microbial Cell Walls
and Membranes; Chapman & Hall: London, 1980.
(37) Wong, K. K.; Pompliano, D. L. Adv. Exp. Med. Biol. 1998, 456,
197.
(38) Silver, L. L. Curr. Opin. Microbiol. 2003, 6, 431.
(39) van Heijenoort, J. Glycobiology 2001, 11, 25R.
(40) van Heijenoort, J. In Escherichia coli and Salmonella Cellular
and Molecular Biology; Neidhardt, F. C., Ed.; ASM Press:
Washington, D.C., 1996.
Walker et al.
(41) Park, J. T. In Escherichia coli and Salmonella Cellular Biology;
Neidhardt, F. C., Ed.; ASM Press: Washington, D.C., 1996.
(42) Ghuysen, J.-M. New Comprehensive Biochemistry, Bacterial Cell
Wall; Elsevier: Amsterdam, 1994.
(43) Bugg, T. D. H.; Walsh, C. T. Nat. Prod. Rep. 1992, 199.
(44) Mengin-Lecreulx, D.; Flouret, B.; van Heijenoort, J. J. Bacteriol.
1982, 151, 1109.
(45) Mengin-Lecreulx, D.; Parquet, C.; Desviat, L. R.; Pla, J.; Flouret,
B.; Ayala, J. A.; van Heijenoort, J. J. Bacteriol. 1989, 171, 6126.
(46) Marquardt, J. L.; Siegele, D. A.; Kolter, R.; Walsh, C. T. J.
Bacteriol. 1992, 174, 5748.
(47) Benson, T. E.; Marquardt, J. L.; Marquardt, A. C.; Etzkorn, F.
A.; Walsh, C. T. Biochemistry 1993, 32, 2024.
(48) Betina, V. The Chemistry and Biology of Antibiotics; Elsevier
Scientific Publishing Co.: New York, 1983.
(49) Benson, T. E.; Harris, M. S.; Choi, G. H.; Cialdella, J. I.; Herberg,
J. T.; Martin, J. P., Jr.; Baldwin, E. T. Biochemistry 2001, 40,
2340.
(50) Bertrand, J. A.; Auger, G.; Fanchon, E.; Martin, L.; Blanot, D.;
van Heijenoort, J.; Dideberg, O. EMBO J 1997, 16, 3416.
(51) Gordon, E.; Flouret, B.; Chantalat, L.; van Heijenoort, J.;
Mengin-Lecreulx, D.; Dideberg, O. J. Biol. Chem. 2001, 276,
10999.
(52) Mol, C. D.; Brooun, A.; Dougan, D. R.; Hilgers, M. T.; Tari, L.
W.; Wijnands, R. A.; Knuth, M. W.; McRee, D. E.; Swanson, R.
V. J. Bacteriol. 2003, 185, 4152.
(53) Schonbrunn, E.; Sack, S.; Eschenburg, S.; Perrakis, A.; Krekel,
F.; Amrhein, N.; Mandelkow, E. Structure 1996, 4, 1065.
(54) Yan, Y.; Munshi, S.; Leiting, B.; Anderson, M. S.; Chrzas, J.;
Chen, Z. J. Mol. Biol. 2000, 304, 435.
(55) Ikeda, M.; Wachi, M.; Jung, H. K.; Ishino, F.; Matsuhashi, M.
J. Bacteriol. 1991, 173, 1021.
(56) Bouhss, A.; Crouvoisier, M.; Blanot, D.; Mengin-Lecreulx, D. J.
Biol. Chem. 2004, 279, 29974.
(57) Taku, A.; Fan, D. P. J. Biol. Chem. 1976, 251, 6154.
(58) Ha, S.; Walker, D.; Shi, Y.; Walker, S. Protein Sci. 2000, 9, 1045.
(59) Hu, Y.; Chen, L.; Ha, S.; Gross, B.; Falcone, B.; Walker, D.;
Mokhtarzadeh, M.; Walker, S. Proc. Natl. Acad. Sci. U.S.A. 2003,
100, 845.
(60) Holtje, J. V. Microbiol. Mol. Biol. Rev. 1998, 62, 181.
(62) Goffin, C.; Ghuysen, J. M. Microbiol. Mol. Biol. Rev. 1998, 62,
1079.
(63) Young, K. D. Biochimie 2001, 83, 99.
(64) Suzuki, H.; Nishimura, Y.; Hirota, Y. Proc. Natl. Acad. Sci.
U.S.A. 1978, 75, 664.
(65) Tamaki, S.; Nakajima, S.; Matsuhashi, M. Proc. Natl. Acad. Sci.
U.S.A. 1977, 74, 5472.
(66) Kahan, F. M.; Kahan, J. S.; Cassidy, P. J.; Kropp, H. Ann. N. Y.
Acad. Sci. 1974, 235, 364.
(67) Ghuysen, J.-M. Annu. Rev. Microbiol. 1991, 45, 37.
(68) Ghuysen, J.-M.; Dive, G. In Bacterial Cell Wall; Ghuysen, J.M., Hakenbeck, R., Eds.; Elsevier: Amsterdam, 1994.
(69) Somner, E. A.; Reynolds, P. E. Antimicrob. Agents Chemother.
1990, 34, 413.
(70) Reynolds, P. E.; Somner, E. A. Drugs Exp. Clin. Res. 1990, 16,
385.
(71) Anderson, J. S.; Matsuhashi, M.; Haskin, M. A.; Strominger, J.
(72) Anderson, J. S.; Strominger, J. L. Biochem. Biophys. Res.
Commun. 1965, 21, 516.
(73) Schrader, W. P.; Fan, D. P. J. Biol. Chem. 1974, 249, 4815.
(74) Bupp, K.; van Heijenoort, J. J. Bacteriol. 1993, 175, 1841.
(75) Pless, D. D.; Neuhaus, F. C. J. Biol. Chem. 1973, 248, 1568.
(76) van Heijenoort, Y.; Gomez, M.; Derrien, M.; Ayala, J.; van
Heijenoort, J. J. Bacteriol. 1992, 174, 3549.
(77) Brotz, H.; Bierbaum, G.; Leopold, K.; Reynolds, P. E.; Sahl, H.
G. Antimicrob. Agents Chemother. 1998, 42, 154.
(78) Brotz, H.; Josten, M.; Wiedemann, I.; Schneider, U.; Gotz, F.;
Bierbaum, G.; Sahl, H. G. Mol. Microbiol. 1998, 30, 317.
(79) Hechard, Y.; Sahl, H. G. Biochimie 2002, 84, 545.
(80) Hoffmann, A.; Pag, U.; Wiedemann, I.; Sahl, H. G. Farmaco
2002, 57, 685.
(81) Lo, M.-C.; Men, H.; Branstrom, A.; Helm, J.; Yao, N.; Goldman,
R.; Walker, S. J. Am. Chem. Soc. 2000, 122, 3540.
(82) Branstrom, A. A.; Midha, S.; Goldman, R. C. FEMS Microbiol.
Lett. 2000, 191, 187.
(83) Auger, G.; Crouvoisier, M.; Caroff, M.; van Heijenoort, J.; Blanot,
D. Lett. Pept. Sci. 1997, 4, 371.
(84) Men, H. B.; Park, P.; Ge, M.; Walker, S. J. Am. Chem. Soc. 1998,
120, 2484.
(85) Ye, X.-Y.; Lo, M.-C.; Brunner, L.; Walker, D.; Kahne, D.; Walker,
S. J. Am. Chem. Soc. 2001, 123, 3155.
(86) VanNieuwenhze, M. S.; Mauldin, S. C.; Zia-Ebrahimi, M.; Aikins,
J. A.; Blaszczak, L. C. J. Am. Chem. Soc. 2001, 123, 6983.
(87) Ha, S.; Chang, E.; Lo, M. C.; Men, H.; Park, P.; Ge, M.; Walker,
S. J. Am. Chem. Soc. 1999, 121, 8415.
(88) Chen, L.; Men, H.; Ha, S.; Ye, X. Y.; Brunner, L.; Hu, Y.; Walker,
S. Biochemistry 2002, 41, 6824.

(89) Cudic, P.; Behenna, D. C.; Yu, M. K.; Kruger, R. G.; Szewczuk,
L. M.; McCafferty, D. G. Bioorg. Med. Chem. Lett. 2001, 11, 3107.
(90) Auger, G.; van Heijenoort, J.; Mengin-Lecreulx, D.; Blanot, D.
FEMS Microbiol. Lett. 2003, 219, 115.
(91) Liu, H.; Ritter, T. K.; Sadamoto, R.; Sears, P. S.; Wu, M.; Wong,
C.-H. ChemBiochem 2003, 4, 603.
(92) Li, J. J.; Bugg, T. D. Chem. Commun. 2004, 182.
(93) Helm, J. S.; Chen, L.; Walker, S. J. Am. Chem. Soc. 2002, 124,
13970.
(94) Hu, Y.; Helm, J. S.; Chen, L.; Ginsberg, C.; Gross, B.; Kraybill,
B.; Tiyanont, K.; Fang, X.; Wu, T.; Walker, S. Chem. Biol. 2004,
11, 703.
(95) Schwartz, B.; Markwalder, J. A.; Wang, Y. J. Am. Chem. Soc.
2001, 123, 11638.
(96) VanNieuwenhze, M. S.; Mauldin, S. C.; Zia-Ebrahimi, M.;
Winger, B. E.; Hornback, W. J.; Saha, S. L.; Aikins, J. A.;
Blaszczak, L. C. J. Am. Chem. Soc. 2002, 124, 3656.
(97) Breukink, E.; van Heusden, H. E.; Vollmerhaus, P. J.;
Swiezewska, E.; Brunner, L.; Walker, S.; Heck, A. J.; de Kruijff,
B. J. Biol. Chem. 2003, 278, 19898.
(98) Chen, L.; Walker, D.; Sun, B.; Hu, Y.; Walker, S.; Kahne, D.
(99) Barrett, D. S.; Chen, L.; Litterman, N. K.; Walker, S. Biochemistry 2004, 43, 12375.
(100) Segel, I. H. Enzyme Kinetics; John Wiley & Sons: New York,
1975.
(101) Hu, Y. N.; Helm, J. S.; Chen, L.; Ye, X. Y.; Walker, S. J. Am.
Chem. Soc. 2003, 125, 8736.
(102) Puyang, X.; Fan, J.; Schumacher, T.; Borsari, B.; Davey, M.; Ling,
L. 43rd Interscience Conference on Antimicrobial Agents and
Chemotheraphy (ICAAC), Chicago, IL, 2003.
(103) Lo, M.-C. Ph.D. Thesis, Princeton University, 2000.
(104) Cudic, P.; Kranz, J. K.; Behenna, D. C.; Kruger, R. G.; Tadesse,
H.; Wand, A. J.; Veklich, Y. I.; Weisel, J. W.; McCafferty, D. G.
Proc. Nat. Acad. Sci. U.S.A. 2002, 99, 7384.
(105) Lo, M.-C.; Helm, J. S.; Sarngadharan, G.; Pelczer, I.; Walker, S.
J. Am. Chem. Soc. 2001, 123, 8640.
(106) Ghadiri, M. R.; Granja, J. R.; Milligan, R. A.; McRee, D. E.;
Khazanovich, N. Nature 1993, 366, 324.
(107) Bong, D. T.; Ghadiri, M. R. Angew. Chem., Int. Ed. 2001, 40,
2163.
(108) Karle, I.; Gopi, H. N.; Balaram, P. Proc. Natl. Acad. Sci. U.S.A.
2002, 99, 5160.
(109) Jiang, W.; Wanner, J.; Lee, R. J.; Bounaud, P.-Y.; Boger, D. L.
J. Am. Chem. Soc. 2003, 125, 1877.
(110) Rew, Y.; Shin, D.; Hwang, I.; Boger, D. L. J. Am. Chem. Soc.
2004, 126, 1041.
(111) Ciabatti, R.; Maffioli, S.; Checchia, A.; Romano, G.; Candiani,
G.; Panzone, G. WO 2003076460, 2003; Chem. Abstr. 2003, 139,
261566.
(112) Ciabatti, R.; Cavalleri, B. Eur. Patent 321696, 1989; Chem. Abstr.
1989, 112, 77952.
(113) Chen, L.; Yuan, Y.; Helm, J. S.; Hu, Y.; Rew, Y.; Shin, D.; Boger,
D. L.; Walker, S. J. Am. Chem. Soc. 2004, 126, 7462.
(114) Schoellkopf, U. Tetrahedron 1983, 39, 2085.
(115) Evans, D. A.; Sjogren, E. B.; Weber, A. E.; Conn, R. E.
Tetrahedron Lett. 1987, 28, 39.
(116) Evans, D. A.; Weber, A. E. J. Am. Chem. Soc. 1986, 108, 6757.
(117) Reno, D. S.; Lotz, B. T.; Miller, M. J. Tetrahedron Lett. 1990,
31, 827.
(118) Seebach, D.; Juaristi, E.; Miller, D. D.; Schickli, C.; Weber, T.
Helv. Chim. Acta 1987, 70, 237.
(119) Belokon, Y. N.; Bulychev, A. G.; Vitt, S. V.; Struchkov, Y. T.;
Batsanov, A. S.; Timofeeva, T. V.; Tsyryapkin, V. A.; Ryzhov,
M. G.; Lysova, L. A.; Bakhmutov, V. I.; Belikov, V. M. J. Am.
Chem. Soc. 1985, 107, 4252.
(120) Jung, M. E.; Jung, Y. H. Tetrahedron Lett. 1989, 30, 6637.
(121) Pons, D.; Savignac, M.; Genet, J. P. Tetrahedron Lett. 1990, 31,
5023.
(122) Woltering, T. J.; Weitz-Schmidt, G.; Wong, C.-H. Tetrahedron
Lett. 1996, 37, 9033.
(123) Shao, H.; Goodman, M. J. Org. Chem. 1996, 61, 2582.
(124) Li, G.; Chang, H.-T.; Sharpless, K. B. Angew. Chem., Int. Ed.
Engl. 1996, 35, 451.
(125) Li, G.; Angert, H. H.; Sharpless, K. B. Angew. Chem., Int. Ed.
Engl. 1997, 35, 22813.

(126) Noyori, R.; Ikeda, T.; Ohkuma, T.; Widhalm, M.; Kitamura, M.;
Takaya, H.; Akutagawa, S.; Sayo, N.; Saito, T.; Taketomi, T.;
Kumobayashis, H. J. Am. Chem. Soc. 1989, 111, 9134.
(127) Genet, J. P.; Pinel, C.; Mallart, S.; Juge, S.; Thorimbert, S.;
Laffitte, J. A. Tetrahedron: Asymmetry 1991, 2, 555.
(128) Kuwano, R.; Okuda, S.; Ito, Y. J. Org. Chem. 1998, 63, 3499.
(129) Morell, J. L.; Fleckenstein, P.; Gross, E. J. Org. Chem. 1977,
42, 355.
(130) Maurer, P. J.; Takahata, H.; Rapoport, H. J. Am. Chem. Soc.
1984, 106, 1095.
(131) Roemmele, R. C.; Rapoport, H. J. Org. Chem. 1989, 54, 1866.
(132) Beaulieu, P. L. Tetrahedron Lett. 1991, 32, 1031.
(133) Blaskovich, M. A.; Lajoie, G. A. Tetrahedron Lett. 1993, 34, 3837.
(134) Saeed, A.; Young, D. W. Tetrahedron 1992, 48, 2507.
(135) Lotz, B. T.; Gasparski, C. M.; Peterson, K.; Miller, M. J. J. Am.
Chem. Soc. 1990, 112, 7.
(136) Vassilev, V. P.; Uchiyama, T.; Kajimoto, T.; Wong, C.-H.
(137) Kimura, T.; Vassilev, V. P.; Shen, G.-J.; Wong, C.-H. J. Am.
Chem. Soc. 1997, 119, 11734.
(138) Tominaga, F.; Hiwaki, C.; Maekawa, T.; Voshida, H. J. Biochem.
Tokyo 1963, 53, 227.
(139) Singerman, A.; Liwschitz, Y. Tetrahedron Lett. 1968, 4733.
(140) Okai, H.; Izumiya, N. Bull. Chem. Soc. Jpn. 1969, 42, 3550.
(141) Hanessian, S.; Vanasse, B. Can. J. Chem. 1993, 71, 1401.
(142) Cho, G. Y.; Ko, S. Y. J. Org. Chem. 1999, 64, 8745.
(143) Cardillo, G.; Gentilucci, L.; Tolomelli, A.; Tomasini, C. Synlett
1999, 1727.
(144) Deng, J.; Hamada, Y.; Shioiri, T. J. Am. Chem. Soc. 1995, 117,
7824.
(145) Tohdo, K.; Hamada, Y.; Shioiri, T. Synlett 1994, 247.
(146) Boger, D. L.; Lee, R. J.; Bounaud, P.-Y.; Meier, P. J. Org. Chem.
2000, 65, 6770.
(147) Boger, D. L. Med. Res. Rev. 2001, 21, 356.
(148) Jiang, W.; Wanner, J.; Lee, R. J.; Bounaud, P.-Y.; Boger, D. L.
J. Am. Chem. Soc. 2002, 124, 5288.
(149) Maplestone, R. A.; Cox, J. P. L.; Williams, D. H. FEBS Lett. 1993,
326, 95.
(150) Brady, S. F.; Varga, S. L.; Freidinger, R. M.; Schwenk, D. A.;
Mendlowski, M.; Holly, F. W.; Veber, D. F. J. Org. Chem. 1979,
44, 3101.
(151) Rich, D. H.; Bhatnagar, P.; Mathiaparanam, P.; Grant, J. A.;
Tam, J. P. J. Org. Chem. 1978, 43, 296.
(152) Beaulieu, P. L.; Schiller, P. W. Tetrahedron Lett. 1988, 29, 2019.
(153) Weinreb, S. M.; Demko, D. M.; Lessen, T. A.; Demers, J. P.
(154) Huang, J.; Widlanski, T. S. Tetrahedron Lett. 1992, 33, 2657.
(155) Fan, C.-X.; Hao, X.-L.; Ye, Y.-H. Synth. Commun. 1996, 26, 1455.
(156) Li, H.; Jiang, X.; Ye, Y.-H.; Fan, C.; Romoff, T.; Goodman, M.
Org. Lett. 1999, 1, 91.
(157) Holdrege, C. T. U.S. Patent 3646024, 1972; Chem. Abstr. 1972,
77, 34543.
(158) Boger, D. L.; Chen, J.-H. J. Am. Chem. Soc. 1993, 115, 11624.
(159) Boger, D. L.; Chen, J.-H.; Saionz, K. W. J. Am. Chem. Soc. 1996,
118, 1629.
(160) Boeckman, R. K., Jr.; Potenza, J. C. Tetrahedron Lett. 1985, 26,
1411.
(161) Boger, D. L.; Miyazaki, S.; Kim, S. H.; Wu, J. H.; Loiseleur, O.
R.; Castle, S. L. J. Am. Chem. Soc. 1999, 121, 3226.
(162) Boger, D. L.; Miyazaki, S.; Kim, S. H.; Wu, J. H.; Castle, S. L.;
Loiseleur, O.; Jin, Q. J. Am. Chem. Soc. 1999, 121, 10004.
(163) Boger, D. L.; Ledeboer, M. W.; Kume, M. J. Am. Chem. Soc. 1999,
121, 1098.
(164) Boger, D. L.; Ledeboer, M. W.; Kume, M.; Searcey, M.; Jin, Q.
J. Am. Chem. Soc. 1999, 121, 11375.
(165) Kettenring, J. K.; Ciabatti, R.; Winters, G.; Tamborini, G.;
Cavalleri, B. J. Antibiot. 1989, 42, 268.
(166) Wanner, J.; Tang, D.; McComas, C. C.; Crowley, B. M.; Jiang,
W.; Moss, J.; Boger, D. L. Bioorg. Med. Chem. Lett. 2003, 13,
1169.
(167) Ciabatti, R.; Cavalleri, B. Eur. Patent 337203, 1989; Chem. Abstr.
1990, 112, 179893.
CR030106N
Chem. Rev. 2005, 105, 477497
477
Molecular Insights into Aminoglycoside Action and Resistance

Sophie Magnet and John S. Blanchard*
Department of Biochemistry, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, New York 10461
Received April 20, 2004
Contents
1. Introduction
2. Aminoglycosides
2.1. Chemical Structures
2.2. Ribosome Binding Site and Translation
Effects
2.3. Secondary and Bactericidal Effects
2.4. Properties and Clinical Use
3. Aminoglycoside Resistance
3.1. Decreased Intracellular Concentration of the
Drug
3.2. Target Modification
3.2.1. 16S rRNA Methylation
3.2.2. Ribosomal Mutations
3.3. Enzymatic Drug Modification
3.3.1. Aminoglycoside Adenylyltransferases
3.3.2. Aminoglycoside Phosphotransferases
3.3.3. Aminoglycoside Acetyltransferases
3.4. Origin and Prevalence
4. Resisting Resistance
5. Acklnowledgment
6. References
477
478
478
478
482
482
482
482
485
485
485
486
486
487
489
493
494
495
495
1. Introduction
Following the discovery of penicillin, which was
inactive against Mycobacterium tuberculosis, Waksman discovered the first antituberculosis agent,
streptomycin, by systematic screening of bacterial
culture supernatants for the presence of M. tuberculosis inhibitory activity.1 Several years after the
introduction of this aminoglycoside in human antibacterial chemotherapy, organisms resistant to streptomycin began to appear. Subsequently, other antituberculosis agents, including isoniazid, rifampicin,
and ethambutol, were discovered and replaced streptomycin in the treatment of tuberculosis. However,
resistance to these drugs also appeared rapidly. To
reduce the emergence of resistant organisms, a sixmonth-long, multidrug chemotherapy regimen was
adopted. More recently, streptomycin has regained
interest and significance for the treatment of multidrug-resistant (isoniazid- and rifampicin-resistant)
M. tuberculosis infections. In this interval, a large
number of other aminoglycosides were isolated from
various Actinomycetes producers. Among them, gen* Corresponding author. Phone: (718) 430-3096. Fax: (718) 4308565. E-mail: blanchar@aecom.yu.edu.
tamicin, isolated from Micromonospora in 1963,

constituted a significant advance in the treatment of
Gram-negative bacterial infections. Several other
molecules, such as dibekacin and amikacin, were
semisynthesized by modification of natural compounds with the aims of extending their antibacterial
spectrum and potency, reducing their nephro- and
ototoxicity, and evading the resistance mechanisms.
The most recent aminoglycoside introduced into
human antibacterial therapy was arbekacin, a kanamycin B derivative used in Japan since 1990. Today,
the family of aminoglycosides includes a large number of compounds. However, the number of resistance
mechanisms developed by microorganisms has increased in parallel with the number of drugs available and the frequency of their use.
The mechanisms of bacterial resistance to aminoglycosides have been the subject of numerous
genetic and biochemical studies and have been the
focus of many recent reviews.2-4 In some species,
broad spectrum and low level resistance can be
achieved by decreasing the intracellular drug concentration. However, decreased affinity of the drug
for its target, the bacterial ribosome, by modification
of the drug or ribosome is the major cause of aminoglycoside resistance, and among these mechanisms
the enzymatic inactivation of the drug is by far the
most clinically significant. There are three classes of
aminoglycoside-modifying enzymes: the aminoglycoside nucleotidyltransferases (ANTs), the aminoglycoside phosphotransferases (APHs), and the aminoglycoside acetyltransferases (AACs). The reactions
catalyzed by these enzymes are usually regioselective, and the site of modification is indicated in
parentheses. A roman numeral and a letter are then
added to the nomenclature to distinguish the enzymes according to the pattern of aminoglycoside
resistance that they confer and to their primary
sequence, respectively. Nevertheless, in some bacterial species other mechanisms of resistance that
involve unique features of the bacterium are predominant. This is the case for aminoglycoside resistance due to impermeability in Pseudomonas aeruginosa or due to ribosomal modification in M. tuberculosis.
In the past decade, several high-resolution structural studies have been performed to identify the
molecular nature of the interactions between aminoglycosides and the ribosome or the proteins involved in their inactivation via enzymatic modification. This work has brought considerable new insight
10.1021/cr0301088 CCC: $53.50 2005 American Chemical Society

Magnet and Blanchard
in antibiotic use. Finally, we will present the different

strategies used to design inhibitors of the resistant
determinants or new compounds that can escape
from known mechanisms of resistance.
2. Aminoglycosides
2.1. Chemical Structures
Sophie Magnet was born in Montelimar, France, in 1973. She studied

Biology at the Pierre et Marie Curie University in Paris and received her
Masters degree in Microbiology in 1997. She then did her graduate
research on aminoglycoside resistance in the Unite des Agents Antibacteriens at the Institut Pasteur in Paris. After receiving her Ph.D. in 2001,
she joined the laboratory of Dr. John S. Blanchard at the Albert Einstein
College of Medicine, New York, for a Postdoctoral fellowship in
Biochemistry, during which time she performed mechanistic and kinetic
studies of aminoglycoside acetyltransferases and DNA ligases. She is
returning to France to work on enzymes involved in cell wall synthesis in
Gram-positive bacteria at INSERM.
Aminoglycosides are hydrophilic molecules, consisting of a characteristic, central aminocyclitol linked

to one or more amino sugars by pseudoglycosidic
bond(s). For the majority of clinically useful compounds, referred to in this review as typical aminoglycosides, the aminocyclitol is the 2-deoxystreptamine ring, and it can be monosubstituted in
position 4, as is the case for neamine, or disubstituted
in positions 4 and 5, or 4 and 6. The accepted
nomenclature usually used refers to ring I as the
primed ring and corresponds to the amino sugar at
position 4 of the deoxystreptamine ring. Ring II is
unnumbered and corresponds to the central aminocyclitol, while ring III is referred to as the doubly
primed ring and has the substituent in position 5 or
6 of the deoxystreptamine ring. Ring IV (triply
primed numbering) corresponds to any additional
ring attached to ring III (Figure 1). However, a
number of active aminoglycosides are structurally
atypical according to the above description. For
instance, streptomycin possesses a streptidine ring
as the central aminocyclitol, and spectinomycin consists of three fused rings. The structures of the more
common atypical aminoglycosides are shown in Figure 2.
2.2. Ribosome Binding Site and Translation

Effects
John S. Blanchard was born in Waterbury, CN. He received his B.S. in

Chemistry from Lake Forest College and did his graduate research in
biochemistry in the laboratory of Dr. W. W. Cleland at the University of
Wisconsin. After a three-year Postdoctoral NIH-supported Fellowship, he
was appointed Assistant Professor of Biochemistry at the Albert Einstein
College of Medicine, New York, in 1983 and became the Dan Dancinger
Professor of Biochemistry in 1998. His early interests in enzyme kinetics
and isotope effects focused on flavoenzyme catalyzed reactions. His
research efforts into the mechanism of action of isoniazid in the human
bacterial pathogen Mycobacterium tuberculosis led to his current interest
in antibiotic resistance. He was supported in 2000 by a Fogarty
International fellowship in the laboratory of Dr. Patrice Courvalin at the
Institut Pasteur in Paris, where he studied aminoglycoside N-acetyltransferases and met Dr. Magnet. He is the author of over 100 peer-reviewed
manuscripts and 20 reviews and has been awarded six United States
patents. His work has been generously supported by the National Institutes
of Health for the last 21 years.
into the mechanism of action of various compounds

and into the mechanisms by which bacteria become
resistant to them.
In this review we will describe the current molecular understanding of aminoglycoside action and
resistance, focusing on recent structural advances.
The emergence of new mechanisms of resistance and
the evolution of the distribution of resistance determinants will be presented in parallel with changes
The primary target of aminoglycosides is the bacterial small ribosomal subunit. Aminoglycoside binding
to the 16S rRNA, at the tRNA acceptor A site
(Aminoacyl site), inhibits the translation process by
causing misreading and/or hindering the translocation step. Two crystal structures of the 30S ribosomal
subunit of Thermus thermophilus were solved by
X-ray diffraction methods in 2000.5,6 These studies
brought considerable insights into the components
and the function of the ribosomal A site. It is believed
that the fidelity of translation depends on two steps,
an initial recognition between the codon of the mRNA
and the anticodon of a charged tRNA, and subsequent proofreading. The A site includes portions of
the 530 loop, helix 34, and the base of helix 44. The
tRNA anticodons bind within a cleft formed between
the individual domains, and relative movements of
these domains are likely to be involved in both the
decoding and translocation processes.
The earliest structural studies of complexes containing aminoglycosides were performed using a 27nucleotide-long RNA stem loop which mimicked the
conserved helix 44 moiety of the 16S rRNA A site
that was shown to bind the 2-deoxystreptaminecontaining aminoglycosides. A stoichiometric 1:1
complex was generated with the 4,5-disubstituted
2-deoxystreptamine, paromomycin, and its three-
Figure 1. Structures of clinically useful typical aminoglycosides: A, 4,5-disubstituted deoxystreptamine aminoglycosides;

B, 4,6-disubstituted deoxystreptamine aminoglycosides. The deoxystreptamine ring is shown in red.
dimensional structure was solved using transfer

nuclear Overhauser effect-derived distance constraints and simulated annealing methods.7 The data
revealed that the antibiotic binds in the major groove
of the RNA, where it adopts an L-shaped conformation. The primed ring was bound in a pocket formed
by the non-Watson-Crick A1408-A1493 base pair
and the unpaired A1492, which generates a bulge in
the helical structure. The majority of the intermolecular contacts were made with the central 2-deoxystreptamine ring and the primed 2,6-dideoxy-2,6diamino-glucose ring substituents. Comparison of the
unliganded RNA oligonucleotide with the drug-RNA
complex structure showed that the most dramatic
change occurred at the totally conserved triplet
adenine pocket (A1408, A1492, A1493), which is

displaced toward the minor groove upon paromomycin binding.8 In another report, the same authors
described the structures of three additional aminoglycoside-RNA complexes,9 one containing the
structurally simplest aminoglycoside, neamine, and
two others containing the 4,5-disubstituted 2-deoxystreptamine ribostamycin or neomycin. The conformation of the bound drug, the binding site, and the
intermolecular contacts made with the RNA were
similar for ribostamycin and neomycin to those
observed in the paromomycin complex. In the case
of neamine, two possible orientations were proposed
involving opposite interactions of the 1- and 3-amino
groups of the deoxystreptamine ring. In 2000, Carter
Figure 2. Structures of clinically useful atypical aminoglycosides. The aminocycitol ring is shown in red.
et al. described the crystal structures of various

complexes of aminoglycoside bound to the intact 30S
ribosomal subunit.10 The structure of the paromomycin complex revealed, as expected, that the drug
binds at the A site in the major groove of helix 44.
The interactions reported in the NMR structures
were largely confirmed. In addition to a stacking
interaction between the primed ring and G1491, the
hydroxyl groups in positions 4 and 6 make hydrogen
bonds with the N1 of A1408 and the phosphate
oxygen of A1493, respectively (Figure 3, top). The two
amino groups of the deoxystreptamine ring in positions 1 and 3 interact directly with N7 of G1494 and
O4 of U1495, respectively. Besides an intramolecular
hydrogen bond observed between the 2-amino group
and the 5-hydroxyl group, the double primed ribose
ring makes a single interaction with the N7 group of
G1491. Finally, the triply primed 2,6-dideoxy-2,6diamino-glucose ring, which was disordered in the
NMR structure, was clearly observed in the 30S
subunit structure interacting with backbone phosphates from both sides of helix 44, including C1490
and G1405. In the 30S-paromomycin complex, A1492
and A1493 are flipped out compared to the free 30S
subunit and to a greater extent than previously
observed in the NMR structure. The structures of two
4,6-disubstituted aminoglycosides, gentamicin C1a11
and tobramycin,12 bound to a single or a dimeric
synthetic A site RNA, respectively, were also solved
by X-ray crystallography. Both complexes were very
similar, and only small differences were observed in
the position of the double primed ring. The interactions with specific nucleotides as well as with phosphate backbone oxygen atoms and the central and
primed rings were, as expected, very similar to those
observed for the 4,5-disubstituted deoxystreptamines.
The 2-hydroxyl group and 3-amino group of ring
III make additional specific contacts with O6 and N7
of G1405, respectively, compared to the 4,5-disubstituted aminoglycosides (Figure 3, bottom). Two additional intramolecular hydrogen bonds were also
Figure 3. Interactions between typical aminoglycosides

and the 30S ribosome. Top: Interactions between paromomycin and the 30S ribosome (Reprinted with permission
from Nature (http://wwww.nature.com), ref 10. Copyright
2000 Nature Publishing Group.). Bottom: Interactions
between tobramycin and the 30S ribosome (Reprinted with
permission from ref 12. Copyright 2002 Elsevier). The
superscripts refer to the nucleotide functional group that
interacts with the bound aminoglycoside. The nucleotide
numbering is for the homologous E. coli ribosome.
revealed between the 5-hydroxyl group and the ring

III oxygen and between the 1-amino and the 2hydroxyl groups of the drug.
The selection of aminoacyl-tRNA during translation occurs by formation of a minihelix between the
codon of the mRNA and the anticodon of the cognate

tRNA. When a tRNA-mRNA complex is formed,
A1492 and A1493 form a hydrogen-bonding network
with the ribose 2-hydroxyl groups of the two first
bases of both the codon and the anticodon, allowing
discrimination between different base pairing geometries. When a cognate mRNA-tRNA complex is so
recognized, the two adenines flip out from the helix
with the net effect of increasing the affinity for the
cognate tRNA and so stabilizing the complex. The
structural data presented above showed that the
binding of a disubstituted 2-deoxystreptamine in the
decoding center causes a similar conformational
change to the formation of a cognate mRNA-tRNA
complex. These data confirm biochemical results that
show that this class of aminoglycosides both decreases the rate of A site tRNA dissociation13 and
increases the tRNA binding affinity.14 As a consequence, the affinity of the A site for a noncognate
mRNA-tRNA complex is increased upon drug binding, preventing the ribosome from efficiently discriminating between noncognate and cognate complexes. This provides a dramatic atomic level rationalization of the well-known miscoding effect of the
disusbtituted 2-deoxystreptamines antibiotics.
The structures of three structurally atypical aminoglycosides, streptomycin, spectinomycin, and hygromycin B, bound to the 30S subunit, have also been
reported. The structure of the streptomycin-30S
complex revealed that the drug binds tightly to the
A site, but its binding site is adjacent to the one of
the disubstituted deoxystreptamines.10 In contrast to
paromomycin, which interacts only with residues
contained within helix 44, streptomycin makes interactions with the backbone phosphates and ribose
hydroxyl groups of residues from four different
domains of the 16S rRNA molecule, including U14
in helix 1, C526 and G527 of the 530 loop in helix
18, A913 and A914 from helix 27 and 28, respectively,
and C1490 and G1491 from helix 44 (Figure 4, top).
In addition, streptomycin is the only aminoglycoside
that interacts with ribosomal protein side chains.
Indeed, the structure of the complex showed that the
-amino group of K45 of the S12 ribosomal protein
contacts both the 4- and 5-hydroxyl groups of the
streptamine ring. This structure has provided considerable insights into the dynamic functions that
occur during aminoacyl-tRNA binding to the A site.
It was proposed that H27, which interacts with H44,
can have two alternative base pairing schemes during
translationsone which leads to a ribosomal ambiguity (ram) conformation of the decoding center, with
high affinity for tRNA which results in increased
miscoding, and a second that leads at the opposite
to a restrictive state with low tRNA affinitysand the
balance of theses two states could be involved in the
proofreading process.15 The binding site of streptomycin revealed by the structural data suggests that
the antibiotic affects the dynamic equilibrium of the
two ribosomal conformations by stabilizing the ram
state, providing an explanation for the error prone
effect of this drug.10 However, a confirmation of this
model would require an atomic resolution structure
of the restrictive form of the ribosome.
Figure 4. Interactions between atypical aminoglycosides

and the 30S ribosome. Top: Interactions between streptomycin and the 30S ribosome (Reprinted with permission
from Nature (http://wwww.nature.com), ref 10. Copyright
2000 Nature Publishing Group.). Bottom: Interactions
between hygromycin B and the 30S ribosome (Reprinted
with permission from ref 16. Copyright 2000 Elsevier). The
superscripts refer to the nucleotide functional group that
interacts with the bound aminoglycoside. The nucleotide
numbering is for the homologous E. coli ribosome.
The structure of spectinomycin bound to the 30S

subunit was also described in the same paper.10 The
rigid molecule binds in the minor groove of the 16S
rRNA, at the end of helix 34 near to the pivot point
of the head region. The drug makes four hydrogen
bonds with the bases G1064, C1066, and C1192 of
helix 34 in addition to the interaction with the 2ribose hydroxyl group of G1068 contained within
helix 36. These observations lead the authors to
propose that the binding of spectinomycin, known to
inhibit the EF-G catalyzed GTP-dependent translocation of the peptidyl-tRNA from the A site to the P
site (Peptidyl site), prevents the movement of the
head region and helix rearrangements necessary for
the translocation step.
Finally, the structure of the 30S-hygromycin B
complex has been reported by the same group in a
separate paper.16 Hygromycin B is active against both
prokaryotic and eucaryotic ribosomes. It acts primarily by inhibiting the translocation step but also, to
a lesser extent than streptomycin or the disubstituted
2-deoxystreptamines, causes miscoding. Hygromycin
B binds just above the binding site of paromomycin
at the very top of helix 44. The ring I substituents
interact with both the backbone phosphate oxygen
atoms of G1494 and the bases G1405 G1494, U1495,
and C1496 (Figure 4, bottom). The rest of the
molecule makes only weak base specific hydrogen
bonds with C1403 and U1498. Ring IV is positioned
very close to the P site. The binding of hygromycin

B between the A site and the P site explains its dual
effect on translation accuracy and translocation.
Moreover, the absence of specific contact between
hygromycin B and A1408, a residue totally conserved
among eubacteria, can certainly account for the
affinity of this molecule for eucaryotic rRNA, which
possesses a G at the corresponding position.17
2.3. Secondary and Bactericidal Effects

Whereas spectinomycin and kasugamycin, two
inhibitors of the translocation step, act bacteriostatically, streptomycin and the disubstituted 2-deoxystreptamines are bactericidal antibiotics. Although
the mechanism of action of aminoglycosides at the
translational level was substantially clarified by the
structural data detailed above, the origin of the
bactericidal effect of most of these compounds remains enigmatic. Comparison of the effect of aminoglycosides on translation and on cell viability
suggests that lethality is correlated with the production of mistranslated proteins, which could have fatal
secondary effects. It was shown that aminoglycosides
that cause misreading, like streptomycin, increase
the passive permeability of the cell membranes to
several small molecules, as originally probed by the
release of K+ ions.18 This permeabilization presumably occurs as a result of the incorporation of truncated or incorrectly folded proteins in the membranes, since it has not been observed in mutant
strains in which protein synthesis is insensitive to
streptomycin by mutation in the rpsL gene.19-21 This
phenomenon results in an increased aminoglycoside
uptake that leads to an irreversible accumulation of
the drug inside the cell. It was proposed that following the rapid degradation of the membrane-associated mistranslated proteins the drug becomes trapped
inside the cell.18 The resulting high intracellular drug
concentration is believed to play a critical role in the
bactericidal effect.21 The saturation of the ribosomes
with aminoglycosides, potentially coupled to the
inhibition of new ribosome synthesis and assembly,22
could result in the complete inhibition of protein
synthesis, leading to bacterial death. Alternatively,
the production of mistranslated proteins might affect
another vital cellular processes. For instance, it has
been reported that aminoglycosides that cause misreading inhibit the initiation of DNA replication,20
but this finding requires more detailed investigations
that have not been pursued.
2.4. Properties and Clinical Use

Because of their poor oral absorption, aminoglycosides are most often administered parenterally in
order to achieve therapeutically adequate serum
concentrations. As a consequence of their polar
structure, the drugs inefficiently cross biological
membranes and thus their intracellular tissue concentration is low except in the proximal renal tubule,
where they are concentrated.23,24 Whereas the conventional dosing for gentamicin, tobramycin, or netilmicin is around 5 (mg/kg)/day, 3-fold higher doses are
used for other compounds such as streptomycin or
amikacin.23,25 For patients with normal renal activity,

a single daily dosing was shown to be more favorable
than the traditionally recommended multiple daily
dosing regimens. Because of the concentration-dependent activity and postantibiotic effect of aminoglycosides, a similar efficacy can be obtained by
single daily dosing with a lower cost, reduced toxicity,
and reduced emergence of an adaptive resistant
population.25-31 In addition, aerosolized or liposomeencapsulated aminoglycosides have been shown to be
advantageous for the treatment of respiratory tract
infections32-34 or against intracellular bacteria like
Mycobacterium avium,35-38 respectively. Finally,
therapy strategies such as alternating between different classes of antibiotics, switch therapy, or the
rotational use of different aminoglycosides are increasingly encouraged to prevent therapeutic failure
due to resistance.31,39,40
Aminoglycosides are active against a wide range
of aerobic Gram-negative bacilli, many staphylococci,
mycobacteria, and some streptococci.4,23,24,41,42 They
are particularly useful for the treatment of neutropenic patients and serious infections caused by
aerobic Gram-negative bacteria. Gentamicin continues to be the aminoglycoside of choice to treat
hospital acquired enterobacteriaceae and P. aeruginosa infections. Most often it is used in combination
with a -lactam, which results in a synergistic
bactericidal effect due to an enhanced uptake of
aminoglycosides.43 Other inhibitors of bacterial cell
wall synthesis can also be coadministered with aminoglycosides to treat infections due to bacterial
species that are naturally resistant to aminoglycosides because of impaired uptake, including the
enterococci.41,42 For the treatment of urinary tract
infections, aminoglycosides can be used alone in
monotherapy. More recently introduced aminoglycosides, such as amikacin or arbekacin, which are not
substrates for a number of aminoglycoside-modifying
enzymes23,44 or which retain their antibacterial activity after modification,45 are used to treat infections
due to gentamicin-resistant organisms, including
methicillin-resistant Staphylococcus aureus.46-50 Finally, streptomycin is still used in multidrug chemotherapy to treat multidrug-resistant M. tuberculosis.23,39
3. Aminoglycoside Resistance
3.1. Decreased Intracellular Concentration of the
Drug
Decreased aminoglycoside concentration inside a
target cell, by reduction of drug uptake, activation
of drug efflux, or both, will affect the susceptibility
of the strain to the whole class of aminoglycoside
compounds and can be the cause of intrinsic or
acquired resistance. Although the exact mechanism
of aminoglycoside uptake remains unknown, it is
accepted that the process consists of three consecutive steps.51,52 The first step is the adsorption of the
cationic compounds to the surface of bacteria by
electrostatic interactions with the negatively charged
lipopolysaccharides of the outer membrane of Gramnegative bacteria. The next two steps are dependent
on the transmembrane potential generated by the

respiratory chain, with the second being characterized by a faster rate of uptake. As a result, anaerobic
bacteria are intrinsically resistant to aminoglycosides
due to impermeability.53 Similarly, respiratory chain
mutants or strains containing functional mutations
in their ATP synthases were shown to exhibit decreased susceptibility to aminoglycosides.51,54 Such
mutants have been isolated from clinical or experimental endocarditis caused by infection with Escherichia coli, S. aureus, or P. aeruginosa.55 Changes
in membrane components involved in the initial
electrostatic binding of aminoglycosides have also
been associated with increased levels of resistance,
especially in the case of P. aeruginosa.56 Clinical
strains exhibiting low level resistance to gentamicin
were shown to have a modified, less negatively
charged, lipopolysaccharide that exhibits a lower
affinity for gentamicin.57 In addition, the extracellular alginate produced by mucoid strains of P. aeruginosa, which both inhibits phagocytosis by monocytes
and neutrophils and enhances bacterial adherence
to the respiratory epithelia, was shown to decrease
the uptake and early bactericidal effect of aminoglycosides. It was proposed that the viscous polyanionic
alginate gel acts as a physical and ionic trap for the
drug.58
Energy-dependent bacterial efflux is now recognized as a major cause of antibiotic resistance. This
is particularly true for the multidrug-resistant opportunist pathogens responsible for nosocomial infections, which have to counter the environmental
pressure exerted by the constant presence of antibiotics. Bacterial species constitutively expressing such
transporters are intrinsically resistant to low levels
of various antibiotics. Moreover, mutations in the
regulatory genes of the pumps, or induction of
expression in the presence of substrate, can lead to
the overexpression of the originally constitutive, or
silent, pump genes. It had been thought that multidrug transporters were specific for hydrophobic or
amphiphilic compounds and, thus, that aminoglycosides would not be affected by this mechanism of
resistance. However, in the last several years, aminoglycosides have been shown to be substrates for a
number of multidrug efflux pumps, including members of the five superfamilies of bacterial transporters.59
Recently, crystallographic structural studies on
different components of the tripartite transporters of
the RND (resistance nodulation cell division) superfamily, which play a particularly important role in
Gram-negative bacteria, have been revealing.60,61 The
transporters of the RND superfamily use the membrane proton-motive force as energy source. They are
localized in the cytoplasmic membrane, and in Gramnegative bacteria they interact with a membrane
fusion protein (MFP), located in the periplasmic
space, and an outer membrane protein (OMP) to form
a continuous, tripartite channel able to export substrates directly out of the cell. E. coli AcrB, which
exhibits broad substrate specificity but does not efflux
aminoglycosides, has served as the structural prototype of a bacterial RND protein involved in antibiotic
resistance. AcrB interacts with the MFP, AcrA, and

the OMP, TolC. The structures of the AcrB apoprotein,60 AcrB in complex with four substrates (rhodamine 6G, ethidium bromide, dequalinium, and ciprofloxacin),62 and TolC63 have been solved. Both
membrane proteins exist as homotrimers. Trimeric
TolC forms a barrel composed of 12 -strands that
span the outer membrane and 12 R-helices that
extend into the periplasmic space over 100 . The
internal cavity is open to the external medium and
provides solvent access. Each monomer of AcrB
contains 12 transmembrane domains and two large
periplasmic domains. The transmembrane domains
of the three protomers are arranged in a ringlike
manner, creating a 30 diameter cavity, where
substrates bind. This cavity is connected to the
periplasmic funnel by a very narrow pore, and the
dimensions of the funnel are compatible with a direct
interaction with TolC. Between the AcrB monomers,
an opening formed by residues of the periplasmic
domaims is observed that links the central channel
to the periplasm. It was proposed that these openings
(vestibules) provide a way for substrates selected
from the outer leaflet of the cell membrane or from
the periplasmic space to gain access to the channel
and be exported60 (Figure 5). Such a model explains
the structurally broad substrate specificity of this
type of transporter, since both amphiphilic compounds, which can partially penetrate the lipid
bilayer, and polycationic molecules such as aminoglycosides, which interact electrostatically with the
phospholipids of the membrane, can be captured at
the entrance of the vestibule.61 The selection of efflux
substrates would then be determined by the nature
of the residues at the entrance of the openings. This
is supported by domain-swapping experiments between two RND proteins of P. aeruginosa, MexY and
MexB,64 and two RND proteins of E. coli, AcrB and
AcrD,65 indicating that the periplasmic domains are
responsible for efflux specificity. The presence of
many more acidic residues at the entrance of the
vestibules of AcrD, which does export aminoglycosides, compared to AcrB, which does not, also supports this model.61 The structures of AcrB in complex
with various substrates have shown than once they
reach the central cavity, the structurally distinct
ligands bind to different sites, with a stoichiometry
of one per protomer.62 Three conserved charged
residues, located in transmembrane helices 4 and 10,
might constitute the transmembrane proton translocation site.60 Protonation of these residues would
disrupt the ion pair and trigger a conformational
change leading to the pore opening. A preliminary
structure of AcrA by electron crystallography reveals
that the protein has an elongated shape of 100-200
overall length,66 but the nature of the contacts
made with AcrB and TolC, as well as its role in the
system, is still very poorly understood.
Several RND proteins have been shown to be
involved in intrinsic and/or acquired, proton motive
force-dependent, aminoglycoside resistance in various
Gram-negative pathogens, including P. aeruginosa,
Burkholderia pseudomallei, Acinetobacter baumannii, and E. coli. The disruption or deletion of the
Figure 5. Schematic representation of a model proposed for drug capture at the outer leaflet of the cytoplasmic membrane
by a trimeric RND protein. Hydrophilic and positively charged compounds such as aminoglycosides may bind to the acidic
outer surface of the membrane, and amphiphilic compounds such as fluoroquinolones or chloramphenicol may partially
diffuse into the lipid bilayer before being recognized by specific interactions at the periplasmic vestibules of the RND and
drawn into the central cavity: OMP, outer membrane protein; RND, resistance nodulation cell division.
genes encoding the RND proteins MexY from P.

aeruginosa,67,68 AmrB from B. pseudomallei,69 and
AcrD from E. coli70 resulted in the hypersusceptibility
of the strains to aminoglycosides, indicating that
these proteins contribute to the intrinsic resistance
of these species to aminoglycosides. Whereas the
AmrAB system seems to be specific for aminoglycosides and macrolides, others exhibit a very broad
spectrum of antibacterial substrate specificity. Although no clear correlation has been made between
the level of expression of MexXY and the MIC values
of aminoglycosides in clinical isolates of P. aeruginosa, overexpression of the mexXY structural genes
has been shown to be responsible for high-level
acquired resistance in some cases.71 The MexXY
system also appears to be involved, in combination
with membrane impermeability, in the ability of P.
aeruginosa to develop adaptative resistance in response to exposure to inhibitory concentrations of
aminoglycosides.72 The AdeABC system of A. baumannii is another example of an RND pump mediating acquired aminoglycoside resistance. It was shown
that the inactivation of the adeB gene from a multidrug-resistant clinical isolate of A. baumannii
restored the susceptibility of the strain to various
drugs, including aminoglycosides (from 2- to 32-fold),
fluoroquinolones, cefotaxime, erythromycin, tetracycline, chloramphenicol, and trimethoprim.73
Gene(s) encoding putative transcriptional regulator(s) are frequently present in the immediate environment of the structural genes of the efflux system.
The genes amrR69 and mexZ,68 and the two component regulatory gene homologues adeRS,73 have been
identified upstream from the corresponding efflux
genes. While the regulator proteins of some efflux
systems belonging to distinct superfamilies of transporters, such as BmR from Bacillus subtilis or QacR
from S. aureus, have been the subject of detailed

studies showing that they modulate the transcription
of the adjacent genes following drug binding,74,75 very
little is known about the regulators of the RND
family transporters. Although mutations in mexZ
have been associated with increased expression of
mexXY, other results suggest that the regulation of
the system is a more complex process involving
various components, including yet uncharacterized
factors.71 On another hand, the expression of the
OMP component could play an important role in the
functional expression of these efflux systems. In the
case of P. aeruginosa, no gene encoding a cognate
OMP was found in the closed environment of mexXY,
suggesting that the pump can use independently
encoded protein(s) to form an efficient tripartite
system. Although it has been shown that MexXY can
act together with OprM, a more recent report identified two other OMPs, OpmG and OpmI, that can
contribute to aminoglycoside resistance in this species.76
Members of the major facilitator superfamily (MFS)
of transporters have also been shown to decrease
aminoglycoside susceptibility in strains harboring the
structural gene on a multicopy plasmid. The MFS
proteins are implicated in the active transport of both
sugars and antibiotics. They contain 12 or 14 transmembrane segments and use the proton motive force
as energy source. MdfA from E. coli was the first
putative MFS protein reported to have an effect on
aminoglycoside resistance.77 Althought this effect,
observed in a strain of E. coli harboring the gene on
a plasmid, was very modest and not clinically relevant (2- to 3-fold increases in the MIC values of
kanamycin and neomycin), this finding led to other
investigations on the role of MFS proteins in other
species. The tap and P55 genes, isolated from Myco-
bacterium fortuitum78 and Mycobacterium bovis,79

respectively, encode putative MFS proteins and
conferred resistance to aminoglycosides, including
streptomycin, when cloned in Mycobacterium smegmatis. Genes and proteins homologous to P55/P55
were detected by polymerase chain reaction (PCR)
or by western blot analysis, in many Mycobacterium
species, including M. tuberculosis (Rv1410c). These
observations, together with the identification of a
total of 16 open reading frames that encode putative
MFS proteins in the genome of M. tuberculosis,80 may
account for streptomycin-resistant clinical isolates of
M. tuberculosis that cannot be assigned to mutations
in rpsL or rrs genes or enzymatic modification of the
drug.
3.2. Target Modification

3.2.1. 16S rRNA Methylation
The lack of post-transcriptional methylation of
A1518 and A1519 16S rRNA nucleotides, as a result
of mutations in the ksgA gene, which encodes an
S-adenosylmethionine (SAM)-dependent RNA methylase, was associated with resistance to kasugamycin,
an inhibitor of the initiation step of translation, in
E. coli and Bacillus stearothermophilus.81-83 Although this example is not clinically relevant, it
illustrates the influence of the 16S rRNA methylation
pattern on the interactions with aminoglycosides.
Members of the Actinomycetes produce inactive
aminoglycosides, including acetylated- or phosphorylated-forms, which are cleaved during or after their
export out from the cell by specific enzymes, to form
the active antibiotics.84-86 To further resist the
secreted active compounds, many aminoglycosideproducing organisms also express rRNA methylases
capable of modifying the 16S rRNA molecule at
specific positions critical for the tight binding of the
drug.87 A number of genes encoding such enzymes
have been identified from several aminoglycoside
producers.88-95 The corresponding rRNA methyltransferases form the Agr family of methyltransferases (for aminoglycoside resistance).96 Some of
these enzymes have been characterized. KamA from
Streptomyces tenjimariensis and KamB from Streptomyces tenebrarius catalyze the modification of
A1408 at the N1 position and confer high-level
resistance to kanamycin, tobramycin, sisomicin, and
apramycin but not gentamicin.87,88 GmrA from the
gentamicin producer Micromonospora purpurea and
KgmB from S. tenebrarius catalyze the modification
of G1405 at the N7 position, conferring high-level
resistance only to the 4,6-disubstituted deoxystreptamines, including gentamicin.87,89 Methylation of
these nucleotides presumably abolishes the intermolecular contacts that they make with the drug
(discussed previously in section 2.2 and Figure 3).
The specific interaction observed between ring III of
the 4,6-disubstituted deoxystreptamines and G1405
is consistent with the resistance pattern conferred
by G1405 methylation. The presence of a methyl
group at the 6 position of some compounds of the
gentamicin mixture could modify the interaction with
A1408 compared to those observed with tobramycin,
explaining why modification of this residue does not

confer resistance to gentamicin.
Until recently, genes encoding a 16S rRNA methyltransferase had been restricted to the aminoglycoside producers, but three reports in 2003 and 2004
described the characterization of similar genes in
clinical isolates of human Gram-negative pathogens.
The rmtA and rmtB genes, located on plasmid-borne
transposons, were found in clinical isolates of P.
aeruginosa and Serratia marcescens, respectively.97,98
These strains were all isolated in Japan, where
arbekacin has been used extensively since 1990. The
two genes share 82% sequence identity, and the
encoded Rmt enzymes confer high-level resistance
(MICs > 1024 g/mL) to almost all clinically useful
aminoglycosides, including arbekacin, but not streptomycin. The considerable primary sequence similarity observed between the Rmt proteins and the 16S
rRNA methylases of Actinomycetes, as well as the
high G-C content of the gene (55%), suggests a
possible gene transfer from the producing organisms
to Gram-negative pathogens. Another 16S rRNA
methylase was characterized from Klebsiella pneumoniae. The structural gene, armA, was located on
a plasmid containing several other resistant genes,
including those conferring resistance to -lactams,
trimethoprim, sulfonamides, and other aminoglycoside resistance determinants. In contrast to rmtA and
rmtB, the low G-C content of armA (30%) does not
suggest a direct or recent acquisition from the Actinomycetes. However, the ArmA methylase is able to
confer high-level resistance to essentially all aminoglycosides except streptomycin.99 The site specificity of the modification catalyzed by these two enzymes was not determined, but the pattern of
resistance associated with ArmA, which includes all
4,6-disubstituted deoxystreptamines but not apramycin, led the authors to propose that the N7 of
G1405 is the locus of modification. The armA gene
has been detected by PCR in several other Enterobacteriaceae isolated from different European countries.99 Because these methylases can modify all
copies of the 16S rRNA and lead to high-level
resistance to an extremely wide range of compounds,
the emergence of this resistance mechanism in human pathogens is of concern for the future, especially
considering that the structural genes can apparently
be easily disseminated.
3.2.2. Ribosomal Mutations

Resistance to aminoglycosides by mutation of the
ribosomal target is clinically relevant only for streptomycin in M. tuberculosis. Mycobacterium is the only
genus of eubacteria with species that contain a single
copy of the ribosomal operon, which implies that a
single mutation can lead to the production of a
homogeneous population of mutant ribosomes and
thus can result in resistance regardless of the recessive nature of the mutation involved. The mutations
in the rrs gene, encoding the 16S rRNA and associated with streptomycin resistance in M. tuberculosis,
affect two highly conserved regions, the 530 loop and
the region around nucleotide 912, according to E. coli
numbering,100-105 and result in decreased affinity for
Figure 6. Reaction catalyzed by aminoglycoside O-adenylyltransferases. The reaction shown is that catalyzed by ANT(4)
catalyzing the MgATP-dependent 4-O-adenylylation of kanamycin A.
streptomycin.106 The structure of the streptomycin30S ribosomal subunit complex revealed specific
interactions between the drug and the backbone
phosphate or ribose hydroxyl groups of nucleotides
C526 and G527 of helix 18 and A913 and A914 of
helixes 27 and 28, respectively,10 providing a rationale for the location of mutations previously identified and their effect on streptomycin binding. Apart
from clinically significant streptomycin resistance in
M. tuberculosis, a few reports have described 16S
rRNA mutations associated with aminoglycoside
resistance in clinical isolates of microorganisms
containing a single or low copy number of rrs genes.
Mutations at positions 1400 and 1401 were found in
kanamycin-resistant isolates of M. tuberculosis,107 the
mutation A1408G (corresponding to the eucaryotic
allele of this position) was identified in the unique
rRNA operon of Mycobacterium abscessus and Mycobacterium chelonae isolates resistant to 2-deoxystreptamine-containing aminoglycosides, and mutations affecting the base pair G1064-C1192 of helix
34 were found in the three copies of the 16S rRNA of
spectinomycin-resistant isolates of Neisseria.108
The introduction of a single rrs gene on a multicopy
plasmid has been used for many years to study the
effect of 16S rRNA mutants on the activity of aminoglycosides in a heterogeneous population of ribosomes or homogeneous purified mutant ribosomes.
These studies have shown that at least half of the
ribosomes must be in the mutant form to confer
aminoglycoside resistance.106,109,110 More recently, to
circumvent the problem of the recessive nature of
ribosomal mutations, strains containing a single copy
of the rrs gene have been genetically engineered.111-113
Synthetic oligonucleotides mimicking the A site have
also been used for similar in vitro studies.114 These
studies have confirmed the effect of the naturally
occurring mutations previously shown to affect streptomycin activity,115 and they revealed other mutations associated with spectinomycin,116,117 hygromycin
B,118,119 or 4,6-disubstitued 2-deoxystreptamine113,120-122
resistance. These studies have shown that mutations
leading to a steric or allosteric change in the drugbinding pocket can be more deleterious for antibiotic
activity than mutations abolishing direct contacts
between the drug and the 16S rRNA.
Mutations in genes encoding ribosomal proteins
can also alter the activity of aminoglycosides. Notably, mutations in protein S12 are the other major
cause of streptomycin resistance in M. tuberculosis100,101,105,123 and other species.124,125 Mutations occurred in two regions, located around residues 42 and
87 (E. coli numbering), that contact helixes 18, 27,
and 44 of the 16S rRNA. The most frequent substitu-
tions were found to occur at residues P41, K42, K87,

and P90. Although structural data showed that S12
makes direct contact with streptomycin,10 the effect
of such mutations appears likely to be conformational
changes in the rRNA that prevent drug binding. This
conclusion, highlighted by structural data and chemical protection experiments, was first predicted by the
observation that different mutations in S12 lead to
different phenotypes: streptomycin resistance or
streptomycin dependence.115,126-129 In addition, S12
mutations can be compensated for by other mutations
in the rRNA or in other ribosomal proteins.128-131
With the exception of K42R, mutations in S12 are
associated with a hyperaccurate phenotype that can
lead to dependence on streptomycin, which stabilizes
the ram state of the A site. The streptomycindependent phenotype can be relieved by ram mutations in proteins S4 and S5 and located at the
interface of the two proteins.127-129,132 Other mutations altering rRNA, EF-Tu, or S12 itself can also
compensate for streptomycin dependence,128 indicating that all three components are involved in the
conformational stability of the A site. Restrictive
mutations located in 50S subunit components, including truncation of the C terminus of L6 or substitution of G2661, have also been associated with
resistance to various aminoglycosides.126,131 Such
mutations do not affect the binding of aminoglycosides but likely compensate for the ram phenotypic
effect of these drugs by increasing the affinity of the
EF-Tu for 50S. Finally, mutations in the N-terminal
half of S5, which contact helix 34, can confer resistance to spectinomycin by destabilizing the network
of interactions in the 30S subunit, allowing the head
region to move even in the presence of the drug.10,133
3.3. Enzymatic Drug Modification

3.3.1. Aminoglycoside Adenylyltransferases
Aminoglycoside adenylyltransferases, with only 10
enzymes identified to date, include examples of both
chromosomally encoded and plasmid-encoded enzymes. In Gram-negative organisms, the ant(2) and
ant(3) genes encoding adenylyltransferases are often
identified on mobile genetic elements in resistant
organisms. In Gram-positive organisms, the ant(4),
ant(6), and ant(9) genes are also found on plasmids
or integrated into transposons.134 These enzymes
catalyze the reaction between Mg-ATP and aminoglycoside to form the O-adenylylated aminoglycoside and the magnesium chelate of inorganic pyrophosphate (Figure 6). Enzymes that regioselectively
adenylylate the 6 and 3 positions and the 9 and 3
positions, of the atypical aminoglycosides streptomy-
cin and spectinomycin, respectively, have been identified. From a clinical perspective, the reactions
catalyzed by the ANT(2) and ANT(4) are of most
significance and have been the most thoroughly
mechanistically and structurally studied.
The earliest mechanistic studies of the adenylyltransferases were those of Lombardini135 and
Northrop136-139 on the E. coli ANT(2). Both found
the kinetic mechanism to be sequential, requiring
both substrates to be present before catalysis could
take place, and Lombardini suggested an ordered
mechanism of substrate binding, with nucleotide
(ATP) binding before aminoglycoside. In a detailed
series of kinetic investigations, Northrop argued for
this same order of substrate addition but added the
ordered release of inorganic pyrophosphate followed
by the rate-limiting release of the adenylylated
aminoglycoside. The slow release of this final product
makes the kinetic mechanism Theorell-Chance. The
enzyme exhibits comparable activity with all nucleotide triphosphates and even their deoxy derivatives.
It also shows activity with a broad array of 4,6disubstituted substrates, but the relative V/K values
for these substrates vary by a factor of 4000. Interestingly, there is a significant positive correlation
between the enzymatic V/K values for aminoglycoside
substrates and the MIC values for these aminoglycosides in strains expressing ANT(2).140
The S. aureus ANT(4) has also been studied in
significant detail. The enzyme was initially identified
from a kanamycin- and gentamicin-resistant clinical
strain141,142 and subsequently shown to have activity
with a large number of aminoglycosides possessing
either 4- or 4-hydroxyl substituents.143 The kinetic
mechanism is sequential, and the stereochemistry at
the R-phosphorus atom of ATP has been shown to
undergo inversion during turnover,144 suggesting that
adenylyl transfer occurred via a direct displacement
of the leaving group, inorganic pyrophosphate, by the
nucleophilic hydroxyl group of the aminoglycoside.
In 1993, Holden reported the three-dimensional
crystal structure of the enzyme145 and subsequently
reported the structure of the ANT(4)-kanamycinAMg-AMPCPP (R,-methylene-ATP) ternary complex.146 These structures were determined using a
thermostable mutant of the wild-type enzyme, and
this was both the first structure of an aminoglycosidemodifying enzyme and the first structure of an
aminoglycoside in complex with a modifying enzyme.
The structure of the complex revealed an unusual
dimeric arrangement of monomers with obvious Nand C-terminal domains of approximately equal size
(Figure 7). In general, residues from the N-terminal
domain interact with the nucleotide, while those of
the C-terminal domain interact with the aminoglycoside. The two bound nucleotides are far apart, and
the majority of interactions are between the triphosphate moiety and the enzyme, suggesting a
structural basis for the lack of nucleotide triphosphate specificity. The two bound kanamycin A molecules are as close as 3.5 . There are at least four
negatively charged side chains of aspartate and
glutamate residues that interact with the aminoglycoside, and one of these, glutamate 145, appears
Figure 7. Structure of the Staphylococcus aureus

ANT(4)-AMP-PNP-kanamycin A complex. The monomers making up the active dimer are shown as blue and
yellow ribbons. AMP-PNP is shown in stick representation
and colored by atom type (C, gray; N, blue; O, red; P, pink).
Kanamycin is shown in stick representation and colored
by a different atom type (C, green; N, blue; O, red).
Coordinates were obtained from the Protein Data Bank
(1KNY).
positioned to act as a general base to deprotonate the

4-hydroxyl group. The distance from the 4-hydroxyl
to the R-phosphorus atom of AMPCPP is 5.0 ,
suggesting to these authors that a precatalytic conformational change must occur to allow for the
nucleophilic attack on the nucleotide. Alternatively,
the use of the nonreactive nucleotide might not allow
for the appropriate positioning of the nucleotide, or
contacts in the crystal do not allow this positioning
to be obtained. The 4-hydroxyl group and pyrophosphate moiety are positioned for a direct, in-line
attack at the R-phosphorus atom of AMPCPP. Due
to the symmetry of kanamycin A and the resolution
at which the structure was determined (2.5 ), it was
not possible to unambiguously distinguish between
two conformations in which either the 4- or 4hydroxyl groups would be adenylylated.
This last problem was recently solved by the
assignment of the regioselectivity of the reaction
using NMR.147 In this report, the exclusive monoadenylylation of kanamycin A was demonstrated, as
was the exclusive 4-regiospecificity. A priori, both the
4- and 4-hydroxyl groups in this symmetric molecule could have been adenylylated, and there are
examples of ANT(4,4) isozymes that, when presented with a substrate lacking the 4-hydroxyl group
(dibekacin), adenylylate at the 4 position.148 Finally,
using sensitive 18O kinetic isotope effect methods and
a poor substrate, m-nitrobenzyl triphosphate, the
transition state for the enzymatic 4-adenylylation of
kanamycin A was shown to be associative.149
3.3.2. Aminoglycoside Phosphotransferases

Aminoglycoside phosphotransferases catalyze the
regiospecific transfer of the -phosphoryl group of
ATP to one of the hydroxyl substituents present on
aminoglycoside (Figure 8). They represent a large
Figure 8. Reaction catalyzed by aminoglycoside O-phosphotransferases. The reaction shown is that catalyzed by APH(3)
catalyzing the MgATP-dependent 4-O-phosphorylation of kanamycin A.
class of aminoglycoside-modifying enzymes and are

particularly relevant to clinical resistance to aminoglycosides in Enterococcal and Staphylococcal species. The vast majority of the biochemical and structural work on this family has focused on the plasmidencoded APH(3)-IIIa enzyme from Enterococcus
faecalis, as will this discussion. The 264 amino acid
protein was overexpressed, purified, and shown to
catalyze phosphorylation of a broad spectrum of
aminoglycosides,150 as expected by the broad range
of aminoglycosides to which the original clinical
strain was resistant (tobramycin, a 4,6-disubstituted
aminoglycoside lacking a 3-hydroxyl substituent, is
not a substrate for the enzyme, but rather a potent
inhibitor). The enzyme binds substrates tightly, with
steady-state affinities in the low micromolar range,
and the kinetic parameter for aminoglycosides which
correlated with the MIC values was V/K. Enzymecatalyzed generation of phosphorylated kanamycin
A followed by one- and two-dimensional NMR analysis revealed that the proposed 3-regioselectivity,
based on the resistance phenotype, was correct.
Several extensions of these regioselectivity studies
have been performed. On the basis of the resistance
phenotypes observed for a battery of aminoglycosides,
it was proposed that for aminoglycosides lacking a
3-hydroxyl substituent, such as the 4,5-substituted
lividomycin A, phosphorylation could occur on the 5hydroxyl substituents of the ribose ring.151 Using ,bidentate Cr-ATP as a paramagnetic probe, the
conformations of amikacin, a 4,6-disubtituted aminoglycoside, and butirosin A, a 4,5-disubstituted
aminoglycoside, bound to the APH(3)-IIIa were
probed.152 The conformations of the two types of
aminoglycosides were quite different, and the modeled conformations allowed only the 3-hydroxyl substituents of amikacin to approach the -phosphate
of ATP, whereas the conformation of enzyme-bound
butirosin allowed both the 3- and 5-hydroxyl substituents of butirosin to approach the -phosphate of
ATP. These results were confirmed and extended,
and the exclusive monophosphorylation of the 3hydroxyl substituents of 4,6-disubstituted aminoglycosides was demonstrated. Those 4,5-disubstituted
aminoglycosides containing both 3- and 5-hydroxyl
substituents were shown to be rapidly monophosphorylated and subsequently bis-phosphorylated.153
These results reveal that the APH(3)-IIIa has a
remarkable ability to bind a large number of structurally distinct aminoglycosides and catalyze their
phosphorylation with a range of regioselectivities.
The steady-state kinetic mechanism was shown to
be sequential on the basis of the intersecting pattern
of lines observed in the reciprocal plot, arguing

against mechanisms invoking a phosphoenzyme intermediate.154 ATP binds first to the enzyme followed
by aminoglycoside, since tobramycin, a potent deadend inhibitor, exhibits linear uncompetitive inhibition
versus ATP and linear competitive inhibition versus
kanamycin A. On the basis of product inhibition
studies and the uncompetitive substrate inhibition
exhibited by aminoglycosides versus ATP, the ordered release of phosphorylated drug followed by the
slow rate-limiting release of ADP was proposed. This
corresponds to a Theorell-Chance-type kinetic mechanism, where the binding of the aminoglycoside to
the E-ADP complex results in the observed uncompetitive substrate inhibition. Strong support for this
Theorell-Chance kinetic mechanism came from a
subsequent study, employing viscosity variation,
solvent kinetic isotope effect measurements, and the
use of -thio-ATP.155
The three-dimensional structure of the APH(3)IIIa-ADP complex was reported in 1997.156 Although
the selenomethionine substituted enzyme had been
crystallized in the presence of Mg-ATP, the electron
density due to the bound nucleotide was only compatible with a bis-magnesium-chelated-ADP complex.
This suggests that, during the month-long crystallization, hydrolysis of ATP occurred and that the
more tightly bound ADP product complex was observed. The enzyme existed as a doubly disulfidebonded dimer in the crystal, confirming earlier
results showing the enzyme to be active as both a
monomer and dimer, and with this equilibrium being
affected by the presence of disulfide reducing agents.150
Each 263 amino acid monomer, lacking only the
N-terminal methionine residue, folded into a 94residue N-terminal domain and a 157 residue Cterminal domain connected by a flexible 12-residue
linker (Figure 9). The bound ADP molecule was
observed in the cleft between the two domains. The
overall architecture of the APH(3)-IIIa, and in
particular the predominantly -strand N-terminal
domain, was similar to the structures previously
reported for serine/threonine protein kinases, including the catalytic subunit of the cAMP-dependent
protein kinase. Residues at the active site included
two conserved lysine residues, K33 and K44, E60,
D190, N195, and D208. K33 and D190 make no
interactions with bound ADP, but all others make
direct or water-mediated interactions with the pyrophosphate moiety of the nucleotide. These results
drove a series of site-directed mutagenesis studies
that confirmed an important role for K44 in nucleotide binding and an essential role for D190, which
Figure 9. Structure of the Enterococcus faecalis APH(3)Mg2-ADP complex. The monomer is shown composed of red
R-helices and yellow -strands. ADP is shown in stick
representation and colored by atom type (C, gray; N, blue;
O, red; P, pink), and the two magnesium atoms are shown
as purple spheres. Kanamycin is shown in stick representation and colored by a different atom type (C, green; N,
blue; O, red). Coordinates were obtained from the Protein
Data Bank (1L87).
was proposed to function as the general base in the

deprotonation of the 3- (or 5-) hydroxyl group of
bound aminoglycoside.
The three-dimensional structure of the APH(3)IIIa in complex with ADP and either kanamycin A
or neomycin B157 provided a molecular explanation
for many of the above-referenced studies and others
which probed the importance of residues in the
carboxyl terminal domain using site-directed mutagenesis.158 As expected, both aminoglycosides make
the majority of interactions with residues of the
C-terminal domain and the overwhelming majority
of these interactions are electrostatic. Thus, the
positively charged aminoglycosides interact with
E157, E160, D190, E230, D262, E262, and the
carboxyl terminus of F264. Aspartate 190 interacts
with the 3-hydroxyl, supporting its role as the
general base responsible for initiating catalysis by
deprotonating the 3-hydroxyl substituent. In the
neomycin B complex, the ribose 5-hydroxyl is pointed
toward D190, but the distance between the two is not
consistent with D190 acting as a base, in catalysis.
The majority of the interactions are between the
enzyme and the central deoxystreptamine and the
primed, 6-deoxy-6-aminoglucose ring. Two additional
subsites were identified that provide interactions to
the doubly primed rings of 4,6-substituted and 4,5substituted aminoglycosides. Although there are no
large domain movements accompanying aminoglycoside binding, residues 147-170 close over the
bound aminoglycoside and constitute a binding loop
that firmly fixes the conformation of the aminoglycoside. The conformations of the bound aminoglycosides were compared to those observed for structurally related aminoglycosides bound to the A site and
to the entire 30S ribosomal subunit. Remarkably for
these conformationally flexible tricyclic compounds,
the structures were quite similar, with root mean
squared (rms) deviations of 1.7 between neomycin
B bound to APH(3)-IIIa and paromomycin bound to

the 16S rRNA of the 30S subunit. The authors
suggest that these are low energy conformations and
that, while of obviously different character, the
kinase active site has evolved to mimic the rRNA
binding site and tempt the drug into this binding
pocket before it can find its true target-binding site.
While eukaryotic protein kinases and APH(3)-IIIa
do not exhibit high degrees of overall sequence
homology, the presence of highly conserved residues
at the nucleotide binding site and the structural
similarity of the proteins suggested that the two
proteins were evolutionarily related. Whether these
similarities extended to the functional ability of the
bacterial APH(3)-IIIa to phosphorylate proteins was
demonstrated in 1999.159 APH(3)-IIIa was shown to
be capable of phosphorylating several, but not all,
basic peptides, including the MARCKS (myristolated
alanine-rich C-kinase substrates) K and R peptides
as well as protamine and myelin basic protein at
rates that were 10-20% that of kanamycin A phosphorylation. The specificity of the activity was limited, and the enzyme was incapable of phosphorylating kemptide, histone 1, and peptide substrates for
tyrosine protein kinases. Phosphoamino acid analysis
of the product phosphopeptides revealed that the
phosphorylation occurred on the serine residue of the
MARCKS K peptide, although this peptide contains
neither a threonine nor tyrosine residue (Ac-FKKSFKL-NH2).
Although this section has focused on the E. faecalis
APH(3)-IIIa, in S. aureus a bifunctional enzyme
containing both a kinase domain and an acetyltransferase domain (see below) is of major clinical significance. This enzyme is termed AAC(6)-Ie-APH(2)Ia and, because of its dual kinase-acetyltransferase
activities, can provide resistance to the majority of
clinically useful aminoglycosides. The kinetic mechanism of the kinase reaction of the bifunctional
enzyme differs from that of APH(3)-IIIA, being rapid
equilibrium random.160 APH(2) is inactivated by the
lipid kinase inhibitor, wortmannin, but contrary to
APH(3)-IIIa is not inactivated by the ATP analogue,
5-[p-(fluorosulfonyl)benzoyl]adenosine.161,162 However, like APH(3)-IIIa, the enzyme can use protein
kinase substrates to perform phosphorylation on
serine residues.159
3.3.3. Aminoglycoside Acetyltransferases

Well over four dozen unique sequences exist for
aminoglycoside acetyltransferases. These enzymes
catalyze the acetyl-CoA-dependent N-acetylation of
one of the four amino groups of typical aminoglycosides (Figure 10). They include enzymes that acetylate the 1- and 3-amino groups of the central deoxystreptamine ring and enzymes that acetylate the 2and 6-amino groups of the primed, 6-deoxy-6-aminoglucose ring. Two distinct AAC(1) activities have been
identified in E. coli and Actinomycete strains,163,164
but their importance is minor, because the E. coli
enzyme does not modify the clinically useful aminoglycosides and the other, which exhibits a broader
substrate specificity, is not found in human pathogens.
Figure 10. Reaction catalyzed by aminoglycoside N-acetyltransferases. The reaction shown is that catalyzed by AAC(6)
catalyzing the acetyl-CoA-dependent 6-N-acetylation of kanamycin A.
There are five aminoglycoside acetyltransferases

that catalyze regioselective acetylation of the 2amino group, and they all are chromosomally encoded
and species specific. The first to be identified was the
178 amino acid AAC(2)-Ia from Providencia stuartii.165 In wild-type strains, the low level of expression
of the aac(2)-Ia gene is not sufficient to confer
aminoglycoside resistance, but the transcription of
the structural gene was shown to be subject to a
complex regulation.166 Mutations in the aac(2)-Ia
gene resulted in altered levels of peptidoglycan
O-acetylation and cell morphology,167 suggesting that
peptidoglycan acetylation might be a physiologic
function of the enzyme. The expressed and purified
enzyme was shown biochemically to exhibit broad
specificity for aminoglycosides containing 2-amino
substituents. The relative V/K values for aminoglycosides did not correlate with in vivo MIC values,168
supporting the idea that the enzyme functions physiologically in other reactions, possibly peptidoglycan
acetylation. All other members of the AAC(2) family
have been identified in mycobacterial species. They
were originally identified in rapidly growing species
of mycobacteria.169 The aac(2)-Ib gene identified in
M. fortuitum appeared to be transcriptionally silent
in its natural host, but its expression in M. smegmatis
resulted in increased MIC values for all 2-aminosubstituted aminoglycosides.170 Subsequently, other
aac(2) genes were identified in both M. smegmatis
and M. tuberculosis. The corresponding enzymes
were shown to bear 60-70% sequence identity with
the M. fortuitum enzyme but only 30-40% identity
with the P. stuartii AAC(2)-Ia. None of the encoded
proteins bore significant sequence homology (<10%)
to other proteins, including other aminoglycoside
acetyltransferases.
The M. tuberculosis AAC(2)-Ic, chromosomally
encoded by the Rv0262 gene, was expressed and
purified and shown to catalyze the acetyl-CoAdependent acetylation of a broad range of aminoglycoside substrates. In contrast to several other aminoglycoside acetyltransferases, the AAC(2)-Ic activity,
even though highest with aminoglycosides containing
a 2-amino substitutent, could also be detected with
kanamycin A and amikacin, both of which contain a
2-hydroxyl substituent,171 suggesting that this enzyme can catalyze O-acetylation. Both standard
steady-state and alternative substrate kinetics supported the ordered addition of acetyl-CoA followed
by aminoglycoside to generate a ternary complex
from which acetyltransferase chemistry ensued. Very
modest solvent kinetic isotope effects were observed,
suggesting that chemistry was not rate limiting. The
Figure 11. Structure of the Mycobacterium tuberculosis

AAC(2)-CoA-ribostamycin complex. The monomers making up the active dimer are shown as ribbons that are
colored by sequence position (N-terminal, green > yellow
> red > blue, C-terminal). CoA is shown in stick representation and colored by atom type (C, gray; N, blue;
O, red; P, pink). Ribostamycin is shown in stick representation and colored by atom type (C, green; N, blue; O, red).
(1M4G).
appearance of substrate activation at high aminoglycoside concentrations suggested that acetylated

aminoglycoside dissociated first, followed by the slow
release of CoA, and that binding of aminoglycoside
to the CoA binary complex could increase this products rate of dissociation.
The three-dimensional structures of the full-length
181 amino acid apo form of the enzyme, plus three
different ternary complexes containing bound CoA
and tobramycin, kanamycin A, or ribostamycin, were
recently reported at resolutions between 1.5 and 1.8
.172 Although lacking any sequence homology to
other aminoglycoside N-acetyltransferases, the monomer fold was nearly identical to those of the S.
marcescens AAC(3) and E. faecalis AAC(6)-Ii, identifying AAC(2)-Ic as a member of the Gcn5-related
N-acetyltransferase (GNAT) superfamily.173 The conserved features of this fold are an N-terminal R-helix,
a central, antiparallel three-stranded -sheet, and a
helix-sheet-helix at the C-terminus of the fold
(Figure 11). This fold serves to bind and orient the
phosphopantothenoyl arm of acetyl-CoA, while very
few interactions are made between the enzyme and
the adenine ring. The majority of interactions between the enzyme and aminoglycoside occur between
acidic residues in the active site (D35, D40, E82,
D152, D179, and the carboxyl terminus of W181) and

the hydroxyl and amino substituents of the central
deoxystreptamine and primed rings, either directly
or via intervening water molecules. The 2-amino
group of the bound substrate is positioned 3.8 from
the sulfur atom of bound CoA, confirming the regioselective nature of the acetylation. The C-terminal
carboxylate of W181 is hydrogen-bonded through a
series of water molecules to the 2-amino group of
bound ribostamycin, suggesting that this residue
functions as the general base responsible for activating the amine and deprotonating the zwitterionic
tetrahedral intermediate. V84 and G83 are positioned
in the beta bulge that is conserved in GNAT
superfamily members, but a model of the complex
containing acetyl-CoA suggests that a single hydrogen bond between the thioester carbonyl of the
substrate and the backbone amide nitrogen of V84
can be accommodated. The phenolic hydroxyl group
of Y126 is appropriately positioned to function as a
general acid, protonating the leaving thiolate of CoA.
Together, these structures suggest that the enzyme
binds the two substrates tightly, in an orientation
that facilitates catalysis, and provides enzyme side
chains that can function as both general base (W131)
and general acid (Y126). Thus, in spite of the strongly
favorable thermodynamics of the reaction, the enzyme assists in carbonyl polarization, base-assisted
amine protonation, and general acid-assisted breakdown of the tetrahedral intermediate.
The AAC(3) family of aminoglycoside acetyltransferases is one of the largest and includes four major
types, I-IV, based on the pattern of aminoglycoside
resistance that they confer. The first aminoglycosidemodifying enzyme to be purified to homogeneity was
the E. coli R-plasmid-encoded gentamicin acetyltransferase.174 This allowed for the first studies of the
substrate specificity of these enzymes and the establishment of a correlation between measured MIC
values and the kinetic parameter V/K, indicative of
the efficiency of substrate utilization. These early
studies revealed that the kinetic mechanism was
sequential but was dependent on the identity of the
substrate: with good substrates such as sisomicin,
the mechanism was random, while, with poor substrates such as tobramycin, the kinetic mechanism
was rapid equilibrium random, with chemistry being
rate-limiting for the overall reaction.175 The demonstration of uncompetitive substrate inhibition suggested that aminoglycosides could bind to the product
binary E-CoA complex. A subsequent detailed structure-activity relationship allowed important interactions involving the 2-, 6-, 3-, and 3-amino groups
to be defined.140 Finally, these studies made it clear
that the kinetic parameter V/K for the aminoglycoside was positively correlated with in vivo activity
against strains expressing the enzyme.
The AAC(3)-I from S. marcescens, originally identified in 1991,176 was the first aminoglycoside acetyltransferase whose three-dimensional structure was
determined.177 The enzyme-CoA complex structure
was determined at a resolution of 2.3 , allowing the
interactions between the enzyme and the product to
be identified. The monomer fold was typical of the
Figure 12. Structure of the Serratia marcescens AAC(3)CoA complex. The monomers making up the active dimer
are shown as ribbons that are colored by sequence position
(N-terminal, green > yellow > red > blue, C-terminal). CoA
is shown in stick representation and colored by atom type
(C, gray; N, blue; O, red; P, pink). Coordinates were
obtained from the Protein Data Bank (1BO4).
GNAT superfamily, with the characteristic central

antiparallel -sheet covered by the two amino terminal helices on one side and the two carboxy
terminal helices on the other (Figure 12). While the
oligomeric property of the enzyme in solution was not
discussed, the enzyme crystallizes with a dimer in
the asymmetric unit, suggesting that, like most
GNAT superfamily members, the dimer is functionally active. Unfortunately, no structures containing
bound aminoglycosides were reported, leaving the
discussion of how these enzymes regioselectively
catalyze acetylation open. The recent studies of the
AAC(6) subgroup of acetyltransferases have allowed
this issue to be resolved.
As probed by the structural data of bound aminoglycosides to the 30S ribosomal subunit,5,7,9,10 the
6-amino group plays an important role in target
binding and the subsequent antibacterial activity of
the drug. This substituent is thus not surprisingly
the target of one of the major classes of aminoglycoside-modifying enzymes, the AAC(6) class, which
includes more than 25 members. The most common
pattern of resistance associated with the production
of these enzymes (type I) includes the majority of the
useful aminoglycosides except the mixture of gentamicins. Among this subclass, three enzymes have
been extensively studied, the two species specific and
chromosomally encoded AAC(6)-Ii and AAC(6)-Iy
from E. faecium and Salmonella enterica, respectively, and the widespread, plasmid-encoded, bifunctional AAC(6)-Ie-APH(2).
In E. faecium the chromosomally encoded AAC(6)Ii is, in part, responsible for the intrinsic low-level
resistance to aminoglycosides of the species. The first
structural studies of aminoglycoside binding to
AAC(6)-Ii involved the use of NMR spectroscopy and
molecular modeling to determine the conformations
of two aminoglycosides, the 4,6-disubstituted deoxystreptamine isepamicin and the 4,5-disubstituted
butirosin A, in the active site of the enzyme.178 These
data showed that, in the ternary AAC(6)-Ii-CoAisepamicin complex, the drug can adopt two distinct
conformations, which are able to interconvert. The
structure of AAC(6)-Ii with either the substrate
acetyl-CoA179 or the product CoA180 bound at the
active site was subsequently reported. The overall
Figure 13. Structure of the Enterococcus feacium

AAC(6)-CoA complex. The monomers making up the
active dimer are shown as ribbons that are colored by
sequence position (N-terminal, green > yellow > red >
blue, C-terminal). CoA is shown in stick representation and
colored by atom type (C, gray; N, blue; O, red; P, pink).
(1N71).
fold of the AAC(6)-Ii monomer revealed that the

enzyme was a member of the GNAT superfamily
characterized by a structurally conserved core (see
above). The shape of the monomer can be likened to
a V, with the acetyl-CoA binding site positioned
between the two arms. Most of the interactions
formed between the enzyme and acetyl-CoA involved
the main chain atoms, and only the side chains of
K149 and T89 interact with the substrate. Although
the monomer fold was quite similar to those of other
GNAT superfamily members, the structure of the
AAC(6)-Ii dimer revealed a significant diversity in
the dimer assembly (Figure 13). The dimeric enzyme
exhibits broad substrate specificity for acyl donors
and aminoglycosides containing a 6-amino group.181
Initial velocity and inhibition studies performed with
desulfo-CoA are consistent with an ordered sequential kinetic mechanism with acetyl-CoA binding first
and CoA released last.182 This result is consistent
with the location of the acetyl-CoA binding site in
the bottom of the active site. The rate-limiting steps
of the reaction were explored by solvent viscosity and
solvent isotope effects. The results showed that
diffusion-controlled events (substrate binding and/
or product release) were the rate-limiting steps of the
reaction rather than chemistry.182 The potential
catalytic roles of several residues located in the
AAC(6)-Ii active site were investigated kinetically
using different mutant forms of the enzyme.183 These
studies showed that none of the residues mutagenized (Q72, H74, L76, and Y147) perturb the
structural integrity of the enzyme and none are
involved in general base or acid catalysis. However,
the results suggest that Q72 may play a role in
aminoglycoside recognition and orientation in the
active site and that the amide group of L76 is
involved in transition state stabilization. Another
mutant form of AAC(6)-Ii, W104A, which does not
form a dimer in solution, was also produced to
investigate subunit cooperativity in the AAC(6)-Ii
dimer that was suggested by the partial and mixed
inhibition kinetic patterns.182 The unusual inhibition
patterns were alleviated when the mutant monomeric form of the enzyme was used, and isothermal
calorimetry (ITC) analysis of aminoglycoside binding
to WT AAC(6)-Ii revealed that two nonequivalent
binding sites exist in the dimer, supporting a subunit
cooperativity. The structural homology between eucaryotic histone acetyltransferases and AAC(6)-Ii, as
well as the relatively inefficient aminoglycoside
acetyltransferase activity displayed by AAC(6)-Ii
(kcat/Km 104 M-1 s-1) and the lack of correlation
between V/K values for aminoglycosides and
MIC values, has led the authors to investigate the
ability of the enzyme to acetylate other substrates.
AAC(6)-Ii is capable of acetylating small basic proteins such as calf histones, myelin basic protein, or
ribonuclease A.179
The S. enterica AAC(6)-Iy has been shown to confer
broad aminoglycoside resistance to a clinical isolate
in which a chromosomal deletion lead to the expression of the usually silent structural gene.184 The
purified recombinant AAC(6)-Iy was expressed in E.
coli and shown to exist as a dimer in solution. The
enzyme exhibits a high preference for acetyl-CoA as
the acyl donor (Km ) 10 M) and a strict specificity
for the aminoglycosides containing a 6-amino group;
lividomycin A, which possess a 6- hydroxyl substituent, is a powerful inhibitor of the reaction. On the
basis of their kinetic behavior, the aminoglycoside
substrates can be divided into two classes, one that
exhibits Michaelis-Menten kinetics and a second
that displays substrate activation.185 The thermodynamic parameters of substrate binding, obtained
from both fluorescence spectroscopy and ITC, showed
that both aminoglycosides and acyl-CoAs can bind
to the free enzyme and that aminoglycoside binding
to AAC(6)-Iy is strongly enthalpically driven.186
Kinetic and thermodynamic studies performed using
the wild type or mutant forms of the enzyme suggest
that C70 is essential for drug binding at the active
site. Steady-state kinetics and alternative antibiotic
kinetics indicated that the enzyme displays a sequential kinetic mechanism. Dead-end inhibition performed with desulfo-CoA and lividomycin A together
with the dependence of V and V/Kacetyl-CoA on the
identity of the aminoglycoside used argued for the
random order of substrate binding to the enzyme. The
inequality of the solvent isotope effects on D2OV and
D2O
V/K suggests that release of CoA is rate-limiting.185
The three-dimensional structure of the enzyme,
solved at 2.4 resolution by multiwavelength anomalous diffraction methods, placed AAC(6)-Iy in the
Gcn5-related N-acetyltransferase (GNAT) superfamily.187 While the tertiary structure of the monomer
is very similar to those observed for other members
of the superfamily, the structure of the active dimer
consists of a continuous 12-stranded -sheet characterized by a carboxyl terminal strand exchange
(Figure 14). This exchange has not been previously
observed with aminoglycoside N-acetyltransferases
but has been observed in the dimeric yeast histone
acetyltransferase Hpa2, which is the closest structural homologue of AAC(6)-Iy. Although not added
in the crystallization solution, CoA was found in both
active sites of the enzyme formed by a negatively
Figure 14. Structure of the Salmonella enterica AAC(6)CoA-ribostamycin complex. The monomers making up the
active dimer are shown as ribbons that are colored by
sequence position (N-terminal, green > yellow > red >
blue, C-terminal). CoA is shown in stick representation and
colored by atom type (C, gray; N, blue; O, red; P, pink).
Kanamycin is shown in stick representation and colored
by atom type (C, green; N, blue; O, red). Coordinates were
obtained from the Protein Data Bank (1S3Z).
charged channel at the dimer interface. In the

AAC(6)-Iy-CoA-ribostamycin ternary complex, described in the same paper, ribostamycin is stacked
between the aromatic rings of W22 and Y66, and the
primed and central rings make contacts with the side
chains of E79, D115, and E136. The 6-amine of the
antibiotic is 3.4 away from the sulfur atom of CoA,
consistent with a direct nucleophilic attack of the
amine on the thioester. D115 was proposed to function as the general base via an intervening water
molecule between D115 and the 6-amino group.
Similarly, it was proposed that the protonation of the
thiolate after collapse of the tetrahedral intermediate
occurs using a water molecule accessible to bulk
solvent. The comparison of the two ternary complexes
of CoA and ribostamycin bound to AAC(6)-Iy or
AAC(2)-Ic provided a molecular basis for the regioselective activity of these enzymes. In the AAC(6)Iy-ribostamycin complex, the central and primed
rings make contacts with N-terminal elements of the
enzyme and the 2-amine makes an intramolecular
hydrogen bond with the 5-hydroxyl group of the
ribose ring orientating the 6-amino group toward
acetyl-CoA. In the AAC(2)-Ic complex, the primed
and central rings make contacts with C-terminal
elements of the enzyme and the 6-amino group
makes an intramolecular hydrogen bond with the 2hydroxyl group of the ribose ring, which is rotated
almost 180 relative to that of the AAC(6)-Iy complex. This change results in the positioning of the 2amino group toward acetyl-CoA. Finally, intimate
contacts between two dimers of recombinant AAC(6)-Iy were seen in the crystal structure, such that
the extended affinity tag-containing N-terminal extension of one dimer is bound into the active site of
an adjacent dimer. This observation led to the
demonstration that AAC(6)-Iy is also capable of
autoacetylation, acetylation of eucaryotic histones
and the human histone H3 N-terminal peptide in a
regioselective manner.187
The bifunctional AAC(6)-Ie-APH(2)-Ia, which
confers broad spectrum, high-level aminoglycoside
resistance in Enterococci and Staphylococci, differs
from the two AAC(6)s described above in its genetic
localization and catalytic mechanism.188 The struc-
tural gene of the enzyme is generally found on

transposable elements, frequently carried on R plasmids. These mobile supports account for the intergenus transfer of the resistant determinant, originally isolated from E. faecalis. The enzyme is
monomeric in solution, and the acetyltransferase
activity exhibits exceptionally broad substrate specificity for aminoglycosides including fortimcin A and
aminoglycosides possessing a hydroxyl group at the
6-position. The O-acetylated paromomycin product
was identified on the basis of its lability under basic
conditions and infrared spectrometry analysis.188 The
ability of AAC(6)-Ie to catalyze both N- and Oacetylation was attributed to the presence of a
general base that would assist in hydroxyl deprotonation. The solvent kinetic isotope effect, pH studies,
and site-directed mutagenesis identified D99 as the
active site base required for aminoglycoside O-acetylation and N-acetylation of fortimicin A.189 Finally,
a mutant form of this enzyme, exhibiting an arbekacin 4-N-acetyltransferase activity, was identified in
a methicillin-resistant strain of S. aureus in Japan.
It was proposed that the G80D mutation affects the
conformation of the protein and leads to a change in
its enzymatic regiospecificity from 6- to 4-acetylation.190 The bifunctional AAC(6)-APH(2) enzyme
has been proposed to arise by gene fusion to confer
resistance to a wider spectrum of aminoglycosides to
its bacterial hosts than either enzyme alone, and it
can develop an AAC(4) activity to more efficiently
modify arbekacin, which is widely used in Japan.
This example illustrates the remarkable ability of
bacteria to rapidly adapt to changes in antibiotic use
and selective pressure.
3.4. Origin and Prevalence

Acquired aminoglycoside resistance is primarily
due to the expression of enzymes that catalyze the
chemical modification of the drug, thus preventing,
or substantially weakening, their interaction with the
ribosome. Whereas some enzymes are encoded by
chromosomal genes specific for a bacterial genus or
species, the majority of the structural genes encoding
inactivating enzymes, both in Gram-negative and
Gram-positive bacteria, are located on transferable
genetic elements including plasmids, transposons,
and integrons, which allowed their dissemination.191-195 For these modifying enzymes, a positive
correlation between the V/K values for aminoglycosides and the MIC values obtained in vivo for the
same aminoglycosides in strains expressing one of
these enzyme is observed. Northrop has argued that,
for the greatest efficiency, an aminoglycoside-modifying enzyme would be expected to exhibit such a
correlation.140 On the contrary, there is usually no
correlation between the V/K values for aminoglycosides and the MIC values for strains expressing a
originally chromosomally encoded modifying enzyme.
This observation suggests that enzymes carried on
mobile genetic elements have evolved to be efficient
aminoglycoside-modifying enzymes and have been
disseminated for this purpose. Their presence may
thus be the result of the dissemination of the resistant determinants of producer microorganisms. On
another hand, chromosomally encoded and species

specific enzymes likely have other physiological functions in their bacterial host and are selectively
recruited to counter antibacterial agents. The possible physiological functions of some AAC(2) and
AAC(6) acetyltransferases have been discussed in
the literature;167,172,179,187 however, the physiological
substrates for such bacterial GNAT proteins have not
been unambiguously identified.
As a consequence of the influence of selective
pressure on the acquisition of transferable resistance
genes, the distribution of specific inactivating enzymes varies depending on the geographic area and
on specific aminoglycoside use.196,197 For example, the
incidence of resistance due to expression of AAC(6)-I
enzymes capable of inactivating amikacin is significantly higher in countries where this antibiotic is
used extensively, including France, Belgium, and
Greece, but is less frequent in other European
countries or in the United States.196 Because mobile
genetic supports usually harbor several antibioticresistant genes, their acquisition often results in a
multidrug resistance phenotype. Besides this genetic
linkage, the simultaneous selective pressure of various antibiotics can also be the origin of multidrug
resistance acquisition, illustrated by the higher frequency of aminoglycoside resistance in organisms
simultaneously resistant to another class of drugs.
For example, a study performed under the European
SENTRY program indicated that while the percentage of gentamicin resistance was only 7% in strains
of methicillin-susceptible S. aureus, it reached 80%
in methicillin-resistant strains.198 The emergence of
multidrug-resistant organisms expressing efflux
pumps and responsible for nosocomial infections is
another consequence of the multidrug pressure.
Finally, mechanisms of resistance such as permeability alteration and target mutation are not horizontally transferable, but they remain important and
can account for a high percentage of the resistant
population of a particular species.197 The recent
characterization of genes encoding methyltransferases that catalyze the modification of the 16S rRNA
and are located on transferable elements portends a
wide and rapid dissemination of these enzymes,
responsible for broad spectrum and high-level aminoglycoside resistance, soon.
4. Resisting Resistance
On the basis of the therapeutic revival of the
-lactam class of antibacterials upon the introduction
of formulations containing a -lactamase inhibitor
and the parent -lactam, interest in the development
of inhibitors of aminoglycoside-modifying enzymes
has increased sharply. The first report of such an
inhibitor was by Northrop, who semisynthetically
prepared the bisubstrate analogue of kanamycin B
and CoA (Figure 15).199 This exerted powerful inhibition versus the aminoglycoside acetyltransferase,
exhibiting a Ki value of 9 nM, and did not restore
sensitivity to aminoglycosides in strains expressing
the N-acetyltransferase, undoubtedly because the
compound was incapable of penetrating the bacterial
envelope. The ability of the natural product 7-hy-
Figure 15. Structure of the bisubstrate analogue described by Williams and Northrop.
Figure 16. Structure of the bis-neamine dimer described

by Wong et al.
droxytropolone to inhibit the ANT(2) enzyme and

to restore sensitivity to aminoglycosides in E. coli
strains harboring the ant(2) gene was reported in
1982.200 A more recent report has attempted to
synthesize compounds that have high affinity for the
bacterial ribosomal A site but are only poorly recognized and acetylated by aminoglycoside-modifying
enzymes. An example of a successful approach used
variably cross-linked dimers of the bicyclic aminoglycoside neamine.201 This series of compounds were
shown to bind to an A site rRNA oligoribonucleotide
with high affinity (40 nM for the diaminobutane
cross-linked compound, Figure 16) and displayed
antibacterial activity (MIC ) 6.25 M). This compound was a poor substrate for both acetyltransferases AAC(6)-Ii and AAC(6)-APH(2), and in fact,
it was a powerful inhibitor of the kinase activity
(Ki ) 0.7 M) of the bifunctional enzyme. Another
approach that found some success was based on the
highly anionic nature of the aminoglycoside binding
site in the structure of the AAC(6)-Ii. Reasoning that
cationic peptides that by themselves exhibited antibacterial properties might bind to this enzyme, a
series of relatively short (12-24 residues) and highly
cationic (charge +4 to +9) peptides were synthesized
and tested as inhibitors. These peptides did indeed
inhibit the AAC(6)-Ii and both the APH (3)-IIa and
APH(2)-Ia with micromolar affinity.189
Mobashery and colleagues have synthesized a
number of aminoglycoside analogues with the potential to be poor substrates for aminoglycoside phosphotransferases, to inactivate the modifying enzyme,
modified, but yet exhibit good antibacterial properties, either alone or in combination with extant
aminoglycosides, is a good one. Another example
illustrating this concept are the neamine derivatives,
substituted in positions 1 and 6, that exhibit a higher
antibacterial activity against both neamine-sensitive
and neamine-resistant strains than the parental
compounds.205
5. Acknowledgment
Figure 17. Proposed reactions of 2-nitrokanamycin with
the APH(3) phosphotransferase. (Reprinted with permission from ref 202. Copyright 1995 American Chemical
Society).
The authors would like to thank Dr. Matt W.

Vetting for assistance in preparing Figures 7, 9, 11,
12, 13, and 14. This work has been supported by the
Unites States National Institutes of Health.
6. References
Figure 18. Proposed reaction of 3-ketokanamycin with the

APH(3) phosphotransferase. (Reprinted with permission
from ref 203. Copyright 1999 American Chemical Society.)
or to regenerate themselves after enzymatic modification. The first of these was a 2-nitro-substituted
aminoglycoside. Upon phosphorylation by APH(3),
the adjacent nitro group reduces the pK of the 2proton sufficiently that elimination of the 3-phosphate occurs. The vinylogous product is a Michael
acceptor that can be captured by an enzymic nucleophile, resulting in a novel mechanism-based inhibition of the kinase (Figure 17).202 A second strategy
involved the synthesis of a 3-keto aminoglycoside
derivative.203 In solution, the ketone is hydrated and
mimics the 3-equatorial hydroxyl group that is the
locus of phosphorylation. In fact, the compound is
phosphorylated by APH(3) but decomposes with the
elimination of phosphate to regenerate the 3-keto
group (Figure 18). The compound itself exhibits weak
antibiotic activity, as determined by its MIC value
(250 M), and does modestly sensitize bacteria harboring APH(3) to other aminoglycosides (4-8-fold
decreases in MIC values). Finally, the synthesis and
evaluation of 4,4-difluorokanamycin A and neamine
derivatives have recently been reported.204 Because
of the presence of the highly electron withdrawing
fluorine substituent adjacent to the 3-hydroxy group,
the nucleophilicity of the hydroxy group is significantly diminished. The turnover numbers for the 3phosphorylation of the difluoroaminoglycosides are
decreased by almost 3 orders of magnitude. While the
MIC values of these difluoroaminoglycosides are not
impressive, the compounds are as effective against
strains expressing APH(3) as against wild-type
strains. This is a clear demonstration that the
concept of synthesizing aminoglycoside derivatives
that are slowly modified, or incapable of being
(1) Schatz, A.; Bugie, E.; Waksman, S. A. Proc. Soc. Exp. Biol. Med.
1944, 55, 66-69.
(2) Kotra, L. P.; Haddad, J.; Mobashery, S. Antimicrob. Agents
Chemother. 2000, 44, 3249-56.
(3) Vicens, Q.; Westhof, E. Biopolymers 2003, 70, 42-57.
(4) Vakulenko, S. B.; Mobashery, S. Clin. Microbiol. Rev. 2003, 16,
430-50.
(5) Wimberly, B. T.; Brodersen, D. E.; Clemons, W. M., Jr.; MorganWarren, R. J.; Carter, A. P.; Vonrhein, C.; Hartsch, T.; Ramakrishnan, V. Nature 2000, 407, 327-39.
(6) Schluenzen, F.; Tocilj, A.; Zarivach, R.; Harms, J.; Gluehmann,
M.; Janell, D.; Bashan, A.; Bartels, H.; Agmon, I.; Franceschi,
F.; Yonath, A. Cell 2000, 102, 615-23.
(7) Fourmy, D.; Recht, M. I.; Blanchard, S. C.; Puglisi, J. D. Science
1996, 274, 1367-71.
(8) Fourmy, D.; Yoshizawa, S.; Puglisi, J. D. J. Mol. Biol. 1998, 277,
333-45.
(9) Fourmy, D.; Recht, M. I.; Puglisi, J. D. J. Mol. Biol. 1998, 277,
347-62.
(10) Carter, A. P.; Clemons, W. M.; Brodersen, D. E.; Morgan-Warren,
R. J.; Wimberly, B. T.; Ramakrishnan, V. Nature 2000, 407,
340-8.
(11) Yoshizawa, S.; Fourmy, D.; Puglisi, J. D. EMBO J. 1998, 17,
6437-48.
(12) Vicens, Q.; Westhof, E. Chem. Biol. 2002, 9, 747-55.
(13) Karimi, R.; Ehrenberg, M. Eur. J. Biochem. 1994, 226, 355-60.
(14) Pape, T.; Wintermeyer, W.; Rodnina, M. V. Nat. Struct. Biol.
2000, 7, 104-7.
(15) Puglisi, J. D.; Blanchard, S. C.; Green, R. Nat. Struct. Biol. 2000,
7, 855-61.
(16) Brodersen, D. E.; Clemons, W. M., Jr.; Carter, A. P.; MorganWarren, R. J.; Wimberly, B. T.; Ramakrishnan, V. Cell 2000,
103, 1143-54.
(17) Lynch, S. R.; Puglisi, J. D. J. Mol. Biol. 2001, 306, 1037-58.
(18) Busse, H. J.; Wostmann, C.; Bakker, E. P. J. Gen. Microbiol.
1992, 138 (Pt 3), 551-61.
(19) Yoneyama, H.; Sato, K.; Nakae, T. Chemotherapy 1991, 37, 23945.
(20) Bakker, E. P. J. Gen. Microbiol. 1992, 138 (Pt 3), 563-9.
(21) Davis, B. D. Microbiol. Rev. 1987, 51, 341-50.
(22) Mehta, R.; Champney, W. S. Antimicrob. Agents Chemother.
2002, 46, 1546-9.
(23) Edson, R. S.; Terrell, C. L. Mayo Clin. Proc. 1999, 74, 519-28.
(24) Gonzalez, L. S., 3rd.; Spencer, J. P. Am. Fam. Physician 1998,
58, 1811-20.
(25) Freeman, C. D.; Nicolau, D. P.; Belliveau, P. P.; Nightingale, C.
H. J. Antimicrob. Chemother. 1997, 39, 677-86.
(26) Craig, W. A.; Vogelman, B. Ann. Intern. Med. 1987, 106, 9002.
(27) Zhanel, G. G.; Hoban, D. J.; Harding, G. K. DICP, Ann.
Pharmocother. 1991, 25, 153-63.
(28) Gilbert, D. N. Antimicrob. Agents Chemother. 1991, 35, 399405.
(29) Karlowsky, J. A.; Zhanel, G. G.; Davidson, R. J.; Hoban, D. J. J.
Antimicrob. Chemother. 1994, 33, 937-47.
(30) Preston, S. L.; Briceland, L. L. Pharmacotherapy 1995, 15, 297316.
(31) Santucci, R. A.; Krieger, J. N. J. Urol. 2000, 163, 1076-84.
(32) Ratjen, F.; Doring, G.; Nikolaizik, W. H. Lancet 2001, 358, 9834.
(33) Moss, R. B. Chest 2002, 121, 55-63.
(34) Couch, L. A. Chest 2001, 120, 114S-117S.
(35) Schiffelers, R.; Storm, G.; Bakker-Woudenberg, I. J. Antimicrob.
Chemother. 2001, 48, 333-44.

(36) Ashtekar, D.; Duzgunes, N.; Gangadharam, P. R. J. Antimicrob.
Chemother. 1991, 28, 615-7.
(37) Gangadharam, P. R.; Ashtekar, D. R.; Flasher, D. L.; Duzgunes,
N. Antimicrob. Agents Chemother. 1995, 39, 725-30.
(38) Majumdar, S.; Flasher, D.; Friend, D. S.; Nassos, P.; Yajko, D.;
Hadley, W. K.; Duzgunes, N. Antimicrob. Agents Chemother.
1992, 36, 2808-15.
(39) Gillespie, S. H. Antimicrob. Agents Chemother. 2002, 46, 26774.
(40) Gerding, D. N. Infect. Control Hosp. Epidemiol. 2000, 21, S127.
(41) Graham, J. C.; Gould, F. K. J. Antimicrob. Chemother. 2002,
49, 437-44.
(42) Shanson, D. C. J. Antimicrob. Chemother. 1998, 42, 292-6.
(43) Nakamura, A.; Hosoda, M.; Kato, T.; Yamada, Y.; Itoh, M.;
Kanazawa, K.; Nouda, H. J. Antimicrob. Chemother. 2000, 46,
901-4.
(44) Inoue, M.; Nonoyama, M.; Okamoto, R.; Ida, T. Drugs Exp. Clin.
Res. 1994, 20, 233-9.
(45) Hotta, K.; Sunada, A.; Ikeda, Y.; Kondo, S. J. Antibiot. (Tokyo)
2000, 53, 1168-74.
(46) Cordeiro, J. C.; Reis, A. O.; Miranda, E. A.; Sader, H. S. Braz.
J. Infect. Dis. 2001, 5, 130-5.
(47) Kono, K.; Takeda, S.; Tatara, I.; Arakawa, K. Jpn. J. Antibiot.
1994, 47, 710-9.
(48) Suzuki, K. Pediatr. Int. 2003, 45, 301-6.
(49) Hamilton-Miller, J. M.; Shah, S. J. Antimicrob. Chemother. 1995,
35, 865-8.
(50) Ida, T.; Okamoto, R.; Shimauchi, C.; Okubo, T.; Kuga, A.; Inoue,
M. J. Clin. Microbiol. 2001, 39, 3115-21.
(51) Taber, H. W.; Mueller, J. P.; Miller, P. F.; Arrow, A. S. Microbiol.
Rev. 1987, 51, 439-57.
(52) Hancock, R. E. Annu. Rev. Microbiol. 1984, 38, 237-64.
(53) Bryan, L. E.; Kowand, S. K.; Van Den Elzen, H. M. Antimicrob.
Agents Chemother. 1979, 15, 7-13.
(54) Miller, M. H.; Edberg, S. C.; Mandel, L. J.; Behar, C. F.;
Steigbigel, N. H. Antimicrob. Agents Chemother. 1980, 18, 7229.
(55) Balwit, J. M.; van Langevelde, P.; Vann, J. M.; Proctor, R. A. J.
Infect. Dis. 1994, 170, 1033-7.
(56) Bryan, L. E.; OHara, K.; Wong, S. Antimicrob. Agents Chemother. 1984, 26, 250-5.
(57) Kadurugamuwa, J. L.; Lam, J. S.; Beveridge, T. J. Antimicrob.
(58) Hatch, R. A.; Schiller, N. L. Antimicrob. Agents Chemother. 1998,
42, 974-7.
(59) Putman, M.; van Veen, H. W.; Konings, W. N. Microbiol. Mol.
Biol. Rev. 2000, 64, 672-93.
(60) Murakami, S.; Nakashima, R.; Yamashita, E.; Yamaguchi, A.
Nature 2002, 419, 587-93.
(61) Yu, E. W.; Aires, J. R.; Nikaido, H. J. Bacteriol. 2003, 185, 565764.
(62) Yu, E. W.; McDermott, G.; Zgurskaya, H. I.; Nikaido, H.;
Koshland, D. E., Jr. Science 2003, 300, 976-80.
(63) Koronakis, V.; Sharff, A.; Koronakis, E.; Luisi, B.; Hughes, C.
Nature 2000, 405, 914-9.
(64) Eda, S.; Maseda, H.; Nakae, T. J. Biol. Chem. 2003, 278, 20858.
(65) Elkins, C. A.; Nikaido, H. J. Bacteriol. 2002, 184, 6490-8.
(66) Avila-Sakar, A. J.; Misaghi, S.; Wilson-Kubalek, E. M.; Downing,
K. H.; Zgurskaya, H.; Nikaido, H.; Nogales, E. J. Struct. Biol.
2001, 136, 81-8.
(67) Westbrock-Wadman, S.; Sherman, D. R.; Hickey, M. J.; Coulter,
S. N.; Zhu, Y. Q.; Warrener, P.; Nguyen, L. Y.; Shawar, R. M.;
Folger, K. R.; Stover, C. K. Antimicrob. Agents Chemother. 1999,
43, 2975-83.
(68) Aires, J. R.; Kohler, T.; Nikaido, H.; Plesiat, P. Antimicrob.
(69) Moore, R. A.; DeShazer, D.; Reckseidler, S.; Weissman, A.;
Woods, D. E. Antimicrob. Agents Chemother. 1999, 43, 465-70.
(70) Rosenberg, E. Y.; Ma, D.; Nikaido, H. J. Bacteriol. 2000, 182,
1754-6.
(71) Sobel, M. L.; McKay, G. A.; Poole, K. Antimicrob. Agents
Chemother. 2003, 47, 3202-7.
(72) Hocquet, D.; Vogne, C.; El Garch, F.; Vejux, A.; Gotoh, N.; Lee,
A.; Lomovskaya, O.; Plesiat, P. Antimicrob. Agents Chemother.
2003, 47, 1371-5.
(73) Magnet, S.; Courvalin, P.; Lambert, T. Antimicrob. Agents
Chemother. 2001, 45, 3375-80.
(74) Grkovic, S.; Brown, M. H.; Skurray, R. A. Microbiol. Mol. Biol.
Rev. 2002, 66, 671-701, table of contents.
(75) Schumacher, M. A.; Brennan, R. G. Mol. Microbiol. 2002, 45,
885-93.
(76) Jo, J. T.; Brinkman, F. S.; Hancock, R. E. Antimicrob. Agents
Chemother. 2003, 47, 1101-11.
(77) Edgar, R.; Bibi, E. J. Bacteriol. 1997, 179, 2274-80.
(78) Silva, P. E.; Bigi, F.; de la Paz Santangelo, M.; Romano, M. I.;
Martin, C.; Cataldi, A.; Ainsa, J. A. Antimicrob. Agents Chemother. 2001, 45, 800-4.

(79) Ainsa, J. A.; Blokpoel, M. C.; Otal, I.; Young, D. B.; De Smet, K.
A.; Martin, C. J. Bacteriol. 1998, 180, 5836-43.
(80) De Rossi, E.; Arrigo, P.; Bellinzoni, M.; Silva, P. A.; Martin, C.;
Ainsa, J. A.; Guglierame, P.; Riccardi, G. Mol. Med. 2002, 8,
714-24.
(81) Van Buul, C. P.; Damm, J. B.; Van Knippenberg, P. H. Mol. Gen.
Genet. 1983, 189, 475-8.
(82) van Buul, C. P.; van Knippenberg, P. H. Gene 1985, 38, 65-72.
(83) Vila-Sanjurjo, A.; Squires, C. L.; Dahlberg, A. E. J. Mol. Biol.
1999, 293, 1-8.
(84) Walsh, C. Antibiotics: actions, origins, resistance; ASM Press:
Washington, DC, 2003.
(85) Lacalle, R. A.; Tercero, J. A.; Vara, J.; Jimenez, A. J. Bacteriol.
1993, 175, 7474-8.
(86) Tercero, J. A.; Espinosa, J. C.; Lacalle, R. A.; Jimenez, A. J. Biol.
Chem. 1996, 271, 1579-90.
(87) Cundliffe, E. Annu. Rev. Microbiol. 1989, 43, 207-33.
(88) Skeggs, P. A.; Thompson, J.; Cundliffe, E. Mol. Gen. Genet. 1985,
200, 415-21.
(89) Thompson, J.; Skeggs, P. A.; Cundliffe, E. Mol. Gen. Genet. 1985,
201, 168-73.
(90) Kelemen, G. H.; Cundliffe, E.; Financsek, I. Gene 1991, 98, 5360.
(91) Holmes, D. J.; Cundliffe, E. Mol. Gen. Genet. 1991, 229, 22937.
(92) Demydchuk, J.; Oliynyk, Z.; Fedorenko, V. J. Basic Microbiol.
1998, 38, 231-9.
(93) Kojic, M.; Topisirovic, L.; Vasiljevic, B. J. Bacteriol. 1992, 174,
7868-72.
(94) Holmes, D. J.; Drocourt, D.; Tiraby, G.; Cundliffe, E. Gene 1991,
102, 19-26.
(95) Ohta, T.; Hasegawa, M. Gene 1993, 127, 63-9.
(96) Bujnicki, J. M.; Rychlewski, L. Acta Microbiol. Pol. 2001, 50,
7-17.
(97) Yokoyama, K.; Doi, Y.; Yamane, K.; Kurokawa, H.; Shibata, N.;
Shibayama, K.; Yagi, T.; Kato, H.; Arakawa, Y. Lancet 2003,
362, 1888-93.
(98) Doi, Y.; Yokoyama, K.; Yamane, K.; Wachino, J.; Shibata, N.;
Yagi, T.; Shibayama, K.; Kato, H.; Arakawa, Y. Antimicrob.
(99) Galimand, M.; Courvalin, P.; Lambert, T. Antimicrob. Agents
Chemother. 2003, 47, 2565-71.
(100) Finken, M.; Kirschner, P.; Meier, A.; Wrede, A.; Bottger, E. C.
Mol. Microbiol. 1993, 9, 1239-46.
(101) Honore, N.; Cole, S. T. Antimicrob. Agents Chemother. 1994, 38,
238-42.
(102) Honore, N.; Marchal, G.; Cole, S. T. Antimicrob. Agents Chemother. 1995, 39, 769-70.
(103) Bottger, E. C. Trends Microbiol. 1994, 2, 416-21.
(104) Musser, J. M. Clin. Microbiol. Rev. 1995, 8, 496-514.
(105) Meier, A.; Kirschner, P.; Bange, F. C.; Vogel, U.; Bottger, E. C.
Antimicrob. Agents Chemother. 1994, 38, 228-33.
(106) Powers, T.; Noller, H. F. EMBO J. 1991, 10, 2203-14.
(107) Suzuki, Y.; Katsukawa, C.; Tamaru, A.; Abe, C.; Makino, M.;
Mizuguchi, Y.; Taniguchi, H. J. Clin. Microbiol. 1998, 36, 12205.
(108) Galimand, M.; Gerbaud, G.; Courvalin, P. Antimicrob. Agents
Chemother. 2000, 44, 1365-6.
(109) Sigmund, C. D.; Ettayebi, M.; Borden, A.; Morgan, E. A. Methods
Enzymol. 1988, 164, 673-90.
(110) Sparling, P. F.; Davis, B. D. Antimicrob. Agents Chemother. 1972,
1, 252-8.
(111) Asai, T.; Zaporojets, D.; Squires, C.; Squires, C. L. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 1971-6.
(112) Pfister, P.; Hobbie, S.; Vicens, Q.; Bottger, E. C.; Westhof, E.
ChemBioChem 2003, 4, 1078-88.
(113) Sander, P.; Prammananan, T.; Bottger, E. C. Mol. Microbiol.
1996, 22, 841-8.
(114) Recht, M. I.; Douthwaite, S.; Puglisi, J. D. EMBO J. 1999, 18,
3133-8.
(115) Springer, B.; Kidan, Y. G.; Prammananan, T.; Ellrott, K.;
Bottger, E. C.; Sander, P. Antimicrob. Agents Chemother. 2001,
45, 2877-84.
(116) Brink, M. F.; Brink, G.; Verbeet, M. P.; de Boer, H. A. Nucleic
Acids Res. 1994, 22, 325-31.
(117) Johanson, U.; Hughes, D. Nucleic Acids Res. 1995, 23, 464-6.
(118) Pfister, P.; Risch, M.; Brodersen, D. E.; Bottger, E. C. Antimicrob.
(119) De Stasio, E. A.; Dahlberg, A. E. J. Mol. Biol. 1990, 212, 12733.
(120) Recht, M. I.; Douthwaite, S.; Dahlquist, K. D.; Puglisi, J. D. J.
Mol. Biol. 1999, 286, 33-43.
(121) Recht, M. I.; Puglisi, J. D. Antimicrob. Agents Chemother. 2001,
45, 2414-9.
(122) Prammananan, T.; Sander, P.; Brown, B. A.; Frischkorn, K.;
Onyi, G. O.; Zhang, Y.; Bottger, E. C.; Wallace, R. J., Jr. J. Infect.
Dis. 1998, 177, 1573-81.
(123) Dobner, P.; Bretzel, G.; Rusch-Gerdes, S.; Feldmann, K.; Rifai,
M.; Loscher, T.; Rinder, H. Mol. Cell Probes 1997, 11, 123-6.

(124) Stuy, J. H.; Walter, R. B. J. Bacteriol. 1992, 174, 5604-8.
(125) Bonny, C.; Montandon, P. E.; Marc-Martin, S.; Stutz, E. Biochim.
Biophys. Acta 1991, 1089, 213-9.
(126) Toivonen, J. M.; Boocock, M. R.; Jacobs, H. T. Mol. Microbiol.
1999, 31, 1735-46.
(127) Allen, P. N.; Noller, H. F. J. Mol. Biol. 1989, 208, 457-68.
(128) Gregory, S. T.; Cate, J. H.; Dahlberg, A. E. J. Mol. Biol. 2001,
309, 333-8.
(129) Bjorkman, J.; Samuelsson, P.; Andersson, D. I.; Hughes, D. Mol.
Microbiol. 1999, 31, 53-8.
(130) Ramakrishnan, V.; White, S. W. Nature 1992, 358, 768-71.
(131) Melancon, P.; Tapprich, W. E.; Brakier-Gingras, L. J. Bacteriol.
1992, 174, 7896-901.
(132) Zuurmond, A. M.; Zeef, L. A.; Kraal, B. Microbiology 1998, 144
(Pt 12), 3309-16.
(133) Davies, C.; Bussiere, D. E.; Golden, B. L.; Porter, S. J.; Ramakrishnan, V.; White, S. W. J. Mol. Biol. 1998, 279, 873-88.
(134) Shaw, K. J.; Rather, P. N.; Hare, R. S.; Miller, G. H. Microbiol.
Rev. 1993, 57, 138-63.
(135) Lombardini, J. B.; Cheng-Chu, M. Int. J. Biochem. 1980, 12,
427-31.
(136) Van Pelt, J. E.; Northrop, D. B. Arch. Biochem. Biophys. 1984,
230, 250-63.
(137) Gates, C. A.; Northrop, D. B. Biochemistry 1988, 27, 3834-42.
(140) Williams, J. W.; Northrop, D. B. J. Biol. Chem. 1978, 253, 590814.
(141) Le Goffic, F.; Martel, A.; Capmau, M. L.; Baca, B.; Goebel, P.;
Chardon, H.; Soussy, C. J.; Duval, J.; Bouanchaud, D. H.
(142) Le Goffic, F.; Baca, B.; Soussy, C. J.; Dublanchet, A.; Duval, J.
Ann. Microbiol. (Paris) 1976, 127, 391-9.
(143) Davies, J.; Smith, D. I. Annu. Rev. Microbiol. 1978, 32, 469518.
(144) Van Pelt, J. E.; Iyengar, R.; Frey, P. A. J. Biol. Chem. 1986,
261, 15995-9.
(145) Sakon, J.; Liao, H. H.; Kanikula, A. M.; Benning, M. M.;
Rayment, I.; Holden, H. M. Biochemistry 1993, 32, 11977-84.
(146) Pedersen, L. C.; Benning, M. M.; Holden, H. M. Biochemistry
1995, 34, 13305-11.
(147) Gerratana, B.; Cleland, W. W.; Reinhardt, L. A. Biochemistry
2001, 40, 2964-71.
(148) Schwotzer, U.; Kayser, F. H.; Schwotzer, W. FEMS Microbiol.
Lett. 1978, 3, 29-33.
(149) Gerratana, B.; Frey, P. A.; Cleland, W. W. Biochemistry 2001,
40, 2972-7.
(150) McKay, G. A.; Thompson, P. R.; Wright, G. D. Biochemistry 1994,
33, 6936-44.
(151) Trieu-Cuot, P.; Courvalin, P. Gene 1983, 23, 331-41.
(152) Cox, J. R.; McKay, G. A.; Wright, G. D.; Serpersu, E. H. J. Am.
Chem. Soc. 1996, 118, 1295-1301.
(153) Thompson, P. R.; Hughes, D. W.; Wright, G. D. Biochemistry
1996, 35, 8686-95.
(154) McKay, G. A.; Wright, G. D. J. Biol. Chem. 1995, 270, 2468692.
(155) McKay, G. A.; Wright, G. D. Biochemistry 1996, 35, 8680-5.
(156) Hon, W. C.; McKay, G. A.; Thompson, P. R.; Sweet, R. M.; Yang,
D. S.; Wright, G. D.; Berghuis, A. M. Cell 1997, 89, 887-95.
(157) Fong, D. H.; Berghuis, A. M. EMBO J. 2002, 21, 2323-31.
(158) Thompson, P. R.; Schwartzenhauer, J.; Hughes, D. W.; Berghuis,
A. M.; Wright, G. D. J. Biol. Chem. 1999, 274, 30697-706.
(159) Daigle, D. M.; McKay, G. A.; Thompson, P. R.; Wright, G. D.
Chem. Biol. 1999, 6, 11-8.
(160) Martel, A.; Masson, M.; Moreau, N.; Le Goffic, F. Eur. J.
Biochem. 1983, 133, 515-21.
(161) McKay, G. A.; Robinson, R. A.; Lane, W. S.; Wright, G. D.
Biochemistry 1994, 33, 14115-20.
(162) Boehr, D. D.; Lane, W. S.; Wright, G. D. Chem. Biol. 2001, 8,
791-800.
(163) Lovering, A. M.; White, L. O.; Reeves, D. S. J. Antimicrob.
Chemother. 1987, 20, 803-13.
(164) Sunada, A.; Nakajima, M.; Ikeda, Y.; Kondo, S.; Hotta, K. J.
Antibiot. (Tokyo) 1999, 52, 809-14.
(165) Rather, P. N.; Orosz, E.; Shaw, K. J.; Hare, R.; Miller, G. J.
Bacteriol. 1993, 175, 6492-8.
(166) Macinga, D. R.; Rather, P. N. Front. Biosci. 1999, 4, D132-40.
(167) Payie, K. G.; Rather, P. N.; Clarke, A. J. J. Bacteriol. 1995, 177,
4303-10.

(168) Franklin, K.; Clarke, A. J. Antimicrob. Agents Chemother. 2001,
45, 2238-44.
(169) Udou, T.; Mizuguchi, Y.; Wallace, R. J., Jr. Am. Rev. Respir. Dis.
1987, 136, 338-43.
(170) Ainsa, J. A.; Martin, C.; Gicquel, B.; Gomez-Lus, R. Antimicrob.
(171) Hegde, S. S.; Javid-Majd, F.; Blanchard, J. S. J. Biol. Chem.
2001, 276, 45876-81.
(172) Vetting, M. W.; Hegde, S. S.; Javid-Majd, F.; Blanchard, J. S.;
Roderick, S. L. Nat. Struct. Biol. 2002, 9, 653-8.
(173) Dyda, F.; Klein, D. C.; Hickman, A. B. Annu. Rev. Biophys.
Biomol. Struct. 2000, 29, 81-103.
(174) Williams, J. W.; Northrop, D. B. Biochemistry 1976, 15, 12531.
(175) Williams, J. W.; Northrop, D. B. J. Biol. Chem. 1978, 253, 59027.
(176) Javier Teran, F.; Alvarez, M.; Suarez, J. E.; Mendoza, M. C. J.
(177) Wolf, E.; Vassilev, A.; Makino, Y.; Sali, A.; Nakatani, Y.; Burley,
S. K. Cell 1998, 94, 439-49.
(178) DiGiammarino, E. L.; Draker, K. A.; Wright, G. D.; Serpersu,
E. H. Biochemistry 1998, 37, 3638-44.
(179) Wybenga-Groot, L. E.; Draker, K.; Wright, G. D.; Berghuis, A.
M. Structure Fold Des. 1999, 7, 497-507.
(180) Burk, D. L.; Ghuman, N.; Wybenga-Groot, L. E.; Berghuis, A.
M. Protein Sci. 2003, 12, 426-37.
(181) Wright, G. D.; Ladak, P. Antimicrob. Agents Chemother. 1997,
41, 956-60.
(182) Draker, K. A.; Northrop, D. B.; Wright, G. D. Biochemistry 2003,
42, 6565-74.
(183) Draker, K. A.; Wright, G. D. Biochemistry 2004, 43, 446-54.
(184) Magnet, S.; Courvalin, P.; Lambert, T. J. Bacteriol. 1999, 181,
6650-5.
(185) Magnet, S.; Lambert, T.; Courvalin, P.; Blanchard, J. S. Biochemistry 2001, 40, 3700-9.
(186) Hegde, S. S.; Dam, T. K.; Brewer, C. F.; Blanchard, J. S.
Biochemistry 2002, 41, 7519-27.
(187) Vetting, M. W.; Magnet, S.; Nieves, E.; Roderick, S. L.; Blanchard, J. S. Chem. Biol., in press.
(188) Daigle, D. M.; Hughes, D. W.; Wright, G. D. Chem. Biol. 1999,
6, 99-110.
(189) Boehr, D. D.; Jenkins, S. I.; Wright, G. D. J. Biol. Chem. 2003,
278, 12873-80.
(190) Fujimura, S.; Tokue, Y.; Takahashi, H.; Kobayashi, T.; Gomi,
K.; Abe, T.; Nukiwa, T.; Watanabe, A. FEMS Microbiol. Lett.
2000, 190, 299-303.
(191) Deverstein-van Hall, M. A.; Blok, H. E.; Donders, A. R.; Paauw,
A.; Fluit, A. C.; Verhoef, J. J. Infect. Dis. 2003, 187, 251-9.
(192) White, P. A.; McIver, C. J.; Rawlinson, W. D. Antimicrob. Agents
Chemother. 2001, 45, 2658-61.
(193) Culebras, E.; Martinez, J. L. Front. Biosci. 1999, 4, D1-8.
(194) Derbise, A.; Dyke, K. G.; el Solh, N. Plasmid 1996, 35, 174-88.
(195) Simjee, S.; Gill, M. J. J. Hosp. Infect. 1997, 36, 249-59.
(196) Miller, G. H.; Sabatelli, F. J.; Hare, R. S.; Glupczynski, Y.;
Mackey, P.; Shlaes, D.; Shimizu, K.; Shaw, K. J. Clin. Infect.
Dis. 1997, 24 Suppl 1, S46-62.
(197) Schmitz, F. J.; Verhoef, J.; Fluit, A. C. Eur. J. Clin. Microbiol.
Infect. Dis. 1999, 18, 414-21.
(198) Schmitz, F. J.; Fluit, A. C.; Gondolf, M.; Beyrau, R.; Lindenlauf,
E.; Verhoef, J.; Heinz, H. P.; Jones, M. E. J. Antimicrob.
Chemother. 1999, 43, 253-9.
(199) Williams, J. W.; Northrop, D. B. J. Antibiot. (Tokyo) 1979, 32,
1147-54.
(200) Allen, N. E.; Alborn, W. E., Jr.; Hobbs, J. N., Jr.; Kirst, H. A.
(201) Sucheck, S. J.; Wong, A. L.; Koeller, K. M.; Boehr, D. D.; Draker,
K.; Sears, P.; Wright, G. D. J. Am. Chem. Soc. 2000, 122, 52305231.
(202) Roestamadji, J.; Grapsas, I. L.; Mobashery, S. J. Am. Chem. Soc.
1995, 117, 80-84.
(203) Haddad, J.; Vakulenko, S.; Mobashery, S. J. Am. Chem. Soc.
1999, 121, 11922-11923.
(204) Kim, C.; Haddad, J.; Vakulenko, S. B.; Meroueh, S. O.; Wu, Y.;
Yan, H.; Mobashery, S. Biochemistry 2004, 43, 2373-83.
(205) Haddad, J.; Kotra, L. P.; Llano-Sotelo, B.; Kim, C.; Azucena, E.
F.; Liu, M.; Vakulenko, S. B.; Chow, C. S.; Mobashery, S. J. Am.
Chem. Soc. 2002, 124, 3229-37.
CR0301088
Chem. Rev. 2005, 105, 499527
499
Translation and Protein Synthesis: Macrolides

Leonard Katz* and Gary W. Ashley
Kosan Biosciences, Incorporated, 3832 Bay Center Place, Hayward, California 94545
Contents
1. Introduction
2. Classes of Macrolides
2.1. Twelve-Membered Macrolides
2.2. Fourteen-Membered Macrolides
2.3. Sixteen-Membered Macrolides
2.3.1. Tylactone Group
2.3.2. Platenolide Group
2.3.3. Mycinamicin
2.3.4. ChalcomycinNeutramycin Group
3. Clinical Uses of Macrolides
3.1. First-Generation Macrolides
3.2. Second-Generation Macrolides
3.3. Third-Generation Macrolides: Ketolides
3.4. Side Effects of Macrolides and Ketolides
4. Mode of Action
4.1. Inhibition of Translation
4.2. MacrolideRibosome Structural Studies
4.3. Inhibition of Ribosome Assembly
5. Macrolide Resistance
5.1. MLSB Resistance
5.1.1. Inducible Resistance
5.1.2. Constitutive Resistance
5.1.3. Inducible vs Constitutive
5.2. Efflux
5.3. Mutations in Ribosomal RNA
5.4. Mutations in Ribosomal Proteins
5.5. Enzymatic Inactivation of Macrolides
5.5.1. Hydrolysis of the Macrolactone
5.5.2. Phosphorylation
5.5.3. Glucosylation
6. Biosynthesis of Macrolides
6.1. Biosynthesis of the Aglycone: Modular
Polyketide Synthases
6.1.1. Erythromycin
6.1.2. Lankamycin and Oleandomycin
6.1.3. Methymycin and Pikromycin
6.1.4. Tylosin
6.1.5. Platenolide
6.1.6. Chalcomycin
6.2. Biosynthesis of Deoxysugars
6.3. Post-Polyketide Modification
6.3.1. Erythromycin and Megalomicin
499
500
500
500
501
501
501
501
501
501
501
504
506
507
507
507
507
509
509
510
510
511
511
512
512
513
513
513
513
513
514
514
514
516
516
517
517
517
518
519
519
520
* To whom correspondence should be addressed. Phone: 510-7315264. Fax: 510-732-8400. E-mail: katz@kosan.com.
6.3.3. Tylosin
6.3.4. Other Macrolides
6.4. Regulation of Macrolide Biosynthesis
7. New Macrolides and Ketolides
7.1. Chemistry
7.2. Genetic Engineering
8. Conclusions
9. Acknowledgments
10. References
520
521
522
522
522
522
523
524
524
1. Introduction
The term macrolide was originally proposed by
R. B. Woodward in 1957 as an abbreviation of
macrolactone glycoside antibiotics, a class of natural
products composed of macrocyclic lactones to which
were attached one or more deoxysugar residues.1,2
Macrolides are produced as secondary metabolites
largely from the actinomycete family of bacteria,
organisms that inhabit the soil. The first macrolide
discovered was pikromycin in 1950, followed shortly
thereafter by erythromycin, the first macrolide introduced for clinical use in human medicine.3,4 Macrolide antibiotics have been used to treat infections in
humans and animals for more than 50 years. Interest
in derivatization of erythromycin to improve its
properties started in the 1960s and has continued to
the present time. A recent chemical derivative of
erythromycin, telithromycin, was approved for clinical use in the United States in 2004.
Macrolides can be classified in a number of ways.
From a chemical viewpoint they are divided into
groups based on the number of atoms in the macrocyclic rings: 12, 14, 16, or larger, as outlined in
section 2. Each group is subdivided further on the
basis of the general structure of the lactone moiety
or sugar substitutions. From a clinical point of view
the compounds are described as first-, second-, or
third-generation macrolides, as discussed in section
3. The first-generation molecules are the natural
products that were introduced as drugs in the 1950s,
followed by the semisynthetic second-generation
compounds in the 1990s, and the semisynthetic thirdgeneration molecules in the early 2000s.
Macrolides act as antibiotics by binding to ribosomes and consequently blocking protein synthesis.
The high affinity to bacterial ribosomes, together
with the highly conserved structure of ribosomes
across virtually all of the bacterial families, gives
macrolides broad-spectrum activity. The mode of
10.1021/cr030107f CCC: $53.50 2005 American Chemical Society

Katz and Ashley
of molecules. At the genetic level, and corresponding

biochemical level, biosynthesis of the polyketide and
deoxysugar components of macrolides is now understood well enough to account not only for the structure of macrolides, but also for the structural diversity seen among this family of compounds. In section
6 we will describe our level of understanding of
biosynthesis and discuss briefly the changes to the
structure of erythromycin and other macrolides produced from manipulation of the genes responsible for
their syntheses.
2. Classes of Macrolides
Leonard Katz received his B.Sc. degree at McGill University in Microbiology
and his Ph.D. degree in Molecular Genetics at the University of California
at Santa Barbara. After a postdoctoral fellowship at the University of
California at La Jolla studying plasmid replication, he did a brief stint as
a faculty member of the Biology Department at New York University. He
entered industry in 1977 at Schering-Plough in New Jersey. In 1979 he
went to Abbott Laboratories in Illinois, where he stayed for 19 years. For
the past 6 years Dr. Katz has been at Kosan Biosciences in Hayward,
CA, where he currently holds the title Vice President of Biological Sciences.
His interest in macrolides began in 1979. At Abbott he lead the group
that isolated and sequenced the biosynthesis genes for the antibiotic
erythromycin in S. erythraea and produced the first rationally determined
genetically engineered erythromycin analogues.
This section is limited to macrolides that have been

isolated as natural products. Some are congeners of
the parent compound. In general, the congeners are
late-pathway intermediates to the final product.
2.1. Twelve-Membered Macrolides

Only two macrolide antibiotics have been identified
that contain 12-membered rings: methymycin [1]
and neomethymycin [2]. They differ in the position
of a single OH group: C-12 in methymycin vs C-14
in neomethymycin. Both contain the deoxyaminosugar D-desosamine.
Gary W. Ashley obtained his S.B. degree in Chemistry from the

Massachusetts Institute of Technology. He obtained his Ph.D. degree in
Chemistry from the University of California, Berkeley, under the guidance
of Paul A. Bartlett. After a postdoctoral fellowship in biochemistry with
JoAnne Stubbe at the University of Wisconsin, Madison, he taught
chemistry at Northwestern University. In 1993 he returned to California,
where he established the chemistry program at Kosan Biosciences, Inc.
His interests include bioorganic and natural products chemistry as well
as patent law. He was licensed to practice before the U.S. Patent and
Trademark Office in 2002 and currently divides his time at Kosan between
scientific and legal duties.
action of macrolides will be discussed in section 4

within the framework of recent structural information on macrolide-ribosome interaction. Clinical
resistance to macrolides in bacterial pathogens and
self-resistance to macrolides in the macrolide-producing actinomycetes have been well characterized and
found to share many common mechanisms. Resistance will be discussed in section 5 as a basis for the
discovery of novel, more potent compounds.
The aglycone components of macrolides are complex polyketides, partially reduced acyl chains formed
from the condensation of thioester-containing precursors in a manner common to all members of this class
Five compound families have been identified in this

class: erythromycin A [3] and its B, C, and D
congeners [4-6], pikromycin [7] and its 12-deoxy
congener narbomycin [8], megalomicin A [9] and its
congeners, oleandomycin [10], and lankamycin [11].
Erythromycin A is commonly referred to simply as
erythromycin. Megalomicin and erythromycin share
a common aglycone, 6-deoxyerythronolide B (6-dEB).
The aglycone of oleandomycin, 8,8a-deoxyoleandolide,
differs by the absence of the methyl group that is
present at C-15 in 6-dEB. All 14-membered macrolides except lankamycin contain desosamine at C-5;
lankamycin contains the neutral sugar chalcose at
that position. The neutral sugar in erythromycin A
at C-3 is L-cladinose, the 3-O-methyl derivative of
L-mycarose present at C-3 in megalomicin and in
erythromycins D and C. Megalomicin also contains
a second aminosugar, megosamine, at the 6-OH
position. Megalomicin is less potent as an antibiotic
than erythromycin A but has antiparasitic activity
through its inhibition of vesicular transport between
the medial- and trans-Golgi, resulting in the undersialylation of cellular proteins.5 The C-3 sugars in
oleandomycin and lankamycin are L-oleandrose and
L-arcanose, respectively. Pikromycin and narbomycin
contain only desosamine. The oxygen atom present
at C-3 is in the form of the ketone. Pikromycin,
discovered in 1950, therefore, is a natural ketolide,
a term first applied in the mid-1990s to describe the
new 3-descladinosyl-3-oxo derivatives of clarithromycin (6-O-methylerythromycin) that were found to
have increased antibacterial potency over erythromycin. Pikromycin, however, has weak antibacterial
activity. Erythromycin A has hydroxyl groups at both
C-6 and C-12 that are introduced by cytochrome
P450-type hydroxylases. Erythromycin congeners
Translation and Protein Synthesis
lacking the 6-OH group are weaker antibiotics.

Pikromycin lacks the 6-OH group. Oleandomycin and
lankamycin also lack the 6-OH group but are hydroxylated (lankamycin) or epoxidated (oleandomycin) at C-8. The presence of the 6- and 12-OH groups
in erythromycin A is a major source of instability
(Scheme 1). In protic solvents erythromycin A exists
as a mixture of the 9-keto form [3], the 9,12hemiketal form [3a], and the 6,9-hemiketal form [3b].
Under acidic conditions the hemiketal forms dehydrate to form enol ether derivatives [3c] and [3d],
respectively, which further degrade by reaction to
form the spiroketal derivative [3e]. Further degradation involves acid-catalyzed hydrolysis of the cladinose residue from [3e] to form erythralosamine. The
keto form [3] is the only species to have significant
antibacterial activity, whereas enol ether [3d] is a
potent agonist of the motilin receptor and is the main
cause of the gastrointestinal distress associated with
erythromycin therapy. The 12-membered macrolide
methymycin and the 14-membered macrolide pikromycin are made in the same host, Streptomyces
venezuelae.6 With the exception of the additional two
carbons in the aglycone component of pikromycin, the
two compounds are identical in structure.

Sixteen-membered macrolides represent the largest
group of macrolides. We have subdivided this group
into four subgroups on the basis of the structure of
the polyketide backbone that forms the macrolactone
after release from the corresponding polyketide synthase and before any further modification, e.g.,
glycosylation, hydroxylation, etc., takes place. Some
of these aglycones are inferred from our current
understanding of the biochemistry of complex polyketide synthesis, which is described in detail below.
2.3.1. Tylactone Group

The most commercially important member of this
group is tylosin [12], produced from the bacterium
Streptomyces fradiae and used in veterinary medicine. Tylosin contains the disaccharide D-mycaminosyl-L-mycarose at C-5 and the monosaccharide D-mycinose at C-23. Tylosin carries the C-20 aldehyde
group: oxidation of the 6(S)-ethyl side chain of the
aglycone of tylosin takes place after macrolactone is
formed. Similarly, hydroxylation of the 14(R)-methyl
side chain (to enable subsequent glycosylation) is a
post-polyketide processing step. S. fradiae also produces tylosin D [13] (formerly named relomycin) in
which the aldehyde is reduced to the alcohol. Tylosin
D is much less potent than tylosin. Conversion of
tylosin to tylosin D is carried out by an adventitious
reductase that is not associated with the tylosin
biosynthesis gene cluster.7 Tylosin has undergone
extensive chemical derivatization, and the genes for
its biosynthesis have been characterized. Another
compound in this group includes rosamicin [14],
which carries only a single sugar, desosamine, at C-5,
and the 12,13-epoxide was in human clinical trials
but not further developed into a drug. Additional
members include cirramycin and the juvenimicins.
2.3.2. Platenolide Group

This represents the largest group of 16-membered
macrolides. All members carry the 6(S)-CH3CH2CHO
side chain that is essential for antibiotic potency and
the mycaminosyl-mycarose disaccharide at C-5.
Platenolide does not have a C-14 methyl side chain
and thus offers no possibility of glycosylation on the
left side of the macrolactone. Fully elaborated compounds in this group may also contain various
acylations at C-3 and at the 4-hydroxyl of mycarose;
hence, families rather than single species of molecules are often produced from a single organism. We
have subdivided the platenolide-based group on the
basis of additional modification to the aglycone
moiety. The carbomycin B series contains no further
modifications. An example of a compound of this
subgroup is niddamycin [15]. The carbomycin A
series contains the 12,13-epoxide. The leucomycin
series is characterized by reduction of the C-9 keto
group and includes midecamycin A1 [16] and the
spiramycins (a series of three congeners) [17-19].
Midecamycin A1 and spiramycin were commercialized for human use. The spiramycins carry the
aminosugar forosamine at C-9 and various acyl
groups at C-3 or C4. Other members of the leucomycin series include the maridomycins, which carry
the 12,13-epoxide.
2.3.3. Mycinamicin
This group consists of one series of molecules, the
mycinamicins, produced by Micromonospora griseorubida. The aglycone contains a 2,3-trans double
bond, 4(R)-Me, 6(S)-Me, 14(R)-Me, 15(S)-Et. The
mycinamicins all contain the sugars desosamine at
C-5 and D-mycinose at C-21. The mycinamicins differ
from each other in the presence or absence of the 12,13-epoxide and 14(S)-OH group. An example is mycinamicin I [20]. Mycinamicins were not developed
for human use.
2.3.4. ChalcomycinNeutramycin Group

The aglycones of chalcomycin and mycinamicin
differ by the presence of a methyl group at C16 in
the latter compound. Neutramycin differs from chalcomycin [21] by the substitution of the C6-methyl
group in chalcomycin for a hydrogen atom. All
compounds in the group contain D-mycinose at C-20.
Chalcomycin has the neutral sugar D-chalcose at C-5,
the 12,13-epoxide, and an 8-hydroxyl group.
3. Clinical Uses of Macrolides

3.1. First-Generation Macrolides
The first-generation macrolides developed for clinical use were the natural products erythromycin A,
spiramycin, midecamycin A1, leucomycin, and carbomycin. These were isolated as fermentation products and required purification. Specifications of the
drugs allowed for small amounts of congeners; spiramycin was a mixture of 17-19. In general, the
compounds displayed excellent activity against Grampositive bacteria and were used initially to treat skin
caused by Staphylococcus aureus and Staphylococcus
Katz and Ashley
Figure 1. Structures of macrolides and ketolides.
epidermidis and soft tissue infections caused by S.

aureus. In the 1960s these compounds began to be
used to treat upper and lower respiratory infections
caused by Streptococcus pneumoniae or Streptococcus
pyogenes and found lesser use against staphylococci.
The enterococci are much less susceptible to macrolides. These compounds have also been used for the
treatment of Legionnaires Disease (Legionella pneumophila), Lyme Disease (Borrelia burgdorferi), syphilis (Treponema pallidum), diphtheria (Corynebacte-
Katz and Ashley
Scheme 1
rium diphtheriae), pertussis (Bordatella pertussis),

and respiratory infections caused by Moraxella cattarhalis and Mycoplasma pneumoniae. Chlamydia
pneumoniae is also susceptible to macrolides. Erythromycin has only modest activity against Gramnegative enterobacteria (e.g., Escherichia coli, Klebsiella) and no activity against Pseudomonas strains.
Sixteen-membered macrolides are somewhat more
potent against Gram negatives. Tylosin was developed to treat respiratory infections in animals, largely
caused by the Gram negatives Pasteurella multocida,
Mannheimia haemolytica, and various species of
Haemophilus. Haemophilus influenzae, a common
intracellular respiratory pathogen in children, is
treatable with macrolides.
The first-generation macrolides proved to be effective and fairly well tolerated. The most prominent
side effects of erythromycin were bitter taste and
stomach cramps, which was later found to be due to
the ability of the 8,9-anhydro-6,9-hemiketal form
([3e], Scheme 1) to mimic the effects of the hormone
motilin and stimulate gastrointestinal contractions.8
The most important drawbacks to the use of first-
generation macrolides were their short half-life and

poor oral bioavailability, prompting the requirement
for dosing three or four times a day. Despite these
weaknesses, these compounds were used successfully
for more than 25 years and were important first-line
agents for individuals with respiratory infections who
were hypersensitive to penicillin and its derivatives.
Because of their relatively low cost of production, they
are still used in Latin America, Africa, and some
parts of Asia.
3.2. Second-Generation Macrolides

The generally poor bioavailability, acid instability,
and unpredictable pharmacokinetics of the firstgeneration macrolides prompted the search for new
derivatives with improved properties. Five derivatives of erythromycin were developed and commercialized: clarithromycin (Biaxin; Abbott) [22],
dirithromycin (Dynebac; Sanofi) [23], roxithromycin
(Rulide; Aventis) [24], flurithromycin (Pierrel) [25],
and azithromycin (Zithromax; Pfizer) [26]. Miokamycin (Meiji) [27] and rokitamycin [28] were the only
16-membered second-generation compounds devel-
Scheme 2
oped for human use. Tilmicosin (Elanco) [29], a

semisynthetic derivative of tylosin, was developed for
veterinary use. Clarithromycin and azithromycin are
marketed worldwide; dirithromycin, flurithromycin,
and roxithromycin have much more limited distribution.
Clarithromycin is prepared from erythromycin A
in a short sequence of chemical transformations. The
propensity of the 6- and 12-OH groups to form
hemiketal derivatives with the 9-carbonyl together
with the higher reactivity of the 2- and 4-OH groups
on the glycosyl residues precludes efficient direct
alkylation of the 6-OH group. In a typical synthetic
sequence, erythromycin A is converted into the
9-oxime, which is then protected as an oxime ether.
The use of acetal groups to protect the oxime has
been found to be particularly convenient. Subsequent
blocking of the glycosyl hydroxyls, most simply as
trimethylsilyl ethers, provides protected derivatives
that can be efficiently methylated on the 6-OH under
basic conditions. The selectivity for alkylation of the
tertiary 6-OH group over the secondary 11-OH or
tertiary 12-OH groups is not entirely understood but
appears to be related to the unusually high acidity
of the 6-OH in erythromycin oxime derivatives.
Subsequent hydrolysis of the oxime acetal and trimethylsilyl ethers and deoximation provides clarithromycin. This six-step sequence produces clarithromycin in high yields yet significantly increases the cost
of the drug relative to erythromycin.
Azithromycin is prepared from erythromycin A
oxime by Beckmann rearrangement, for example, by
treatment with a sulfonyl chloride buffered with
aqueous sodium bicarbonate. This reaction is dependent upon trapping of the reactive Beckmann
intermediate by the 6-OH group rather than solvent
water to provide an isolable isoamide, which is
subsequently reduced to provide an intermediate
ring-expanded azalide. N-Methylation completes the
synthesis of azithromycin.
The second-generation erythromycin derivatives all
contain modifications at C6 or C9, preventing formation of the enol ether [3e] and thereby imparting
greater resistance to acid-catalyzed inactivation.
Clarithromycin is still degraded under acidic conditions to form derivatives analogous to [3d] and
descladinosyl derivatives, albeit at reduced rates
relative to erythromycin A.9-11 The five analogues
each had improved oral bioavailability and extended
half-life in plasma, enabling them to be taken orally
once (azithromycin) or twice (clarithromycin) a day.12
These compounds also exhibited enhanced tissue
penetration due to their increased lipophilicities over
the parent compound erythromycin A and hence were
effective for treatment of intracellular pathogens such
as H. influenzae.13,14 Although the search for secondgeneration macrolides was predicated on the desire
to discover compounds with expanded spectra and
improved activity, the compounds selected did not
exhibit improved activity against Gram-positive bacteria, and some, in fact, such as azithromycin, had
reduced potency.15,16 Nevertheless, they were selected
for development mainly because of their enhanced
pharmacokinetic profiles, in particular the ability to
accumulate to high levels in lung tissue. Clarithromycin is also used, generally in combination with
other antibiotics, for the treatment of gastric ulcers

caused by Helicobacter pylori and for AIDS-related
respiratory infections caused by Mycobacterium avium complex.
The second-generation 16-membered macrolides
miokamycin and rokitamycin did not show enhanced
potencies in vitro over their parent compounds,
midecamycin A1 or leucomycin A5, but did show
improved in vivo potencies in experimental animals.
These compounds are not marketed for use in the
United States.
3.3. Third-Generation Macrolides: Ketolides

Whereas the search for second-generation macrolides in the 1970s and 1980s was driven by the need
for improved stability and pharmacokinetics, the
basis for the search for third-generation compounds
shifted to macrolide resistance that had arisen suddenly and rapidly in the 1980s and 1990s. A 2001
report indicated that 23% of the S. pneumoniae
strains in the United States were resistant to macrolides.17 Macrolide resistance is described in some
detail below. The only third-generation macrolide in
clinical use as of 2004 is telithromycin (Ketek;
Aventis) [30], a 14-membered ketolide that employs
clarithromycin as the starting material. The term
ketolide is used to indicate the presence of the
3-keto group in place of the L-cladinose present in
clarithromycin and other second-generation compounds and was adopted in the early 1990s to
describe new, semisynthetic series.18,19 Removal of
the cladinosyl group from erythromycin could be
accomplished by acid treatment (after protection of
the 9-keto group), but the resulting 3-OH derivative
was found to have lost much of its potency. Moreover,
oxidation of the 3-OH to the ketone was not practical
because of subsequent 3,6-cyclization. Pikromycin,
the first natural ketolide, exhibited weak potency.
Hence, generation of the potent semisynthetic ketolides awaited the creation of clarithromycin, which
occurred in the mid-1980s. Replacement of the Lcladinose moiety with the 3-keto group in clarithromycin rendered the resulting compound a noninducer
of MLSB resistance (described below), but it also
exhibited decreased potency, likely through loss of
binding interactions to the cladinose group and/or
increased flexibility of the macrolactone. Addition of
the fused 11,12-cyclic carbamate made the macrolactone more rigid, adding potency against some
strains, and addition of the aryl alkyl side chain to
the N-11a position compensated for loss of binding
interactions to cladinose and imparted 2-10-fold
enhanced in-vitro activity against macrolide-susceptible streptococci and staphylococci over clarithromycin. Telithromycin is at least as potent in vitro as
clarithromycin against H. influenzae and the atypical
respiratory pathogens M. cattharalis, L. pneumophila, M. pneumoniae, and C. pneumoniae.20 Telithromycin is not as potent as azithromycin against H.
influenzae but accumulates in lung tissue well enough
to be clinically useful against this organism.
Telithromycin is prepared from clarithromycin
using a sequence of eight chemical steps (Scheme 2).21
Acid hydrolysis of clarithromycin provides the 3-des-
Katz and Ashley
cladinosyl derivative, which is protected at the 2OH by acetylation with acetic anhydride in the
absence of added base. Under such conditions the 2OH is unusually reactive toward acylating (but not
alkylating or silylating) reagents due to the adjacent
dimethylamino functionality. Presumably the amine
reacts with the anhydride to form an acylammonium
salt, which transfers the acyl group to the 2-OH. This
is suggested by the observation that use of acid
halides rather than anhydrides results in formation
of N-acyl-N-demethyl derivatives rather than O-acyl
derivatives. Subsequent oxidation of the 3-OH to a
ketone is followed by introduction of the 11,12-cyclic
carbamate according to the method of Baker, using
an amine prepared in several steps from 4-(3-pyridyl)imidazole and 4-bromobutylphthalimide. Treatment of the product with methanol results in removal
of the 2-acetate group and production of telithromycin. This rather lengthy sequence starts from clarithromycin, and so is 14 steps removed from erythromycin
A. This adds substantially to the cost of the drug,
and indeed, telithromycin may represent the economic limit of what is feasible in the antibacterial
market.
The most important benefit of telithromycin is its
unprecedented in-vitro potency against macrolideresistant S. pneumoniae.22,23 Resistance to telithromycin in S. pneumoniae has not yet been reported
over the 2 years that the drug has been in clinical
use in Europe. As will be described in more detail
below, the two most prominent mechanisms of acquired macrolide resistance are efflux and ribosome
methylation. Unlike azithromycin and clarithromycin, telithromycin evades the efflux pumps found in
S. pneumoniae and S. pyogenes and does not induce
ribosomal methylation associated with inducible
MLSB resistance in streptococci and staphylococci.
However, staphylococcal and S. pyogenes strains that
carry methylated ribosomes are not susceptible to
telithromycin. The differences between S. pyogenes
and S. pneumoniae with regard to ribosomal methylation and telithromycin susceptibility are further
discussed below.
A second ketolide, cethromycin (ABT-773; Abbott)
[31], also designated ABT-773 (developed by Abbott
Laboratories and not yet FDA approved), carries the
11,12-cyclic carbamate and the 3-keto group present
in telithromycin, but the aryl alkyl side chain is
attached in an ether linkage to the 6-hydroxyl group.
Synthesis of cethromycin and other 6-O-arylalkyl
ketolides has been described previously.24 As with
telithromycin, cethromycin is prepared through a
lengthy series of chemical transformations. Erythromycin A is converted into the 9-oxime and protected
as the 9,2,4-tribenzoate. This derivative is allylated
on the 6-OH using the tert-butyl carbonate of 1-(3quinolyl)-2-propenol with palladium catalysis. Subsequent deblocking of the oxime and deoximation
provides the 9-ketone, which is subjected to 11,12cyclic carbamate formation in a one-pot, four-step
sequence. Subsequent cladinose hydrolysis requires
rather forcing conditions, as 4-O-acylated cladinose
is rather resistant toward hydrolysis. Oxidation of
the resulting 3-OH group provides the 3-ketone. Final
debenzoylation of the 2-OH provides cethromycin.

Cethromycin has similar potencies as telithromycin
against the macrolide-susceptible organisms, does not
induce MLSB resistance, and is active against effluxmediated resistant and constitutive MLSB-resistant
S. pneumoniae and, unlike telithromycin, S. pyogenes.
In addition to their activity against macrolideresistant streptococci, the ketolides also have the
unexpected and unprecedented property of bactericidal activity against S. pneumoniae. All macrolides
exhibit time-dependent (12-24 h after administration), concentration-independent killing of bacteria
and are classified as bacteriostatic rather than
bactericidal agents. Ketolides, on the other hand,
exhibit concentration-dependent killing of S. pneumoniae but not S. pyogenes, S. aureus, or H. influenzae.25 Thus, although the basis is not understood,
the ketolides are considered bactericidal for S. pneumoniae only. This desirable property may forestall
the development of resistance to ketolides in these
organisms.
Although some differences between cethromycin
and telithromycin in individual pharmacokinetic
parameters have been demonstrated, the two compounds are quite comparable overall in efficacy in
experimental animals and according to initial reports
in humans as well. Telithromycin is administered
once per day, albeit at 800 mg dosing; the dosing of
cethromycin to humans has not yet been reported.
At present, the only significant reported difference
between the two compounds is the lack of efficacy in
vitro of telithromycin against constitutive MLSBbased macrolide-resistant S. pyogenes.26
3.4. Side Effects of Macrolides and Ketolides

Until recently the most significant side effects of
macrolides reported have been their ability to induce
stomach cramps in some individuals and the bitter
aftertaste of some of the compounds. High-level
interest is now focused on the occurrence of torsades
de pointes upon treatment with macrolides. Torsades
de pointes is a rare but potentially fatal ventricular
arrhythmia associated with delayed repolarization
and prolongation of the QT interval. Interactions
between macrolide antibiotics and other drugs that
prolong the QT interval have been known to cause
torsades de pointes, but recent studies have demonstrated that clarithromycin itself may induce prolongation of the QT interval and may lead directly to
ventricular arrhythmia. Azithromycin alone does not
appear to have any effect on the QT interval in rats,
but reports of QT prolongation associated with azithromycin in combination with other drugs have appeared recently.27-29 Telithromycin also induces a
modest increase in the QT interval, although smaller
than that induced by erythromycin or clarithromycin.
Subsequent studies and clinical use have also suggested unexpectedly frequent cases of temporary
visual disturbances.30 Concerns were voiced over both
potential side effects at the FDA Anti-infective Drugs
Advisory Committee hearing on telithromycin in
2001.31
4. Mode of Action
4.1. Inhibition of Translation
It has been known since their discovery that
macrolides block protein synthesis, but the molecular
details of how they arrest translation has been
uncovered only very recently. Footprinting experiments and detailed studies of macrolide resistance
over a period of more than 20 years indicated that
these compounds bind to the 50S component of
bacterial ribosomes and make specific interactions
with the 23S RNA. Early studies employing biochemical assays of the individual activities associated with
the translation processsinitiation, peptide bond formation, and translocationsled to the following conclusions: all macrolides bind in the region of domain
V of the ribosome in the peptidyltransferase center;
carbomycin and other 16-membered macrolides that
carried acyl extensions on the mycarose moiety were
found to block peptidyltransferase activity (peptide
bond formation) by binding the A site and blocking
the binding of aminoacyl tRNA; erythromycin and
other 14-membered macrolides were found to have
no effect on peptidyltransferase activity. Treatment
of bacterial cells with macrolides were found to cause
accumulation of peptidyl-tRNA, prompting workers
to suggest that the primary mechanism of action
common to all macrolides was premature ejection of
peptidyl tRNA from the ribosomes.32
Determination of the nucleotide sequences of ribosomal RNA and proteins enabled identification of the
sites on the ribosome with which macrolides interacted. Footprinting experiments (protection of nucleotides in ribosomal RNA from chemical modification
due to binding of added compounds to purified
ribosomes) demonstrated direct interaction between
macrolides and 23S rRNA.33 All macrolides, ketolides,
lincosamides, and streptogramin B protected nucleotides 2058-2062 (in domain V), but tylosin also
protected nucleotide A752 (in domain II).34 Telithromycin and cethromycin also protected A752.35,36
Erythromycin, on the other hand, protected the
domain V region but made A752 more susceptible to
chemical modification.34 These experiments, along
with determinations of the sites in the 23S ribosomal
RNA that conferred resistance by mutation or enzymatic modification, identified the precise locations on
the ribosome where macrolides were bound. Less was
known about the atoms on the macrolides themselves
that interacted with the ribosomal RNA.
4.2. MacrolideRibosome Structural Studies

Solution of the ribosomal structure at atomic
resolution with macrolides bound has clarified some
of the enigmas that have arisen associated with
macrolide action yet has raised new issues as well.
X-ray crystal structures of the 50S subunits of
ribosomes from both Haloarcula morismortui and
Deinococcus radiodurans in the presence of macrolides, ketolides, or the streptogramins were determined in the laboratories of Tom Steitz and Ada
Yonath, respectively.37-41 The structures of the Haloarcula ribosomal subunit with bound macrolides
Figure 2. View of erythromycin A bound to the 50S

subunit of the D. radiodurans ribosome, looking down the
peptide exit tunnel toward the peptidyl transferase center.
The macrolide binding site is composed of a purine-rich
pocket formed by residues from domain V (blue) with
contributions from domains II (red) and IV (magenta).
Binding of erythromycin blocks peptide formation by closing the peptide exit tunnel some distance from the peptidyl
transferase center. Residue A2058 (blue) is critical to
binding the desosamine sugar and is the site of methylation
in erm-based resistance. Residue 752 (red) is protected by
ketolide binding.
were obtained even though the Haloarcula ribosome

is not expected to be macrolide susceptible due to the
presence of G rather than A at position 2058 (E. coli
numbering). As revealed by the structure of erythromycin A bound to the Deinococcus subunit (Figure
2), the macrolide binding pocket consists of RNA from
domains II, IV, and V, with the majority of the pocket
being composed of residues from domain V. There are
some stabilizing contributions to the binding pocket
from ribosomal proteins L3, L4, L22, and L34, but
there appear to be no direct contacts between the
macrolide and these ribosomal proteins. The more
highly conserved structural region of the macrolides,
from C1 to C8, lies against a wall of mostly purine
residues from domain V. This interaction with domain V includes tight, specific interactions between
the desosamine residue and a binding pocket containing A2058 (Figure 3). The remaining region of
erythromycin interacts rather loosely with a pyrimidine-rich side of the tunnel composed of residues from
domains II and IV. The dearth of specific contacts
(seven H-bonds) between the macrolide and the
ribosome make it difficult to rationalize the very high
binding affinities observed. Nonetheless, RNA (Figure 4) and ribosomal protein mutations previously
known to affect macrolide susceptibility lie within or
near this binding pocket, thus adding confidence in
the relevance of these crystal structures. The binding
pocket lies in the peptide exit tunnel 10-15 distal
from the peptidyltransferase site; macrolide binding
appears to block progression of peptide chain upon
contact between the growing peptide chain and the
macrolide, which occurs after a small number of
elongation steps. This is in agreement with biochemical data showing the formation of very short peptides
in the presence of erythromycin. The cladinose residue of erythromycin points along the tunnel toward
Katz and Ashley
Figure 3. View of the erythromycin binding pocket on the

50S subunit of the D. radiodurans ribosome, showing the
close interactions with the desosamine residue. Regions of
high negative charge are colored red; a primary interaction
appears between the phosphate of G2505 (far right) and
the desosamine amino group. A2058 lies at the bottom of
the pocket in this view.
Figure 4. Position of 23S RNA residues at the macrolide

binding site (red) where mutation is known to lead to
erythromycin (yellow) resistance.
the peptidyltransferase site, in agreement with biochemical experiments, indicating that derivatives of
erythromycin acylated on the 4-OH of the cladinose
residue and thus extending further toward the peptidyltransferase site may interfere with peptidyltransferase activity.
While much of the macrolide binding pocket appears loose and rather devoid of specific contacts,
quite specific contacts are observed between the
desosamine sugar and the RNA in the region of
A2058, including a probably crucial charge interaction with the phosphate of G2505 (Figure 3). Not
surprisingly, quinupristin, a streptogramin B compound, also makes specific interactions with A2058.
As described in more detail in section 5, alterations
to A2058 result in macrolide resistance; methylation
at N6 of A2058 is a common mode of bacterial
resistance to macrolides, lincosamides, and the streptogramin B compounds as is mutation of A2058 to
G. Both alterations to A2058 result in loss of specific
contacts between A2058 and the 2-hydroxyl and 3dimethylamino groups of desosamine. Similarly,
chemical modifications to either the 2-hydroxyl or
the 3-dimethylamino group of macrolides has been
found to greatly reduce or destroy antibacterial
activity.
Crystal structures of ketolides bound to these 50S

subunits have also been revealing. Removal of the
3-O-cladinosyl group in the ketolides results in a
dramatic loss in potency that is compensated for by
the addition of heteroaryl groups either at the 11position (telithromycin) or the 6-position (cethromycin). In agreement with ribosomal footprinting experiments, which indicated protection of residue
A752 in domain II, the heteroaryl groups of both
telithromycin and cethromycin were found to bind to
a region of domain II adjacent to the ribosomal
binding pocket.42 In both cases, the position of the
macrolactone portion of the ketolides was observed
to be slightly shifted relative to that seen for erythromycin A, leading to the suggestion that the ketolides may tolerate some perturbations to the macrolide binding site while compensating for lost
interactions by picking up new binding from the
heteroaryl groups. However, the observed shift is
roughly within the resolution of the structures, and
such findings should not be overinterpreted at this
stage of refinement. More specific interactions were
observed between the telithromycin heteroaryl group
and the ribosome than for the cethromycin heteroaryl
group; as cethromycin has generally better in-vitro
activity against a wide range of organisms, it is clear
that such apparently improved ribosomal binding
does not necessarily translate to improved antibacterial activity.
The current X-ray crystal structures of macrolides
bound to 50S ribosomal subunits offer a snapshot of
macrolide action at the ribosome, and it is important
to remember that ribosomes are dynamic machines
and that the complete picture of macrolide activity
is likely to be significantly more complex. Macrolides
are known to act during translation, for example,
with the actual inhibited complex consisting of a
macrolide bound to a complete ribosome having a
partially completed peptide in the exit tunnel. There
may well be more specific interactions between the
macrolide and the complete ribosome-peptide complex than are observed in the current X-ray crystal
structures.
The structural studies have led to the conclusion
that binding of the macrolide to the ribosome is
sufficient to block the progression of peptide synthesis beyond the di- to hexapeptide stage. Hence,
binding alone may be sufficient for the antibiotic
action of these compounds, and the additional effects
of macrolides observed in vitro may not be required
for efficacy. On the other hand, the information
obtained from the structural work on two ribosomes
that are from clinically nonrelevant organisms does
not, at this point, provide answers to all effects seen
by macrolides on different pathogenic strains or by
different macrolides on individual strains. Azithromycin and claithromycin appear to bind in a fashion
similar to the ribosome, but azithromycin has better
potency against H. influenzae and is less potent
against S. pneumoniae and S. pyogenes.43,44 Telithromycin binds E. coli and S. pneumoniae ribosomes
with Kd ) 2-10 nM; the Kd of clarithromycin is 3050 nM, yet the two compounds have equal potencies
against S. aureus and S. pneumoniae in vitro.45
Finally, the structural studies do not themselves

provide any clues as to why the ketolides are bactericidal to S. pneumoniae but only bacteriostatic to S.
pyogenes, S. aureus, and H. influenzae. Are the
differences among the ribosomes from these different
organisms sufficient to account for the different
effects of these compounds? Is the mode of action of
these compounds entirely explained by their binding?
It is likely that differences in intracellular accumulation among the various bacteria, rather than differences in ribosomal structure, may account for all or
most of the observed differences. Nonetheless, it
would be interesting to see the molecular details of
interaction of macrolides and ketolides with ribosomes from E. coli, H. influenzae, and Gram-positive
pathogens.
4.3. Inhibition of Ribosome Assembly

Champney and co-workers have shown that
macrolides and ketolides inhibit the assembly of
the 50S ribosome unit in a number of organisms
including S. aureus, S. pneumoniae, E. coli, and H.
influenzae.46-51 Assembly of the 30S ribosome was
unaffected by these compounds. Using sucrose density gradient sedimentation analysis of ribosomes
prepared from cells pulse-labeled with 3H-uridine and
chased with an excess of the unlabeled nucleoside,
they showed that the addition of macrolides and
ketolides promotes accumulation of a 32S particle
which degrades upon continued exposure to the drug.
Using pulse labeling to analyze translation, they
determined that the IC50s for ketolides for arresting
translation and inhibiting ribosome assembly in S.
aureus were the same, ca. 10 nM, the approximate
Kd of ketolide-ribosome interactions in vitro. Not
surprisingly, inhibition of 50S subunit assembly
requires macrolide/ketolide binding. Assembly of the
50S subunit in bacteria takes place unobstructed in
cells that carry MLSB resistance in the presence of
macrolides/ketolides, suggesting that these compounds can interact with subribosomal particles.
These data also indicate that in the assembly of 50S
ribosomes, if methylation of A2058 does take place,
it must occur before the assembly of a ribonucleoprotein particle that can interact with macrolides.
Champney proposed that macrolide binding to such
a particle directly prevents the addition of one or
more ribosomal proteins to the maturing particle and
leaves segments of the rRNA in the particle exposed
to the action of cellular RNases. Whether cessation
of ribosome assembly is sufficient to explain the
bactericidal effect of ketolides in S. pneumoniae
remains to be seen.
5. Macrolide Resistance
Resistance to erythromycin was first reported in
1952, the same year erythromycin was introduced
into clinical practice, in two strains of S. aureus, the
first organism targeted by the drug.52 Resistance also
developed in most of the other organisms against
which erythromycin and other macrolides were used,
but accurate estimates of macrolide resistance in
different countries and different locations within
Katz and Ashley
countries have been difficult to determine accurately.

Though there is uncertainty about the exact extent
of resistance, there is no doubt, however, that that
resistance to macrolides is an important basis for
clinical failure of macrolide therapy.
Genes associated with macrolide resistance have
been found in all the Gram-positive pathogens for
which erythromycin and other macrolides were prescribed as well as in strains that were not targeted
by macrolides. Resistance genes are present in the
microorganisms that produce macrolides. The two
most common resistance mechanisms in the bacterial
pathogens are (1) reduced binding of the drug due to
modification of the bacterial ribosome, either through
the acquisition of a resistance gene or through
mutation in the host, and (2) efflux of macrolides
from the bacterial cell, through acquisition of a
resistance gene. Less common mechanisms include
direct inactivation of the antibiotic itself. Clinical
strains have been uncovered that carry more than a
single type of resistance. Most of the genes that
confer self-resistance in the macrolide-producing
actinomyctes are counterparts of the resistance genes
found in clinical isolates.
S-adenosyl-methionine as methyl donor and all enzymes have signature sequences characteristic of
S-AdoMet binding sites. Structures of ErmAM and
ErmC have been solved.57-59 The erm genes have
been found on high and low copy plasmids and within
transposons, often in association with other antibioticresistance genes. They are also found in the chromosomes of macrolide-producing organisms, clustered
among the genes for macrolide biosynthesis. The
ermE gene from the erythromycin-producer Saccharopolyspora erythraea has been found in commercial
preparations of the drug, causing one to wonder
whether resistance in clinical isolates originated from
the producing strain and whether it was spread
directly from use of the drug.60-62
Erm-mediated resistance exists in two forms: inducible and constitutive. In the inducible form resistance and hence ribosome methylation develop only
after the macrolide is administered to the cells. In
hosts that are constitutively resistant to macrolides,
Erm-catalyzed methylation of the ribosomes does not
require the presence of macrolides. Both inducible
and constitutive MLSB resistance require an intact
coding sequence of the erm gene.
5.1. MLSB Resistance
5.1.1. Inducible Resistance
Gram-positive cells (and E. coli) can acquire a gene

that confers high-level resistance to macrolides, lincosamides (e.g., clindamycin), and members of the
streptogramin B class of antibiotics (e.g., pristinamycin I).53 The basis for this type of resistance is
either N6-mono- or N6,N6-dimethylation of nucleotide A2058 (E. coli numbering) in 23S ribosomal
RNA. Genetic, biochemical, and structural data have
shown that the MLSB phenotype is conferred from
the overlapping binding of these molecules to domain
V making contact with A2058. It is believed, but has
yet to be proven conclusively, that either methylation
of A2058 changes the structure of the site sufficiently
so that macrolides no longer bind or the bulky methyl
groups interfere directly with the binding of the drug.
The enzyme class was named Erm for erythromycin
resistance methylase, and the genes that determine
these enzymes were designated ermA, ermB, ermC,
etc. At present, 21 classes of erm genes, some
containing six or more members, have been identified.54 These proteins are approximately 29 KDa and
show very high degrees of sequence conservation. In
vitro, Erm-mediated methylation uses 23S rRNA as
substrate and does not take place on intact ribosomes
or the 50S subunit.55 The actual substrate for methylation in bacteria has not been determined conclusively. The Erm enzymes do not appear to be
specific for their cognate substrates: all Erm enzymes tested use 23S rRNA obtained from many
species of bacteria as well as 23S RNA generated by
in-vitro transcription. Some of the enzymes, such as
ErmN, catalyze only monomethylation, whereas others, such as ErmE and ErmC, catalyze dimethylation,
but it is not known whether dimethylation takes
place through a concerted two-step process. These
latter enzymes can use monomethylated RNA as a
substrate.56 ErmAM (also called ErmB) catalyzes
either mono- or dimethylation. Methylation employs
The best-studied mechanism of inducible MLSB

resistance involves the ermC gene found in S. aureus
and was based on the initial observations that cells
resistant to erythromycin and susceptible to 16membered macrolides, lincomycins, and pristinamycin I could become resistant to the latter three classes
if treated first with small doses of erythromycin.63
The basis of inducible MLSB resistance has emerged
over the past 30 years and is summarized here.53 The
ermC-coding region is preceded by a sequence which
encodes a 19-amino acid leader peptide and the two
genes, separated by a segment consisting of 81
nucleotides, form an operon. Each gene has its own
ribosome binding site (RBS). The mRNA segment
corresponding to the leader peptide contains several
overlapping inverted repeats and, theoretically, can
assume a number of secondary structures, including
one in which the ribosome binding site of ermC is
sequestered, resulting in the inability of the ribosomes to enter the site and translate the mRNA
corresponding to the ermC gene. Under such conditions the ribosomes would not be methylated and the
cells would be susceptible to macrolides. The gene
for the leader peptide, however, whose RBS is exposed, is expressed in these cells. In the presence of
erythromycin, the model proposes that the mRNA
corresponding to the leader peptide undergoes reorganization wherein the RBS of the ermC gene is
exposed so that it can be translated, producing the
methylase that acts to generate methylated ribosomes and thereby conferring resistance to erythromycin and other MLSB antibiotics. A fascinating
model of the induction process has been developed
and is reviewed in detail by Weisblum.53 Briefly
summarized, molecules of erythromycin enter the
cells, bind to ribosomes engaged in synthesis of the
leader peptide, and cause the translation process to
stall after the ninth amino acid is introduced into the
nascent peptide, generating the peptide MGIFSIFVI

attached to tRNA. The induction model proposed an
association among erythromycin, the stalled leader
peptide on the ribosome, and the mRNA into a yet
to be understood complex that results in the change
in the secondary structure of the leader region to
permit ribosomes to bind to the RBS and translate
the ermC gene. A requirement for the stalled leader
peptide in the induction process was based on the
findings that mutations affecting the leader sequence
after Ileu-9 had no effect on inducibility, but mutations resulting in translation termination of the
leader before Ileu-9 resulted in noninducibility (failure of erythromycin to confer resistance). Mutations
further into leader segment in the region surrounding
the RBS, which themselves would destabilize the
secondary structure of the mRNA in that region,
resulted in constitutive resistance, i.e., expression of
ermC in the absence of erythromycin. Separation of
the leader peptide from the ermC gene, or introduction of a nonsense codon at a position corresponding
to residue 10, also resulted in noninducibility. Finally, within the first nine residues of the leader
peptide, amino acid substitutions of some of the
residues did not affect inducibility, but substitutions
at other residues resulted in noninducibility. These
findings demonstrated that induction required the
first nine amino acids of the leader peptide, that the
structure of the peptide was important, and that the
leader peptide must located cis to the erm gene and
be interrupted in its translation. Moreover, and most
importantly, induction depended upon the presence
of the antibiotic with the correct structuresa 14- or
15-membered macrolide that contained the neutral
sugar at C-3; 16-membered macrolides and (14membered)-ketolides are not inducers. The lincosamide celesticetin was later determined to be an
inducer.64 Derivatives of erythromycin that are devoid of antibiotic activity are also not inducers.
Within the current framework of ribosome structure and macrolide binding, it is difficult imagine the
role of the macrolide in the induction process. Erythromycin, binding in the polypeptide exit tunnel, could
allow the stalling of translation to generate the
9-residue leader peptidyl-tRNA, but other than the
tRNA component of the leader peptidyl-tRNA, neither the peptide itself nor erythromycin is in contact
with the mRNA, in particular the segment 70 nucleotides downstream that contain the RBS. If erythromycin does make direct contact with the mRNA,
it must employ different atoms than those used for
binding to rRNA in domain V. The cladinosyl moiety
is a likely candidate for such interactions since it is
required for induction. On the other hand, it has not
been ruled out that the noninducers, such as the
ketolides and 16-membered macrolides, cause the
ribosome to stall in the leader at a site different from
that caused by erythromycin so that the correct
inducer peptide is not produced. A structure of the
induction complex at atomic resolution is needed
to enable fuller understanding of inducible resistance.
Other examples of inducible MLSB resistance have
been reported. TlrA (also named ErmSF and ErmS)
in Streptomyces fradiae, the tylosin producer, is an
A2058-dimethyltransferase that is induced by tylosin

(or a precursor in the biosynthesis pathway) not
erythromycin.64 ErmSV in Streptomyces viridochromogenes NRRL 2860 is induced by either tylosin or
erythromycin.65 Interestingly, the S. fradiae host also
contains two additional 23S rRNA methyltransferases, TlrD (ErmN), an A2058 monomethyltransferase that is induced by tylosin but not erythromycin, and TlrB, a constitutive methyltransferase that
acts on G748 in domain II. Methylation by either
TlrB or TlrD alone does not confer tylosin resistance;
resistance is conferred by the two methylations acting
synergistically.66 Induction of each of these A2058
methyltransferases is believed to occur through a
translational attenuation process analogous to that
described for ErmC with different structural requirements for the leader peptide and macrolide.
An interesting variation on the mechanism for
inducible ErmK-mediated MLSB resistance in Bacillus lichenoformis has been reported. In addition to
translational attenuation observed for ErmC production, in the absence of inducer, transcription is halted
in the leader region through a rho-independent
transcription terminator. In the presence of inducer,
transcription proceeds through ermK.67
5.1.2. Constitutive Resistance

In this class the erm genes are constitutively
expressed in their hosts and thus confer resistance
to all MLSB antibiotics without the need for prior
exposure to one or another macrolide. Both MLSBinducible (resistant to erythromycin but susceptible
to tylosin) and MLSB-constitutive (resistant to erythromycin and tylosin) strains have been found in
clinical isolates of S. aureus harboring ermC. Most
of the isolates in the latter class carry mutations,
deletions, or duplications in the leader region that
are thought to destabilize the secondary structure
and allow expression of the ermC gene in the absence
of inducer. Mutation from MLSB inducible to MLSB
constitutive can also be accomplished in the laboratory by simply plating inducible cells in the presence
of tylosin and selecting for survivors.68
5.1.3. Inducible vs Constitutive

S. aureus cells that carry erm genes exhibit either
fully MLSB-inducible or MLSB-constitutive phenotypes. In inducible strains methylation of ribosomal
RNA could not be detected prior to exposure of the
cells to erythromycin.69 Hence, MLSB-inducible S.
aureus strains are almost fully susceptible to noninducers such as 16-membered macrolides and ketolides. In clinical isolates of S. pneumoniae carrying
ermAM, a wide range of susceptibility to noninducers
has been observed. The degree of resistance (minimum inhibitory concentration) to the noninducing
macrolide and ketolides has been correlated with the
degree of dimethylation of A2058 in these strains
determined before exposure to the drug.69 Addition
of erythromycin to all strains promoted increased
resistance to tylosin, resulting from additional dimethylation of A2058. Thus, the high-level resistance
of all clinical isolates of S. pneumoniae containing
ermAM to clarithromycin and azithromycin is most
likely due to full induction of the methyltransferase

by the drugs resulting in production of fully dimethylated ribosomes. The differential response of the
same strains to the noninducers tylosin and telithromycin can also be rationalized. Resistance to tylosin,
where seen, is likely due to the constitutive presence
of sufficient numbers of dimethylated ribosomes to
allow a level of protein synthesis necessary for
survival. Susceptibility to telithromycin, on the other
hand, may be due to the unique bactericidal effects
that take place upon binding of the drug to the
unmethylated (or monomethylated) ribosomes present
in these hosts.
S. pyogenes strains carrying ermAM are much more
resistant to telithromycin than their ermAM-containing S. pneumoniae counterparts but are still very
susceptible to cethromycin. It is not yet known
whether the difference between the two species is due
to the fact that telithromycin (but not cethromycin)
can induce ermAM-mediated resistance in S. pyogenes but not in S. pneumoniae. A more compelling
explanation would be that the levels of dimethylation
are greater in S. pyogenes than in S. pneumoniae and
that dimethylated S. pyogenes ribosomes, while refractory to telithromycin binding, can still bind
cethromycin with clinical efficacy.
5.2. Efflux
Decreased accumulation due to efflux in a macrolide-resistant isolate was first reported in the 1980s
in S. epidermidis and in the 1990s in S. pyogenes and
S. pneumoniae and presently accounts for a significant proportion of the macrolide-resistant S. pneumoniae strains identified.70-75 Efflux-mediated resistance is still relatively rare in S. aureus. In streptococci, macrolide efflux is mediated by the gene
product encoded by mefA, the name denoting a group
of genes encoding proteins that share >90% identity.
MefA confers resistance to 14- and 15-membered
macrolides but not 16-membered macrolides, lincosamides, or streptogramin B. Furthermore, Mef-mediated resistance is induced by the presence of clarithromycin and azithromycin but not by 16-membered
macrolides. Ketolides are poor inducers of MefA and
hence are still very potent antibacterials against
streptococci carrying this gene. Since the Mef proteins do not contain recognizable ATP binding sites
and resistance to macrolides in mefA-containing hosts
takes place in the presence of ATP-associated energy
uncouplers, Mef-mediated transport of the macrolide
is believed to be driven by a proton motive force.
Efflux in staphylococci is mediated by MsrA, a
member of the ABC superfamily that employs ATP
as the energy source for transport and is thought to
work in concert with a membrane-associated host
protein to confer resistance. MsrA confers high-level
resistance to 14- and 15-membered macrolides and
streptogramin B and weak resistance to ketolides,
and it does not confer resistance to 16-membered
macrolides and lincosamides. MsrA is induced by
clarithromycin, azithromycin, and telithromycin but
not by streptogramin B, even though the latter
compound is a substrate for MsrA-mediated transport. Nucleotide sequencing of the region upstream
Katz and Ashley
of msrA revealed a leader sequence reminiscent of

the leader upstream of ermC, suggesting that MsrA
is induced by a translational attenuation process.76
ABC transporters have also been found in some
Streptomyces hosts that produce macrolides. These
genes are located at the edges of their cognate
biosynthesis gene cluster and confer resistance to 16or 14-membered macrolides when expressed in heterologous hosts. Their roles in conferring selfresistance or in export of the macrolide during
production are not yet known.
A number of transport systems not specific for
macrolides have been identified in Gram-negative
bacteria. These tripartite pumps are members of the
RND family and are composed of an inner membrane
component, which extrudes the macrolide in exchange for a proton, a protein in the outer membrane
that may form a gated channel (pore), and a periplasmic protein that links the two membrane-associated efflux proteins. Examples include the MexABOprM system in Pseudomonas aeruginosa, the
AcrAB-TolC system in E. coli, and the acrAB-Omp2
system in H. influenzae. These systems, encoded in
the chromosomes of Gram negatives, are the primary
bases for intrinsic resistance to membered macrolides
as well as many other compounds including antibiotics such as rifampicin, novobiocin, and tetracycline.
In some hosts the RND pump genes are expressed
constitutively; in others, a mutation is required RNDmediated resistance.
5.3. Mutations in Ribosomal RNA

Bacteria contain between one and seven copies of
the operons that encode the genes for ribosomal RNA.
Mutations in domain V encoding resistance to
clarithromycin have been reported in clinical isolates
of a number of organisms, including H. pylori, S.
aureus, S. pneumoniae, H. influenzae, Neisseria gonorrheae, Mycobacterium tuberculosis, Mycobacterium
avium, and Treponenum pallidum.77-87 Several patterns of resistance were seen. Deletion of A2058 in
S. pneumoniae conferred high-level resistance to
macrolides and increased the MIC to 4 g/mL of
telithromycin, but it is not clear whether the increase
in MIC translates to resistance to the ketolide in a
clinical setting. A2058G or A2058T mutations conferred high-level resistance to all three classes of
MLSB antibiotics. A2059G mutations conferred intermediate-level resistance to macrolides and lincosamides but did not confer resistance to streptogramin B. In N. gonorrheae, a C2611T mutation (in
domain V) was identified. In H. pylori, the particular
mutation was found in each of the two copies of the
rrl gene (23S rRNA) present in the chromosome. In
S. aureus, the mutation was present in a minimum
of four of the six rrl genes present in the host and in
N. gonorrheae three of the four rrl genes. In T.
pallidum the A2058G mutation was present in both
copies of rrl. In no cases did a resistant strain carry
more than a single type of mutation, suggesting that
each mutation was introduced into a single copy of
the rrl genes, and through selection in the presence
of the drug, the mutant allele replaced all or most of
the wild-type copies of the gene in the host, likely
via a process involving recombination.
siella, and Enterobacter species) as well as in clinical

isolates of S. aureus.99,100 Currently, esterase-mediated resistance to erythromycin is rare in S. aureus
and has yet to be detected in streptococci. These
enzymes are specific for 14-membered macrolide
substrates. Two streptogramin B hydrolases, VgbA
and VgbB, have recently been identified in S. aureus.101 These enzymes do not employ macrolides as
substrates.
Mutations in domain V in E. coli strains selected

for resistance to macrolides have been mapped to
nucleotides, 2058, 2059, and 2612.88 A U2609C mutation in domain V was found in an E. coli strain
selected for resistance to telithromycin or cethromycin.89 Interestingly, this mutation increased slightly
the susceptibility of the host to erythromycin and
azithromycin.
Mutations in domain II in the vicinity of A748
conferring resistance to macrolides or ketolides have
not been reported in clinical isolates, but the U754A
mutation in the hairpin 35 segment of domain II was
found in an E. coli host selected for resistance to
telithromycin.90 A different class of laboratoryselected strains of E. coli that exhibited increased
resistance to erythromycin was found to carry mutations within a hairpin structure between nucleotides
1198 and 1247 in domain II of the 23S rRNA, close
to a segment of the RNA that encodes a pentapeptide
(E-peptide) that confers resistance to erythromycin.91
It is believed that mutation in this region of domain
II increases expression of the segment encoding the
E-peptide. Interestingly, it was found that the peptide acted only cis on ribosomes carrying the 23S
RNA harboring the domain II mutation and conferred
resistance only to erythromycin and not ketolides or
16-membered macrolides.92 By site-directed mutagenesis of the E-peptide coding region, the sequence
could be changed to permit the production of different
peptides that conferred resistance to ketolides or 16membered macrolides.93 Although this mechanism of
macrolide resistance in E. coli is not clinically relevant at the present time, these findings raise the
possibility that short peptides produced from the
rRNA as well as segments of the ribosomal RNA itself
may play a role in the binding of macrolides to
ribosomes to stop translation or, perhaps, to promote
expression of an erm gene.
Enzymes that transfer phosphate from ATP to the

2-OH of erythromycin were originally discovered in
E. coli. Members of the MphA group employ 14- and
15-membered macrolides as substrates exclusively.102,103 The MphB enzyme can phosphorylate
both 14- and 16-membered macrolides.104 Macrolide
2-phosphotransferase activity, related to MphA, was
recently detected in two clinical isolates of P. aeruginosa from hospital patients in Japan, where macrolides are used for long-term chemotherapy of P.
aeruginosa panbronchiolitis.105 A related enzyme,
MphC, has also been identified in a clinical isolate
of S. aureus.106 Expression of the mphA gene in E.
coli is regulated by an adjacent gene, mphR, whose
gene product binds to the operator-promoter region
of mphA and represses transcription. Transcription
of mphA takes place in the presence of erythromycin,
which enters the cell, binds to MphR, and removes
it from the operator-promoter.107 In this system
erythromycin is the inducer of (self-)resistance. Although the MphA-MphR resistance system has thus
far been found only in E. coli, it is reasonable to
suggest that it originated in a macrolide-producing
bacterium and that a 2-phosphatase, which would
restore antibacterial activity to 2-phosphoerythromycin, would also be uncovered in a macrolideproducing host.
5.4. Mutations in Ribosomal Proteins
5.5.3. Glucosylation
A number of clinical isolates of H. influenzae, S.

aureus, and S. pneumoniae resistant to macrolides
have been characterized to carry mutations in genes
for 50S ribosomal proteins L4 or L22.81,82,87 As
described above, these two proteins border the
polypeptide exit tunnel. Mutations in E. coli conferring resistance to erythromycin were also determined
to reside in the genes for L4 and L22 proteins.
Ribosomes from L4 mutants exhibited reduced binding of erythromycin, but the L22 mutant ribosomes
could still bind drug, indicating that the mutation
affected the structure of the tunnel such that binding
of macrolide did not block translation.94-96
Macrolide resistance mediated through 2-glucosylation has not been reported in a bacterial pathogen
but has been found in Streptomyces antibioticus, the
producer of oleandomycin.108 Extracts of several other
streptomycetes were found to contain activities that
transferred the glucose moiety from UDP-glucose to
a number of 12-, 14-, and some 16-membered macrolides, suggesting that the resistance gene spread
from a macrolide producer.109-111 In their natural
locations in the chromosome the mgt genes confer
weak resistance to macrolides on their hosts. In S.
antibioticus, the MGT gene, oleI, is accompanied by
the gene oleR, which encodes an enzyme that removes the glucose residue from 2-glucosyloleandomycin, restoring the antibacterial activity to the
compound.108 Both oleI and oleR are located in the
oleandomycin biosynthesis cluster. OleI is thought
to confer self-resistance to the host while the compound is produced intracellularly, and OleR restores
its activity during or prior to transport from the host.
It is interesting to note that the oleandomycin biosynthesis cluster does not contain an erm gene;
hence, the host makes oleandomycin-employing ribosomes that are susceptible to the drug.
5.5. Enzymatic Inactivation of Macrolides

5.5.1. Hydrolysis of the Macrolactone
Two unrelated genes, ereA and ereB, each conferring resistance to erythromycin in E. coli, were
identified on separate plasmids and shown to encode
esterases that opened the macrolactones of erythromycin and oleandomycin.97,98 These genes were subsequently identified in a number of members of other
Gram-negative bacteria (Citrobacter, Proteus, Kleb-
5.5.2. Phosphorylation
6. Biosynthesis of Macrolides
Biosynthesis of macrolides follows discrete biochemical pathways but can be viewed as taking place
in three stages: synthesis of the aglycone, synthesis
of the sugars and attachment to the aglycone, tailoring steps to produce the completed product. The
genes for the biosynthesis of the aglycone and deoxysugars and the genes for the tailoring steps are
generally clustered. Genes that confer self-resistance
are located within the biosynthesis cluster. Much of
our current understanding of macrolide biosynthesis
has come from the determination of the nucleotide
sequence of the biosynthesis genes in the 1990s.
However, the biochemical pathways for erythromycin
and tylosin were largely understood well before this
period from the analysis of compounds produced in
fermentations of mutants blocked at different steps
of the synthesis.112-114 In addition, early feeding
experiments indicated that macrolides were produced
from acetate, propionate, and butyrate, but the key
experiments demonstrating the bioconversion of compounds 5-9 carbons in length with structures representing intermediates in aglycone biosynthesis into
the aglycones of erythromycin and tylosin indicated
that biosynthesis of the macrolactone takes place
through a stepwise process.115,116
6.1. Biosynthesis of the Aglycone: Modular

Polyketide Synthases
The aglycones of macrolides are complex polyketides
that are assembled through successive decarboxylative condensations of small carboxyacyl thioesters
(e.g., malonyl CoA, methylmalonyl CoA) in a manner
resembling fatty acid biosynthesis. Each step of the
synthesis is programmed to determine the acyl unit
incorporated into the growing chain (e.g., malonyl
CoA, methylmalonyl CoA, etc.) and the degree to
which the resulting -carbonyl generated from the
condensation is reduced. In addition, the stereochemistry of the R-side chain (if present) is also programmed. Programming is carried out by the polyketide synthase (PKS) that catalyzes all the steps in
assembly of the aglycone. In general, each enzymatic
step is conducted by a discrete component of the PKS,
and as in fatty acid biosynthesis, all steps take place
with the growing acyl chain tethered to the enzyme
in a thioester linkage. Macrolide PKSs are large,
multifunctional polypeptides that can contain more
than 30 enzymatic functions, but the functions associated with a single condensation and -carbonyl
reduction cycle are present in an uninterrupted linear
sequence, commonly referred to as a module, hence
the term modular PKS. Each module is similar in
overall organization to type I fatty acid synthases.
The enzymatic functions within each module are
called domains. The domains are arranged in a linear
sequence and separated by interdomain spacer regions. The KS domain, approximately 550 amino
acids in length, encodes the -ketoacyl ACP synthase
that catalyzes the condensation between the growing
acyl chain (attached in thioester linkage to the Cys173 residue of the KS) and the extender unit tethered
to the ACP domain (acyl carrier protein) through a
Katz and Ashley
thioester linkage with the 4-phosphopantotheine

prosthetic group.117 The AT domain, ca. 300 amino
acids, encodes the acyltransferase, the component
that binds the extender acyl-CoA unit via an ester
linkage with the Ser residue in the GHSxG active
site, and transfers it to the ACP for condensation
with the nascent acyl chain. Each AT domain is
selective for the extender unit it binds and transfers
to its cognate ACP. Comparisons of the sequences of
AT domains showed that malonyl- and methylmalonyl-transferring domains each clustered with members of the same group strongly, indicating structuredetermined selectivity.118
All modules in macrolide PKSs contain KS, AT, and
ACP domains. The remaining domains determine the
extent to which the -carbonyl produced through
condensation is reduced. If the KR domain (-ketoreductase) is absent or mutated, the -keto group will
not be processed further. If the KR is present, the
-keto group is reduced to the hydroxyl. The stereochemistry of the hydroxyl group is determined by
the KR domain.119,120 The KR domains have the
GxGxxAxxxA motif for NADPH binding.121 The DH
(dehydratase) domain removes the -OH group and
a proton from the R-carbon to leave an R,-double
bond. It is not known if the DH domains remove 3(R)OH and pro-2(S) hydrogen in syn eliminations as
observed in fatty acid synthase.122 All double bonds
found in macrolides are trans. The ER (enoylreductase) domain reduces a trans double bond to the
-methylene center. ER domains contain a NADPH
binding motif.
All macrolide PKSs contain a TE (thioesterase)
domain at the C terminus of the last module that acts
to release the polyketide chain from the PKS and
cyclize it. These are referred to as TE-I domains. The
TE-I domains of the erythromycin and pikromycin
PKSs have been crystallized.123,124 Macrolide biosynthesis clusters also contain a discrete gene encoding
a short-chain thioesterase (TE-II) that play a role in
macrolide production by removing aberrant intermediates produced from improper decarboxylation of the
extender molecule.125-127
6.1.1. Erythromycin
The erythromycin PKS, 6-dEB synthase, or DEBS,
was the first modular PKS identified through sequencing of the corresponding genes.128,129 DEBS
consists of three proteins though each is thought to
exist as a head-to-head dimer in the holoenzyme.130
6-dEB is made from the successive condensations of
one propionate molecule and six molecules of methylmalonate.
The predicted domain organization of DEBS and
biosynthetic intermediates at the end of each cycle
of condensation and -carbonyl reduction is shown
in Figure 5. DEBS1 contains the loading module and
modules 1 and 2. The AT domain of the loading
domain binds propionyl CoA and transfers it to the
adjacent ACP [a]. All ACP domains of DEBS are
phosphopantetheinylated by the phosphopantetheinyltransferase SePptII, whose gene is not found in
the erythromycin biosynthesis cluster.131 The propionyl residue is then transferred to the KS domain of
Figure 5. Domain organization of DEBS and structures of proposed intermediates at the end of each condensation cycle.
Linear sequences of polypetides are shown as open arrows. Domains are shown as spheres. Color-coding indicates
components of the nascent polyketide chain programmed by particular modules. Abbreviations: ACP, acyl carrier protein;
AT, acyltransferase; DH, dehydratase; ER, enoylreducase; KR, -ketoreductase; KS, -keto acyl-CoA synthase; TE,
thioesterase.
module 1. It has been shown that propionyl CoA, and

not methylmalonyl CoA, binds directly to the loading
domain and that propionyl CoA can bind directly to
the KS1 domain in the absence of a loading module,
albeit very inefficiently.132-134 All AT domains of
modules 1-6 bind and transfer the 2(S) enantiomer
of methylmalonyl CoA to their cognate ACPs; hence,
epimerization of the R-methyl group produced after
the second, fifth, and sixth condensations must take
place, but it is not yet known how these epimerizations are controlled by the PKS.135 After the first
condensation, reduction of the -carbonyl is catalyzed
by the KR domain of module 1 to generate the
diketide intermediate b. As seen in Figure 5 the
carbon atoms of the propionyl starter and first
extender ultimately become C 11-15 of the completed aglycone. The acyl chain of b is transferred to
the KS of module 2, and condensation with the
methylmalonyl CoA extender on ACP2 generates a
triketide whose -carbonyl is reduced by the KR2
domain [c]. Direct evidence for the activities of
associated with DEBS 1 comes from the production
of the predicted triketide lactone both in vivo and in
vitro from a DEBS construct in which the TE domain
was moved from the end of module 6 to the C
terminus of DEBS 1.136-139
The next step requires interpolypeptide transfer of
the nacent acyl chain from the ACP2 of DEBS1 to
KS3 of DEBS2. Recognition sequences (linker regions) at the ends and beginnings of PKS subunits
ensure proper associations to prevent aberrant nascent chain passage.140,141 Module 3 contains a sequence that resembles a KR domain, but the conserved NADP(H) binding site is not present and,
hence, is not functional. The -carbonyl of the formed
tetraketide [d] is not further processed and becomes

the C-9 keto group in 6-dEB. After the fourth
condensation the KR, DH, and ER domains process
the -carbonyl to the methylene [e] found at C7 in
6-dEB. After the fifth and sixth condensations only
ketoreductions are programmed to take place to
produce the OH groups at C-5 and C-3 of 6-dEB.
After reduction of the -carbonyl of the heptaketide,
the TE domain acts to release f from the PKS and
promotes the nucleophilic attack of the C-13 hydroxyl
on the C-1 carbanion formed, resulting in the production of the macrolactone. How the PKS is programmed to avoid premature release of the chain
prior to the last -ketoreduction is not yet understood.
The genes that determine DEBS have been expressed in a number of heterologous hosts, including
Streptomyces coelicolor, Streptomyces lividans, and
E. coli.142-145 The DEBS proteins have been purified
and used to make 6-dEB, intermediates, or derivatives in vitro.138,139,146,147 The specificities of the various KS domains of DEBS have been examined using
N-acetylcysteamine thioesters of the syn or anti
diastereomers of 2-methyl-3-hydroxyl-containing acyl
chains for direct loading onto the KS2, KS5, or KS6
domains for single or multiple chain extensions in
vivo or in vitro.117,148-151 It was found that all three
domains utilized only the syn diastereomers and that
whereas KS2 and KS5 could use either enantiomer
KS6 showed high preference for the (2S,3R) enantiomer. It should be pointed out that KS5 normally
does not utilize a 2-methyl-3-hydroxy-containing
substrate for elongation; its substrate is fully reduced
at C3.
Katz and Ashley
Figure 6. Domain organization of the Pik PKS, and structures of proposed intermediates at the end of the condensation
cycle. Polypeptides and domains as in Figure 5. Abbreviations: KSQ, KS domain carrying Cys173Ala mutation; all others
as in Figure 5.
6.1.2. Lankamycin and Oleandomycin

The lankamycin PKS is identical in both module
and domain organization the DEBS but differs in
amino acid sequence.152 The only structural difference
between the untailored aglycones is the replacement
of the 13-ethyl side chain in 6-dEB with the 1(S)methyl-2(S)-hydroxypropyl group in the aglycone of
lankamycin. The discovery that the lankamycin PKS
contained only six modules (and a loading module)
suggests that the starter is either 2-methylbutyryl
CoA (which is hydroxylated at C-2 after polyketide
synthesis) or 2-methyl-3-hydroxybutyryl CoA.
The aglycone of oleandomycin is built from an
acetate starter and six molecules of methylmalonyl
CoA. The Ole PKS is organized identically to that
seen for DEBS with a single difference, a KSQ domain
in the loading module, which is discussed immediately below.

The aglycones of methymycin and pikromycin differ
in structure only with respect to the additional two
carbons in the ring of pikromycin. Methymycin is
produced from one propionate, one malonate, and
four methylmalonate residues. Pikromycin requires
an additional methylmalonate. Both compounds are
produced in S. venezuelae from a single PKS (Figure
6); thus, the nascent chain to the end of the fifth
module is the same for both compounds.153 10Deoxymethynolide is released after the fifth condensation and narbonolide after the sixth. The Pik PKS
is similar to DEBS in overall organization, with a
number of interesting differences. At the N-terminus
of the loading module is a domain labeled KSQ in

which the Cys173 residue at the active site is
replaced by Gln.154 This domain, therefore, cannot
make a thioester linkage with an acyl chain and
hence cannot participate in a condensation reaction.
The domain is still capable of the decarboxylation
event that is required for chain elongation. Hence,
loading modules that carry KSQ domains use starters
that require decarboxylation such as malonyl CoA or
methylmalonyl CoA to yield the required acetyl or
propionyl moieties found in the side chains of the
completed aglycones.155,156 Reduction of the resulting
carbanion is likely conducted by the KSQ as well. The
Pik PKS, therefore, uses methylmalonyl CoA as the
starter and decarboxylates it to propionyl-ACP. The
Ole PKS uses malonyl CoA as the starter and
decarboxylates it to acetyl-CoA. The first, third, and
fourth condensations and -carbonyl-processing events
resemble those seen for 6-dEB. The second condensation employs malonyl CoA rather than methylmalonyl CoA as the extender unit, and the -carbonyl of
the triketide is reduced and then dehydrated by the
KR and DH domains in module 2. The 2,3-double
bond of c thus becomes the 8,9- or 10,11-trans double
bond of methymycin or pikromycin, respectively. The
most interesting differences from DEBS are the
events that take place after the fifth condensation.
Modules 5 and 6 in the Pik PKS are split into
separate polypeptides, PikAIII and PikAIV, respectively. Under conditions that favor the production of
methymycin, nascent chain growth terminates after
the fifth condensation event to release and cyclizes
the acyl chain to produce 10-deoxymethynolide. It has
Figure 7. Domain organization of the Tyl PKS, and structures of proposed intermediates at the end of the condensation
cycle. Polypeptides, domains, and abbreviations as in Figure 5.
been proposed that this is accomplished by the

transfer of intermediate f to the ACP of module 6
without chain elongation, a process referred to as
skipping.157-159 Once attached to ACP6, the adjacent TE domain can release and cyclize the acyl
chain. Under conditions that favor 14-membered ring
production, normal transfer of intermediate f to the
KS of module 6 would take place. How the organism
regulates the production of one macrolide over another is still not fully understood. The pik PKS genes
have been expressed, in whole or in part, in heterologous hosts.144,160 Employing various 2-methyl-3hydroxypentanoyl-S-NACs, the specificities of KS5
and KS6 of the Pik PKS were shown to be similar to
those found for the corresponding DEBS KS domains,
although KS5 was found to have high preference for
the syn (2S,3R)-enantiomer.161,162
6.1.4. Tylosin
The PKS-encoding genes from at least one member
of each of the four groups of 16-membered macrolides
have been sequenced. All contain seven modules and
are organized as shown in Figure 7 for the tylosin
PKS.163 The PKS is composed of five polypeptides:
polypeptide Isload, modules 1 and 2; IIsmodule 3;
IIIsmodules 4 and 5; IVsmodule 6; Vsmodule 7.
The aglycone tylactone is made from the precursors
malonyl CoA, methylmalonyl CoA, and ethylmalonyl
CoA. The presence of the KSQ domain in the loading
module suggests that the starter is methylmalonyl
CoA, which is decarboxylated to propionyl-S-ACP.
The first, second, fourth, and sixth condensations
employ methylmalonate extenders; the third and
seventh use malonyl CoA. The fourth extension uses
ethylmalonyl CoA, which is produced in the cell
through the 2-carboxylation of butyryl CoA. Butyryl
CoA may be produced from the degradation of fatty
acids or through a single round of fatty acid synthesis

from acetyl-CoA. A gene for crotonyl CoA reductase,
which catalyzes conversion of crotonyl CoA to butyryl
CoA, is present in the tylosin biosynthesis cluster.164
The specificities of the KS domains of the Tyl PKS
have not been examined; thus, it remains to be
determined whether the KS2 domain, which is normally presented with the anti-2-methyl-3-hydroxypentanoyl-S-ACP, has preference for one enantiomer
over the other or whether the syn diastereomer can
also be extended.
6.1.5. Platenolide
The predicted domain organization and biosynthetic intermediates of platenolide synthase, which
has been sequenced from the spiramycin and niddamycin producers, is shown in Figure 8.165,166 The
domains are identical to that of the tylosin PKS with
two exceptions: the ATs of the loading module and
module 2 transfer malonyl CoA rather than methylmalonyl CoA; the AT of module 6 transfers methoxylmalonate-thioester rather than methylmalonyl CoA.
In the platenolide cases it is not known whether the
thioester moiety of methoxymalonate is CoA, but it
is thought that methoxymalonyl-ACP is the precursor employed for biosynthesis of the complex polyketides ansimitocin and ascomycin.167,168
6.1.6. Chalcomycin
The PKS of chalcomycin is shown in Figure 9.
Although chalcomycin contains a 2,3-trans double
bond, the Chm PKS does not contain the required
KR and DH domains in module 7 to catalyze its
formation.169 A gene that could encode a ketoreductase was identified 3 kb downstream of the PKS, but
a DH gene was not found. Expression of the chm PKS
Katz and Ashley
Figure 8. Domain organization of the platenolide PKS, and structures of proposed intermediates at the end of the
condensation cycle. Polypeptides, domains, and abbreviations as in Figure 5.
Figure 9. Domain organization of the Chm PKS, and structures of proposed intermediates at the end of the condensation
cycle. Polypeptides, domains, and abbreviations as in Figure 5.
genes in an S. fradiae host that had been deleted of

the tyl PKS genes resulted in the production of the
predicted macrolactone containing a 3-keto group
(chalconolide) but which contained mycaminose at
C-5, indicating that the mycaminosyltransferase used
for tylosin production could utilize chalconolide as
well.169,170 The basis for the introduction of the 2,3double bond in chalcomycin is not yet understood. In
contrast, the seventh module of the mycinamicin PKS
contains the KR and DH domains, which indicates
formation of the double bond on the nascent polyketide.152
6.2. Biosynthesis of Deoxysugars

Genes for the biosynthesis of the deoxysugar
components of macrolides have been identified in the
erythromycin, pikromycin, tylosin, megalomicin,
chalcomycin, oleandomycin, and lankamycin
clusters.152,153,164,169,171-178 Verification of the pathways
have come from (a) transfer of the genes to a
Figure 10. Composite biochemical pathways of deoxysugar biosynthesis in macrolide-producing strains. Proposed enzymes
for given steps are shown.
heterologous host and production of a macrolide

containing the corresponding deoxysugar, in some
cases a novel macrolactone-sugar combination, or (b)
loss of synthesis of the sugar component or change
of the structure of the sugar through the introduction
of a mutation in the corresponding genes.172,174,179-183
A compilation of the proposed pathways of seven
deoxysugars present in macrolides is shown in Figure
10 along with the proposed genes involved in the
particular steps from the corresponding antibioticproducing strains. The proposed pathway for the
synthesis of L-arcanose, the neutral sugar of lankamycin, is not shown. Genes from different organisms
involved in a particular step of a pathway, e.g.,
eryBIV, tylM, show highest similarity scores to each
other of all matches in the sequence databases and
are assigned the given step on the basis of proposed
function. Because they have not been determined
experimentally, the absolute order of reactions for
pathways involving more than two steps are not
certain. The nucleotide carrier thymidine diphosphate has been identified only for the deoxysugars
of tylosin, erythromycin, and oleandomycin; hence,
Figure 10 shows the generic NDP as the carrier. All

deoxysugars are made from the common intermediate 4-keto-6-deoxyglucose, which is itself made in two
steps from glucose-1-phosphate. Genes believed to
determine the enzymes for these steps have been
found in all of the macrolide biosynthesis gene
clusters examined except erythromycin, which uses
the enzymes involved in the synthesis of the deoxysugars of the cell wall.184
6.3. Post-Polyketide Modification

Following their synthesis, the aglycones are modified through glycosylation, oxidation, reduction, and
acylation. The deoxysugars also may be modified.
Each macrolide has an order sequence of reactions
to assemble the final compound, but it is often the
case that various steps may be substituted or bypassed.
6.3.1. Erythromycin and Megalomicin

Pathways for the formation of erythromycin and
megalomicin from the aglycone 6-dEB in S. erythraea
Katz and Ashley
Figure 11. Biochemical pathways of erythromycin and megalomicin biosyntheses. Proposed enzymes for given steps are
shown.
and Micromonospora megalomicea, respectively, are

shown in Figure 11.177 The erythromycin pathway
was determined from the identification of compounds
produced in mutants blocked in different steps of the
pathway.185 The aglycone is hydroxylated at C-6 by
the product of the eryF or megF gene to produce
erythronolide B (EB), which is then glycosylated at
the C-3 OH with NDP-L-mycarose to produce 3-Rmycarosyl EB (MEB) by the mycarosyltransferases
EryBV or MegBV. MEB is glycosylated at the C-5
OH with NDP-desosamine by the desosaminyltransferases EryCIII or MegCIII to yield erythromycin
D.186 In S. erythraea the 6-hydroxylation step can be
bypassed in strains defective in EryF and 6-deoxyerythromycin is formed.187 Hydroxylation of erythromycin D at C12 by EryK or MegK produces
erythromycin C, the last common intermediate in the
pathways of erythromycin A and megolamicin. In S.
erythraea the 3-OH of the mycarosyl residue is
methylated by EryG, converting the residue to Lcladinose and the compound to erythromycin A.
EryG-mediated methylation of erythromycin D can
also take place to produce the side product erythromycin, but this compound is only poorly converted
to erythromycin A by EryK.188-190 In M. megalomicea
erythromycin C is glycosylated at the C-6 OH with
NDP-megosamine by MegDI to produce megalomicin
A. The 3- and 4-OH groups of the mycarose
residue can be acylated with acetate or propionate
in various combinations to produce megalomicins B,
C1, and C2. Acylations are thought to be catalyzed
by MegY.177 The cytochrome P450 hydroxylase EryF

has been crystallized.191,192

The pathways from 10-deoxymethynolide and narbonolide to methymycin and pikromycin, respectively,
are shown in Figure 12. Each is converted to its final
product in two steps: glycosylation at C-5 or C-3 with
desosamine catalyzed by DesII followed by hydroxylation by PikC (also called PicK) to produce the final
compound.174,193-195 It should be noted that PikC
utilizes both YC-17 and narbomycin, two different
size macrolides as substrates, and produces two
different products from YC-17.196
6.3.3. Tylosin
The pathway for the formation of tylosin is shown
in Figure 13 and has been formulated from identification of the compounds produced in mutants blocked
at various steps.113 Unlike erythromycin, glycosylation at C-5 precedes the first oxidation step that
produces the C20 aldehyde. This is followed by a
second oxidation to add the hydroxyl at C-23 for
subsequent glycosylation. The next step is the addition of D-allose to the 23-OH to produce the diglycoside, followed by addition of L-mycarose to the
mycaminose moiety. Glycosylation of OMT by D-allose
can be bypassed in tylD, tylJ, or tylN mutants to
produce the compound desmycinosyltylosin (DMT),
Figure 12. Biochemical pathways of pikromycin and methymycin biosyntheses. Proposed enzymes for given steps are
shown.
Figure 13. Biochemical pathways of tylosin biosynthesis. Proposed enzymes for given steps are shown.
OMT containing the mycarose residue, as the end

product of the pathway.197,198
6.3.4. Other Macrolides

The biosynthesis of oleandomycin [10] follows a
pathway similar to that described for erythromycin.199-201 Three subsequent post-polyketide modi-
fications take place after biosynthesis of the aglycone

in the following order: oxidation mediated by the
P450-enzyme OleP to produce the aglycone containing the 8,8a-epoxide, attachment of the neutral sugar
L-oleandose at C-3, attachment of desosamine to C-5.
Because the biosynthetic products have not been
identified, the pathway to lankamycin is less clear.
The pathway from the aglycone requires hydroxylations at C8 and C12, glycosylations employing an
NDP-neutral sugar at C3 and NDP-chalcose at C5,

and acetylation of the 11- and 4- hydroxyls. The
order of these reactions has not been established. It
is not known if the neutral sugar that is attached at
C-3 is L-arcanose or L-olivose, which is converted to
L-arcanose through 4-O-methylation after the sugar(s) is attached to the aglycone, as in the case of
erythromycin. Furthermore, as described in section
6.1.2, if the starter for the synthesis of the aglycone
is 2-methylbutyryl CoA, hydroxylation of C15 would
also be required to complete the synthesis of lankamycin.
The full complement of genes for the biosynthesis
of any platenolide-based macrolide has not been
reported; thus, little is known about the pathways
of synthesis of these compounds beyond the point of
the aglycones. In cases where reduction of the C-9
carbonyl takes place, a post-polyketide reductase is
thought to be involved. Complete sets of genes for
the biosynthesis of chalcomycin and mycinamicin
have been reported.169,202 For chalcomycin, the order
of reactions (8-hydroxylation, 12,13-epoxidation, and
glycosylations at C-20, following C-20 hydroxylation,
and at C-5) has not been established. The order of
reactions from the aglycone of mycinamicin to final
products is not known, but it has been established
that a cytochrome P450 enzyme catalyzes both the
12,13-epoxidation and 14-hydroxylation steps as the
final steps in the synthesis of mycinamicins.203
6.4. Regulation of Macrolide Biosynthesis

Though macrolides are considered to be secondary
metabolites, little is known of how their biosynthesis
is controlled to initiate toward the end of the logarithmic phase of growth and to stay on through the
stationary phase. Specific regulatory genes that
regulate expression of the PKS and other macrolide
biosynthesis genes have been identified and studied
in the pikromycin and tylosin biosynthesis clusters.
The gene pikD, present in the pik cluster, encodes a
DNA binding protein that is required for the expression of the pik PKS and desosamine genes but not
for expression of pikC.204 Inactivation of PikD leads
to loss of pikromycin and methymycin production.
Tylosin biosynthesis appears to be regulated in a
cascade fashion.205 The gene tylP appears to encode
a repressor that represses expression of tyQ, a
transcriptional activator.206 Repression is relieved by
the presence of a yet to be identified -butyrolactone,
similar to the A factor that regulates production of
streptomycin in Streptomyces griseus.207 TylQ is a
transcriptional repressor of tylR, global regulator
required for tylosin biosynthesis, and a transcriptional activator of tylS, which encodes a tylosin
pathway specific activator and is classified as a
member of the SARP (Streptomyces antibiotic regulatory proteins) family.208,209 TylS also appears to
regulate tylR.210 In addition, it has been found that
intermediates in the pathway beyond tylactone which
contain the deoxysugar mycaminose stimulate production of tylactone, but the mechanism of this
regulation is not yet understood.211
Katz and Ashley
7. New Macrolides and Ketolides

7.1. Chemistry
Erythromycin derivatives wherein the 3-O-cladinosyl moiety has been replaced with an acyl functionality, termed acylides, have been reported.212 Of
particular interest is the 3-O-(4-nitrophenyl)acetyl
derivative of clarithromycin, TEA0777, which shows
potent activity not only against macrolide-susceptible
and efflux-resistant S. pneumoniae, but also against
inducible-MLSB-resistant S. aureus as well. Recent
efforts have led to TEA0929 [32], which shows good
in-vitro activity against macrolide-susceptible and
MLSB-inducible S. aureus and S. pneumoniae and
against H. influenzae and also shows in-vivo activity
equivalent to clarithromycin.213 Ketolides bridged
across the 6-O and 11-O positions, such as EP-13417
[33], have been found to possess high in-vitro and invivo activity against typical respiratory pathogens.214
7.2. Genetic Engineering

Following the discovery of modular macrolide
PKSs, efforts commenced to alter the specificities and
activities of the domains for the purpose of changing
the structure of the corresponding aglycone. This was
enabled by the development of genetic tools for
streptomycetes that permitted DNA to be introduced
into the macrolide producers and recombination to
be selected. Hence, to create desired changes to the
structure of aglycones, the following has been accomplished: to reduce the extent of -carbonyl reduction, KR, DH, or ER domains have been inactivated
through mutation (or deletion); to increase the extent
of reduction, these domains have been introduced
into modules where not present originally; to change
the extender unit incorporated into the nascent
polyketide chain, AT domains have been exchanged.
These exchanges have been performed in the macrolide producers or in hosts into which the PKS genes
were introduced, such as E. coli or S. coelicolor.
Replacement of DEBS AT1 or AT2, AT3, AT5, and
AT6 with a malonyl-transferring AT domain in S.
erythraea or in strains of S. coelicolor or S. lividans
that carried the DEBS genes resulted in the production of the expected erythromycin analogues: AT112-desmethylerythromycin B [34]; AT2-10-desmethylerythromycin A [35] and 10-desmethylerythromycin B, AT3-8-desmethylerythromycin A
[36]; AT5-4-desmethylerythromycin A [37]; AT62-desmethylerythromycin A [38].170,215-217 In the AT1
exchange 12-desmethylerythromycin A was not produced, indicating that the EryK hydroxylase could
not use 12-desmethylerythromycin D as a substrate,
but the EryG methyltransferase utilized the intermediate to some extent (see Figure 11). In the other
cases the A congener was found but was not the most
predominant form of the product. In addition, 38 was
detected after an uncharacterized (and unrecovered)
segment of DNA from the oleandomycin producer was
introduced into a strain of S. erythraea that carried
a mutation in the DEBS PKS.218 Although the basis
of the production of 38 has not been determined,
sequencing of the host used for introduction of the

DNA revealed an in-frame deletion of AT6 (Katz, L.
et al. Unpublished results). It is likely that the
incoming DNA carried a malonyl-transferring AT
domain that acted trans to provide malonyl-AT
function to module 6. trans AT domains have recently
been discovered in nonmacrolide modular PKSs.219,220
Replacement of the AT4 domain of DEBS with an
ethylmalonyl-transferring domain resulted in production of 6-desmethyl-6-ethylerythromycin A [39],
but the host required the addition of a ccr gene
encoding crotonyl CoA reductase.221 In the absence
of ccr, the host containing the exchanged AT domain
produced a small amount of erythromycin. Replacement of AT4 in DEBS in S. erythraea with a malonyltransferring AT was done at Biotica Technology, Ltd.,
and Kosan Biosciences, Inc., with different results.
Using the soil isolate NRRL 2338 and a malonyltransferring AT from the rapamycin PKS, the Biotica
group found that the engineered strain produced
6-desmethylerythromycin D [40], indicating that both
EryK and EryG could not utilize 40 as substrate.222
The Kosan approach employed introducing two sitespecific mutations into DEBS AT4 to alter the
specificity of the domain. In the S. lividans host
carrying the altered DEBS, the expected 6-desmethylerythronolide B was produced.223 When the same
mutations were introduced into an industrially optimized S. erythraea host, the strain produced a small
amount of 6-desmethyl-6-deoxy-7-hydroxyerythromycin D [41].224 The production of a D congener in
the Kosan strain coincides with the findings at
Biotica. The finding of a 7-OH group in 40 is difficult
to explain. Either the EryF hydroxylase had changed
its specificity to hydroxylate the substrate at C7
rather than C6 in the Kosan host or the host contains
an adventitious enzyme that hydroxylates 6-dEB at
C7 to produce a compound that cannot be hydroxylated by EryF.
Exchanges of the loading module have also been
reported. Exchange of the loading AT domain of
DEBS with the loading AT domain from the avermectin PKS in S. erythraea resulted in the production
of 14-methylerythromycin A [42] and 14-ethylerythromycin A [43] along with their B and D congeners.225
Changes were also introduced into the reductive
domains of DEBS to produce novel compounds. Two
examples of such changes in S. erythraea that produced fully elaborated molecules include the inactivation of the ER4 domain to produce 6,7-anhydroerythromycin C [44] and the replacement of the KR2
domain with a DH4/ER4/KR4 domain from the rapamycin PKS to produce 11-deoxyerythromycin
[45].170,226 Multiple changes in DEBS have been done
employing DEBS genes that had been engineered to
contain unique restriction sites at the edges of the
various domains.217 These compounds were produced
in S. lividans that carried the modified DEBS genes;
hence, the compounds were not elaborated beyond
the aglycone.
Hybrid PKSs carrying at least one module of two
different PKSs have also been made. The loading
domain of the spiramycin PKS (Figure 8) was re-
placed with the loading domain of the Tyl PKS in a

Streptomyces ambofaciens host that carried a deletion
of the spiramycin sugar biosynthesis genes. The
resulting compound was the expected 15-methylplatenolide.227 DEBS-Pik, DEBS-Ole, and Tyl-Pik
PKS hybrids yielding predicted compounds have also
been reported.144,228
Of the dozen or so fully elaborated novel macrolides
produced by PKS genetic engineering, most retained
some measure of bioactivity but none showed enhanced potencies over their parent compounds. The
only example of an engineered compound that showed
improved properties was 6-deoxyerythromycin [46],
produced by targeted disruption of the eryF gene in
S. erythraea.187 The compound was less potent than
erythromycin in vitro but showed improved in-vivo
activity in experimental infections due to enhanced
acid stability.187
The most promising new molecules originate from
a combination of genetics, chemistry, and fermentation development. Jacobsen et al. demonstrated that
an S. coelicolor strain carrying DEBS that contained
a C173A replacement (KS1null) could be fed SNAC
diketides in which the C5 methyl group could be
replaced with a number of substitutions (Figure 14:
[47]) including H atoms and phenyl rings to produce
6-dEB analogues that contained the corresponding
substitutions at C13 [48].117,148,149,229 These novel
aglycones could be converted into erythromycin analogues [49] after purification and feeding to an S.
erythraea strain carrying a mutation in the one of
the DEBS genes (e.g., KS1 null host). This technology
was employed by Kosan in collaboration with J & J
Pharmaceutical Research Institute to produce a
number of novel 6-O-arylalkyl ketolides [50-55].
Preliminary studies reported that a number of these
compounds displayed in-vitro and in-vivo activities
comparable to telithromycin or cethromycin.230,231
8. Conclusions
The advancements in the isolation and crystallization of ribosomes have allowed a fuller understanding
of how macrolides and ketolides exert their antibiotic
effects. Whereas it was formerly thought that these
compounds block a specific event during the initiation
or elongation cycle of protein synthesis, it is currently
believed that their binding in the exit tunnel is
sufficient to prevent elongation of the nascent polypeptide chain. It is not yet known if the efficacy of a
compound is directly related to its strength of binding. The ketolides, which bind to domains V and II
of 23S rRNA and so may bind more tightly to
ribosomes, may become preferred as antibiotics over
macrolides, which only bind in domain V. The current
limitations of telithromycin, the only currently approved ketolide, is its modest activity against H.
influenzae, prompting the need for administration of
800 mg/day and the lack of efficacy against MLSBresistant S. pyogenes and constitutive MLSB-resistant
S. aureus. Ribosome binding studies have shed light
on the basis of macrolide resistance, but they do not
as yet enable an understanding of why, in cases of
Katz and Ashley
Figure 14. Schemes showing production of 15-R erythromycin analogues. (A) Pathway to produce 15-R erythromycin. 47
is fed to S. coelicolor DEBS (KS1null) to produce 48, which is fed to S. erythraea KS1null to produce 49. (B) Production of
15-methyl ketolides. 50 is produced using scheme A and converted to 51-55 as described in the text.
macrolide-resistant strains, the ketolides are effective

as antibiotics against some but not effective against
others. The contribution of the effect on 50S ribosome
assembly by macrolides to the overall bacteriostatic
or bactericidal activities of these molecules also
requires further clarification. The ability to manipulate PKSs provided great promise initially that novel
macrolides could be made, including ones that could
not be obtained by conventional chemical modification, and which would contain enhanced properties.
Other than the small number of compounds made by
the feeding of short-chain thioesters to a genetically
engineered host, as only a first step in a three-part
process, the few fully elaborated novel macrolides
produced by genetic engineering have not yet fulfilled
the original promise. It is still too early to tell
whether this avenue of discovery will prove effective.
The findings that many of engineered PKSs either
do not produce the expected compounds or do so at
levels too low to be useful indicate that greater
understanding of the biochemical details of polyketide
biosynthesis is required before full exploitation of
their chemical potential can be realized.
9. Acknowledgments
We thank Scott Blanchard for generating figures
of the ribosome showing bound macrolide. We are
grateful to our former and present colleagues for their
dedication and effort over the many years that we
have engaged in this research.
10. References
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Woodward, R. B. Angew. Chem. 1957, 57, 50.

Woodward, R. B. Angew. Chem. 1957, 69, 585.
Brockmann, H.; Henckel, W. Chem. Ber. 1951, 84, 284.
McGuire, J. M.; Bunch, R. L.; Anderson, R. C.; Boaz, H. E.;
Flynn, E. H.; Powell, H. M.; Smith, J. W. Antibiot. Chemother.
1952, 2, 281.
Bonay, P.; Munro, S.; Fresno, M.; Alarcon, B. J. Biol. Chem.
1996, 271, 3719.
Maezawa, I.; Kinumaki, A.; Suzuki, M. J. Antibiot. (Tokyo) 1974,
27, 84.
Huang, S. L.; Hassell, T. C.; Yeh, W. K. J. Biol. Chem. 1993,
268, 18987.
Itoh, Z.; Nakaya, M.; Suzuki, T.; Arai, H.; Wakabayashi, K. Am.
J. Physiol. 1984, 247, G688.
Erah, P. O.; Goddard, A. F.; Barrett, D. A.; Shaw, P. N.; Spiller,
R. C. J. Antimicrob. Chemother. 1997, 39, 5.
Gill, C. J.; Abruzzo, G. K.; Flattery, A. M.; Smith, J. G.; Jackson,
J.; Kong, L.; Wilkening, R.; Shankaran, K.; Kropp, H.; Bartizal,
K. J. Antibiot. (Tokyo) 1995, 48, 1141.

(11) Mordi, M. N.; Pelta, M. D.; Boote, V.; Morris, G. A.; Barber, J.
J. Med. Chem. 2000, 43, 467.
(12) Rodvold, K. A. Clin. Pharmacokinet. 1999, 37, 385.
(13) Nakagawa, Y.; Itai, S.; Yoshida, T.; Nagai, T. Chem. Pharm. Bull.
(Tokyo) 1992, 40, 725.
(14) Doucet-Populaire, F.; Capobianco, J. O.; Zakula, D.; Jarlier, V.;
Goldman, R. C. J. Antimicrob. Chemother. 1998, 41, 179.
(15) Barry, A. L.; Fuchs, P. C.; Brown, S. D. Eur. J. Clin. Microbiol.
Infect. Dis. 2001, 20, 494.
(16) Barry, A. L.; Jones, R. N.; Thornsberry, C. Antimicrob. Agents
Chemother. 1988, 32, 752.
(17) Sahm, D. F.; Karlowsky, J. A.; Kelly, L. J.; Critchley, I. A.; Jones,
M. E.; Thornsberry, C.; Mauriz, Y.; Kahn, J. Antimicrob. Agents
Chemother. 2001, 45, 1037.
(18) Denis, A.; Agouridas, C. Bioorg. Med. Chem. Lett. 1998, 8, 2427.
(19) Agouridas, C.; Denis, A.; Auger, J. M.; Benedetti, Y.; Bonnefoy,
A.; Bretin, F.; Chantot, J. F.; Dussarat, A.; Fromentin, C.;
DAmbrieres, S. G.; Lachaud, S.; Laurin, P.; Le Martret, O.;
Loyau, V.; Tessot, N. J. Med. Chem. 1998, 41, 4080.
(20) Zhanel, G. G.; Walters, M.; Noreddin, A.; Vercaigne, L. M.;
Wierzbowski, A.; Embil, J. M.; Gin, A. S.; Douthwaite, S.; Hoban,
D. J. Drugs 2002, 62, 1771.
(21) Denis, A.; Agouridas, C.; Auger, J. M.; Benedetti, Y.; Bonnefoy,
A.; Bretin, F.; Chantot, J. F.; Dussarat, A.; Fromentin, C.;
DAmbrieres, S. G.; Lachaud, S.; Laurin, P.; Le Martret, O.;
Loyau, V.; Tessot, N.; Pejac, J. M.; Perron, S. Bioorg. Med. Chem.
Lett. 1999, 9, 3075.
(22) Bonnefoy, A.; Guitton, M.; Delachaume, C.; Le Priol, P.; Girard,
A. M. Antimicrob. Agents Chemother. 2001, 45, 1688.
(23) Bonnefoy, A.; Le Priol, P. J. Antimicrob. Chemother. 2001, 47,
471.
(24) Ma, Z.; Clark, R. F.; Brazzale, A.; Wang, S.; Rupp, M. J.; Li, L.;
Griesgraber, G.; Zhang, S.; Yong, H.; Phan, L. T.; Nemoto, P.
A.; Chu, D. T.; Plattner, J. J.; Zhang, X.; Zhong, P.; Cao, Z.;
Nilius, A. M.; Shortridge, V. D.; Flamm, R.; Mitten, M.; Meulbroek, J.; Ewing, P.; Alder, J.; Or, Y. S. J. Med. Chem. 2001, 44,
4137.
(25) Yassin, H. M.; Dever, L. L. Expert Opin. Investig. Drugs 2001,
10, 353.
(26) Shortridge, V. D.; Zhong, P.; Cao, Z.; Beyer, J. M.; Almer, L. S.;
Ramer, N. C.; Doktor, S. Z.; Flamm, R. K. Antimicrob. Agents
Chemother. 2002, 46, 783.
(27) Ohtani, H.; Taninaka, C.; Hanada, E.; Kotaki, H.; Sato, H.;
Sawada, Y.; Iga, T. Antimicrob. Agents Chemother. 2000, 44,
2630.
(28) Rubinstein, E. Int. J. Antimicrob. Agents 2001, 18 (Suppl 1), S71.
(29) Samarenda, P.; Kumari, S.; Evans, S. J.; Sacchi, T. J.; Navarro,
V. Pacing Clin. Electrophysiol. 2001, 24, 1572.
(30) Prescrire Int. 2003, 12, 8.
(31) Johnson, A. P. Curr. Opin. Investig. Drugs 2001, 2, 1691.
(32) Menninger, J. R.; Otto, D. P. Antimicrob. Agents Chemother.
1982, 21, 811.
(33) Moazed, D.; Noller, H. F. Biochimie 1987, 69, 879.
(34) Poulsen, S. M.; Kofoed, C.; Vester, B. J. Mol. Biol. 2000, 304,
471.
(35) Hansen, L. H.; Mauvais, P.; Douthwaite, S. Mol. Microbiol. 1999,
31, 623.
(36) Douthwaite, S.; Hansen, L. H.; Mauvais, P. Mol. Microbiol. 2000,
36, 183.
(37) Ban, N.; Nissen, P.; Hansen, J.; Moore, P. B.; Steitz, T. A. Science
2000, 289, 905.
(38) Hansen, J. L.; Ippolito, J. A.; Ban, N.; Nissen, P.; Moore, P. B.;
Steitz, T. A. Mol. Cell 2002, 10, 117.
(39) Schlunzen, F.; Zarivach, R.; Harms, J.; Bashan, A.; Tocilj, A.;
Albrecht, R.; Yonath, A.; Franceschi, F. Nature 2001, 413, 814.
(40) Auerbach, T.; Bashan, A.; Harms, J.; Schluenzen, F.; Zarivach,
R.; Bartels, H.; Agmon, I.; Kessler, M.; Pioletti, M.; Franceschi,
F.; Yonath, A. Curr. Drug Targets Infect. Disord. 2002, 2, 169.
(41) Harms, J. M.; Schlunzen, F.; Fucini, P.; Bartels, H.; Yonath, A.
BMC Biol. 2004, 2, 4.
(42) Berisio, R.; Harms, J.; Schluenzen, F.; Zarivach, R.; Hansen, H.
A.; Fucini, P.; Yonath, A. J. Bacteriol. 2003, 185, 4276.
(43) Amsden, G. W. Clin. Ther. 1996, 18, 56.
(44) Jacobs, M. R.; Bajaksouzian, S.; Zilles, A.; Lin, G.; Pankuch, G.
A.; Appelbaum, P. C. Antimicrob. Agents Chemother. 1999, 43,
1901.
(45) Douthwaite, S. Clin. Microbiol. Infect. 2001, 7 (Suppl 3), 11.
(46) Champney, W. S.; Burdine, R. Curr. Microbiol. 1998, 36, 119.
(47) Champney, W. S. Curr. Top Med. Chem. 2003, 3, 929.
(48) Champney, W. S.; Tober, C. L. Curr. Microbiol. 2001, 42, 203.
(49) Champney, W. S. Curr. Drug Targets Infect. Disord. 2001, 1,
19.
(50) Champney, W. S.; Pelt, J. Curr. Microbiol. 2002, 45, 328.
(51) Champney, W. S.; Chittum, H. S.; Tober, C. L. Curr. Microbiol.
2003, 46, 453.
(52) Haight, T. H.; Finland, M. Proc. Soc. Exp. Biol. Med. 1952, 81,
183.
(53) Weisblum, B. Antimicrob. Agents Chemother. 1995, 39, 797.

(54) Roberts, M. C.; Sutcliffe, J.; Courvalin, P.; Jensen, L. B.; Rood,
J.; Seppala, H. Antimicrob. Agents Chemother. 1999, 43, 2823.
(55) Skinner, R.; Cundliffe, E.; Schmidt, F. J. J. Biol. Chem. 1983,
258, 12702.
(56) Katz, L.; Brown, D.; Boris, K.; Tuan, J. Gene 1987, 55, 319.
(57) Bussiere, D. E.; Muchmore, S. W.; Dealwis, C. G.; Schluckebier,
G.; Nienaber, V. L.; Edalji, R. P.; Walter, K. A.; Ladror, U. S.;
Holzman, T. F.; Abad-Zapatero, C. Biochemistry 1998, 37, 7103.
(58) Schluckebier, G.; Zhong, P.; Stewart, K. D.; Kavanaugh, T. J.;
Abad-Zapatero, C. J. Mol. Biol. 1999, 289, 277.
(59) Yu, L.; Petros, A. M.; Schnuchel, A.; Zhong, P.; Severin, J. M.;
Walter, K.; Holzman, T. F.; Fesik, S. W. Nat. Struct. Biol. 1997,
4, 483.
(60) Webb, V.; Davies, J. Trends Biotechnol. 1994, 12, 74.
(61) Webb, V.; Davies, J. Antimicrob. Agents Chemother. 1993, 37,
2379.
(62) Davies, J. Science 1994, 264, 375.
(63) Weaver, J. R.; Patee, P. A. J. Bacteriol. 1964, 88, 574.
(64) Kamimiya, S.; Weisblum, B. J. Bacteriol. 1988, 170, 1800.
(65) Kamimiya, S.; Weisblum, B. Antimicrob. Agents Chemother.
1997, 41, 530.
(66) Liu, M.; Douthwaite, S. Proc. Natl. Acad. Sci. U.S.A. 2002, 99,
14658.
(67) Choi, S. S.; Kim, S. K.; Oh, T. G.; Choi, E. C. J. Bacteriol. 1997,
179, 2065.
(68) Hahn, J.; Grandi, G.; Gryczan, T. J.; Dubnau, D. Mol. Gen. Genet.
1982, 186, 204.
(69) Capobianco, J. O.; Cao, Z.; Shortridge, V. D.; Ma, Z.; Flamm, R.
K.; Zhong, P. Antimicrob. Agents Chemother. 2000, 44, 1562.
(70) Lampson, B. C.; von David, W.; Parisi, J. T. Antimicrob. Agents
Chemother. 1986, 30, 653.
(71) Jenssen, W. D.; Thakker-Varia, S.; Dubin, D. T.; Weinstein, M.
P. Antimicrob. Agents Chemother. 1987, 31, 883.
(72) Ross, J. I.; Farrell, A. M.; Eady, E. A.; Cove, J. H.; Cunliffe, W.
J. J. Antimicrob. Chemother. 1989, 24, 851.
(73) Ross, J. I.; Eady, E. A.; Cove, J. H.; Cunliffe, W. J.; Baumberg,
S.; Wootton, J. C. Mol. Microbiol. 1990, 4, 1207.
(74) Seppala, H.; Nissinen, A.; Yu, Q.; Huovinen, P. J Antimicrob.
Chemother. 1993, 32, 885.
(75) Sutcliffe, J.; Tait-Kamradt, A.; Wondrack, L. Antimicrob. Agents
Chemother. 1996, 40, 1817.
(76) Matsuoka, M.; Janosi, L.; Endou, K.; Nakajima, Y. FEMS
Microbiol. Lett. 1999, 181, 91.
(77) Versalovic, J.; Osato, M. S.; Spakovsky, K.; Dore, M. P.; Reddy,
R.; Stone, G. G.; Shortridge, D.; Flamm, R. K.; Tanaka, S. K.;
Graham, D. Y. J. Antimicrob. Chemother. 1997, 40, 283.
(78) Versalovic, J.; Shortridge, D.; Kibler, K.; Griffy, M. V.; Beyer,
J.; Flamm, R. K.; Tanaka, S. K.; Graham, D. Y.; Go, M. F.
(79) Wang, G.; Taylor, D. E. Antimicrob. Agents Chemother. 1998,
42, 1952.
(80) Prunier, A. L.; Malbruny, B.; Tande, D.; Picard, B.; Leclercq, R.
(81) Tait-Kamradt, A.; Davies, T.; Cronan, M.; Jacobs, M. R.; Appelbaum, P. C.; Sutcliffe, J. Antimicrob. Agents Chemother. 2000,
44, 2118.
(82) Peric, M.; Bozdogan, B.; Jacobs, M. R.; Appelbaum, P. C.
(83) Ng, L. K.; Martin, I.; Liu, G.; Bryden, L. Antimicrob. Agents
Chemother. 2002, 46, 3020.
(84) Lucier, T. S.; Heitzman, K.; Liu, S. K.; Hu, P. C. Antimicrob.
(85) Nash, K. A.; Inderlied, C. B. Antimicrob. Agents Chemother.
1995, 39, 2625.
(86) Lukehart, S. A.; Godornes, C.; Molini, B. J.; Sonnett, B. S.;
Hopkins, S.; Mulcahy, F.; Engelman, J.; Mitchell, S. J.; Rompala,
A. M.; Marra, C. M.; Klausner, J. D. N. Engl. J. Med. 2004, 351,
154.
(87) Canu, A.; Malbruny, B.; Coquemont, M.; Davies, T. A.; Appelbaum, P. C.; Leclercq, R. Antimicrob. Agents Chemother. 2002,
46, 125.
(88) Nakajima, Y. J. Infect. Chemother. 1999, 5, 61.
(89) Garza-Ramos, G.; Xiong, L.; Zhong, P.; Mankin, A. J. Bacteriol.
2001, 183, 6898.
(90) Xiong, L.; Shah, S.; Mauvais, P.; Mankin, A. S. Mol. Microbiol.
1999, 31, 633.
(91) Dam, M.; Douthwaite, S.; Tenson, T.; Mankin, A. S. J. Mol. Biol.
1996, 259, 1.
(92) Tenson, T.; DeBlasio, A.; Mankin, A. Proc. Natl. Acad. Sci. U.S.A.
1996, 93, 5641.
(93) Tenson, T.; Xiong, L.; Kloss, P.; Mankin, A. S. J. Biol. Chem.
1997, 272, 17425.
(94) Pardo, D.; Rosset, R. Mol. Gen. Genet. 1977, 153, 199.
(95) Wittmann, H. G.; Stoffler, G.; Apirion, D.; Rosen, L.; Tanaka,
K.; Tamaki, M.; Takata, R.; Dekio, S.; Otaka, E. Mol. Gen. Genet.
1973, 127, 175.
(96) Davydova, N.; Streltsov, V.; Wilce, M.; Liljas, A.; Garber, M. J.
Mol. Biol. 2002, 322, 635.
(97) Ounissi, H.; Courvalin, P. Gene 1985, 35, 271.

(98) Arthur, M.; Autissier, D.; Courvalin, P. Nucleic Acids Res. 1986,
14, 4987.
(99) Arthur, M.; Andremont, A.; Courvalin, P. Antimicrob. Agents
Chemother. 1987, 31, 404.
(100) Wondrack, L.; Massa, M.; Yang, B. V.; Sutcliffe, J. Antimicrob.
(101) Alignet, J.; Liassine, N.; el Solh, N. Antimicrob. Agents Chemother. 1998, 42, 1794.
(102) OHara, K.; Kanda, T.; Ohmiya, K.; Ebisu, T.; Kono, M. Antimicrob. Agents Chemother. 1989, 33, 1354.
(103) Noguchi, N.; Emura, A.; Matsuyama, H.; OHara, K.; Sasatsu,
M.; Kono, M. Antimicrob. Agents Chemother. 1995, 39, 2359.
(104) Noguchi, N.; Katayama, J.; OHara, K. FEMS Microbiol. Lett.
1996, 144, 197.
(105) Nakamura, A.; Miyakozawa, I.; Nakazawa, K.; OHara, K.;
Sawai, T. Antimicrob. Agents Chemother. 2000, 44, 3241.
(106) Matsuoka, M.; Endou, K.; Kobayashi, H.; Inoue, M.; Nakajima,
Y. FEMS Microbiol. Lett. 1998, 167, 221.
(107) Noguchi, N.; Takada, K.; Katayama, J.; Emura, A.; Sasatsu, M.
J. Bacteriol. 2000, 182, 5052.
(108) Quiros, L. M.; Aguirrezabalaga, I.; Olano, C.; Mendez, C.; Salas,
J. A. Mol. Microbiol. 1998, 28, 1177.
(109) Cundliffe, E. Antimicrob. Agents Chemother. 1992, 36, 348.
(110) Sasaki, J.; Mizoue, K.; Morimoto, S.; Omura, S. J. Antibiot.
(Tokyo) 1996, 49, 1110.
(111) Morisaki, N.; Hashimoto, Y.; Furihata, K.; Yazawa, K.; Tamura,
M.; Mikami, Y. J. Antibiot. (Tokyo) 2001, 54, 157.
(112) Queener, S. W.; Sebek, O. K.; Vezina, C. Annu. Rev. Microbiol.
1978, 32, 593.
(113) Baltz, R. H.; Seno, E. T. Ann. Rev. Microbiol. 1988, 42, 547.
(114) Weber, J. M.; Wierman, C. K.; Hutchinson, C. R. J. Bacteriol.
1985, 164, 425.
(115) Yue, S.; Duncan, J. S.; Yamamoto, Y.; Hutchinson, C. R. J. Am.
Chem. Soc. 1987, 109, 1253.
(116) Cane, D. E.; Yang, C. J. Am. Chem. Soc. 1987, 109, 1255.
(117) Jacobsen, J. R.; Hutchinson, C. R.; Cane, D. E.; Khosla, C.
Science 1997, 277, 367.
(118) Haydock, S. F.; Aparicio, J. F.; Molnar, I.; Schwecke, T.; Khaw,
L. E.; Konig, A.; Marsden, A. F.; Galloway, I. S.; Staunton, J.;
Leadlay, P. F. FEBS Lett. 1995, 374, 246.
(119) Reid, R.; Piagentini, M.; Rodriguez, E.; Ashley, G.; Viswanathan,
N.; Carney, J.; Santi, D. V.; Hutchinson, C. R.; McDaniel, R.
(120) Caffrey, P. Chembiochem 2003, 4, 654.
(121) Scrutton, N. S.; Berry, A.; Perham, R. N. Nature 1990, 343, 38.
(122) Anderson, V. E.; Hammes, G. G. Biochemistry 1984, 23, 2088.
(123) Tsai, S. C.; Lu, H.; Cane, D. E.; Khosla, C.; Stroud, R. M.
(124) Tsai, S. C.; Miercke, L. J.; Krucinski, J.; Gokhale, R.; Chen, J.
C.; Foster, P. G.; Cane, D. E.; Khosla, C.; Stroud, R. M. Proc.
Natl. Acad. Sci. U.S.A. 2001, 98, 14808.
(125) Heathcote, M. L.; Staunton, J.; Leadlay, P. F. Chem. Biol. 2001,
8, 207.
(126) Hu, Z.; Pfeifer, B. A.; Chao, E.; Murli, S.; Kealey, J.; Carney, J.
R.; Ashley, G.; Khosla, C.; Hutchinson, C. R. Microbiology 2003,
149, 2213.
(127) Kim, B. S.; Cropp, T. A.; Beck, B. J.; Sherman, D. H.; Reynolds,
K. A. J. Biol. Chem. 2002, 277, 48028.
(128) Donadio, S.; Staver, M. J.; McAlpine, J. B.; Swanson, S. J.; Katz,
L. Science 1991, 252, 675.
(129) Cortes, J.; Haydock, S. F.; Roberts, G. A.; Bevitt, D. J.; Leadlay,
P. F. Nature 1990, 348, 176.
(130) Staunton, J.; Caffrey, P.; Aparicio, J. F.; Roberts, G. A.; Bethell,
S. S.; Leadlay, P. F. Nat. Struct. Biol. 1996, 3, 188.
(131) Weissman, K. J.; Hong, H.; Oliynyk, M.; Siskos, A. P.; Leadlay,
P. F. Chembiochem 2004, 5, 116.
(132) Aparicio, J. F.; Caffrey, P.; Marsden, A. F.; Staunton, J.; Leadlay,
P. F. J. Biol. Chem. 1994, 269, 8524.
(133) Weissman, K. J.; Bycroft, M.; Staunton, J.; Leadlay, P. F.
(134) Pereda, A.; Summers, R. G.; Stassi, D. L.; Ruan, X.; Katz, L.
Microbiology 1998, 144 (Pt 2), 543.
(135) Marsden, A. F.; Caffrey, P.; Aparicio, J. F.; Loughran, M. S.;
Staunton, J.; Leadlay, P. F. Science 1994, 263, 378.
(136) Kao, C. M.; Luo, G.; Katz, L.; Cane, D. E.; Khosla, C. J. Am.
Chem. Soc. 1994, 116, 11612.
(137) Cortes, J.; Wiesmann, K. E.; Roberts, G. A.; Brown, M. J.;
(138) Wiesmann, K. E.; Cortes, J.; Brown, M. J.; Cutter, A. L.;
Staunton, J.; Leadlay, P. F. Chem. Biol. 1995, 2, 583.
(139) Pieper, R.; Gokhale, R. S.; Luo, G.; Cane, D. E.; Khosla, C.
(140) Gokhale, R. S.; Tsuji, S. Y.; Cane, D. E.; Khosla, C. Science 1999,
284, 482.
(141) Wu, N.; Cane, D. E.; Khosla, C. Biochemistry 2002, 41, 5056.
(142) Kao, C. M.; Katz, L.; Khosla, C. Science 1994, 265, 509.
(143) Xue, Q.; Ashley, G.; Hutchinson, C. R.; Santi, D. V. Proc. Natl.
Acad. Sci. U.SA. 1999, 96, 11740.
(144) Tang, L.; Fu, H.; McDaniel, R. Chem. Biol. 2000, 7, 77.
Katz and Ashley

(145) Pfeifer, B. A.; Admiraal, S. J.; Gramajo, H.; Cane, D. E.; Khosla,
C. Science 2001, 291, 1790.
(146) Pieper, R.; Luo, G.; Cane, D. E.; Khosla, C. Nature 1995, 378,
263.
(147) Pieper, R.; Ebert-Khosla, S.; Cane, D.; Khosla, C. Biochemistry
1996, 35, 2054.
(148) Jacobsen, J. R.; Keatinge-Clay, A. T.; Cane, D. E.; Khosla, C.
Bioorg. Med. Chem. 1998, 6, 1171.
(149) Jacobsen, J. R.; Cane, D. E.; Khosla, C. Biochemistry 1998, 37,
4928.
(150) Weissman, K. J.; Bycroft, M.; Cutter, A. L.; Hanefeld, U.; Frost,
E. J.; Timoney, M. C.; Harris, R.; Handa, S.; Roddis, M.;
Staunton, J.; Leadlay, P. F. Chem. Biol. 1998, 5, 743.
(151) Cane, D. E.; Kudo, F.; Kinoshita, K.; Khosla, C. Chem. Biol. 2002,
9, 131.
(152) Mochizuki, S.; Hiratsu, K.; Suwa, M.; Ishii, T.; Sugino, F.;
Yamada, K.; Kinashi, H. Mol. Microbiol. 2003, 48, 1501.
(153) Xue, Y.; Zhao, L.; Liu, H. W.; Sherman, D. H. Proc. Natl. Acad.
Sci. U.S.A. 1998, 95, 12111.
(154) Shah, S.; Xue, Q.; Tang, L.; Carney, J. R.; Betlach, M.; McDaniel,
R. J. Antibiot. (Tokyo) 2000, 53, 502.
(155) Witkowski, A.; Joshi, A. K.; Lindqvist, Y.; Smith, S. Biochemistry
1999, 38, 11643.
(156) Bisang, C.; Long, P. F.; Cortes, J.; Westcott, J.; Crosby, J.;
Matharu, A. L.; Cox, R. J.; Simpson, T. J.; Staunton, J.; Leadlay,
P. F. Nature 1999, 401, 502.
(157) Xue, Y.; Sherman, D. H. Nature 2000, 403, 571.
(158) Beck, B. J.; Yoon, Y. J.; Reynolds, K. A.; Sherman, D. H. Chem.
Biol. 2002, 9, 575.
(159) Thomas, I.; Martin, C. J.; Wilkinson, C. J.; Staunton, J.; Leadlay,
P. F. Chem. Biol. 2002, 9, 781.
(160) Tang, L.; Fu, H.; Betlach, M. C.; McDaniel, R. Chem. Biol. 1999,
6, 553.
(161) Yin, Y.; Lu, H.; Khosla, C.; Cane, D. E. J. Am. Chem. Soc. 2003,
125, 5671.
(162) Beck, B. J.; Aldrich, C. C.; Fecik, R. A.; Reynolds, K. A.; Sherman,
D. H. J. Am. Chem. Soc. 2003, 125, 12551.
(163) DeHoff, B. S.; Sutton, K. L.; Rosteck, P. R., Jr. GenBank
Accession No. U78289, 1996.
(164) Gandecha, A. R.; Large, S. L.; Cundliffe, E. Gene 1997, 184, 197.
(165) Kakavas, S. J.; Katz, L.; Stassi, D. J. Bacteriol. 1997, 179, 7515.
(166) Burgett, S. G.; Kuhstoss, S. A.; Rao, R. N.; Richardson, M. A.;
Rosteck, P. R., Jr. US Patent No. 5,945,320, 1999.
(167) Carroll, B. J.; Moss, S. J.; Bai, L.; Kato, Y.; Toelzer, S.; Yu, T.
W.; Floss, H. G. J. Am. Chem. Soc. 2002, 124, 4176.
(168) Reeves, C. D.; Chung, L. M.; Liu, Y.; Xue, Q.; Carney, J. R.;
Revill, W. P.; Katz, L. J. Biol. Chem. 2002, 277, 9155.
(169) Ward, S. L.; Hu, Z.; Schirmer, A.; Reid, R.; Revill, W. P.; Reeves,
C. D.; Petrakovsky, O. V.; Dong, S. D.; Katz, L. Antimicrob.
(170) Rodriguez, E.; Hu, Z.; Ou, S.; Volchegursky, Y.; Hutchinson, C.
R.; McDaniel, R. J. Ind. Microbiol. Biotechnol. 2003, 30, 480.
(171) Haydock, S. F.; Dowson, J. A.; Dhillon, N.; Roberts, G. A.; Cortes,
J.; Leadlay, P. F. Mol. Gen. Genet. 1991, 230, 120.
(172) Gaisser, S.; Bohm, G. A.; Cortes, J.; Leadlay, P. F. Mol. Gen.
Genet. 1997, 256, 239.
(173) Summers, R. G.; Donadio, S.; Staver, M. J.; Wendt-Pienkowski,
E.; Hutchinson, C. R.; Katz, L. Microbiology 1997, 143 (Pt 10),
3251.
(174) Tang, L.; McDaniel, R. Chem. Biol. 2001, 8, 547.
(175) Bate, N.; Butler, A. R.; Smith, I. P.; Cundliffe, E. Microbiology
2000, 146 (Pt 1), 139.
(176) Bate, N.; Cundliffe, E. J. Ind. Microbiol. Biotechnol. 1999, 23,
118.
(177) Volchegursky, Y.; Hu, Z.; Katz, L.; McDaniel, R. Mol. Microbiol.
2000, 37, 752.
(178) Aguirrezabalaga, I.; Olano, C.; Allende, N.; Rodriguez, L.; Brana,
A. F.; Mendez, C.; Salas, J. A. Antimicrob. Agents Chemother.
2000, 44, 1266.
E.; Hutchinson, C. R.; Katz, L. Microbiology 1997, 143, 3251.
(180) Borisova, S. A.; Zhao, L.; Sherman, D. H.; Liu, H. W. Org. Lett.
1999, 1, 133.
(181) Trefzer, A.; Blanco, G.; Remsing, L.; Kunzel, E.; Rix, U.; Lipata,
F.; Brana, A. F.; Mendez, C.; Rohr, J.; Bechthold, A.; Salas, J.
A. J. Am. Chem. Soc. 2002, 124, 6056.
(182) Trefzer, A.; Salas, J. A.; Bechthold, A. Nat. Prod. Rep. 1999, 16,
283.
(183) Rodriguez, L.; Aguirrezabalaga, I.; Allende, N.; Brana, A. F.;
Mendez, C.; Salas, J. A. Chem. Biol. 2002, 9, 721.
(184) Linton, K. J.; Jarvis, B. W.; Hutchinson, C. R. Gene 1995, 153,
33.
(185) Katz, L.; Donadio, S. Annu. Rev. Microbiol. 1993, 47, 875.
(186) Majer, J.; Martin, J. R.; Egan, R. S.; Corcoran, J. W. J. Am.
Chem. Soc. 1977, 99, 1620.
(187) Weber, J. M.; Leung, J. O.; Swanson, S. J.; Idler, K. B.; McAlpine,
J. B. Science 1991, 252, 114.
(188) Paulus, T. J.; Tuan, J. S.; Luebke, V. E.; Maine, G. T.; DeWitt,
J. P.; Katz, L. J. Bacteriol. 1990, 172, 2541.

(189) Weber, J. M.; Schoner, B.; Losick, R. Gene 1989, 75, 235.
(190) Lambalot, R. H.; Cane, D. E.; Aparicio, J. J.; Katz, L. Biochemistry 1995, 34, 1858.
(191) Cupp-Vickery, J. R.; Poulos, T. L. Nat. Struct. Biol. 1995, 2, 144.
(192) Cupp-Vickery, J. R.; Poulos, T. L. Steroids 1997, 62, 112.
(193) Xue, Y.; Wilson, D.; Zhao, L.; Liu, H.; Sherman, D. H. Chem.
Biol. 1998, 5, 661.
(194) Xue, Y.; Wilson, D.; Sherman, D. H. Gene 2000, 245, 203.
(195) Lambalot, R. H.; Cane, D. E. J. Antibiot. (Tokyo) 1992, 45, 1981.
(196) Graziani, E. I.; Cane, D. E.; Betlach, M. C.; Kealey, J. T.;
McDaniel, R. Bioorg. Med. Chem. Lett. 1998, 8, 3117.
(197) Baltz, R. H.; Seno, E. T. Antimicrob. Agents Chemother. 1981,
20, 214.
(198) Okamoto, R.; Kiyoshima, K.; Yamamoto, M.; Takada, K.; Ohnuki,
T.; Ishikura, T.; Naganawa, H.; Tatsuta, K.; Takeuchi, T.;
Umezawa, H. J. Antibiot. (Tokyo) 1982, 35, 921.
(199) Rodriguez, A. M.; Olano, C.; Mendez, C.; Hutchinson, C. R.;
Salas, J. A. FEMS Microbiol. Lett. 1995, 127, 117.
(200) Olano, C.; Rodriguez, A. M.; Michel, J. M.; Mendez, C.; Raynal,
M. C.; Salas, J. A. Mol. Gen. Genet. 1998, 259, 299.
(201) Rodriguez, L.; Rodriguez, D.; Olano, C.; Brana, A. F.; Mendez,
C.; Salas, J. A. J. Bacteriol. 2001, 183, 5358.
(202) Anzai, Y.; Saito, N.; Tanaka, M.; Kinoshita, K.; Koyama, Y.; Kato,
F. FEMS Microbiol. Lett. 2003, 218, 135.
(203) Inouye, M.; Takada, Y.; Muto, N.; Beppu, T.; Horinouchi, S. Mol.
Gen. Genet. 1994, 245, 456.
(204) Wilson, D. J.; Xue, Y.; Reynolds, K. A.; Sherman, D. H. J.
Bacteriol. 2001, 183, 3468.
(205) Bate, N.; Butler, A. R.; Gandecha, A. R.; Cundliffe, E. Chem.
Biol. 1999, 6, 617.
(206) Stratigopoulos, G.; Gandecha, A. R.; Cundliffe, E. Mol. Microbiol.
2002, 45, 735.
(207) Horinouchi, S.; Beppu, T. Mol. Microbiol. 1994, 12, 859.
(208) Stratigopoulos, G.; Cundliffe, E. Chem. Biol. 2002, 9, 71.
(209) Wietzorrek, A.; Bibb, M. Mol. Microbiol. 1997, 25, 1181.
(210) Bate, N.; Stratigopoulos, G.; Cundliffe, E. Mol. Microbiol. 2002,
43, 449.
(211) Butler, A. R.; Flint, S. A.; Cundliffe, E. Microbiology 2001, 147,
795.
(212) Tanikawa, T.; Asaka, T.; Kashimura, M.; Misawa, Y.; Suzuki,
K.; Sato, M.; Kameo, K.; Morimoto, S.; Nishida, A. J. Med. Chem.
2001, 44, 4027.
(213) Tanikawa, T.; Asaka, T.; Kashimura, M.; Suzuki, K.; Sugiyama,
H.; Sato, M.; Kameo, K.; Morimoto, S.; Nishida, A. J. Med. Chem.
2003, 46, 2706.
(214) Arya, A.; Scorneau, B.; Polemeropoulous, A.; Lillard, M.; Han,
F.; Amsler, K.; Wang, G.; Wang, Y.; Peng, Y.; Phan, L. T.; Or,
(215)
(216)
(217)
(218)
(219)
(220)
(221)
(222)
(223)
(224)
(225)
(226)
(227)
(228)
(229)
(230)
(231)
Y. S. 43rd Interscience Conference on Antimicrobial Agents and

Chemotherapy, 2003; Abstract F-1190, Chicago, IL.
Oliynyk, M.; Brown, M. J.; Cortes, J.; Staunton, J.; Leadlay, P.
F. Chem. Biol. 1996, 3, 833.
Ruan, X.; Pereda, A.; Stassi, D. L.; Zeidner, D.; Summers, R.
G.; Jackson, M.; Shivakumar, A.; Kakavas, S.; Staver, M. J.;
Donadio, S.; Katz, L. J. Bacteriol. 1997, 179, 6416.
McDaniel, R.; Thamchaipenet, A.; Gustafsson, C.; Fu, H.;
Betlach, M.; Ashley, G. Proc. Natl. Acad. Sci. U.S.A. 1999, 96,
1846.
McAlpine, J. B.; Tuan, J. S.; Brown, D. P.; Grebner, K. D.;
Whittern, D. N.; Buko, A.; Katz, L. J. Antibiot. (Tokyo) 1987,
40, 1115.
Piel, J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 14002.
Cheng, Y. Q.; Tang, G. L.; Shen, B. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 3149.
Stassi, D. L.; Kakavas, S. J.; Reynolds, K. A.; Gunawardana,
G.; Swanson, S.; Zeidner, D.; Jackson, M.; Liu, H.; Buko, A.;
Katz, L. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 7305.
Petkovic, H.; Lill, R. E.; Sheridan, R. M.; Wilkinson, B.; McCormick, E. L.; McArthur, H. A.; Staunton, J.; Leadlay, P. F.;
Kendrew, S. G. J. Antibiot. (Tokyo) 2003, 56, 543.
Reeves, C. D.; Murli, S.; Ashley, G. W.; Piagentini, M.; Hutchinson, C. R.; McDaniel, R. Biochemistry 2001, 40, 15464.
Starks, C. M.; Rodriguez, E.; Carney, J. R.; Desai, R. P.; Carreras,
C.; McDaniel, R.; Hutchinson, R.; Galazzo, J. L.; Licari, P. J. J.
Antibiot. (Tokyo) 2004, 57, 64.
Marsden, A. F.; Wilkinson, B.; Cortes, J.; Dunster, N. J.;
Donadio, S.; McAlpine, J. B.; Sheldon, P. J.; Jackson, M.; Katz,
Kuhstoss, S.; Huber, M.; Turner, J. R.; Paschal, J. W.; Rao, R.
N. Gene 1996, 183, 231.
Yoon, Y. J.; Beck, B. J.; Kim, B. S.; Kang, H. Y.; Reynolds, K.
A.; Sherman, D. H. Chem. Biol. 2002, 9, 203.
Jacobsen, J. R.; Khosla, C. Curr. Opin. Chem. Biol. 1998, 2, 133.
Macielag, M.; Abbanat, D.; Ashley, G.; Foleno, B.; Fu, H.; Li,
Y.; Wira, E.; Bush, K. 42nd Interscience Conference on Antimicrobial Agents and Chemotherapy, 2002; Abstract F-1662, San
Diego, CA.
Abbanat, D. R.; Ashley, G.; Foleno, B.; Fu, H.; Hilliard, J.; Li,
Y.; Licari, P.; Macielag, M.; Melton, J.; Stryker, S.; Wira, E.;
Bush, K. 43rd Interscience Conference on Antimicrobial Agents
and Chemotherapy, 2003; Abstract F-1203, Chicago, IL.
CR030107F
Chem. Rev. 2005, 105, 529542
529
Streptogramins, Oxazolidinones, and Other Inhibitors of Bacterial Protein

Synthesis
Tariq A. Mukhtar and Gerard D. Wright*
Antimicrobial Research Centre, Department of Biochemistry and Biomedical Sciences, McMaster University, 1200 Main Street West,
Hamilton, Ontario, Canada L8N 3Z5
Contents
1. Introduction
2. Molecular Basis of Protein Translation
3. Streptogramin Antibiotics
3.1. Structures and Biosynthesis
3.2. Synergy and Mode of Action
3.3. Clinical Utility
3.4. Streptogramin Resistance
3.5. New Streptogramins
4. Oxazolidinone Antibiotics
4.1. Discovery and StructureActivity
Relationships
4.2. Mode of Action
4.3. Resistance to Oxazolidinones
4.4. Second-Generation Agents
5. Other Antibacterial Inhibitors of Bacterial
Translation
5.1. Chloramphenicol
5.2. Lincosamides
5.3. Pleuromutilins
6. Conclusions
7. Acknowledgment
8. References
529
531
532
532
534
535
536
537
538
538
538
539
539
539
540
540
540
540
541
541
Tariq Mukhtar was born in Toronto in 1975. He obtained his B.Sc. (Hons)
degree in Biochemistry from McMaster University in 1998, where he
subsequently started postgraduate studies. He is currently a Ph.D. student
in Biochemistry and Biomedical Sciences studying the molecular mechanisms of resistance to type B streptogramins and is supervised by G.
D. Wright.
1. Introduction
The process of protein translation and, in particular, the macromolecular ribozyme that is the ribosome were among the first recognized molecular
targets for antibiotics. A great number of these
antibiotics have since found clinical use or therapeutic promise over the past 50 years (Table 1). Translation and the ribosome remain outstanding drug
targets with numerous efforts directed toward further
mining the potential of this ribozyme and associated
activities in drug design. The recently available highresolution structures of virtually all components of
translation and the intact ribosome now make structure-based drug design across the components of the
whole process a reality.1-4 The chemical diversity of
compounds that can productively interfere with ribosome action is astonishing and includes cationic
aminoglycosides and neutral carbohydrates, mac-
Gerry Wright is Professor and Chair of the Department of Biochemistry

He received his Ph.D. degree in Chemistry in 1990 for work on antifungal
targets and spent 2 years as a Natural Sciences and Engineering Research
Council of Canada postdoctoral fellow in Professor Christopher Walshs
laboratory at the Harvard Medical School researching the molecular
mechanisms of vancomycin resistance. He joined the Department of
Biochemistry at McMaster in 1993 and received a Medical Research
Council of Canada Scholarship. Gerrys research program includes
antibiotic-resistance mechanisms, antibiotic biosynthesis, and discovering
new targets for antibacterial and antifungal agents. He presently holds a
Canada Research Chair in Antibiotic Biochemistry.
* To whom correspondence should be addressed. Phone: (905) 5259140, ext. 22454. Fax: (905) 522-9033. E-mail: wrightge@
mcmaster.ca.
rolides, peptides, and diverse small molecules (Figure

1). These act by exploiting numerous binding sites
10.1021/cr030110z CCC: $53.50 2005 American Chemical Society

Mukhtar and Wright
Table 1. Antibiotics that Target Bacterial Translation

antibiotic
molecular target
aminoglycosides
tetracyclines
macrolides
streptogramins
oxazolidinones
lincosamides
chloramphenicol
edeine
thiostrepton
everninomycins
pleuromutilins
fusidic acid
mupirocin
16S rRNA
16S rRNA
23S rRNA
23S rRNA
23 RNA
23S rRNA
23S rRNA
16S rRNA
23S rRNA
23S rRNA
23S rRNA
EF-G
Ile t-RNA synthetase
binding site,
subunit
A, 30S
A, 30S
P, 50S
P, 50S
P, 50S
P, 50S
P, 50S
P/E, 30S
P, 50S
P, 50S
P, 50S
on the ribosome and ancillary proteins necessary for

translation fidelity.
The availability of 3D structures of the various
protein and rRNA components required for translation represents the beginning of a new era in antibiotic biochemistry.5 Co-structures of ribosometargeted antibiotics with the intact ribosomal subunits
are now available for a growing number of antibiotics,

permitting for the first time examination of structurefunction analysis of this complex structure.1 Already
this work has resulted in a new understanding of
antibiotic-ribosome interaction that has served to
rationalize decades of painstaking biochemical research on antibiotic mode of action and resistance.
Furthermore, the availability of these remarkable
structures, even at the relatively low resolution
presently on hand, has permitted the first forays into
structure-based drug design that no doubt will launch
a new generation of ribosome-directed antibiotics.
The importance of the ribosome as a drug target
can readily be seen in other reviews in this special
issue of Chemical Reviews. This review will address
the structure, mode of action, and resistance to the
streptogramin and oxazolidinone antibiotics, two
distinct classes of antibiotics that have recently been
brought to market for treatment of bacterial infections. In addition, a brief discussion of some new and
old classes of antibiotics that block translation will
be presented.
Figure 1. Structural diversity in inhibitors of bacterial translation.
Streptogramins, Oxazolidinones, and Other Inhibitors
2. Molecular Basis of Protein Translation

Bacterial protein synthesis is an iterative process
consisting of initiation, peptide elongation, and termination events (Figure 2). This process is carried
out by a number of cytoplasmic factors associated
with ribosomes, which are large ribonucleoprotein
assemblies made up of two unequal subunits (30S
and 50S) that associate at the onset of translation
initiation events. The ribosomal architecture is best
viewed in context of the three tRNA binding sites
that span both subunits: the amino-acyl tRNA
binding/decoding A site, the peptidyl-tRNA binding
P site, and the E site, where the uncharged tRNA
exits. The active site of the ribosome, the peptidyltransferase complex (PTC), lies at the interface of the
A and P sites on the 50S subunit, and the growing
peptides exit the ribosome via a 100 hydrophobic
tunnel that opens at the back of the PTC.6
Initiation of translation begins with the formation
of a ternary complex consisting of the 30S subunit,
mRNA, and the initiating tRNA charged with formyl
methionine (fMet).7 The fMet-tRNA binds to the P
site and is the only tRNA to do so as all subsequent
amino acyl-tRNAs (aa-tRNAs) must enter through
the decoding A site (Figure 2, Step 1). The initiation
step is dependent on three initiation factors: IF-1,
IF-2, and IF-3. IF-1 is approximately 70 amino acids
in length, possessing a large S1 RNA binding domain,
and binds in the A site.8 IF-2 promotes GTP-depend-
Figure 2. Overview of bacterial translation. Step 1.

Initiation. Association of initiation factors (IF-1, IF-2, IF3), mRNA, and fMet tRNA (GTP-dependent binding to P
site) with 30S ribosomal subunit. Step 2. Association of
50S ribosomal subunit to the ternary complex, completing
initiation complex. Step 3. Elongation. Ef-TuGTP delivers aa-tRNA into the A site followed by GTP hydrolysis
and accommodation into the A site. Step 4. Peptidyl
transfer occurs, transferring the amino acid from the tRNA
in the P site to the aa-tRNA in the A site. EF-G and GTP
aid in the translocation of the tRNA from the A and P sites
to the P and E sites. Step 5. The free tRNA in the E site
exits the ribosome along with the release of GDP+Pi and
EF-G. This is an iterative process in which the A site is
now unoccupied and prepared to receive an incoming aatRNA. Step 6. Termination. Termination occurs upon
encountering a stop codon. Release factors (RF-1, RF-2, RF3, RRF) dissociate the complex, releasing the polypeptide,
mRNA, and ribosomal subunits, preparing them for recycling.
ent binding of the fMet-tRNA to the 30S subunit,

whereas IF-3 is believed to act as a fidelity factor
during the assembly of this ternary initiation complex and as a means of preventing association of the
two subunits prior to initiation.9 Upon formation of
the ternary initiator complex the 50S subunit associates with the 30S subunit followed by release of the
initiation factors (Figure 2, Step 2).
The next step in protein synthesis is elongation.
This consists of a series of codon-anticodon decoding,
peptide synthesis, and translocation events that have
been the focus of intense research over the past
decade. An elongation factor, EF-Tu bound with GTP,
is responsible for delivering the aa-tRNA to the A site
of the ribosome, where the decoding takes place
(Figure 2, Step 3). Decoding is an elegant molecular
process in which the ribosome ensures the accuracy
of the incoming aa-tRNA into the A site.10 Although
the aa-tRNA contains the anticodon for the respective
codon on the mRNA, this does not solely account for
the fidelity of protein synthesis. In fact, it has been
demonstrated that the energy difference between a
cognate codon-anticodon interaction and a near
cognate interaction is not large enough to account for
the high accuracy with which the ribosomes carry out
their function.11,12 The additional accuracy is contributed by a process of dynamic conformational
changes in the ribosome decoding center that detect
orientation and geometry, resulting in incorporation
of correct amino acids onto the nascent chain.13,14
The decoding process consists of a series of molecular checkpoints that ensure the correct incorporation of amino acids.10 Ensuring that the correct amino
acid is incorporated occurs in two stages: prior to
GTP hydrolysis and after GTP hydrolysis. Decoding
begins with the reversible binding of the anticodon
of an EF-Tuaa-tRNA complex to the codon at the A
site of the ribosome. If the interaction is noncognate
in nature, the complex will be released without
hydrolysis of GTP.15 Although the dissociation constants for cognate and near cognate codon-anticodon
interactions do differ, the initial selection process
does not readily distinguish between the two on the
basis of this difference. In principle, any EF-TuaatRNA complex resulting in near-cognate interactions
can bind the A site. Discrimination between cognate
and near-cognate codon-anticodon pairing then occurs at the second stage of proofreading, in which
geometry and overall fit of the aa-tRNA plays a
significant role in its stability and ultimately the rate
at which it is accommodated into the A site.14,16
Accommodation thus becomes a necessary factor for
a specific amino acid to be incorporated in the
growing peptide. Upon accommodation into the A
site, conformational changes in EF-Tu and the ribosome induce GTP hydrolysis, and EF-Tu release16,17
and peptidyl transferase reaction between the peptidyl-tRNA and aa-tRNA can take place (Figure 2,
Step 4).
A fundamental process in the elongation step is
translocation. Translocation is the ratchet-like movement of the ribosome along the mRNA by one codon,
resulting in the tRNA shifting from the A site to the
P site to the E site. The rate of this particular step
is greatly increased by EF-G, which aids in translocation in a GTP-dependent manner. The use of tRNA
mimics has recently shown this process to be a
complex series of interactions and events in which
the 3 end of the aa-tRNA in the A site undergoes a
180 rotation around a local 2-fold rotation axis in
the PTC, and shift of the rest of the tRNA molecule,
to enter the P site. This event occurs concurrently
with peptidyl transfer.18 Various residues within the
PTC cavity have also been suggested to guide and
rotate the tRNA as it passes from A site to P site
and subsequently guide the nascent peptide through
the exit tunnel.18
Peptidyl transfer occurs in the PTC, which is
comprised only of RNA, and thus the ribosome is
clearly a ribozyme. The details of the chemical
mechanism of acyl transfer remain controversial, and
in particular, the contributions of acid-base catalysis
vs substrate proximity have been reviewed.5,19
The final step of protein synthesis, termination,
occurs when the ribosome encounters a stop codon
in the mRNA at the A site (Figure 2, Step 6). Various
cytoplasmic factors, known as release factors, release
the polypeptide and prepare the ribosome for recycling (RF1, RF2, RF3). RRF in conjunction with EF-G
is responsible for separating the subunits and the
associated components.
3. Streptogramin Antibiotics
3.1. Structures and Biosynthesis
Streptogramin antibiotics are natural products
produced by various members of the Streptomyces
genus. This family of antibiotics consists of two
subgroups, type A and type B, which are simultaneously produced by the same bacterial species in a
ratio of roughly 70:30.20,21 Group A streptogramins
are cyclic polyunsaturated macrolactones that are
comprised of a hybrid peptide/polyketide structure
and are cyclized through an internal ester bond
between the carboxyl of the C-terminal amino acid
(generally Pro) and an internal hydroxyl group
(Figure 3). Structural variations in type A streptogramins can arise from desaturation of the Pro
residue and by its substitution for Ala or Cys.22
Examples of group A streptogramins are pristina-
Figure 3. Structures of type A streptogramin antibiotics.
Mukhtar and Wright
Figure 4. Biosynthetic origins of various components of

group A streptogramin antibiotics.
mycin IIA (same as virginiamycin M1), madumycin

II, and the semisynthetic derivative dalfopristin
(Figure 3).
The biosynthetic origins of the components of the
group A streptogramin virginiamycin M1 have been
investigated through traditional precursor fermentation experiments (Figure 4).23 The C-terminal dehydroproline predictably arises from Pro.23 A twoenzyme system comprised of an FMN-dependent
monooxygenase and an FMN reductase that catalyzes Pro oxidation has been purified and characterized from the pristinamycin IIA producer Streptomyces pristinaespiralis.24,25 The oxazole ring is derived
from Ser,23,26 which cyclizes in a fashion reminiscent
of other natural product antibiotics such as microcin
(reviewed in ref 27). The rest of the molecule is
largely comprised of acetate units derived from a
polyketide synthesis scaffold with the exception of
positions 9 and 10 (and likely the amide nitrogen),
which are derived from Gly, and the isopropyl group
that originates in Val (via crotonyl CoA). Methyl
groups 32 and 33 (Figure 4) originate from different
sources, respectively, Met and the methyl group of
acetate, derived from an uncharacterized decarboxylation mechanism.23
Group B streptogramins are cyclic hepta- or hexadepsipeptides, e.g., pristinamycin IA, virginiamycin
S, the semisynthetic quinupristin (Figure 5). The
nomenclature in this field is highly redundant with
several molecules reported in the literature having
an identical structure but different names; for example, pristinamycin IA t streptogramin B t vernamycin BR t mikamycin IA t ostreogrycin B t
PA114B1.28 The general composition of group B
streptogramins is 3-hydroxypicolinic acid-L-Thr-Daminobutyric acid (or D-Ala)-L-Pro-L-Phe (or 4-N,N(dimethylamino)-L-Phe)-X-L-phenylglycine. Residue
X is most commonly L-4-oxo or 4-hydroxypipecolic
acid but can also be Asp or Pro. The invariant
N-terminal Thr residue is N-acetylated with 3-hydroxypicolinic acid and forms a cyclizing ester linkage
with the C-terminal carboxyl group of the peptide via
its secondary hydroxyl group.
The biosynthesis of the unusual amino acids that
comprise type B streptogramins has not been extensively studied. Labeled precursor feeding experiments have shown that phenylalanine is the source
of phenylglycine29,30 and both 4-oxopipecolic acid and
3-hydroxypicolinic acid are derived from Lys30,31
Figure 5. Structures of type B streptogramin antibiotics.
Figure 6. Biosynthesis of 4-N,N-(dimethylamino)-L-Phe

required for type B streptogramin antibiotics.
during the biosynthesis of virginiamycin S1 by Streptomyces virginiae. A four-gene system for the production of 4-N,N-(dimethylamino)-L-Phe has been cloned
and partially characterized from the pristinamycin
producer Streptomyces pristinaespiralis.32 Three of
the genes are reminiscent of bacterial p-aminobenzoic
acid formation with papA acting as a chorismate
aminotransferase, papB as a mutase, and papC as a
predicted dehydrogenase that catalyzes formation of
4-aminphenylpyruvic acid (Figure 6). An unidentified
transaminase then converts the ketone to 4-aminoPhe. The final steps in the biosynthesis of 4-N,N(dimethylamino)-L-Phe are two successive methylations of the 4-amino group catalyzed by the enzyme
PapM, which has been partially characterized and
confirmed to be S-adenosylmethionine (SAM) dependent (Figure 6).32
The other components of type B streptogramins for
which some biochemical information is known are the
Lys-derived 3-hydroxypicolinic acid and 4-oxopipe-
colic acid. Sequencing of DNA downstream from

known regulatory genes of virginiamycin S biosynthesis in S. virginiae identified four genes, visA-D,
predicted to be involved in amino acid biosynthesis.33
Both visA and visC are predicted to encode enzymes
that generate 1-piperidine 2-carboxylic acid. VisA has
been purified and shown to be a pyridoxal-phosphatedependent Lys 2-aminotransferase capable of generating 1-piperidine 2-carboxylate.33 Inactivation of the
visA gene in S. virginiae completely blocked biosynthesis of virginiamycin S1, and addition of 3-hydroxypicolinic acid but not pipecolic acid to the culture
medium rescued this null phenotype, suggesting that
VisA is required for formation of the former (Figure
7).33 Paradoxically, the intracellular levels of 1-piperidine 2-carboxylic acid in the visA null mutant were
comparable to the wild-type organism. The visC and
visD genes show homology to 1-piperidine 2-carboxylic acid, generating Lys cyclodeaminase and P-450
oxidase, respectively. These genes and their products
have not been well studied, but they are predicted
to be involved in the synthesis of 4-oxopipecolic acid
(Figure 7).34
The presence of nonprotein amino acids in type B
streptogramins predicts the requirement of nonribosomal peptide synthetases (NRPSs) for antibiotic
assembly (see review by Sieber and Marahiel in this
issue). The NRPS-encoding genes responsible for
biosynthesis of pristinamycin I from S. prinstinaespiralis have been purified35 and cloned.36,37 Three
NRPS genes, snbA, snbC, and snbDE, are organized
in a 1,2, and 3 module arrangement (Figure 8). The
first gene encodes SnbA, which activates 3-hydroxypicolinic acid;36 this gene has been cloned and designated visB in S. virginiae.34 An epimerization
domain in the aminobutyric acid module of snbC is
consistent with the D-stereochemistry of this amino
acid in the final product. Similarly, the presence of
Mukhtar and Wright
Figure 7. Biosynthesis of 3-hydroxypicolinic acid and 4-oxopipecolic acid components of type B streptogramin antibiotics.
Figure 8. Organization of the nonribosomal assembly line for the biosynthesis of pristinamycin I.
a methyltransferase domain flanking the 4-N,N(dimethylamino)-L-Phe activation domain of SnbDE

is consistent with N-methylation of the amide bond
(Figure 8).
The biosynthesis of streptogramins is highly regulated by -butyrolactone autoregulators that affect
gene expression at nanomolar levels.38 These diffusible small molecules bind to highly specific transcription factors, e.g., BarX of S. virginiae.39 These quorumdependent autoregulators, analogous to the homoserine lactones of some Gram-negative bacteria,40
have been implicated in regulation of secondary
metabolism and cellular differentiation in a number
of actinomycetes.41,42
3.2. Synergy and Mode of Action

Both the A and B streptogramins bind the P site
of the 50S ribosome, a property that is shared with
such structurally diverse antibiotics as the macrolides, lincosamides, and thiopeptides (Table 1).
Streptogramins are unique in that the two compo-
nents (A and B) are separately bacteriostatic, yet

when working in combination, they act synergistically on inhibition of bacterial growth and can become
bactericidal.43 Group A streptogramins bind to the
PTC only in the absence of aminoacyl-tRNAs and
block substrate attachment to the donor and acceptor
sites, preventing the early events in elongation. In
addition, the binding of type A streptogramins causes
a conformational change in the 50S ribosome that
increases the activity of the type B streptogramins
by 100-fold.43 Type B streptogramins, on the other
hand, prevent the extension of protein chains, cause
the release of incomplete peptides, and can bind to
the ribosomes at any step of protein synthesis.43
The structure of virginiamycin M bound to the
Haloarcula marismortui 50S subunit supports the
inhibition of the PTC activity as the antibiotic bridges
both the A and P sites.44 The structure of the 50S
ribosomal subunit from Deinococcus radiodurans in
complex with both dalfopristin and quinupristin has
also been reported (Figure 9).45 The structure con-
Figure 9. Structure of dalfopristin and quinupristin bound to the large subunit of the bacterial ribosome. (A) Electron
density of the dalfopristin and quinupristin bound to the ribosome. (B) Space-filling model showing the positioning of
dalfopristin, quinupristin, and the peptidyl t-RNA. The peptide exit tunnel is shown to demonstrate the impact of
quinupristin binding. (Reprinted with permission from ref 45. Copyright 2004 BioMed Central.)
firmed many of the biochemical studies performed

with streptogramins and the ribosome and also sheds
new light on the basis of synergy for these antibiotics.
The structure demonstrated that quinupristin binds
at the entrance to the exit tunnel through which the
nascent peptide travels. The antibiotic-ribosome
interaction appears to be dominated by hydrophobic
interactions and also via hydrogen bonds to residues
A2062 and C2586 (Figure 5). Interestingly, the quinuclidinylthio moiety that imparts good solubility
and pharmacological properties to quinupristin occupies an empty space in the subunit. This is
consistent with the fact that various modifications
in this position do not interfere with antibiotic
binding.
Like virginiamycin M, dalfopristin is positioned in
a pocket in the PTC.45 It also appears to be bound by
a network of hydrophobic interactions in addition to
hydrogen bonds with residues G2505 and G2061.
These studies indicate that dalfopristin interferes
with the correct positioning of the substrates for the
A and P sites. Therefore, once the P site is occupied
by the substrate, then binding of dalfopristin can be
hindered. This is consistent with studies demonstrating that translating ribosomes are not susceptible to
dalfopristin.46,47
The synergistic nature of these antibiotics is assisted by hydrophobic interactions between the streptogramins. The structure indicates that both antibiotics contact A2062 through both hydrophobic interactions and hydrogen bonds. This is also consistent
with biochemical data, suggesting that A2062 undergoes conformational changes upon binding of
streptogramins.48,49 Perhaps the most interesting
suggestion of this study was the description of
conformational changes in the PTC upon dalfopristin
binding. These changes are shown to be in residue
U2585. This residue appears to be rotated by 180
in the streptogramin-bound state as compared to the
native, making hydrogen bonds with residues C2606
and G2588. Harms et al. performed simulations that
indicated that if there are no long-range conformational changes upon binding of dalfopristin, then
U2585 in the native and rotated conformations have

nearly identical energy states. The binding of dalfopristin therefore provides sufficient energy to move
the residue through the energy barrier to the rotated
conformation, thereby making new contacts via hydrogen bonds. The spontaneous reversal is predicted
to be slow even after removal of dalfopristin due to
similarity in energy state to the native form. This
observation is consistent with studies on the postantibiotic effect of the drug (retention of antibiotic
activity even after circulating levels of antibiotic have
dropped below the minimal inhibitory concentration).
3.3. Clinical Utility

Despite the fact that the streptogramins were first
discovered in the 1950s, these antibiotics only found
marginal use in Europe in the following decades as
antibacterial agents in the clinic. They did however,
and still do, find use as feed additives in agriculture.
The rise in antibiotic resistance in the 1980s and
1990s, largely as a result of the emergence of vancomycin-resistant enterococci (VRE), spurred renewed interest in antibiotic discovery in pharmaceutical companies. Medicinal chemists at RhonePoulenc worked to improve the drug-like properties
of pristinamycin, resulting in the synthesis and
characterization of the semisynthetic streptogramins
dalfopristin (type A) and quinupristin (type B). Their
combination in a 7:3 ratio resulted in the antibiotic
Synercid, which received FDA approval in 1999 for
the treatment of bacteremia caused by vancomycinresistant Enterococcus faecium and skin and skin
structure infections caused by Staphylococcus aureus
and Streptococcus pyogenes.
A number of studies have been performed that
examine the in-vitro and in-vivo antibacterial activity
of quinupristin and dalfopristin. Bouanchaud reviewed the susceptibility of the quinupristin/dalfopristin combination, illustrating its effectiveness
against a wide range of organisms.50 Synercids
spectrum of activity includes most Gram-positive
pathogens such as species of Staphylococci, Streptococci, and Enterococci, including multi-antibiotic-
Mukhtar and Wright
resistant strains. Enterococcus faecalis, however, is

intrinsically resistant to the effects of Synercid, likely
due to the presence of Lsa, a predicted dalfopristinclindamycin efflux pump.51 Furthermore, important
aerobic Gram negatives including Moraxella catarrhalis, Legionella spp., Mycoplasma spp., Neisseria
spp., and to a lesser degree Haemophilus influenzae
are also susceptible to Synercid. Its antibacterial
activity has also been confirmed against a number
of anaerobic organisms such as Bacteroides spp.,
Lactobacillus spp., Clostridium perfringens, and
Clostridium difficile. The wide spectrum of antibacterial activity for quinupristin/dalfopristin therefore
provides a therapeutic alternative for life-threatening
bacterial infections.
3.4. Streptogramin Resistance

Despite the relatively recent development and
clinical entry of Synercid, multiple mechanisms of
resistance to streptogramins are known. This may
be a reflection of the significant agricultural use of
this class of antibiotic over several decades. There
are three major acquired streptogramin-resistance
mechanisms: active efflux, target modification, and
antibiotic inactivation.
Efflux of type B streptogramin antibiotics has been
found to be associated with ATP-binding transport
pumps that are specific for the 14- and 15-membered
macrolide antibiotics as well as streptogramin B, for
example, MsrA, from S. aureus RN4220.52 Efflux of
type A streptogramins has also been associated with
several proteins including Lsa, which is intrinsic to
E. faecalis,51 and Vga53 and VgaB,54 which have been
found on plasmids in S. aureus.
Another mechanism of streptogramin resistance is
target modification. Covalent modification of the 23S
rRNA by an rRNA methylase encoded by the erm
gene is a highly prevalent mechanism of streptogramin B resistance. Erm-mediated ribosomal monoand dimethylation of the N6 of A2058 (E. coli numbering) of the 23S RNA confers resistance not only
to type B streptogramins but also to macrolides such
as erythromycin and to lincosamides such as clindamycin.55 This gives rise to the so-called MLSBresistance phenotype (M ) macrolide, L ) lincosamide, SB ) type B streptogramin) that is predominantly found in Gram-positive bacteria. There
are a number of different erm genes that have been
identified in a variety of organisms, and resistance
is likely a result of a conformational change occurring
in the ribosome.56 The type A streptogramins are not
affected by Erm-mediated resistance, and as a result,
synergy between the two types of streptogramins is
maintained.
An important aspect of this type of resistance is
the regulation of the respective erm genes. Expression of the MLS resistance in staphylococci may be
inducible or constitutive. The inducible nature of the
resistance determinant depends on regions adjacent
to the gene itself. Specifically, inducible erm gene
transcripts contain an upstream regulatory element
that contain four inverted repeats.55 These give rise
to two stem loop structures that in the absence of
erythromycin sequester both the ribosome binding
Figure 10. VatD as a representative inactivator of type

A streptogramin antibiotics. (A) 3D structure of VatD
cocrystallized with Virginiamycin M1 and CoA.58 (B) Predicted molecular mechanism of acetyltransfer of VatD.
site and initiation codon for the erm gene. When

erythromycin is present, the antibiotic binds the
ribosome, resulting in stalling of translation. This
stalling event likely kinetically displaces the stem
loop structures and allows translation of the methylase. Therefore, gene regulation is maintained by a
feedback mechanism; when all the ribosomes are
methylated, erythromycin cannot bind, in turn stalling does not occur, and the mRNA adopt the stem
loop structures again.56
Recently, Tait-kamradt et al. reported clinical
isolates of S. pneumoniae conferring the MSB
phenotype that did not harbor ermB or mutations in
the 23S rRNA.57 These isolates were found to have
modified L4 ribosomal proteins with either a three
amino acid substitution (G69TG to T69PS) or in one
case a six amino acid insertion in this region.
The final type of acquired resistance to streptogramins include a cadre of antibiotic inactivation
enzymes. These consist of acetyltransferases, which
modify type A streptogramins, and lyases, which
inactivate type B streptogramins. The acetyltransferases transfer an acetyl group from acetyl-CoA on
Figure 11. Predicted molecular mechanism of Vgb-catalyzed inactivation of type B streptogramin antibiotics.
to the secondary hydroxyl of the type A streptogramin. This hydroxyl makes multiple contacts with
G2505 (E. coli numbering) of the 23S rRNA in the
PTC as evidenced by the 3D structure of the antibiotic bound to the ribosome,45 and therefore, acetylation of this hydroxyl results in a steric block of drugtarget interaction. On the other hand, the lyases
cause the cleavage of the ester linkage on type B
streptogramins, resulting in linearization of the
peptide and loss of the bioactive conformation necessary for ribosome binding.
A number of streptogramin A acetyltransferases
have been identified in many pathogenic organisms.
The crystal structure for VatD, an acetyltransferase
from E. faecium, has been solved in the apo form,58,59
bound to substrate acetyl-CoA,58,59 product CoA,58
and the streptogramins virginiamycin M158 and dalfopristin59 (Figure 10A). The enzyme is a homotrimer
with each subunit folded into three domains. These
domains consist of a large coiled LH, an extended
loop domain, and C-terminal domain. Recent mutagenesis studies have supported a predicted mechanism where His82 acts as the general base and is a
major determinant of catalytic rate enhancement by
VatD (Figure 10B).59
The other streptogramin inactivation mechanism
associated with pathogenic bacteria is conferred by
Vgb lyase, which inactivates type B antibiotics. Early
studies on type B streptogramin-inactivating enzymes suggested that this enzyme was a lactonase
in which hydrolysis was responsible for linearizing
the cyclic peptide.60-63 However, in 1996 Bateman et
al. reported type B streptogramin-inactivating activity in crude cell-free extracts of Streptomyces lividans.
The extract inactivated the antibiotic etamycin
through an elimination mechanism as opposed to
hydrolysis.20 Recent studies on Vgb lyase, using
purified recombinant enzyme from S. aureus, and
orthologues encoded on the chromosomes of Bordetella pertussis and Streptomyces coelicolor have demonstrated that this enzyme also inactivates type B
streptogramins through an elimination mechanism
as opposed to hydrolysis (Figure 11).
In 1998, Suzuki et al. reported an enzyme from the
streptogramin-producing organism S. virginiae that
was capable of inactivating type A streptogramins by
a previously unidentified mechanism.64 This inactivation occurred through reduction of the 16-carbonyl
group of virginiamycin M, resulting in 16R-dihydroVM. Although this inactivation mechanism has
not yet been reported in the clinics, the possibility
for its appearance is an event that both researchers
and health practitioners should be aware of.
3.5. New Streptogramins

Synercid is not orally available and is administered
by intravenous routes. Efforts have therefore been
made to generate new orally active streptogramins.
For example, RPR 106972 is a 2:1 (molar) mixture
of pristinamycin IB (RPR 112808) and pristinamycin
IIB (RPR 106950) (Figure 12).65 Pristinamycin IB
differs from a closely related molecule pristinamycin
IA at the N-methyl-4-(dimethylamino)phenylalanine
residue. In this particular case pristinamycin IB
contains a N-methyl-4-(methylamino)phenylalanine
at this position. In addition, pristinamycin IIB differs
from pristinamycin IIA in the dehydroproline residue
in which it is replaced with proline.
Rhone Poulenc Rorer (subsequently Aventis Pharma) also developed additional type B streptogramins
through chemical modification of the 4-oxopipecolic
acid moiety and type A streptogramins through
modification of the dehydroproline residue. In particular, a new oral streptogramin designated XRP
2868, a combination of RPR132552A and RPR 202868
Figure 12. New semisynthetic streptogramin antibiotics: (A) type A streptogramins; (B) type B streptogramins.
(Figure 12), has been shown to be very effective

against a number of Gram-positive and Gram-negative organisms and particularly effective against
pneumococcal and Haemophilus strains.66 A number
of patents have been issued to Aventis Pharma for
the development of novel streptogramin derivatives
which have shown to have activity against Grampositive organisms, particularly streptococci, enterococci, and staphylococci.67 Some examples of these
structures have been summarized in Figure 12.
The growing problem of streptogramin resistance
as well as a desire to improve pharmacological
properties will drive the development of new agents
in the future. The availability of 3D strucutres of the
ribosome with bound antibiotics and of resistance
enzymes will greatly facilitate these efforts.
4. Oxazolidinone Antibiotics
4.1. Discovery and StructureActivity
Relationships
The oxazolidinones are the only new chemical class
of antibiotic that have been discovered and success-
Figure 13. Structures of oxazolidinone antibiotics.
Mukhtar and Wright
fully implemented in the clinic over the past 40 years.

These bacteriostatic molecules find no congeners in
natural product compounds and were first synthesized by chemists at E.I. du Pont de Nemours & Co.
in the mid-1980s (e.g., DuP-721, Figure 13).68,69 While
these compounds showed good antibacterial activity
against Gram-positive bacteria such as Staphylococci,
Streptococci, and Enterococci,68 the compound class
was not pursued as a result of toxicity issues in
animals studies.70 Scientists at Pharmacia subsequently found that a structure-toxicity relationship
could be established for the oxazolidinones and
initiated a campaign to develop these compounds as
novel antibacterial agents (historically reviewed in
ref 70).
Initial structure-activity relationship studies revealed a core structure requirement for the 5-S
configuration at position 5 of the oxizolidinone ring,
an acylaminomethyl group linked to C5 and N-aryl
ring substitution (Figure 13). This work was expanded to explore functionalization of the aryl ring
that resulted in improved activity or expanded antibacterial spectrum. For example, substitution with
azole groups added increased activity toward the
Gram-negative pathogens H. influenzae and Moraxella catarrhalis,71 and compounds incorporating
thiomorpholine resulted in activity against mycobacteria.72 Substitution by piperazine,73 morpholine,74 as
well as fluorination75 was also explored. The details
of this SAR program have recently been reviewed.70
The outcome of this drug discovery program is the
antibiotic linezolid (Figure 13), which received FDA
approval in 2000 and is marketed under the trade
name Zyvox.76 Linezolid is active both in oral and
injectable forms and is indicated for the treatment
of a variety of infections caused by Gram-positive
bacteria.
4.2. Mode of Action

Oxazolidinones bind to the P site of the 50S subunit
with micromolar affinity77,78 and inhibit bacterial
translation.79 NMR studies indicate that the solution
structures and the ribosome-bound structures of the
antibiotics are very similar.78 Despite binding to the
P region of the ribosome in a region that overlaps
with chloramphenicol and lincosamide binding sites,
oxazolidinones have no effect on the PTC activity.77
Instead, inhibition of translation appears to be at the
level of translation initiation.80 In particular, binding
of oxazolidinones at the P site interferes with binding
of the initiator fMet-tRNA to this site during the
formation of the initiating complex (Figure 2).81 A
recent study indicated that oxazolidinones also promote frame shifting and stop codon read through in
an E. coli system.82
Mapping of 23 rRNA mutations that result in

oxazolidinone resistance (see below) have further
localized the binding site to the P site, and kinetic
studies of a number of inhibitors of translation with
oxazolidinone-resistance ribosomes reveal some crossresistance with chloramphenicol but not the A-sitespecific antibiotic sparsomycin.83 Careful work by
Pomplianos group at Bristol-Myers Squibb suggests
that oxazolidinones induce a conformational shift in
the peptidyl transferase region that is specific to the
structure of the individual antibiotic.83 Therefore, one
could imagine a suite of similar compounds with
different resistance patterns; high-resolution crystal
structures of various oxazolidinones in complex with
the ribosome will greatly facilitate interpretation of
the available data and its use in structure-guided
drug design.
Taken together, the available research points to a
mechanism of action where oxazolidinones bind to the
P site of the 50S subunit, in particular adjacent to
the 23S rRNA, preventing fMet-tRNA from binding
in a fashion that permits the formation of the first
peptide bond that begins mRNA translation. Additional effects on fidelity (e.g., frame shifting) may
also contribute to the mechanism of action.
4.3. Resistance to Oxazolidinones

Widespread resistance to linezolid in the clinic is
thus far rare. In a survey of over 40 000 isolates of
Gram-positive cocci between 1998 and 2000, only
eight strains were resistant (MIC g 8 g/mL). These
included species of the genera Enterococcus, Staphylococcus, and Streptococcus and two strains of vancomycin-resistant enterococci. Prolonged use in patients has selected for resistance in methicillinresistant Staphylococcus aureus84,85 and enterococci86
and has been shown to emerge in enterococci87 and
others during in-vitro serial dilution or mutagenesis
studies.88,89 Importantly, resistant bacteria in
patients who have not had prior treatment with
linezolid has been reported as well.90,91
Resistance occurs by site mutations in the domain
V region of the 23S rRNA, consistent with the mode
of action (see above). Until recently, all clinically
derived linezolid resistance was associated with a G
f U mutation at position 2576 of the 23S rRNA.
These strains retain resistance to linezolid even in
the presence of the ribosomal methylation enzyme
Erm(C).92 However, Meka et al. recently reported a
T2500A mutation in S. aureus in addition to loss of
a copy of the 23 RRNA gene in some isolates.93 Most
bacterial species carry multiple copies of the rDNA
genes, and in S. aureus there are 5-6 23S rDNA
genes. Heterozygous mutations in these genes result
in a gradient of linezolid susceptibility. For example,
strains with two of six possible G2576T mutations
gave linezolid MIC of 8 g/mL, while mutation of five
of six gave MIC of 32 g/mL compared to the wildtype, linezolid-susceptible, bacterium with MIC of 2
g/mL.85 Reversion of this mechanism has been
reported in a strain of S. aureus with four of five
copies bearing the G2576T mutation following repeated serial passage in antibiotic-free medium.94
Other ribosomal mutations have been reported

during in vitro selection experiments,82,87-89 and the
addition of mutations in DNA repair mechanisms
increases the frequency of linezolid-resistant mutations95 Sander et al. used Mycobacterium smegmatis,
which only carries one copy of the 23S rDNA, to
explore the mechanisms of linezolid resistance identifiable after serial passage experiments. As expected,
23S rRNA mutations were identified but also some
nonribosomal mutants emerged in the screen, possibly the result of altered uptake or efflux.96 Resistance to all antimicrobial agents is predictable;
however, thus far widespread resistance remains rare
for the oxazolidinones likely for two reasons. First,
the fact that these agents are not natural products,
and therefore, the microbial community has not seen
this chemical scaffold in the past. Second, ribosomal
mutations must be hardwired into the chromosome
and the presence of multiple 23S rDNA genes makes
homozygocity, which is associated with high-level
resistance, rare. This argues well for the longevity
of this class of agent and presents an opportunity to
initiate program development in new members of the
class to be ready if (or rather when) resistance
becomes more of an issue.
4.4. Second-Generation Agents

Since the launch of linezolid there has been great
interest in additional SAR studies directed at the
oxazolidinone class of antibiotics to identify new
agents with improved properties such as bacterial
spectrum and to counter emerging resistance. For
example, combinatorial synthesis of S-oxide, fluoroacetamido analogues of linezolid, such as compounds
1 and 2 in Figure 14, provide increased activity
against H. influenzae and M. catarrhalis.97 Similarly,
indolinyl derivatives such as compounds of general
structure 3 also showed improved activity against
these Gram-negative pathogens. A series of pyrrol
derivatives (4) was shown to have activity against
various mycobacterial species including multidrugresistant strains.98
In a creative approach two groups recently reported
a series of compounds linking oxazolidinone and
fluoroquinolone (ciprofloxacin) moieties, e.g., compound 7 (Figure 14).99-101 The best of these analogues
exhibited a broad spectrum of antibacterial activity,
e.g., including bacterial targets not covered by linezolid such as E. coli, and activity against linezolidand fluoroquinolone-resistant bacteria. This clever
combination approach has the potential to be further
optimized to achieve improved drug-like properties.
5. Other Antibacterial Inhibitors of Bacterial

Translation
The structural diversity of compounds shown in
Figure 1 and Table 1 graphically demonstrate that
inhibitors of bacterial translation encompass a very
broad chemical space that has great potential to be
exploited in the development of new antimicrobial
agents. Furthermore, as the lesson of the clinical
development of the streptogramins demonstrates,
reexamination of old agents also has great potential
Mukhtar and Wright
Figure 14. Sample of some newer oxazolidinone antibiot504ics.
in the development of clinically important drugs. In

this section a brief discussion of some old and new
inhibitors of translation will be presented.
5.1. Chloramphenicol
The antibiotic chloramphenicol (Figure 1) was
discovered in 1947 from extracts of Streptomyces
venezuelae.102 It has excellent broad-spectrum antibacterial activity against Gram-positive and Gramnegative bacteria and was widely used clinically
following its discovery. This antibiotic however is no
longer in significant use as a result of potential fatal,
irreversible aplastic anemia which has been associated with it.103 The basis for this rare side effect is
unknown but is likely genetic. Therefore, a sensitive
screen of genotype could reinvigorate the use of this
agent in a safe fashion.
Chloramphenicol binds to the P site of the ribosome
in an area of the PTC that largely overlaps with
dalfopristin.45,104 A second binding site at the entrance of the peptide exit tunnel has also been
suggested.44,105 The crystal structure of chloramphenicol bound to the ribosome reveals the importance of intermediary Mg2+ ions in antibiotic binding
to the PTC,104 a fact that has the potential to be
exploited in the development of new analogues. One
of the key interactions is between the primary alcohol
at C3 and a Mg2+ ion. The most prevalent mechanism
of resistance to chloramphenicol is via a series of
acetyltransferases (many of which are structurally
similar to the Vat proteins described above) that
modify this important OH group.106
5.2. Lincosamides
The lincosamide antibiotics also bind the ribosome
P site in a region that partially overlaps the chloramphenicol/dalfopristin and erythromycin binding sites.104
The first member of the class, lincomycin, was
discovered as a product of Streptomyces lincolnensis
in 1962,107 and the semisynthetic derivative, clindamycin (Figure 1), continues to find some clinical use,
especially in the treatment of infections caused by
anaerobic bacteria.108 Resistance to clindamycin occurs primarily via the Erm-mediated methylation of
the 23S rRNA as described above as part of the MLSB
phentotype (L ) lincosamide).
5.3. Pleuromutilins
The pleuromutilins (Figure 1) are fungal natural
products discovered over 50 years ago.109 They have
found use in agriculture and veterinary medicine but
as a result of poor pharmacological properties have
not been used to treat infections in humans (see
references in ref 110). The pleuromutilins bind to the
PTC as assessed by chemical footprinting studies.111
While at present these antibiotics are unsuitable for
clinical use, recent medicinal chemistry efforts to
improve stability to human cytochrome P-450s, which
results in rapid degradation of the compounds, has
been reported.110
6. Conclusions
Bacterial translation is reemerging as a front-line
target for the modern discovery of antibiotics as
evidenced by the fact that of the very few new
antibacterial agents to receive regulatory approval
in the past 5 years, three of these are translation
inhibitors: Synercid, linezolid, and the ketolide,
telithromycin. Furthermore, the availability of highresolution crystal structures of the bacterial ribosome
and the ability to solve co-structures with inhibitors
of translation now provides the opportunity to launch
structure-based initiatives to develop new agents that
block translation. At the same time, highly important
biochemical and molecular biological tools such as a
strain of E. coli with all of the rRNA genes inactivated112 will greatly facilitate characterization of
mode of action, dose dependency, and resistance. The
potential of using the ribosome as a platform for new
drug discovery is in fact being pursued by a number
of drug discovery companies. Investigation of the
biochemistry, structure, and clinical application of
inhibitors of bacterial translation is therefore poised
to enter a new golden era, mirroring the discovery
phase during the 1950s. The challenge of resistance
however will remain significant in these efforts.
Overcoming existing and eventual resistance will
require the ongoing and concerted efforts of medicinal
chemists working together with microbiologists, pharmacologists, structural biologists, and biochemists to
generate the next generation of translation inhibitors.
7. Acknowledgments
We thank Arif Mukhtar for assistance in preparing
Figure 2. Research in our lab on inhibitors of bacterial translation and associated resistance has been
supported by the Canadian Institutes for Health
Research. T. Mukhtar is support by an Ontario
Graduate Student Scholarship, and G. Wright is
supported by a Canada Research Chair in Antibiotic
Biochemistry.
8. References
(1) Knowles, D. J.; Foloppe, N.; Matassova, N. B.; Murchie, A. I.
Curr. Opin. Pharmacol. 2002, 2, 501.
(2) Bower, J.; Drysdale, M.; Hebdon, R.; Jordan, A.; Lentzen, G.;
Matassova, N.; Murchie, A.; Powles, J.; Roughley, S. Bioorg. Med.
Chem. Lett. 2003, 13, 2455.
(3) Barluenga, S.; Simonsen, K. B.; Littlefield, E. S.; Ayida, B. K.;
Vourloumis, D.; Winters, G. C.; Takahashi, M.; Shandrick, S.;
Zhao, Q.; Han, Q.; Hermann, T. Bioorg. Med. Chem. Lett. 2004,
14, 713.
(4) Foloppe, N.; Chen, I. J.; Davis, B.; Hold, A.; Morley, D.; Howes,
R. Bioorg. Med. Chem. 2004, 12, 935.
(5) Wilson, D. N.; Nierhaus, K. H. Angew. Chem., Int Ed. Engl. 2003,
42, 3464.
(6) Nissen, P.; Hansen, J.; Ban, N.; Moore, P. B.; Steitz, T. A. Science
2000, 289, 920.
(7) Gualerzi, C. O.; Brnadi, L.; Caserta, E.; La Teana, A.; Spurio,
R.; Tomsic, J.; Pon, C. L. In The Ribosome; Garrett, R. A.,
Douthwaite, S. R., Lijas, A., Metheson, A. T., Moore, P. B.,
Noller, H. F., Eds.; ASM Press: Washington, DC, 2000.
(8) Carter, A. P.; Clemons, W. M., Jr.; Brodersen, D. E.; MorganWarren, R. J.; Hartsch, T.; Wimberly, B. T.; Ramakrishnan, V.
Science 2001, 291, 498.
(9) Pioletti, M.; Schlunzen, F.; Harms, J.; Zarivach, R.; Gluhmann,
M.; Avila, H.; Bashan, A.; Bartels, H.; Auerbach, T.; Jacobi, C.;
Hartsch, T.; Yonath, A.; Franceschi, F. EMBO J. 2001, 20, 1829.
(10) Ogle, J. M.; Carter, A. P.; Ramakrishnan, V. Trends Biochem.
Sci. 2003, 28, 259.
(11) Rodnina, M. V.; Pape, T.; Fricke, R.; Kuhn, L.; Wintermeyer,
W. J. Biol. Chem. 1996, 271, 646.
(12) Pape, T.; Wintermeyer, W.; Rodnina, M. V. EMBO J. 1998, 17,
7490.
(13) OConnor, M.; Dahlberg, A. E. J. Mol. Biol. 1995, 254, 838.
(14) Stark, H.; Rodnina, M. V.; Rinke-Appel, J.; Brimacombe, R.;
Wintermeyer, W.; van Heel, M. Nature 1997, 389, 403.
(15) Pape, T.; Wintermeyer, W.; Rodnina, M. EMBO J. 1999, 18,
3800.
(16) Rodnina, M. V.; Daviter, T.; Gromadski, K.; Wintermeyer, W.
Biochimie 2002, 84, 745.
(17) Valle, M.; Sengupta, J.; Swami, N. K.; Grassucci, R. A.; Burkhardt,
N.; Nierhaus, K. H.; Agrawal, R. K.; Frank, J. EMBO J. 2002,
21, 3557.
(18) Agmon, I.; Auerbach, T.; Baram, D.; Bartels, H.; Bashan, A.;
Berisio, R.; Fucini, P.; Hansen, H. A.; Harms, J.; Kessler, M.;
Peretz, M.; Schluenzen, F.; Yonath, A.; Zarivach, R. Eur. J.
Biochem. 2003, 270, 2543.
(19) Green, R.; Lorsch, J. R. Cell 2002, 110, 665.
(20) Bateman, K. P.; Yang, K.; Thibault, P.; White, R. L.; Vining, L.
C. J. Am. Chem. Soc. 1996, 118, 5335.
(21) Pechere, J. C. Drugs 1996, 51, 13.
(22) Johnston, N. J.; Mukhtar, T. A.; Wright, G. D. Curr. Drug
Targets 2002, 3, 335.
(23) Kingston, D. G. I.; Kolpak, M. X.; LeFevre, J. W.; BorupGrochtmann, I. J. Am. Chem. Soc. 1983, 105, 5106.
(24) Thibaut, D.; Ratet, N.; Bisch, D.; Faucher, D.; Debussche, L.;
Blanche, F. J. Bacteriol. 1995, 177, 5199.
(25) Blanc, V.; Lagneaux, D.; Didier, P.; Gil, P.; Lacroix, P.; Crouzet,
J. J. Bacteriol. 1995, 177, 5206.
(26) Purvis, M. B.; Kingston, D. G. I.; Fujii, N.; Floss, H. G. J. Chem.
Soc., Chem. Commun. 1987, 302.
(27) Roy, R. S.; Gehring, A. M.; Milne, J. C.; Belshaw, P. J.; Walsh,
C. T. Nat. Prod. Rep. 1999, 16, 249.
(28) Cocito, C. Microbiol. Rev. 1979, 43, 145.
(29) Reed, J. W.; Kingston, D. G. J. Nat. Prod. 1986, 49, 626.
(30) Molinero, A. A.; Kingston, D. G.; Reed, G. H. J. Nat. Prod. 1989,
52, 99.
(31) Reed, J. W.; Purvis, M. B.; Kingston, D. G.; Biot, A.; Grossele,
F. J. Org. Chem. 1989, 54, 1161.
(32) Blanc, V.; Gil, P.; Bamas-Jacques, N.; Lorenzon, S.; Zagorec, M.;
Schleuniger, J.; Bisch, D.; Blanche, F.; Debussche, L.; Crouzet,
J.; Thibaut, D. Mol. Microbiol. 1997, 23, 191.

(33) Namwat, W.; Kinoshita, H.; Nihira, T. J. Bacteriol. 2002, 184,
4811.
(34) Namwat, W.; Kamioka, Y.; Kinoshita, H.; Yamada, Y.; Nihira,
T. Gene 2002, 286, 283.
(35) Thibaut, D.; Bisch, D.; Ratet, N.; Maton, L.; Couder, M.;
Debussche, L.; Blanche, F. J. Bacteriol. 1997, 179, 697.
(36) de Crecy-Lagard, V.; Blanc, V.; Gil, P.; Naudin, L.; Lorenzon,
S.; Famechon, A.; Bamas-Jacques, N.; Crouzet, J.; Thibaut, D.
J. Bacteriol. 1997, 179, 705.
(37) de Crecy-Lagard, V.; Saurin, W.; Thibaut, D.; Gil, P.; Naudin,
L.; Crouzet, J.; Blanc, V. Antimicrob. Agents Chemother. 1997,
41, 1904.
(38) Yamada, Y.; Sugamura, K.; Kondo, K.; Yanagimoto, M.; Okada,
H. J. Antibiot. (Tokyo) 1987, 40, 496.
(39) Kawachi, R.; Akashi, T.; Kamitani, Y.; Sy, A.; Wangchaisoonthorn, U.; Nihira, T.; Yamada, Y. Mol. Microbiol. 2000, 36,
302.
(40) Miller, M. B.; Bassler, B. L. Annu. Rev. Microbiol. 2001, 55, 165.
(41) Horinouchi, S. Front. Biosci. 2002, 7, d2045.
(42) Choi, S. U.; Lee, C. K.; Hwang, Y. I.; Kinosita, H.; Nihira, T.
Arch. Microbiol. 2003, 180, 303.
(43) Vannuffel, P.; Cocito, C. Drugs 1996, 51, 20.
(44) Hansen, J. L.; Moore, P. B.; Steitz, T. A. J. Mol. Biol. 2003, 330,
1061.
(45) Harms, J. M.; Schluenzen, F.; Fucini, P.; Bartels, H.; Yonath,
A. E. BMC Biol. 2004, 2, 4.
(46) Vannuffel, P.; Di Giambattista, M.; Cocito, C. Nucleic Acids Res
1994, 22, 4449.
(47) Porse, B. T.; Kirillov, S. V.; Awayez, M. J.; Garrett, R. A. RNA
1999, 5, 585.
(48) Chinali, G.; Di Giambattista, M.; Cocito, C. Biochemistry 1987,
26, 1592.
(49) Cocito, C.; Di Giambattista, M.; Nyssen, E.; Vannuffel, P. J.
Antimicrob. Chemother. 1997, 39 Suppl A, 7.
(50) Bouanchaud, D. H. J. Antimicrob. Chemother. 1997, 39 (Suppl.
A), 15.
(51) Singh, K. V.; Weinstock, G. M.; Murray, B. E. Antimicrob. Agents
Chemother. 2002, 46, 1845.
(52) Ross, J. I.; Eady, E. A.; Cove, J. H.; Cunliffe, W. J.; Baumberg,
S.; Wootton, J. C. Mol. Microbiol. 1990, 4, 1207.
(53) Alignet, J.; Loncle, V.; el Sohl, N. Gene 1992, 117, 45.
(54) Alignet, J.; El Solh, N. Gene 1997, 202, 133.
(55) Weisblum, B. Antimicrob. Agents Chemother. 1995, 39, 577.
(56) Leclercq, R.; Courvalin, P. Antimicrob. Agents Chemother. 1991,
35, 1267.
(57) Tait-Kamradt, A.; Davies, T.; Appelbaum, P. C.; Depardieu, F.;
Courvalin, P.; Petitpas, J.; Wondrack, L.; Walker, A.; Jacobs,
M. R.; Sutcliffe, J. Antimicrob. Agents Chemother. 2000, 44, 3395.
(58) Sugantino, M.; Roderick, S. L. Biochemistry 2002, 41, 2209.
(59) Kehoe, L. E.; Snidwongse, J.; Courvalin, P.; Rafferty, J. B.;
Murray, I. A. J. Biol. Chem. 2003, 278, 29963.
(60) Hou, C. T.; Perlman, D.; Schallock, M. R. J. Antibiot. 1970, 23,
35.
(61) Kim, C. H.; Otake, N.; Yonehara, H. J. Antibiot. 1974, 27, 903.
(62) Le Goffic, F.; Capmau, M. L.; Abbe, J.; Cerceau, C.; Dublanchet,
A.; Duval, J. Ann. Microbiol. (Paris) 1977, 128B, 471.
(63) Allignet, J.; Loncle, V.; Mazodier, P.; el Solh, N. Plasmid 1988,
20, 271.
(64) Suzuki, N.; Lee, C. K.; Nihira, T.; Yamada, Y. Antimicrob. Agents
Chemother. 1998, 42, 2985.
(65) Spangler, S. K.; Jacobs, M. R.; Appelbaum, P. C. Antimicrob.
(66) Pankuch, G. A.; Kelly, L. M.; Lin, G.; Bryskier, A.; Couturier,
C.; Jacobs, M. R.; Appelbaum, P. C. Antimicrob. Agents Chemother. 2003, 47, 3270.
(67) Bonfiglio, G.; Furneri, P. M. Expert Opin. Investig. Drugs 2001,
10, 185.
(68) Slee, A. M.; Wuonola, M. A.; McRipley, R. J.; Zajac, I.; Zawada,
M. J.; Bartholomew, P. T.; Gregory, W. A.; Forbes, M. Antimicrob. Agents Chemother. 1987, 31, 1791.
(69) Gregory, W. A.; Brittelli, D. R.; Wang, C. L.; Wuonola, M. A.;
McRipley, R. J.; Eustice, D. C.; Eberly, V. S.; Bartholomew, P.
T.; Slee, A. M.; Forbes, M. J. Med. Chem. 1989, 32, 1673.
(70) Barbachyn, M. R.; Ford, C. W. Angew. Chem., Int. Ed. 2003, 42,
2010.
(71) Genin, M. J.; Allwine, D. A.; Anderson, D. J.; Barbachyn, M. R.;
Emmert, D. E.; Garmon, S. A.; Graber, D. R.; Grega, K. C.;
Hester, J. B.; Hutchinson, D. K.; Morris, J.; Reischer, R. J.; Ford,
C. W.; Zurenko, G. E.; Hamel, J. C.; Schaadt, R. D.; Stapert, D.;
Yagi, B. H. J. Med. Chem. 2000, 43, 953.
(72) Barbachyn, M. R.; Hutchinson, D. K.; Brickner, S. J.; Cynamon,
M. H.; Kilburn, J. O.; Klemens, S. P.; Glickman, S. E.; Grega,
K. C.; Hendges, S. K.; Toops, D. S.; Ford, C. W.; Zurenko, G. E.
J. Med. Chem. 1996, 39, 680.
(73) Tucker, J. A.; Allwine, D. A.; Grega, K. C.; Barbachyn, M. R.;
Klock, J. L.; Adamski, J. L.; Brickner, S. J.; Hutchinson, D. K.;
Ford, C. W.; Zurenko, G. E.; Conradi, R. A.; Burton, P. S.; Jensen,
R. M. J. Med. Chem. 1998, 41, 3727.

(74) Brickner, S. J.; Hutchinson, D. K.; Barbachyn, M. R.; Manninen,
P. R.; Ulanowicz, D. A.; Garmon, S. A.; Grega, K. C.; Hendges,
S. K.; Toops, D. S.; Ford, C. W.; Zurenko, G. E. J. Med. Chem.
1996, 39, 673.
(75) Barbachyn, M. R.; Toops, D. S.; Grega, K. C.; Hendges, S. K.;
Ford, C. W.; Zurenko, G. E.; Hamel, J. C.; Schaadt, J. D.; Stapert,
D.; Yagi, B. H.; Buysse, J. M.; Demyan, W. F.; Kilburn, J. O.;
Glickman, S. E. Bioorg. Med. Chem. Lett. 1996, 6, 1009.
(76) Barrett, J. F. Curr. Opin. Investig. Drugs 2000, 1, 181.
(77) Lin, A. H.; Murray, R. W.; Vidmar, T. J.; Marotti, K. R.
(78) Zhou, C. C.; Swaney, S. M.; Shinabarger, D. L.; Stockman, B. J.
(79) Shinabarger, D. L.; Marotti, K. R.; Murray, R. W.; Lin, A. H.;
Melchior, E. P.; Swaney, S. M.; Dunyak, D. S.; Demyan, W. F.;
Buysse, J. M. Antimicrob. Agents Chemother. 1997, 41, 2132.
(80) Swaney, S. M.; Aoki, H.; Ganoza, M. C.; Shinabarger, D. L.
(81) Patel, U.; Yan, Y. P.; Hobbs, F. W., Jr.; Kaczmarczyk, J.; Slee,
A. M.; Pompliano, D. L.; Kurilla, M. G.; Bobkova, E. V. J. Biol.
Chem. 2001, 276, 37199.
(82) Thompson, J.; OConnor, M.; Mills, J. A.; Dahlberg, A. E. J. Mol.
Biol. 2002, 322, 273.
(83) Bobkova, E. V.; Yan, Y. P.; Jordan, D. B.; Kurilla, M. G.;
Pompliano, D. L. J. Biol. Chem. 2003, 278, 9802.
(84) Tsiodras, S.; Gold, H. S.; Sakoulas, G.; Eliopoulos, G. M.;
Wennersten, C.; Venkataraman, L.; Moellering, R. C.; Ferraro,
M. J. Lancet 2001, 358, 207.
(85) Wilson, P.; Andrews, J. A.; Charlesworth, R.; Walesby, R.; Singer,
M.; Farrell, D. J.; Robbins, M. J. Antimicrob. Chemother. 2003,
51, 186.
(86) Johnson, A. P.; Tysall, L.; Stockdale, M. V.; Woodford, N.;
Kaufmann, M. E.; Warner, M.; Livermore, D. M.; Asboth, F.;
Allerberger, F. J. Eur. J. Clin. Microbiol. Infect. Dis. 2002, 21,
751.
(87) Prystowsky, J.; Siddiqui, F.; Chosay, J.; Shinabarger, D. L.;
Millichap, J.; Peterson, L. R.; Noskin, G. A. Antimicrob. Agents
Chemother. 2001, 45, 2154.
(88) Kloss, P.; Xiong, L.; Shinabarger, D. L.; Mankin, A. S. J. Mol.
Biol. 1999, 294, 93.
(89) Xiong, L.; Kloss, P.; Douthwaite, S.; Andersen, N. M.; Swaney,
S.; Shinabarger, D. L.; Mankin, A. S. J. Bacteriol. 2000, 182,
5325.
(90) Jones, R. N.; Della-Latta, P.; Lee, L. V.; Biedenbach, D. J. Diagn.
(91) Rahim, S.; Pillai, S. K.; Gold, H. S.; Venkataraman, L.; Inglima,
K.; Press, R. A. Clin. Infect. Dis. 2003, 36, E146.
(92) Sakoulas, G.; Gold, H. S.; Venkataraman, L.; Moellering, R. C.,
Jr.; Ferraro, M. J.; Eliopoulos, G. M. J. Antimicrob. Chemother.
2003, 51, 1039.
Mukhtar and Wright

(93) Meka, V. G.; Pillai, S. K.; Sakoulas, G.; Wennersten, C.;
Venkataraman, L.; DeGirolami, P. C.; Eliopoulos, G. M.; Moellering, R. C., Jr.; Gold, H. S. J. Infect. Dis. 2004, 190, 311.
(94) Meka, V. G.; Gold, H. S.; Cooke, A.; Venkataraman, L.; Eliopoulos, G. M.; Moellering, R. C., Jr.; Jenkins, S. G. J. Antimicrob.
Chemother. 2004.
(95) Willems, R. J.; Top, J.; Smith, D. J.; Roper, D. I.; North, S. E.;
Woodford, N. Antimicrob. Agents Chemother. 2003, 47, 3061.
(96) Sander, P.; Belova, L.; Kidan, Y. G.; Pfister, P.; Mankin, A. S.;
Bottger, E. C. Mol. Microbiol. 2002, 46, 1295.
(97) Singh, U.; Raju, B.; Lam, S.; Zhou, J.; Gadwood, R. C.; Ford, C.
W.; Zurenko, G. E.; Schaadt, R. D.; Morin, S. E.; Adams, W. J.;
Friis, J. M.; Courtney, M.; Palandra, J.; Hackbarth, C. J.; Lopez,
S.; Wu, C.; Mortell, K. H.; Trias, J.; Yuan, Z.; Patel, D. V.;
Gordeev, M. F. Bioorg. Med. Chem. Lett. 2003, 13, 4209.
(98) Sbardella, G.; Mai, A.; Artico, M.; Loddo, R.; Setzu, M. G.; La
Colla, P. Bioorg. Med. Chem. Lett. 2004, 14, 1537.
(99) Gordeev, M. F.; Hackbarth, C.; Barbachyn, M. R.; Banitt, L. S.;
Gage, J. R.; Luehr, G. W.; Gomez, M.; Trias, J.; Morin, S. E.;
Zurenko, G. E.; Parker, C. N.; Evans, J. M.; White, R. J.; Patel,
D. V. Bioorg. Med. Chem. Lett. 2003, 13, 4213.
(100) Hubschwerlen, C.; Specklin, J. L.; Sigwalt, C.; Schroeder, S.;
Locher, H. H. Bioorg. Med. Chem. 2003, 11, 2313.
(101) Hubschwerlen, C.; Specklin, J. L.; Baeschlin, D. K.; Borer, Y.;
Haefeli, S.; Sigwalt, C.; Schroeder, S.; Locher, H. H. Bioorg. Med.
Chem. Lett. 2003, 13, 4229.
(102) Ehrlich, J.; Bartz, Q. R.; Smith, R. M.; Joslyn, D. A.; Burkholder,
P. R. Science 1947, 106, 417.
(103) Yunis, A. A. Am. J. Med. 1989, 87, 44N.
(105) Long, K. S.; Porse, B. T. Nucleic Acids Res. 2003, 31, 7208.
(106) Murray, I. A.; Shaw, W. V. Antimicrob. Agents Chemother. 1997,
41, 1.
(107) Mason, D. J.; Deietz, A.; DeBoer, C. In Antimicrobial Agents and
Chemotherapy-1962; Sylvester, J. C., Ed.; American Society for
Microbiology: Ann Arbor, MI, 1963.
(108) Kasten, M. J. Mayo Clin. Proc. 1999, 74, 825.
(109) Kavanigh, F.; Hervey, A.; Robbins, W. J. Proc. Natl. Acad. Sci.
U.S.A. 1951, 37.
(110) Springer, D. M.; Sorenson, M. E.; Huang, S.; Connolly, T. P.;
Bronson, J. J.; Matson, J. A.; Hanson, R. L.; Brzozowski, D. B.;
LaPorte, T. L.; Patel, R. N. Bioorg. Med. Chem. Lett. 2003, 13,
1751.
(111) Poulsen, S. M.; Karlsson, M.; Johansson, L. B.; Vester, B. Mol.
Microbiol. 2001, 41, 1091.
(112) Asai, T.; Zaporojets, D.; Squires, C.; Squires, C. L. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 1971.
CR030110Z
Chem. Rev. 2005, 105, 543558
543
Genetic Approaches to Polyketide Antibiotics. 1

Robert McDaniel,* Mark Welch, and C. Richard Hutchinson
Kosan Biosciences, 3832 Bay Center Place, Hayward, California 94545
Contents
1. Introduction
2. Biochemistry and Genetics of Polyketide
Biosynthesis
3. Polyketide Antibiotic Gene Clusters
3.3. Ansamycins
4. Overview of Methodologies
4.1. Modular PKSs
4.2. Aromatic PKSs
5. New Developments in Modular PKS Manipulation
5.1. Modeling and Engineering of PKS Domains
5.1.1. Acyltransferases
5.1.2. Ketoreductases
5.2. Type II Thioesterase
5.3. Cyclization/Termination
5.4. Intermodular and Intramodular
Communication
5.5. Precursor Engineering
5.6. Tailoring Pathways
6. Creating and Improving Microbial Production
Systems
7. Conclusions
8. Acknowledgments
9. References
543
544
546
546
547
549
552
552
557
553
553
553
553
554
554
554
555
555
556
556
557
0
1. Introduction
Microbial metabolites have for decades been a rich
source of natural product drug leads and therapeutically important drugs.1,2 The terms engineered biosynthesis and combinatorial biosynthesis encompass techniques aimed at increasing chemical diversity
of natural products by altering the function of the
genes and enzymes that govern the production of
these metabolites. The classes of compounds most
frequently associated with this are polyketides and
nonribosomal peptides.3 In particular, the predictable
relationship between the structure and function of
the modular type of microbial polyketide synthases
(PKSs) has enabled genetic manipulation of the
biosynthetic pathways for production of novel variants of classes of naturally occurring compounds,
such as macrolide antibiotics4,5 and antitumor compounds.6 The goals of the approach resemble those
robert.mcdaniel@codexis.com.
of medicinal chemists who synthesize analogues and

derivatives of lead compounds in an attempt to
improve upon existing drugs or find new ones.
Expression of native or engineered PKS genes, as
well as those that govern precursor supply and postPKS modification of the metabolite, in heterologous
hosts is also an important aspect of developing
commercial systems for drug production. The following review presents an overview of the initial technology enabling studies pertaining to polyketide
manipulation (which have been covered in several
previous reviews4-11) and a comprehensive review of
the more recent advances that have set the stage for
broader use of the technology. Although this review
focuses on antibacterial polyketides, engineered or
combinatorial biosynthesis extends to other classes
of polyketides of therapeutic interest including antitumor, immunosuppressive, antifungal, and other
compounds.
Polyketides are a notable class of natural products
with a number of well-established successes in clinical and agricultural applications. Examples include
the antibiotics erythromycin, tylosin, rifamycin, and
the tetracyclines, immunosuppressants FK506 and
rapamycin, and antitumor agents doxorubicin and
mithramycin. Two early seminal works led to the
promise that the genes for secondary metabolites
could be used to create novel structures. The first was
by Hopwood and co-workers12,13 in which genes from
two different antibiotic gene clusters, medermycin or
granaticin with actinorhodin, were used in combination for the first time to produce a hybrid aromatic
polyketide. The second was work performed by Leadlay and co-workers at Cambridge14 and by Katz and
co-workers at Abbot Laboratories,15 who uncovered
the first genes for a modular (type I) PKS in the
erythromycin gene cluster. These megasynthases
have since been the primary focus of polyketide
engineering, and many of the proof of concept experiments have been performed using the erythromycin
PKS.
Modular PKSs make many large and complex
natural products that are difficult to synthesize or
modify by chemistry. Since the discovery of the
erythromycin PKS, on the order of a few hundred
new compounds have been created through modification or heterologous expression of polyketide gene
clusters. These numbers suggest that, in its current
state, polyketide engineering is best suited for optimization of existing lead compounds by specifically
chosen structural modifications rather than random

Robert McDaniel is a Senior Scientist at Kosan Biosciences, Inc. and

has worked extensively in the field of polyketide biosynthesis and
engineering for more than 12 years. He received his B.S. degree in
Chemical Engineering from the University of Colorado at Boulder and
M.S. and Ph.D. degrees in Chemical Engineering at Stanford University.
At Stanford he began his career in polyketide research in the laboratory
of Professor Chaitan Khosla, collaborating with Professor Sir David
Hopwood at the John Innes Institute. He then joined Kosan during its
genesis in order to transfer and further develop polyketide engineering
technology that had been developed at Stanford. He has worked at Kosan
for the past 8 years on several projects, including anti-infectives, and has
published several research articles on the subject. His current interests
are focused on ways to improve polyketide production from natural and
engineered polyketide producing microorganisms.
Mark Welch has over 10 years experience in RNA and protein engineering
with particular emphasis on directed evolution methods to create novel
biocatalysts. Mark received his B.A. degree in Biology from the University
of California at Santa Cruz with College Honors and Highest Honors in
the Major. He received his doctoral degree in Molecular, Cellular, and
Developmental Biology from the University of Colorado at Boulder in 1996
for studies of the role of 23SrRNA in the peptidyl transferase activity of
the ribosome in the laboratory of Dr. Michael Yarus. In 1998 Mark joined
Maxygen, Inc., a biotechnology company specializing in the application
of DNA shuffling methods for biomolecule-directed evolution. As a scientist
in the New Technology group at Maxygen, Mark managed an interdisciplinary team responsible for development and application of novel highthroughput assays to screen for a variety of target enzyme activities. He
joined the New Technology group at Kosan in April 2003 and is currently
leading a group focused of novel PKS engineering.
generation of large libraries for screening against

targets. Some of the applications for which this
technology is used at Kosan include the following: (i)
production of analogues of natural products to improve activity or pharmacodynamics; (ii) introducing
reactive groups into compounds for further chemical
modification; (iii) generating intermediates for chemical synthesis of natural products; and (iv) increasing
McDaniel et al.
C. Richard (Dick) Hutchinson is a Research Fellow at Kosan Biosciences,

Hayward, CA, following 3 years as the Vice President of New Technologies
after joining the company in March 2000. He spent 26 years at the
University of Wisconsin, Madison, as a Professor of Medicinal Chemistry
and Bacteriology and is now a Professor Emeritus. Prior to this he was
on the faculty of the School of Pharmacy, University of Connecticut (1970
74), conducted postdoctoral research with Sir Alan Battersby (1971), and
received his Ph.D. degree in Organic Chemistry from the University of
Minnesota (1970), working with Edward Leete, and B.S. degree in
Pharmacy from Ohio State University (1966), where he carried out
undergraduate research with Raymond Doskotch and Jack Beal. His
research has involved studies of the biosynthesis of natural products with
special emphasis on the molecular genetics and biochemistry of antibiotic
production in microorganisms. He has published over 235 scientific papers
and patents and received several prestigious awards, including Guggenheim and Fullbright fellowships, the Charles Thom Research Achievement
Award of the Society for Industrial Microbiology, the AACP Paul Dawson
Biotechnology Award, the Research Achievement Award of the American
Society for Pharmacognosy, and a Distinguished Alumni Award from Ohio
State University.
the availability or reducing the cost of natural

products (overproduction). Although the architecture
of PKSs suggests that very large libraries could be
theoretically generated, the challenges to doing this
in the laboratory require a further understanding of
PKS structure, specificity, and protein interactions
as well as technologies to perform genetic manipulation more efficiently.
Other approaches to discovering new drugs from
natural products are not discussed here but typically
rely on high-throughput DNA sequencing. For example, novel sets of secondary metabolism genes are
found by either whole genome sequencing or DNA
libraries enriched for secondary metabolite genes and
subsequently expressed through manipulation of
growth conditions or in heterologous hosts. DNA
libraries generated from unculturable organisms or
environmental samples can also be screened in this
manner or cloned directly into heterologous hosts and
screened for biological activity. Finally, the structural
modification of aromatic polyketides, made by socalled type II PKSs, is less amenable to gene manipulation and therefore is not reviewed here, except
to note recent advances in understanding this area.
2. Biochemistry and Genetics of Polyketide

Biosynthesis
Polyketide synthases evolved the capability of
making a vast number of compounds from the same
classes of substrates used by fatty acid synthases and
Genetic Approaches to Polyketide Antibiotics
Figure 1. Illustration of the mechanism of the type II PKS involved in the biosynthesis of tetracenomycin F1 and C. The
PKS consists of individual protein subunits that act in concert to assemble the acetate starter unit (produced by
decarboxylation of enzyme-bound malonate) and nine chain extender units into a TcmM-bound decaketide by an iterative
process involving a malonyl-CoA:ACP acyltransferase (MCAT, which is shared with fatty acid biosynthesis) and the proteins
TcmJ, TcmK, TcmL, and TcmM. The decaketide is cyclized to Tcm F2 by the TcmN enzyme, with assistance by TcmJ;
then Tcm F2 is cyclized once more by TcmI to form Tcm F1. The latter intermediate is converted to tetracenomycin C by
tailoring enzymes.
Figure 2. Illustration of the mechanism of the type I modular PKS involved in the biosynthesis of 6dEB. Each of the
DEBS subunits is represented by a broad, open arrow containing the relevant domains in each module. Key enzymebound intermediates of carbon-chain assembly are shown bound to the ACP domains. Assembly begins at the loading
didomain of the first DEBS subunit upon attachment of propionate which then reacts with ACP-bound 2-methylmalonate,
obtained from its CoA ester. Further equivalents of 2-methylmalonyl-CoA are used by DEBS to produce 6dEB, as explained
in the text. 6dEB is then converted to the erythromycin A-D glycosides by tailoring enzymes.
largely by the same type of biochemistry. Examples

of two types of PKSs and their associated polyketides
are illustrated in Figures 1 and 2. Type II, or
aromatic, PKSs (Figure 1) consist of a collection of
largely monofunctional proteins that catalyze the
formation of typically polycyclic aromatic compounds,
usually from acetate and malonate only. Type I, or
modular, PKSs (Figure 2), in contrast, use large
multifunctional proteins to make polyoxygenated,
aliphatic compounds from several different kinds of

acyl-Coenzyme A substrates.
The so-called modular PKSs15-17 have led to the
most fruitful genetic engineering route to structural
variants18,19 of polyketides that are important therapeutic drugs, like the antibacterial erythromycin A
(Figure 2) or experimental agents such as 17-allylamino-17-demethoxygeldanamycin (17-AAG) that currently is undergoing clinical trials as an antitumor
drug. [Two other types of PKSs, the fungal nonmodular type I and plant chalcone type III, are not
discussed here; for reviews, cf. refs 20 and 21.] A
modular PKS is a massive complex of large, multifunction proteins. Within each protein are one or
more modules, each with different combinations of
domains that function like the constituent biochemical activities of fatty acid synthases to catalyze a
single cycle of polyketide chain elongation and modification. 6-Deoxyerythronolide B synthase (DEBS) is
the PKS that forms the backbone of the erythromycins and is encoded by the three genes, eryAIIII14,15,22 (Figure 2). DEBS catalyzes formation of
6-deoxyerythronolide B (6dEB) by the successive
condensation of one propionyl and six 2-methylmalonyl molecules in their activated Coenzyme A (CoA)
thioester form. Each of the three subunits of DEBS
have two extender modules, containing the activities
needed for one cycle of polyketide chain elongation,
as illustrated by the structures of the six enzymebound intermediates in Figure 2. In addition, the first
module is preceded by a loading didomain for the
starter unit, and the last is followed by a thioesterase
domain for product release and cyclization. Every
extender module contains a ketosynthase (KS), an
acyltransferase (AT), and an acyl carrier protein
(ACP) domain that together catalyze a two-carbon
extension of the chain. In DEBS, the AT domains of
extender modules are specific for 2-methylmalonylCoA, while the AT in the loading module uses
propionyl-CoA. After each two-carbon unit condensation, the oxidation state of the -carbon is either
retained as a ketone (module 3) or modified to a
hydroxyl, methenyl, or methylene group by the
presence of a ketoreductase (KR) (module 2), a KR
+ a dehydratase (DH), or a KR + DH + an enoyl
reductase (ER) (module 4), respectively.
In effect, the AT specificity and the types of
catalytic domains within a module serve as codes for
the structure of each two-carbon unit; the order of
the modules in a PKS specifies the sequence of the
distinct two-carbon units, and the number of modules
determines the length of the polyketide chain. Variations in the acyl-CoA substrates used by a modular
PKS, the number of domains within a module, and
the number of modules in the PKS are responsible
for establishing the first set of structural characteristics of the polyketide, including the chirality of
hydroxyl- and alkyl-bearing carbon centers. After
this, the kinds of biochemical transformations the
compound produced by the PKS undergoes, such as
glycosylation or oxidation, are dictated by the tailoring enzymes that establish the final structure.
Consequently, engineering a microorganism to produce novel polyketides can involve altering only the
PKS genes or the tailoring genes as well.
3. Polyketide Antibiotic Gene Clusters

The 14-membered macrolides are exemplified by
erythromycin, for which the complete gene cluster
has been sequenced.14,15,22-28 The complete or partial
gene clusters for oleandomycin,29-34 megalomicin,35
McDaniel et al.
and picromycin36 are also known (Figure 3). The

overall DEBS-like architecture of the PKS genes is
conserved with the exception of the picromycin PKS
(PicPKS) (Figure 4), in which the last two modules
are contained on two separate proteins. Both DEBS
and the megalomicin PKS (MegPKS) produce 6dEB
and have high sequence similarity.30 The oleandomycin PKS (OlePKS) produces 8,8a-deoxyoleandolide, which is equivalent to 6dEB derived from a
two-carbon starter unit rather than a three-carbon
starter unit. An important distinction between DEBS
(and MegPKS) and OlePKS is the composition and
mechanism of their loading domains. In DEBS, an
AT loads propionyl-CoA (as well as other acyl CoAs
with lower efficiency) whereas the OlePKS loading
domain contains an AT specific for malonyl-CoA and
a KSq domain, an inactive condensation domain that
serves to decarboxylate the ACP-bound substrate to
acetyl-ACP for use as the starter.30,37
The PicPKS differs from the other three PKSs in
several ways. The loading domain of the PicPKS also
contains a KSq domain but an AT that is specific for
methylmalonyl-CoA, which is then decarboxylated to
propionyl-CoA for use as the starter. Module 2 of the
PicPKS extends using a malonyl-CoA rather than a
methylmalonyl-CoA unit and also contains a DH
domain in addition to the KR -keto modifying
activity leading to the C-10/C-11 alkene. Finally,
module 6 lacks a KR, resulting in the 3-ketone of
narbonolide. The PicPKS also produces approximately stoichiometric amounts of the 12-membered
lactone precursor of methymycin, methynolide, via
a mechanism referred to as domain skipping.10,36,38,39
All four PKSs have been expressed heterologously
in Streptomyces coelicolor and/or Streptomyces lividans and their corresponding aglycones (1-3) produced in good yield.30,35,38,40 The MegPKS contains
two regions longer than 0.5 kb with 100% sequence
identity in modules 2 and 6, leading to plasmid
instability caused by homologous recombination during plasmid-borne heterologous expression.41 Similarly, OlePKS contains over 1.2 kb of identical
sequence in modules 2 and 5. Neither DEBS nor
PicPKS possess such repeats.
The polyketide macrolactones are further modified
by a series of glycosylation(s) and oxidation(s). Synthesis of 6dEB is followed by C6-hydroxylation by the
product of eryF to yield erythronolide B. Addition of
the sugar L-mycarose (via thymidine diphospho(TDP)-mycarose) yields 3-O-R-mycarosylerythronolide B, and the addition of D-desosamine (via TDPdesosamine) yields erythromycin D. The two sugars
are produced by independent sets of genes designated
eryB (mycarose) and eryC (desosamine), which flank
the PKS genes in the cluster. The final steps of
erythromycin biosynthesis are hydroxylation of erythromycin D to yield erythromycin C by a second P450
enzyme encoded by eryK and O-methylation of the
mycarosyl residue by the eryG product to yield the
cladinosyl moiety of erythromycin A and B. Megalomicin tailoring resembles that of erythromycin except
that a third deoxysugar, L-megosamine, is added to
the C-6 hydroxyl and mycarose is acylated at C-3
and C-4. The set of genes for both TDP-L-mego-
Figure 3. Fourteen-membered macrolide antibiotics and gene clusters.
samine formation (megD) and the O-acyl transferase

(megY) have been identified in the meg cluster and
expressed in the erythromycin-producing strain, Saccharopolyspora erythraea, to generate megalomicins
in that strain.35
Oleandomycin contains an epoxide at C-8,C8a,
introduced by OleP, a P-450 oxidase. The oleP gene
has been coexpressed with DEBS to generate 8-hydroxy derivatives of 6dEB and 3-O-glycosylated 6dEB
derivatives.30,42 The two deoxysugars added to oleandolide are D-desosamine and L-oleandrose (3-Omethyl-L-olivose). The genes encoding TDP-L-olivose
biosynthesis (oleW, oleV, oleL, oleS, oleE, and oleU)
and its conversion to L-oleandrose (oleY) flank the
OlePKS genes. These genes were used to generate
3-O-olivosyl and 3-O-oleandrosyl erythronolide B
analogues34 (see below). The genes encoding D-desosamine biosynthesis are also located within the ole
cluster and homologous to those in the ery, meg, and
pic clusters.
Picromycin and methymycin contain only the single
deoxysugar, desosamine, encoded by the des genes.
Gene knockouts of desVI,43 encoding the N-methyltransferase, desV,44 encoding a transaminase, and
desI,45 encoding a putative C-4 dehydrase, resulted
in production of methymycin analogues with altered
glycoside moieties. Attachment of desosamine to both

the 12- and 14-membered lactones is performed by
the same glycosyl transferase, encoded by desVII.
After glycosylation, PicK (also called PicC), a P-450
oxidase, hydroxylates C-12 (picromycin) or C-10
(methymycin).46,47

The 16-membered macrolide antibiotics characterized thus far fall into three different classes of
polyketide backbones. These are represented by
tylactone (4), platenolide (5), and mycinamicin lactone (6)/chalcolactone (7) (Figure 4), the polyketide
components of tylosin, niddamycin/spiramycin, and
mycinamicin/chalcomycin, respectively (Figure 5).
The PKSs encoding each have been sequenced.48-51
The organization of modules within subunits is
conserved across all of the PKSs (Figure 4), comprised of the loading domain, module 1 and module
2 on the first subunit, module 3 on the second
subunit, modules 4 and 5 on the third subunit,
module 6 on the fourth subunit, and module 7 on the
fifth subunit.
The chief difference among the 16-membered macrolide PKSs is the composition of extender units
McDaniel et al.
Figure 4. Organization of 14- and 16-membered macrolide PKSs and polyketide products.
utilized. Mycinamicin lactone and chalcolactone are

derived exclusively from malonyl-CoA and methylmalonyl-CoA and differ only in the starter unit and
C-3 functionality. Modules 5 of the tylactone and the
platenolide PKSs utilize a 2-ethylmalonyl-CoA extender unit. A crotonyl-CoA reductase gene, ccr, is
clustered with the tylosin biosynthetic genes (Figure
5) and presumably contributes to the flux of ethyl-
malonyl-CoA synthesis from four-carbon acyl-CoA

pools during tylosin production.50 Homologues of ccr
are present in other PKS gene clusters requiring
ethylmalonyl-CoA, and a homologue from Streptomyces collinus was expressed in S. erythraea containing a modified DEBS construct, which then incorporated ethylmalonyl-CoA at module 4.52 In addition
to the other three precursors, the platenolide PKS
Figure 5. Sixteen-membered macrolide antibiotics and gene clusters.
also incorporates a 2-methoxymalonyl extender unit

at module 6. Genes encoding the biosynthetic pathway of methoxymalonyl-ACP have been found in the
FK520,53 ansamitocin,53 and geldanamycin gene clusters54 (see below) and have been used to produce this
precursor in other hosts.55 Presumably, these sets of
genes exist in the unsequenced regions of the 16membered macrolide gene clusters that produce
platenolide.
The genes for the biosynthetic pathways for all
three deoxysugars attached to tylosin, D-mycaminose,
L-mycarose, and D-mycinose (added as D-allose and
modified to mycinose after attachment), have been
sequenced from S. fradiae56 (Figure 5). The tylI and
tylH genes, both encoding P-450 oxidases, are responsible for the oxidations that occur on the C-6
ethyl and C-14 methyl branches of tylosin, respectively. Chalcomycin is one of the few macrolides that
does not contain an amino-sugar at C-5 but rather
has the neutral deoxysugar, D-chalcose, at that position (Figure 5). The putative genes for TDP-Dchalcose formation have been identified in the chalcomycin51 and lankamycin57 gene clusters. The genes
for the remaining chalcomycin modifications, including TDP-mycinose biosynthesis/attachment and the
two P-450s which hydroxylate C-8 and oxygenate
C-12/C-13, have also been sequenced in the chalcomycin gene cluster.51
3.3. Ansamycins
The ansamycins are related to the macrolides
biosynthetically but differ in the choice of starter unit
(3-amino-5-hydroxybenzoic acid (AHBA)) and the lack
of glycosylation. Formation of a macrolactam between
the terminal carboxyl and 3-amino group of AHBA,
instead of a macrolactone as above, results in a
characteristic basket with handle molecular conformation.58 Rifamycin, geldanamycin, herbimycin,
McDaniel et al.
Figure 6. Ansamycin family of polyketides and gene clusters.
and the ansamitocins (Figure 6) are ansamycins for

which the biosynthetic genes have been cloned and
characterized. Derivatives of rifamycin are widely
used in the treatment of tuberculosis, whereas geldanamycin analogues are undergoing clinical trials as
antitumor drugs. The ansamitocins and closely related maytansine (found in plant extracts) are among
the most cytotoxic polyketides known and are usually
referred to as maytansinoids.59
Investigations of the rifamycin genes began with
studies of the biosynthesis of AHBA at the enzymatic
level, following the proof through isotopic labeling
experiments that AHBA is the precursor of the
studied ansamycins. The rifK gene was cloned by
reverse genetics using information about the amino
acid sequence of the AHBA synthase, RifK,60 and
then used to obtain the remainder of the AHBA
biosynthesis, PKS and tailoring enzyme genes, by
gene cloning and sequencing experiments.61 Some of
these genes were cloned independently by researchers at Novartis.62 A notable feature of the modular
rifamycin PKS (RifPKS) is its tendency to shed the
enzyme-bound PKS intermediates spontaneously,63,64
quite unlike the macrolide PKSs. This directly demonstrates the processivity of the RifPKS and shows
that the 9,10-dihydronaphthoquinone ring of rifamycin is formed during this process to produce proansamycin X (8, Figure 7). Largely oxidative tailoring
reactions remodel proansamycin X via rifamycin W
(not shown) into the characteristic ansamycin framework of rifamycin B.65 Initial attempts to modify the
function of the RifPKS by the same type of domain
inactivation experiments used for DEBS and other
macrolactone PKSs (see below) have been greatly
complicated by the spontaneous shedding of truncated PKS assembly intermediates (Y. Doi-Katayama, Y. J. Yoon, C. S. Park, and C. R. Hutchinson,
unpublished work).
Figure 7. Organization of ansamycin PKSs and polyketide products.
In the ansamitocin gene cluster three of the AHBA

biosynthesis genes are separated from the rest of the
asm genes by at least 30 kb of intervening DNA.59,66
Disruption of many of the tailoring enzyme genes has
been used to develop a picture of the sequence of
biosynthetic steps,67 and the substrate specificity of
the Asm19 O-acyltransferase has been defined through
studies of the purified enzyme.68 This enzyme distinguishes the ansamitocins (3-O-acylesters) from
maytansine (3-O-(N-acetyl-N-methyl-L-alanine ester).
As noted above, the asm cluster also contains a set
of five genes for formation of 2-methoxymalonate, one
of the substrates used by the asm PKS for chain
elongation. In fact, studies of these genes along with
their orthologs from the FK520-producing streptomycete have elucidated what is known about the
biosynthesis of this atypical substrate.55,69
Geldanamycin and herbimycin are made by distinct Streptomyces hygroscopicus strains, but their
biosynthesis and gene clusters are nearly identical.
Rascher et al.54 described a major portion of the
geldanamycin genes, which, like those of ansamitocin
biosynthesis, occur as two separate clusters. One
cluster (Figure 6) contains the PKS and tailoring
enzyme genes plus one AHBA biosynthesis gene,
gdmO; the rest of the AHBA biosynthesis genes lie
somewhere else in the chromosome yet are essential
for geldanamycin biosynthesis, but not primary me-
tabolism, because their disruption abolishes geldanamycin production (A. Rascher and C. R. Hutchinson,
unpublished results). The corresponding herbimycin
PKS gene cluster is nearly identical, lacking only the
gdmF (amide synthase) and gdmM (monooxygenase)
orthologs (A. Rascher, Z. Hu, and C. R. Hutchinson,
unpublished results). Of the gdm tailoring genes,
gdmM and gdmN have been characterized functionally by gene knockout experiments54 (A. Rascher and
C. R. Hutchinson, unpublished results). Extensive
engineering of the geldanamycin PKS (GdmPKS)
genes (Figure 7), aided by development of convenient
methods for their modification and expression from
integrative vectors, has provided several geldanamycin analogues.70 Some of these are undergoing evaluation as antitumor drugs at Kosan Biosciences in the
quest for analogues with reduced hepatotoxicity and
greater water solubility.
4. Overview of Methodologies
4.1. Modular PKSs
A number of strategies for reprogramming modular
PKSs at the genetic level have emerged over the past
decade, ranging from single point mutations to
multiple module replacements, all resulting in
polyketides with targeted structural modifications.
Most of these strategies have been discussed in
previous reviews,4-6,8,9,11,18,20 and an outline is presented here. Briefly, these strategies include active
site inactivation or replacement, module substitution,
subunit complementation, and precursor-directed
biosynthesis.
The most common method of engineering thus far
is AT substitution, in which the native AT domain
is replaced with an AT encoding a different starter
or extender unit specificity. The AT used for substitution is usually derived from a heterologous modular
PKS. All six of the methyl-malonyl-specific AT domains in DEBS have been successfully replaced by
malonyl-specific AT domains to create 6dEB or
erythromycin analogues lacking a corresponding
methyl branch.4,71-73 Similar replacements have also
been accomplished with the spiramycin,50 FK520,74
rapamycin (J. Kennedy, personal communication),
and geldanamycin PKSs (unpublished data). The
loading AT of DEBS has been replaced by loading
ATs or AT/KSq domains from the avermectin,75,76
tylosin, and oleandomycin and rapamycin PKSs77 for
production of C-13-substituted derivates of erythromycin. Replacing an AT domain in DEBS with an
ethylmalonyl-specific AT52 and a 2-methoxymalonylspecific AT55 domain to generate ethyl and methoxy
branches, respectively, in either erythromycin or
6dEB has also been performed successfully. In each
case, introduction of exogenous genes involved in
precursor metabolism (ethylmalonyl-CoA and 2-methoxymalonyl-ACP) from other PKS gene clusters was
required for production in the engineered host.
Manipulation of -keto processing activities has
been accomplished by inactivation of domains, deletion of domains, and substitution of domains. Nearly
all of these examples have been performed with
DEBS. Examples of inactivation include the ER
McDaniel et al.
domain in module 4 by site-directed mutagenesis to

generate a 6,7-anhydro erythromycin derivative78
and deletion of KR domains in modules 5 and 6 to
generate C-3 and C-5 keto derivatives of 6-dEB,
respectively,15,79 and replacement of the KR domains
in modules 2, 5, and 6 to produce C10-11, C4-5, and
C-2-3 anhydro derivatives of 6dEB.79 Similarly,
substitution of KR domains in modules 2 and 6 with
a DH+ER+KR domain resulted in C-10 and C-3
deoxy 6dEB analogues.79
There is at least one report of a single whole
module substitution, DEBS module 2 with rifamycin
module 5.80 Although the activities of the two modules are identical and production of 6dEB was
reproduced, it suggests a useful alternative route in
situations where standard domain engineering falls
short. Construction of hybrid modules, subunits, and
subunit complementation within families of related
PKSs has been demonstrated in a number of cases.
Examples include DEBS/PicPKS,30,81 DEBS/rapamycin PKS,10,82 DEBS/OlePKS,41 PicPKS/OlePKS,83
PicPKS/TylPKS,81 and chalcomycin PKS/spiramycin
PKS.84 In many of these experiments the production
levels of the polyketide remain relatively high, demonstrating effective communication between noncognate subunits.
Precursor-directed biosynthesis combines the advantages of modern synthetic chemistry with complex
polyketide biochemistry to create a powerful approach to engineering polyketide chains with unique
features. This is achieved by constructing strains
which are blocked at a particular step in the pathway
(e.g., a KS domain) or in the ability to produce a
necessary precursor and supplying synthetically derived precursors in the form of N-acylcysteamine
thioesters.85,86 Several novel 14- and 16-membered
lactones have been produced using a KS1 derivative
of DEBS and supplying synthetic diketide or triketide
precursors.87-89 Side chains harboring halogens and
reactive groups for further modification by synthetic
chemistry have been incorporated into 6dEB. To date,
this has been the most successful approach to making
macrolides with potency equal to or better than
erythromycin.6
4.2. Aromatic PKSs

Although the field of polyketide gene engineering
was catalyzed by work with aromatic PKSs, this class
has not been exploited as intensely as modular PKSs.
Most successful attempts at engineering aromatic
polyketides result from mixing and matching of
separate individual enzyme subunits. For example,
pairs of ketosynthase and chain length factors (KS/
CLF), which together synthesize a length-specific
polyketide backbone, ketoreductases, cyclases, and
aromatases from different aromatic gene clusters
have been combined to manipulate chain length,
hydroxylation pattern, cyclization regiospecificity,
and aromaticity.90-95 A recent advance is the ability
to manipulate the choice of starter unit incorporation
through the use of a separate initiation module,
following identification of such a unit by the study
of the gene clusters for some of the rare aromatic
polyketides that use nonacetate starters.96 Cycliza-
alter the specificity of an AT domain102,103 and

inactivation of an AT, followed by complemention
with a separate trans-acting AT subunit104 (Figure
8). Both methods hold the advantage that structural
modification to a domain is minimized and potentially detrimental perturbations are avoided. In the
former case, the crystal structure of the E. coli fatty
acid malonyl transferase and sequence alignments
were used to identify residues putatively involved in
substrate specificity. These mutations were introduced into the AT4 domain of DEBS to produce
6-desmethyl-6dEB (11),102 although the mutations led
to relaxed specificitys6dEB (1) was also produceds
rather than a complete change in specificity. Actual
structures of modular PKS AT domains may help
refine the residues that are involved in substrate
selection and permit engineering of ATs with more
stringent specificity. In the second example, an AT6null mutant of DEBS was created by mutation of the
active site and the PKS complemented with a type
II fatty acid malonyltransferase (MAT) from S. coelicolor to produce 2-desmethyl-6dEB (12) in E. coli
with yields similar to that of the wild-type PKS.104
5.1.2. Ketoreductases
Figure 8. Examples of recent strategies for AT and KR
engineering. See sections 5.1.1 and 5.1.2 for details.
tion of the polyketide chain can be controlled with

appropriate choice of cyclases and aromatization
enzymes, provided one can be found with the desired
specificity. In the absence of such enzymes, however,
the final product structure(s) is determined by spontaneous cyclization, which may be difficult to predict.
Post-PKS steps such as oxidation, O-methylation, or
glycosylation are more amenable to alteration, as
illustrated by the work of Salas and co-workers with
mithramycinandrelatedchromanequinoneantibiotics.97-100
5. New Developments in Modular PKS

Manipulation
5.1. Modeling and Engineering of PKS Domains
5.1.1. Acyltransferases
Not every attempt at modular PKS AT domain
replacement is successful when a conserved set of
boundaries is used across different PKS modules. At
least one major reason for such unproductive AT
swaps is an apparent disruption in protein structure
so that chain elongation catalyzed by the KS and
ACP domains is severely attenuated.101 Recent experiments suggest that the choice of the domain
boundaries used to create the hybrid enzyme can be
a critical determinant of success. For example, our
group failed to engineer a productive malonyl-AT
domain replacement in module 4 of DEBS using
several different domain junctions.102 However, a set
of alternative junctions used by Leadlay and coworkers did result in a productive AT replacement73
(Figure 8).
Two alternative approaches to wholesale AT swaps
recently described are site-directed mutagenesis to
KR domains in modular PKSs catalyze stereospecific reduction and may be classified into two groups
according to the stereochemical outcome relative to
the polyketide backbone. Many examples of both
types of KRs exist. Inversion of an alcohol stereocenter is possible by replacing a KR of one class with
a KR from the other,79,105 but relatively few examples
have been reported, and all were performed at the
terminal module. Alteration of KR specificity in a
module preceding another module may require concomitant modification of the downstream KS so that
the altered stereocenter is recognized and processed.
Recent studies with sequence alignments of the two
KR types have identified differences in amino acid
residues that correlate with stereospecificity.106,107
The perfect correlation of these residues allows one
to predict stereochemical outcome in cases where a
gene sequence is known but the final product or
absolute stereochemistry is not known. Two models
of substrate binding relative to the NADPH cofactor
have been proposed to explain the relative outcomes
of ketoreduction,106,108 and Caffrey107 has proposed
mechanisms by which these residues may dictate
specificity in either model.
Homology modeling to the short-chain dehydrogenase/reductase family suggested a putative catalytic
triad in modular KRs.106 Point mutation of the
catalytic serine resulted in complete inactivation of
the KR6 domain in DEBS, producing 3-deoxy-3-oxo6dEB (13). Modification of the KR by this method
resulted in only the targeted analogue, whereas
deletion of the entire KR6 domain in DEBS affected
the specificity of the adjacent AT domain and led to
unexpected products (14) as well (Figure 8). This
particular example stresses the benefit of engineering
strategies which seek to minimize structural disturbances. The same mutation has been used to inactivate KR domains in the epothilone109 and geldana-
mycin PKSs (manuscript submitted) and produce

corresponding analogues.
5.2. Type II Thioesterase

Modular PKS gene clusters commonly contain
genes encoding a type II thioesterase (TEII), defined
initially through studies in E. coli as hydrolytic
enzymes acting on long-chain fatty acid thioesters.110
These enzymes are not involved in the terminal event
of PKS assembly like the thioester domains discussed
below and, to some degree, are dispensable because
inactivation of their genes does not always abolish
polyketide formation, although product titers usually
are drastically lowered, as in the case of tylosin111
and picromycin.36 In erythromycin biosynthesis by S.
erythraea, genetic and biochemical studies have
shown that the ery-ORF5 TEII enzyme favors hydrolysis of acetyl groups bound to the loading ACP
domain to ensure formation of 6dEB from a propionate instead of an acetate starter unit.112 Thus, the
most likely role of such enzymes is to edit the process
of PKS assembly by selective hydrolysis of misprimed
or -acylated active site cysteines, as directly shown
by Schwarzer et al.113 in the case of TEII enzymes of
nonribosomal peptide antibiotic biosynthesis. Understanding the exact contributions made by TEIIs will
be critical for optimizing productivity of engineered
modular PKSs.
5.3. Cyclization/Termination
The TE domains attached to the terminal modules
of PKSs are generally tolerant toward polyketide
chain length as well as substitutions at the C-2 and
C-3 positions of the lactone, although with varying
efficiencies. The TE domains from DEBS and PicPKS
have been studied in vitro89,114 and can cyclize lactone
ring sizes in the range of 6-16 carbons. The PicPKS
TE has a much higher preference for 3-keto versus
3-hydroxy substrates as compared to DEBS TE, and
fusion of the PicPKS TE to DEBS module 3 resulted
in an enzyme with greater efficiency than DEBS
module 3 with DEBS TE, indicating that the PicPKS
TE is a better catalyst for cyclizing 3-keto acyl
intermediates.114 Crystal structures of both TE domains from DEBS115 and PicPKS116 have now been
obtained and, in addition to providing clues about the
overall tertiary architecture of modular PKSs, provide a structural basis for potentially altering substrate specificity. In addition to cyclizing lactones
from a variety of macrolide PKSs, DEBS TE was used
recently to generate a novel pentaketide lactone from
the spinosyn PKS.117
The ansamycin PKSs, unlike macrolide PKSs,
utilize a separate protein for cyclization of the linear
polyketide to the corresponding macrolactam. These
amide synthases, RifF (rifamycin), GdmF (geldanamycin), and Asm9 (ansamitocin), are homologous to
N-acetyl CoA transferases. The degree of substrate
flexibility among this class of enzymes is not currently known but if similar to the TEs could be useful
for engineering novel lactams.
McDaniel et al.
5.4. Intermodular and Intramodular

Communication
Accurate programmed polyketide extension depends on several protein-protein interactions to
correctly orient catalytic domains relative to ACPlinked substrates as well as facilitate specific intermodular transfer of the growing polyketide chain. In
addition to such structural communication, proper
extension requires compatibility of enzyme specificities such that each intermediate is accepted through
the reaction sequence. Our ability to reorganize PKS
structure for the synthesis of novel compounds will
depend on our understanding the rules for functional
inter- and intramodular communication and the
flexibility with which we can modify communication
independent of catalytic activity.
Recent work suggests that one reason domain
exchange has generally yielded poorly active PKSs
could be extra-domain intramodular structural perturbation. An engineered DEBS module 6+TE, in
which the AT was replaced with that from RAPS
module 2, was shown to retain nonlimiting malonyl
transferase and KS loading activities but was impaired in substrate condensation.101 These results
suggest that the AT insertion may disrupt correct
interaction between ACP and KS domains necessary
for the C-C bond-forming extension reaction.
Much work of late has focused on the determinants
of intermodular interaction that facilitate the specific
channeling of the growing polyketide chain. There is
clear evidence that a significant role is played by the
intervening protein sequence between the ACP domain and the KS of the downstream module80,104,118-122
(Figure 9). These sequences, which have been termed
intermodular linkers, facilitate both intra- and
interpolypeptide substrate channeling. Intrapolypeptide linkers bridge modules within the same protein.
Interpolypeptide linkers are terminal sequences that
bridge C- and N-terminal modules on separate proteins.
There is considerable evidence that modulemodule interaction can be engineered independently
of module catalytic function. Linkers can be altered
or exchanged without loss of intrinsic module activity,
and several hybrid PKSs have been successfully
constructed using compatible interpolypeptide linkers
to bridge normally incommunicative modules.80,119-122
Thus, engineering of linkers may allow flexibility in
the physical reorganization of multimodular PKSs.
Recent evidence suggests that acyl chain substrate
specificity can be a significant barrier to intermodular
channeling. Specificity in the transfer of the acyl
chain from the upstream ACP to the next module KS,
in the KS-catalyzed condensation reaction, or in
eventual product release by a TE domain may restrict
elongation of novel compounds on hybrid PKSs. A
thorough study of the substrate selectivities of several PKS modules has given interesting insights into
some specificity determinants.123 In particular, it was
observed that modules unable to extend substrates
were blocked not at acyl chain transfer to the KS but
at subsequent condensation, implying that somehow
transition-state stabilization in the rate-limiting
condensation reaction is more stringent than that in
Figure 9. Intra- and intermodular communication in polyketide extension. (A) Some interactions critical to substrate
channeling are shown: 1 and 3, substrate acceptance by KS; 2 and 4, condensation of KS- and ACP-bound intermediates;
5, product hydrolysis by TE. Intermodular transfer via specific interpolypeptide linkers is illustrated in a comparison of
panels A-D. Matched linker pairs (A and D) from DEBS modules 2 and 3 (L2 and L3) or modules 4 and 5 (L4 and L5)
greatly facilitate intermodular transfer of the growing polyketide chain. Data shown are taken from Tsuji et al.119
acyl transfer. Understanding the basis for this discrimination may allow us to change or broaden
specificity, opening new routes to functional hybrid
PKSs and their diverse potential products.
5.5. Precursor Engineering

PKSs occasionally require substrates for polyketide
biosynthesis that are not commonly found in bacterial
cells, for instance, 2-methoxymalonate and AHBA.
Unlike the ubiquitous small branched-chain fatty
acids used as starter units, which are believed to
come from the catabolism of valine, leucine, and
isoleucine,124 formation of 2-methoxymalonate for the
biosynthesis of FK520, ansamitocins, and geldanamycin requires a dedicated set of five genes to convert
some glycolytic intermediate to the ACP-bound form
of 2-methoxymalonate (15), as currently believed
(Figure 10). The acyl ACP dehydrogenase gene
product has been studied in vitro,125 and heterologous
expression of the genes from the ansamitocin and
FK520 producers in S. coelicolor or S. fradiae has
been used to produce, respectively, an erythromycin55
and a midecamycin126 analogue.
E. coli is being developed as a host for polyketide
production,127,128 which also requires the introduction
of metabolic pathways to make PKS substrates.
Pathways to form (2S)-methylmalonyl-CoA (16) from
propionate via propionyl-CoA carboxylase128 or from
succinyl-CoA via methylmalonyl-CoA mutase129 have
been engineered to support polyketide production of
6dEB (1) and analogues in E. coli (Figure 10). A 15methyl-6dEB analogue was produced in E. coli by
overexpressing the acetoacetyl-CoA:acetyl-CoA trans-
Figure 10. Precursor pathways that have been engineered

in S. coelicolor and S. fradiae (15) and E. coli (16 and 17).
ferase (atoAD) to generate butyryl-CoA (17), which

is utilized by the loading AT of DEBS, from fed
butyric acid or butanol.130
5.6. Tailoring Pathways

Desosamine is present on all of the 14-membered
macrolides and is essential for antibiotic activity.
Recent crystallography studies with erythromycin

bound to ribosomes reveal the essential role that
desosamine plays in binding to the ribosome.131
Fourteen-membered lactones lacking desosamine are
completely inactive, and no synthetic substitute or
modification to desosamine has yet been found which
maintains the same level of activity. The entire set
of des genes, including the glycosyl transferase from
the picromycin cluster, were coexpressed with modified DEBS genes which produce analogues of 6dEB
to produce a small library of 14-membered macrolides
containing only desosamine.132 Antibiotic activity was
detected in nearly every case, suggesting desosamine
is sufficient to confer antibiotic activity to macrolactones.
Several experiments have been conducted to introduce modified or unnatural deoxysugars to macrolide
backbones (reviewed in Mendez and Salas133). Two
approaches to producing desosaminylated tylosin
derivatives have been described. In one example, the
TylMII mycaminosyltransferase was expressed in an
engineered S. erythraea strain which does not make
erythromycin and used to bioconvert tylactone to 5-Odesosaminyl-tylactone.134 In the other case, two genes
involved in desosamine biosynthesis from narbomycin were imported into the tylosin producing strain,
Streptomyces fradiae, to convert TDP-D-mycaminose
(TDP-D-4-deoxydesosamine) to TDP-D-desosamine
and produce desosaminylated tylonolide derivatives.135 The genes encoding L-olivose and L-oleandrose biosynthesis were expressed from plasmids in
Streptomyces albus and used to bioconvert erythronolide B to 3-O-olivosyl and 3-O-oleandrosyl erythronolide B analogues.136 These and other experiments
indicate some degree of deoxysugar substrate flexibility and the ability to engineer novel glycosides of
macrolide antibiotics. However, to date, such modifications have not led to compounds with activity
superior to the natural glycosides.
6. Creating and Improving Microbial Production

Systems
In addition to genetically engineering PKSs and
manipulating tailoring enzymes, another genetic
approach to novel metabolites has involved the
development of heterologous bacterial hosts. This
work has had two goals: to expedite structurefunction investigations of PKSs and the exploration
of metabolic pathways and to improve polyketide
titers by faster means than lengthy strain improvement carried out by random mutation and screening.
E. coli is becoming a versatile host for the expression of modular PKS genes, as demonstrated by the
recent work at Stanford University and Kosan.
Pfeiffer et al.128 produced 6dEB and novel analogues
made from aromatic acid starter units, and Kosan
scientists made a wide range of 6dEB analogues with
novel substituents at C-13 derived from either butyric
acid130 or fed N-acetylcysteamine thioesters (J.
Kennedy et al., manuscript in preparation). Recent
work with the rifamycin137,138 and epothilone PKS
genes139 (S. Mutka et al., submitted for publication)
and certain deoxysugar biosynthesis genes (H. Gramajo et al., manuscript in preparation) have shown
McDaniel et al.
Figure 11. Erythromycin analogues produced by engineered DEBS genes expressed in an overproducing S.
erythraea host.
that E. coli is likely to be a quite versatile host for

polyketide pathway engineering.
Another useful approach is to use an industrial
strain whose titer has already been improved as the
host for expression of PKS genes to gain the benefit
of the presumably enhanced levels of gene expression,
substrates, or tailoring enzymes that could be present.
Two examples of this strategy have been described
in which the native PKS genes were deleted from the
high-producing strain and replaced with the genes
from a wild-type producer. In both cases, use of the
wild-type genes did not lower the polyketide titer,
demonstrating that increased titers in the industrial
strains were not due to PKS mutations. Engineered
DEBS genes were then expressed in the high-producing S. erythraea strain to generate erythromycin
analogues at substantially higher levels than those
obtained in S. coelicolor140,141 or the wild-type S.
erythraea strain (Figure 11). More recently, a highproducing S. fradiae strain has been used to produce
a number of novel 16-membered macrolides in high
yield.84,126
7. Conclusions
Since the initial cloning of genes for polyketide
biosynthesis in the late 1980s and early 1990s, a
number of strategies and tools for engineering the
genes and pathways to create new polyketide compounds have been developed. These have been used
to create analogues of structurally complex compounds that would be difficult to obtain through
conventional organic synthesis or semi-synthesis.
Though many of the strategies are empirically based,
they are fairly robust when compared to the success
rate of a typical synthesis reaction as applied to
different substrates and so can be viewed as a
practical and complementary tool in the design and
production of new therapeutic entities. These techniques should be refined through continued practice,
and additional methods will likely emerge to provide

increased versatility. Current and future information
derived from ongoing biochemical and structural
studies of PKSs will certainly help guide these
approaches.
8. Acknowledgments
We thank David Hopwood, Hugo Gramajo, and
Bryan Julien for their helpful comments on the
manuscript.
9. References
(1) Newman, D. J.; Cragg, G. M.; Snader, K. M. J. Nat. Prod. 2003,
66, 1022.
(2) Cragg, G. M.; Newman, D. J.; Snader, K. M. J. Nat. Prod. 1997,
60, 52.
(3) Cane, D. E.; Walsh, C. T.; Khosla, C. Science 1998, 282, 63.
(4) Katz, L.; McDaniel, R. Med. Res. Rev. 1999, 19, 543.
(5) McDaniel, R.; Katz, L. In Development of Novel Antomicrobial
Agents: Emerging Strategies; Lohner, K., Ed.; Horizon Scientific
Press: Wymondham, England, 2000; Vol. 19.
(6) Hutchinson, C. R.; McDaniel, R. Curr. Opin. Investig. Drugs
2001, 2, 1681.
(7) Baltz, R. H. Trends Microbiol. 1998, 6, 76.
(8) McDaniel, R.; Khosla, C. In Enzyme technologies for pharmaceutical and biotechnological applications; Kirst, H. A., Yeh, W.K., Zmijewski, M. J., Jr., Eds.; Marcel Dekker: New York, 2001.
(9) Weissman, K. J.; Staunton, J. In Enzyme technologies for
pharmaceutical and biotechnological applications; Kirst, H. A.,
Yeh, W.-K., Zmijewski, M. J., Jr., Eds.; Marcel Dekker: New
York, 2001.
(10) Rowe, C. J.; Bohm, I. U.; Thomas, I. P.; Wilkinson, B.; M., R. B.
A.; Foster, G.; Blackaby, A. P.; Sidebottom, P. J.; Roddis, Y.;
Buss, A. D.; Staunton, J.; Leadlay, P. F. Chem. Biol. 2001, 89,
1.
(11) Reeves, C. D. Crit. Rev. Biotechnol. 2003, 23, 95.
(12) Hopwood, D. A.; Malpartida, F.; Kieser, H. M.; Ikeda, H.;
Duncan, J.; Fujii, I.; Rudd, B. A.; Floss, H. G.; Omura, S. Nature
1985, 314, 642.
(13) Omura, S.; Ikeda, H.; Malpartida, F.; Kieser, H. M.; Hopwood,
D. A. Antimicrob. Agents Chemother. 1986, 29, 13.
(14) Cortes, J.; Haydock, S. F.; Roberts, G. A.; Bevitt, D. J.; Leadlay,
P. F. Nature 1990, 348, 176.
(15) Donadio, S.; Staver, M. J.; McAlpine, J. B.; Swanson, S. J.; Katz,
L. Science 1991, 252, 675.
(16) Rawlings, B. J. Nat. Prod. Rep. 2001, 18, 190.
(17) Rawlings, B. J. Nat. Prod. Rep. 2001, 18, 231.
(18) Khosla, C.; Gokhale, R. S.; Jacobsen, J. R.; Cane, D. E. Annu.
Rev. Biochem. 1999, 68, 219.
(19) Liou, G. F.; Khosla, C. Curr. Opin. Chem. Biol. 2003, 7, 279.
(20) Staunton, J.; Weissman, K. J. Nat. Prod. Rep. 2001, 18, 380.
(21) Austin, M. B.; Noel, J. P. Nat. Prod. Rep. 2003, 20, 79.
(22) Bevitt, D. J.; Cortes, J.; Haydock, S. F.; Leadlay, P. F. Eur. J.
Biochem. 1992, 204, 39.
E.; Hutchinson, C. R.; Katz, L. Microbiol. 1997, 143, 3251.
(24) Haydock, S. F.; Dowson, J. A.; Dhillon, N.; Roberts, G. A.; Cortes,
J.; Leadlay, P. F. Mol. Gen. Genet. 1991, 230, 120.
(25) Stassi, D.; Donadio, S.; Staver, M. J.; Katz, L. J. Bacteriol. 1993,
175, 182.
(26) Gaisser, S.; Bohm, G. A.; Cortes, J.; Leadlay, P. F. Mol. Gen.
Genet. 1997, 256, 239.
(27) Gaisser, S.; Bohm, G. A.; Doumith, M.; Raynal, M. C.; Dhillon,
N.; Cortes, J.; Leadlay, P. F. Mol. Gen. Genet. 1998, 258, 78.
(28) Weber, J. M.; Leung, J. O.; Swanson, S. J.; Idler, K. B.; McAlpine,
J. B. Science 1991, 252, 114.
(29) Swan, D. G.; Rodriguez, A. M.; Vilches, C.; Mendez, C.; Salas,
J. A. Mol. Gen. Genet. 1994, 242, 358.
(30) Shah, S.; Xue, Q.; Tang, L.; Carney, J. R.; Betlach, M.; McDaniel,
R. J. Antibiotics 2000, 53, 502.
(31) Rodriguez, A. M.; Olano, C.; Mendez, C.; Hutchinson, C. R.
FEMS Microbiol. Lett. 1995, 127, 117.
(32) Quiros, L. M.; Aguirrezabalaga, I.; Olano, C.; Mendez, C.; Salas,
J. A. Mol. Microbiol. 1998, 28, 1177.
(33) Olano, C.; Rodriguez, A. M.; Michel, J. M.; Mendez, C.; Raynal,
M. C.; Salas, J. A. Mol. Gen. Genet. 1998, 259, 299.
(34) Aguirrezabalaga, I.; Olano, C.; Allende, N.; Rodriguez, L.; Brana,
A. F.; Mendez, C.; Salas, J. A. Antimicrob. Agents Chemother.
2000, 44, 1266.
(35) Volchegursky, Y.; Hu, Z.; Katz, L.; McDaniel, R. Mol. Microbiol.
2000, 37, 752.

(36) Xue, Y.; Zhao, L.; Liu, H.-w.; Sherman, D. H. Proc. Natl. Acad.
Sci. U.S.A. 1998, 95, 12111.
(37) Bisang, C.; Long, P. F.; Cortes, J.; Westcott, J.; Crosby, J.;
Matharu, A.-L.; Cox, R. J.; Simpson, T. J.; Staunton, J.; Leadlay,
P. F. Nature 1999, 401, 502.
(38) Tang, L.; Fu, H.; Betlach, M. C.; McDaniel, R. Chem. Biol. 1999,
6, 553.
(39) Beck, B. J.; Yoon, Y. J.; Reynolds, K. A.; Sherman, D. H. Chem.
Biol. 2002, 9, 575.
(40) Kao, C. M.; Katz, L.; Khosla, C. Science 1994, 265, 509.
(41) Hu, Z.; Desai, R. P.; Volchegursky, Y.; Leaf, T.; Woo, E.; Licari,
P.; Santi, D. V.; Hutchinson, C. R.; McDaniel, R. J. Ind.
Microbiol. Biotechnol. 2003, 30, 161.
(42) Gaisser, S.; Lill, R.; Staunton, J.; Mendez, C.; Salas, J.; Leadlay,
P. F. Mol. Microbiol. 2002, 44, 771.
(43) Zhao, L.; Sherman, D. H.; Liu, H.-w. J. Am. Chem. Soc. 1998,
120, 10256.
(44) Zhao, L.; Que, N. L. S.; Xue, Y.; Sherman, D. H.; Liu, H.-w. J.
Am. Chem. Soc. 1998, 120, 12159.
(45) Borisova, S. A.; Zhao, L.; Sherman, D. H.; Liu, H. W. Org. Lett.
1999, 1, 133.
(46) Betlach, M. C.; Kealey, J. T.; Betlach, M. C.; Ashley, G. A.;
McDaniel, R. Biochemistry 1998, 37, 14937.
(47) Xue, Y.; Wilson, D.; Zhao, L.; Liu, H.-w.; Sherman, D. H. Chem.
Biol. 1998, 5, 661.
(48) DeHoff, B. S.; Sutton, K. L.; Rosteck, P. R. GenBank accession
#U78289, 1996.
(49) Kakavas, S. J.; Katz, L.; Stassi, D. J. Bacteriol. 1997, 179, 7515.
(50) Kuhstoss, S.; Huber, M.; Turner, J. R.; Paschal, J. W.; Rao, R.
N. Gene 1996, 183, 231.
(51) Ward, S. L.; Hu, Z.; Schirmer, A.; Reid, R.; Revill, P. W.; Reeves,
C. D.; Petrakovsky, O. V.; Dong, S. D.; Katz, L. Antimicrob.
Agents Chemother. 2004, in press.
(52) Stassi, D. L.; Kakavas, S. J.; Reynolds, K. A.; Gunawardana,
G.; Swanson, S.; Zeidner, D.; Jackson, M.; Liu, H.; Buko, A.;
Katz, L. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 7305.
(53) Wu, K.; Chung, L.; Revill, P. W.; Katz, L.; Reeves, C. D. Gene
2000, 251, 81.
(54) Rascher, A.; Hu, Z.; Viswanathan, N.; Schirmer, A.; Reid, R.;
Nierman, W. C.; Lewis, M.; Hutchinson, C. R. FEMS Microbiol.
Lett. 2003, 218, 223.
(55) Kato, Y.; Bai, L.; Xue, Q.; Revill, W. P.; Yu, T. W.; Floss, H. G.
J. Am. Chem. Soc. 2002, 124, 5268.
(56) Cundliffe, E.; Bate, N.; Butler, A.; Fish, S.; Gandecha, A.;
Merson-Davies, L. Antonie Van Leeuwenhoek 2001, 79, 229.
(57) Mochizuki, S.; Hiratsu, K.; Suwa, M.; Ishii, T.; Sugino, F.;
Yamada, K.; Kinashi, H. Mol. Microbiol. 2003, 48, 1501.
(58) Rinehart, K. L. J.; Shield, L. S. Fortschr. Chem. Org. Naturst.
1976, 33, 231.
(59) Cassady, J. M.; Chan, K. K.; Floss, H. G.; Leistner, E. Chem.
Pharm. Bull. 2004, 52, 1.
(60) Kim, C. G.; Yu, T. W.; Fryhle, C. B.; Handa, S.; Floss, H. G. J.
Biol. Chem. 1998, 273, 6030.
(61) August, P. R.; Tang, L.; Yoon, Y. J.; Ning, S.; Muller, R.; Yu, T.
W.; Taylor, M.; Hoffmann, D.; Kim, C. G.; Zhang, X.; Hutchinson,
C. R.; Floss, H. G. Chem. Biol. 1998, 5, 69.
(62) Schupp, T.; Toupet, C.; Engel, N.; Goff, S. FEMS Microbiol. Lett.
1998, 159, 201.
(63) Stratmann, A.; Toupet, C.; Schilling, W.; Traber, R.; Oberer, L.;
Schupp, T. Microbiol. 1999, 145 (Pt 12), 3365.
(64) Yu, T. W.; Shen, Y.; Doi-Katayama, Y.; Tang, L.; Park, C.; Moore,
B. S.; Richard Hutchinson, C.; Floss, H. G. Proc. Natl. Acad. Sci.
U.S.A. 1999, 96, 9051.
(65) Stratmann, A.; Schupp, T.; Toupet, C.; Schilling, W.; Oberer, L.;
Traber, R. J. Antibiot. 2002, 55, 396.
(66) Yu, T. W.; Bai, L.; Clade, D.; Hoffmann, D.; Toelzer, S.; Trinh,
K. Q.; Xu, J.; Moss, S. J.; Leistner, E.; Floss, H. G. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 7968.
(67) Spiteller, P.; Bai, L.; Shang, G.; Carroll, B. J.; Yu, T. W.; Floss,
H. G. J. Am. Chem. Soc. 2003, 125, 14236.
(68) Moss, S. J.; Bai, L.; Toelzer, S.; Carroll, B. J.; Mahmud, T.; Yu,
T. W.; Floss, H. G. J. Am. Chem. Soc. 2002, 124, 6544.
(69) Carroll, B. J.; Moss, S. J.; Bai, L.; Kato, Y.; Toelzer, S.; Yu, T.
W.; Floss, H. G. J. Am. Chem. Soc. 2002, 124, 4176.
(70) Patel, K.; Piagentini, M.; Rascher, A.; Tian, Z.-Q.; Buchanan,
G. O.; Regentin, R.; Hu, Z.; Hutchinson, C. R.; McDaniel, R.
Chem. Biol. 2004, 11, 1625.
(71) Ruan, X. R.; Pereda, A.; Stassi, D. L.; Zeidner, D.; Summers, R.
G.; Jackson, M.; Shivakumar, A.; Kakavas, S.; Staver, M. J.;
Donadio, S.; Katz, L. J. Bacteriol. 1997, 179, 6416.
(72) Liu, L.; Thamchaipenet, A.; Fu, H.; Betlach, M.; Ashley, G. J.
Am. Chem. Soc. 1997, 119, 10553.
(73) Petkovic, H.; Lill, R. E.; Sheridan, R. M.; Wilkinson, B.; McCormick, E. L.; McArthur, H. A.; Staunton, J.; Leadlay, P. F.;
Kendrew, S. G. J. Antibiot. 2003, 56, 543.
(74) Revill, W. P.; Voda, J.; Reeves, C. R.; Chung, L.; Schirmer, A.;
Ashley, G.; Carney, J. R.; Fardis, M.; Carreras, C. W.; Zhou, Y.;
Feng, L.; Tucker, E.; Robinson, D.; Gold, B. G. J. Pharmacol.
Exp. Ther. 2002, 302, 1278.

(75) Marsden, A. F.; Wilkinson, B.; Cortes, J.; Dunster, N. J.;
(76) Pacey, M. S.; Dirlam, J. P.; Geldart, R. W.; Leadlay, P. F.;
McArthur, H. A.; McCormick, E. L.; Monday, R. A.; OConnell,
T. N.; Staunton, J.; Winchester, T. J. J. Antibiot. 1998, 51,
1029.
(77) Long, P. F.; Wilkinson, C. J.; Bisang, C. P.; Cortes, J.; Dunster,
N.; Oliynyk, M.; McCormick, E.; McArthur, H.; Mendez, C.;
Salas, J. A.; Staunton, J.; Leadlay, P. F. Mol. Microbiol. 2002,
43, 1215.
(78) Donadio, S.; McAlpine, J. B.; Sheldon, P. J.; Jackson, M.; Katz,
(79) McDaniel, R.; Thamchaipenet, A.; Gustafsson, C.; Fu, H.;
Betlach, M.; Betlach, M.; Ashley, G. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 1846.
(80) Gokhale, R. S.; Tsuji, S. Y.; Cane, D. E.; Khosla, C. Science 1999,
284, 482.
(81) Yoon, Y. J.; Beck, B. J.; Kim, B. S.; Kang, K. Y.; Reynolds, K.
A.; Sherman, D. H. Chem. Biol. 2002, 9, 203.
(82) Ranganathan, A.; Timoney, M.; Bycroft, M.; Cortes, J.; Thomas,
I. P.; Wilkinson, B.; Kellenberger, L.; Hanefeld, U.; Galloway, I.
S.; Staunton, J.; Leadlay, P. F. Chem. Biol. 1999, 6, 731.
(83) Tang, L.; Fu, H.; McDaniel, R. Chem. Biol. 2000, 7, 77.
(84) Reeves, C. D.; Ward, S. L.; Revill, P. W.; Suzuki, H.; Marcus,
M.; Petrakovsky, O. V.; Marquez, S.; Fu, H.; Dong, S. D.; Katz,
L. Chem. Biol. 2004, 11, 1465.
(85) Jacobsen, J. R.; Hutchinson, C. R.; Cane, D. E.; Khosla, C.
Science 1997, 277, 367.
(86) Kinoshita, K.; Pfeifer, B. A.; Khosla, C.; Cane, D. E. Bioorg. Med.
Chem. Lett. 2003, 13, 3701.
(87) Jacobsen, J. R.; Cane, D. E.; Khosla, C. J. Am. Chem. Soc. 1998,
120, 9096.
(88) Jacobsen, J. R.; Keatinge-Clay, A. T.; Cane, D. E.; Khosla, C.
Bioorg. Med. Chem. 1998, 6, 1171.
(89) Gokhale, R. S.; Hunziker, D.; Cane, D.; Khosla, C. Chem. Biol.
1999, 6, 117.
(90) McDaniel, R.; Ebert-Khosla, S.; Hopwood, D. A.; Khosla, C.
Nature 1995, 375, 549.
(91) Tang, Y.; Tsai, S. C.; Khosla, C. J. Am. Chem. Soc. 2003, 125,
12708.
(92) Tang, Y.; Lee, T. S.; Kobayashi, S.; Khosla, C. Biochemistry 2003,
42, 6588.
(93) Keatinge-Clay, A. T.; Shelat, A. A.; Savage, D. F.; Tsai, S. C.;
Miercke, L. J.; OConnell, J. D., III; Khosla, C.; Stroud, R. M.
Structure (Cambridge) 2003, 11, 147.
(94) Pan, H.; Tsai, S.; Meadows, E. S.; Miercke, L. J.; Keatinge-Clay,
A. T.; OConnell, J.; Khosla, C.; Stroud, R. M. Structure (Cambridge) 2002, 10, 1559.
(95) Meadows, E. S.; Khosla, C. Biochemistry 2001, 40, 14855.
(96) Tang, Y.; Lee, T. S.; Khosla, C. PLoS Biol. 2004, 2, E31.
(97) Remsing, L. L.; Gonzalez, A. M.; Nur-e-Alam, M.; FernandezLozano, M. J.; Brana, A. F.; Rix, U.; Oliveira, M. A.; Mendez,
C.; Salas, J. A.; Rohr, J. J. Am. Chem. Soc. 2003, 125, 5745.
(98) Trefzer, A.; Blanco, G.; Remsing, L.; Kunzel, E.; Rix, U.; Lipata,
F.; Brana, A. F.; Mendez, C.; Rohr, J.; Bechthold, A.; Salas, J.
A. J. Am. Chem. Soc. 2002, 124, 6056.
(99) Remsing, L. L.; Garcia-Bernardo, J.; Gonzalez, A.; Kunzel, E.;
Rix, U.; Brana, A. F.; Bearden, D. W.; Mendez, C.; Salas, J. A.;
Rohr, J. J. Am. Chem. Soc. 2002, 124, 1606.
(100) Lombo, F.; Kunzel, E.; Prado, L.; Brana, A. F.; Bindseil, K. U.;
Frevert, J.; Bearden, D.; Mendez, C.; Salas, J. A.; Rohr, J. Angew.
Chem., Int. Ed. Engl. 2000, 39, 796.
(101) Hans, M.; Hornung, A.; Dziarnowski, A.; Cane, D. E.; Khosla,
C. J. Am. Chem. Soc. 2003, 125, 5366.
(102) Reeves, C. D.; Murli, S.; Ashley, G. A.; Piagentini, M.; Hutchinson, C. R.; McDaniel, R. Biochemistry 2001, 40, 15464.
(103) Del Vecchio, F.; Petkovic, H.; Kendrew, S. G.; Low, L.; Wilkinson,
B.; Lill, R.; Cortes, J.; Rudd, B. A.; Staunton, J.; Leadlay, P. F.
J. Ind. Microbiol. Biotechnol. 2003, 30, 489.
(104) Kumar, P.; Koppisch, A. T.; Cane, D. E.; Khosla, C. J. Am. Chem.
Soc. 2003, 125, 14307.
(105) Kao, C. M.; McPherson, M.; McDaniel, R.; Fu, H.; Cane, D.;
Khosla, C. J. Am. Chem. Soc. 1998, 120, 2478.
(106) Reid, R.; Piagentini, M.; Rodriguez, E.; Ashley, G.; Viswanathan,
N.; Carney, J.; Santi, D. V.; Hutchinson, C. R.; McDaniel, R.
McDaniel et al.
(107) Caffrey, P. Chembiochem. 2003, 4, 654.
(108) Yin, Y.; Gokhale, R.; Khosla, C.; Cane, D. E. Bioorg. Med. Chem.
Lett. 2001, 11, 1477.
(109) Tang, L.; Ward, S.; Chung, L.; Carney, J. R.; Li, Y.; Reid, R.;
Katz, L. J. Am. Chem. Soc. 2004, 126, 46.
(110) Cronan, J. E., Jr.; Rock, C. O. In Escherichia coli and Salmonella
Typhimurium: Cellular and Molecular Biology; Lin, C. C., Low,
K. B., Magasanik, B., Reznikoff, W. S., Riley, M., Schaechter,
M., Umbarger, H. E., Eds.; American Society for Microbiology:
Washington, DC, 1996.
(111) Butler, A. R.; Bate, N.; Cundliffe, E. Chem. Biol. 1999, 6,
287.
(112) Hu, Z.; Pfeifer, B. A.; Chao, E.; Murli, S.; Kealey, J.; Carney, J.
R.; Ashley, G.; Khosla, C.; Hutchinson, C. R. Microbiology 2003,
149, 2213.
(113) Schwarzer, D.; Mootz, H. D.; Linne, U.; Marahiel, M. A. Proc.
Natl. Acad. Sci. U.S.A. 2002, 99, 14083.
(114) Lu, H.; Tsai, S. C.; Khosla, C.; Cane, D. E. Biochemistry 2002,
41, 12590.
Natl. Acad. Sci. U.S.A. 2001, 98, 14808.
(116) Tsai, S. C.; Lu, H.; Cane, D. E.; Khosla, C.; Stroud, R. M.
(117) Martin, C. J.; Timoney, M. C.; Sheridan, R. M.; Kendrew, S. G.;
Wilkinson, B.; Staunton, J. C.; Leadlay, P. F. Org. Biomol. Chem.
2003, 1, 4144.
(118) Broadhurst, R. W.; Nietlispach, D.; Wheatcroft, M. P.; Leadlay,
P. F.; Weissman, K. J. Chem. Biol. 2003, 10, 723.
(119) Tsuji, S. Y.; Cane, D. E.; Khosla, C. Biochemistry 2001, 40,
2326.
(120) Tsuji, S. Y.; Wu, N.; Khosla, C. Biochemistry 2001, 40,
2317.
(121) Wu, N.; Cane, D. E.; Khosla, C. Biochemistry 2002, 41,
5056.
(122) Wu, N.; Tsuji, S. Y.; Cane, D. E.; Khosla, C. J. Am. Chem. Soc.
2001, 123, 6465.
(123) Watanabe, K.; Wang, C. C.; Boddy, C. N.; Cane, D. E.; Khosla,
C. J. Biol. Chem. 2003, 278, 42020.
(124) Tang, L.; Zhang, Y. X.; Hutchinson, C. R. J. Bacteriol. 1994, 176,
6107.
(125) Watanabe, K.; Khosla, C.; Stroud, R. M.; Tsai, S. C. J. Mol. Biol.
2003, 334, 435.
(126) Rodriguez, E.; Ward, S.; Fu, H.; Revill, W. P.; McDaniel, R.; Katz,
L. Appl. Microbiol. Biotechnol. 2004, 66, 85.
(127) Pfeifer, B. A.; Khosla, C. Microbiol. Mol. Biol. Rev. 2001, 65,
106.
(128) Pfeifer, B. A.; Admiraal, S. J.; Gramajo, H.; Cane, D. E.; Khosla,
C. Science 2001, 291, 1790.
(129) Dayem, L. C.; Carney, J. R.; Santi, D. V.; Pfeifer, B. A.; Khosla,
C.; Kealey, J. T. Biochemistry 2002, 41, 5193.
(130) Murli, S.; Kennedy, J.; Dayem, L. C.; Carney, J. R.; Kealey, J.
T. J. Ind. Microbiol. Biotechnol. 2003, 30, 500.
(132) Tang, L.; McDaniel, R. Chem. Biol. 2001, 8, 547.
(133) Mendez, C.; Salas, J. A. Trends Biotechnol. 2001, 19, 449.
(134) Gaisser, S.; Reather, J.; Wirtz, G.; Kellenberger, L.; Staunton,
J.; Leadlay, P. F. Mol. Microbiol. 2000, 36, 391.
(135) Butler, A. R.; Bate, N.; Kiehl, D. E.; Kirst, H. A.; Cundliffe, E.
Nat. Biotechnol. 2002, 20, 713.
(136) Rodriguez, L.; Aguirrezabalaga, I.; Allende, N.; Brana, A. F.;
Mendez, C.; Salas, J. A. Chem. Biol. 2002, 9, 721.
(137) Admiraal, S. J.; Khosla, C.; Walsh, C. T. Biochemistry 2002, 41,
5313.
(138) Admiraal, S. J.; Khosla, C.; Walsh, C. T. J. Am. Chem. Soc. 2003,
125, 13664.
(139) Boddy, C. N.; Hotta, K.; Tse, M. L.; Watts, R. E.; Khosla, C. J.
Am. Chem. Soc. 2004, 126, 7436.
(140) Rodriguez, E.; Hu, Z.; Ou, S.; Volchegursky, Y.; Hutchinson, C.
R.; McDaniel, R. J. Ind. Microbiol. Biotechnol. 2003, 30, 480.
(141) Starks, C. M.; Rodriguez, E.; Carney, J. R.; Desai, R. P.; Carreras,
C.; McDaniel, R.; Hutchinson, R.; Galazzo, J. L.; Licari, P. J. J.
Antibiot. 2004, 57, 64.
CR0301189
Chem. Rev. 2005, 105, 559592
559
Bacterial Topoisomerase Inhibitors: Quinolone and Pyridone Antibacterial

Agents
Lester A. Mitscher*
Departments of Medicinal Chemistry and Molecular Biosciences and The Chemical Methodologies and Library Development Center of Excellence,
The University of Kansas, 4010 Malott Hall, 1251 Wescoe Hall Drive, Lawrence, Kansas 66045-7582
Contents
1. Introduction: History and Overview
2. Synthetic Chemistry
2.1. GouldJacobs Reaction
2.2. GroheHeitzer Reaction
2.3. GersterHayakawa Syntheses
2.4. ChuMitscher Synthesis
2.5. ChuLi Syntheses
3. Some Quinolone Chemical Reactions of
Significance to Their Medicinal Properties
3.1. Chelation
3.2. AcidBase Character
3.3. Photochemistry
4. In Vitro Antimicrobial Spectra
5. StructureActivity Relationships
5.1. N-1 Ethyl Family
5.2. N-1 Cyclopropyl Family
5.3. N-1 to C-8 Bridged (Tricyclic) Family
5.4. N-1 Aryl Family
5.5. Positions C-2, C-3, and C-4
5.6. C-4a Substituted Analogues
5.7. C-5 Substituents
5.9.1. Piperazinyl and Related Moieties
5.9.2. Pyrrolidinyl and Related Moieties
5.9.3. Cyclobutylaminyl and Related Moieties
5.9.4. Bicycloaminyl Moieties
5.9.5. Carbon-Linked Substituents
5.10. Substituents at C-8
5.11. Resume of StructureActivity Relationships of
Quinolones
5.12. Absolute Configuration
6. StructureToxicity Relationships
6.1. Chemotype Toxicities and Side Effects
6.2. Severe Toxicities Associated with Particular
Quinolones
6.3. DrugDrug Interactions
7. Molecular Mode of Action
7.1. Models of the Ternary Complex
8. Assay Methods
9. Molecular Modes of Resistance
9.1. Uptake Inhibition
9.2. Plasmid-Mediated Resistance
559
561
562
562
562
563
563
564
564
564
565
566
568
568
568
568
569
570
571
571
571
571
572
572
572
572
572
573
573
574
575
575
576
576
576
582
585
586
586
587
* Phone: (785) 864-4562. Fax: (785) 864-5326. E-mail: lmitscher@

ku.edu.
9.3.
9.4.
9.5.
9.6.
9.7.
10.
11.
12.
13.
14.
Enzymatic Alteration of Quinolone Structures

Mutations of DNA Gyrase
Mutations of Topoisomerase IV
Effect of Mutations on Microbial Vitality
Possible Approaches to Dealing with the
Resistance Problem
Clinical Indications
Pharmacokinetics
Future Prospects
Conclusions
References
587
587
587
588
588
589
589
590
590
590
1. Introduction: History and Overview

The quinolone anti-infective agents are of wholly
synthetic origin and are not modeled knowingly after
any natural antibiotic. Several ring systems are or
have been involved. Those of greatest prominence
and their numbering systems are illustrated in
Figure 1. The reader will note that the numbers assigned to analogous positions frequently change when
different quinolone ring systems are considered.
Of all the totally synthetic antimicrobial agents,
the (fluoro)quinolones have proven to be the most
successful economically and clinically. They are orally
and parenterally active, have a broad antimicrobial
spectrum that includes many frequently encountered
pathogens, are bactericidal in clinically achievable
doses, generate comparatively tolerable resistance
levels, possess a fascinating molecular mode of action,
are comparatively easily synthesized, and with a few
notable exceptions are safe. That is not to say that
they are perfect drugs and cannot be improved but
rather that they are important weapons in the
ongoing struggle against morbidity and mortality
caused by microbial pathogens. Consequently, from
a slow beginning as a modest group of urinary tract
disinfectants they have grown to be a group of nearly
two dozen institutional and office practice agents of
which ciprofloxacin and levofloxacin, most notably,
have become billion dollar agents persistently found
among the top 200 most frequently prescribed medications in North America and, indeed, worldwide.
This paper presents an overview of this important
topic with an emphasis on recent developments.
Reflecting their importance and the high interest in
quinolones, they have been the subject of numerous
books1-9 and recent reviews.10-26 These books and
reviews can be consulted for further information and
differing opinions about them.
10.1021/cr030101q CCC: $53.50 2005 American Chemical Society

Mitscher
Figure 2. Origin of the quinolones.

Lester A. Mitscher received his Ph.D. degree in chemistry in 1958 from
Wayne State University, Detroit, where he worked on the structure of
coffee oil diterpenes and on optical rotatory dispersion. He continued his
work on natural product chemistry at Lederle Laboratories, where he rose
to group leader in antibiotic discovery until he accepted an Associate
Professorship in Natural Products Chemistry at The Ohio State University
(1967), rising soon to the position of Professor. In 1975 he accepted a
University Distinguished Professorship and Chairmanship in the Department of Medicinal Chemistry at Kansas University. He returned to the
faculty in 1991, where he remains. His academic studies have centered
around spectroscopy, synthesis, screening, and structure determination
primarily of naturally occurring antimicrobial and antimutagenic agents.
He has published actively in the quinolone field since 1965. He consults
extensively in the pharmaceutical industry and is a member of the National
Institutes of Health Drug Discovery and Antimicrobial Resistance Study
Section. His research awards include the Smissman Award in Medicinal
Chemistry (American Chemical Society), Volweiler Award (American
Association for Pharmaceutical Education), Research Achievement Award
in Natural Products Chemistry (American Pharmaceutical Association),
and Award in Medicinal Chemistry, Medicinal Chemistry Division, American
Chemical Society, and he is an Elected Fellow of the American Association
for the Advancement of Science.
Figure 1. Structures and numbering systems of the most

significant ring systems in the antibacterial quinolone
family.
The first antimicrobial quinolone was discovered

about 50 years ago as an impurity in the chemical
manufacture of a batch of the antimalarial agent
chloroquine (Figure 2).27 It demonstrated anti Gramnegative antibacterial activity, but its potency and
antimicrobial spectrum were not significant enough
to be useful in therapy. Building on this lead, however, subsequently nalidixic acid was commercialized.
Nalidixic acid remains on the market today and
represents the so-called first generation quinolones.

Despite its convenient oral activity, bactericidal
action, and ease of synthesis, its limited antimicrobial
spectrum (primarily activity against Escherichia coli)
and poor pharmacokinetic characteristics limit its use
primarily to treatment of sensitive communityacquired urinary tract infections. For a few years
research on analogues resulted mainly in the introduction of competing products with enhanced though
still moderate activity against Gram-negatives and
a sniff of anti Gram-positive activity, but these agents
were also used primarily for urinary tract infections
of community origin. Sales of this group of agents
never became impressive.
This picture changed significantly with the discovery of norfloxacin, the first of the second-generation
family of quinolones.28 This agent had dramatically
enhanced and broader spectrum anti Gram-negative
activity and possessed significant anti Gram-positive
activity as well. The potency of norfloxacin was in
the same range as that of many fermentation-derived
antibiotics, and its comparative structural simplicity
and synthetic accessibility lead to a very significant
effort to find even more improved analogues. Norfloxacin and its N-methyl analogue pefloxacin ultimately failed to find major use outside of the genitourinary tract because of poor active blood levels and
limited potency against Gram-positives.
Shortly thereafter, ciprofloxacin29-31 and ofloxacin,31,32 as well as its optically active form levofloxacin,33 were introduced. The second-generation agents
have significant broad-spectrum antimicrobial activity including important Gram-positive pathogens.
This is coupled with gratifying safety and pharmacokinetic characteristics. These agents have found
excellent acceptance as office practice anti-infective
agents worldwide, and ciprofloxacin and levofloxacin
are regularly found among the top 100 most frequently prescribed drugs in North America. Gatifloxacin and ofloxacin have also appeared among the
top 200 during the past decade but have subsequently
fallen in popularity.
A wide variety of clinical indications have been
approved for quinolones including many infections
commonly encountered in community practice including upper and lower respiratory infections, gastrointestinal infections, gynecologic infections, sexually transmitted diseases, prostatitis, and some skin,
bone, and soft tissue infections.34
Bacterial Topoisomerase Inhibitors
Recently introduced members of the fluoroquinolone family belong to the third generation. These
include gatifloxacin35 and moxifloxacin,36 which possess further enhanced activity against Gram-positive
infections, and anti-anaerobic coverage is now present
although at present only trovafloxacin37 is approved
for this indication. Among the agents still in preclinical study, clinifloxacin38 is the most promising
anti-anaerobic agent.
When first introduced, there was no idea of the
molecular mode of action of these agents. Indeed, the
availability of nalidixic acid was instrumental in
assisting the discovery of the targeted bacterial type
II topoisomerases.39 Those of importance to the
quinolones are bacterial topoisomerase II,40 also
known as DNA gyrase, and bacterial topoisomerase
IV.41,42 These enzymes are vital for dictating the
proper topology of DNA important for protein biosynthesis, DNA replication and repair, and DNA
decatenation. Fluoroquinolones form a ternary complex consisting of drug, DNA, and enzyme that interferes with DNA transcription, replication, and repair
and promotes its cleavage, leading to rapid bacterial
cell death. They are without apparent significant
action on individual molecules of DNA or topoisomerases alone, but the interaction of DNA with
enzyme creates a binding pocket for the quinolones.
The ternary complex is rapidly bactericidal through
processes that are not completely understood.
As with the aminoglycoside class of antibiotics,
bacterial killing with fluoroquinolones is concentration-dependent rather than dosage-interval-dependent and the fluoroquinolones possess a significant
postantibiotic action lasting for 1 or 2 h.43 Although
the distinction is not precise, generally anti Gramnegative activity is more closely associated with DNA
gyrase inhibition and anti Gram-positive activity is
more closely associated with bacterial topoisomerase
IV inhibition.44 With a number of quinolones activity
is attributed to interference with the function of both
of these enzymes. For example, a survey of the ability
of a collection of quinolones to inhibit the catalytic
action of topoisomerases showed that the ratio of
DNA gyrase to topoisomerase IV action for E. coli was
between about 15 and 27, whereas for Staphylococcus
aureus the ratio was reversed, with topoisomerase
IV inhibition over DNA gyrase inhibition being from
about 1.7 to 21.45 In a later study of 15 quinolones,
they were divided into three groups on the basis of
their relative ability to inhibit S. aureus strains with
a resistance mutant toward one or the other enzyme.
With group I (norfloxacin, enoxacin, fleroxacin, ciprofloxacin, lomefloxacin, trovafloxacin, grepafloxacin,
ofloxacin, and levofloxacin) topoisomerase IV was the
more sensitive target. With group 2 (sparfloxacin and
nadifloxacin) DNA gyrase was the more sensitive
target. With group 3 (gatifloxacin, pazufloxacin,
moxifloxacin, and clinafloxcin) both were equivalently
sensitive. The latter were termed the dual targeting
quinolones.42 This classification holds up better against
the intact microorganisms than it does against the
purified enzymes. Human topoisomerase II is generally not inhibited by these agents at the doses
normally employed since it is often at least 100-1000
times less sensitive to them.46 Despite significant

homologies with the bacterial enzyme, creative analoguing is able to distinguish clearly between them,
and safe agents are readily produced. The novel
molecular mode of action of the quinolones helps to
account for their popularity. Unfortunately, combination of quinolones with other anti-infective agents is
not reliably synergistic.
Resistance in the clinic to this class of anti-infective
agents was comparatively slow to develop but is now
worrisome.47 There are no clear-cut examples of
resistance due to bacterial modification of the chemical structures of fluoroquinolones. Rather, resistance
is most commonly associated with genetic-based
alterations in the topoisomerases, resulting in decreased drug binding48 and, particularly, with efflux
of these agents from bacterial cells before they reach
their intercellular targets.49 Decreased porin presence
also decreases their uptake. Active uptake seems not
to be a significant factor in their absorption although
it is believed to contribute to distribution and excretion.
Absorption of quinolones following oral administration is usually good, mostly 50% or better, and, in
some cases, in excess of 95%.50 Most of the available
literature is consistent with passive absorption. They
are well distributed in the body; however, efflux
pumps protect the central nervous system to some
extent.51 Comparatively little metabolism takes place
with them, and excretion is mostly in active form in
the urine. Active excretion into the urine takes place
to some extent and is modestly enantioselective.52
They are generally safe and well-tolerated drugs.
With few exceptions, their side effects are mostly
annoying rather than severe although the list is
comparatively long. Side effects include GI, CNS,
rashes and photosensitivity, arthropathy, arthralgia
and joint swelling, and interactions with various
drugs. These occur to a greater or lesser extent with
all the members of this family but are more pronounced with certain individual agents.13
Recently, a number of comparatively rare but
severe toxicities have been observed with particular
quinolones. For example, trovafloxacin can cause
severe enough liver toxicity to require transplant or
cause death,53 temafloxacin has been associated with
a collection of severe problems involving kidney
failure, hemolysis, thrombocytopenia, and disseminated intravenous coagulation, leading to a number
of deaths,54 sparfloxacin and grepafloxacin cause
significant prolongation of the cardiac QT interval,55
and the C-8 halogenated members have been associated with an increased incidence of phototoxicity.56
Discovery research has slowed a bit recently, but
many agents are at various stages of clinical development. The field promises to remain active well into
the foreseeable future.
2. Synthetic Chemistry
Intense research in this therapeutic area has now
resulted in the introduction of nearly two dozen
competing agents into the clinic, and these agents
return very significant profits to the firms responsible
for them. Analogues are relatively accessible in
Mitscher
Figure 3. Basic Gould-Jacobs reaction.
substantial variety by short syntheses. As a consequence, tens of thousands of analogues have now
been prepared and tested. The deficiencies being
addressed by synthetic campaigns include the desire
to address worries about drug resistance, the need
to avoid toxicities, the ability to administer both
orally and parenterally, one-a-day administration,
freedom from drug-drug interactions, and the desire
to include anaerobic microorganisms and other presently relatively insensitive pathogens in their activity
spectrum.
2.1. GouldJacobs Reaction

The synthetic chemistry that makes all of this
activity possible consists fundamentally of variants
on a few pathways. The original method was the wellprecedented Gould-Jacobs reaction between suitably
substituted anilines and a substituted ethylenemalonate analogue at high temperature. This is illustrated in Figure 3.57 The initial reaction is an
addition-elimination sequence followed by cycloacylation. The specific nature of the product depends on
the symmetry properties of the starting aniline, and
its facility depends on the degree and position of
maximum electron richness of the ring. Alkylation
follows. This requires an alkyl halide or its equivalent
capable of SN2 displacement. This largely restricted
the products to N-alkyl groups from which ethyl or
bioisosteric analogues thereof (O-methyl and Nmethyl, for example) proved commercially significant.
The synthesis concludes with an ester hydrolysis.
When X ) Cl or F at carbon 7, a nuclear aromatic
displacement reaction with a secondary amine conveniently introduces an amino substituent at the C-7
position because of the activation by the C-4 carbonyl
substituent. It is expedient to perform the hydrolysis
first to avoid an ester-amide exchange. This reaction
sequence allows a wide number of analogues to be
prepared, variously substituted piperazines and pyrrolidines of which are most significant. Substitutions
at other points require variants of this overall
process.
2.2. GroheHeitzer Reaction

Access to novel quinolones was greatly expanded
subsequently by the introduction of the GroheHeitzer cycloacylation synthesis (Figure 4).58 In this
process, the aromatic ring is acylated to begin with
so the positions of substitution are preset in the
starting acid and persist to the end. Commonly, a
suitable benzoic acid derivative is first elaborated
into benzoylmalonate ester. The active methylene
function is then condensed under dehydrating condi-
Figure 4. Basic Grohe-Heitzer reaction.
tions with an ortho ester. The resulting enol ether is

often subjected to an addition-elimination reaction
with a suitable primary amine, and this product can
cyclize in a tandom addition-elimination reaction at
the ortho position. Alternatively, a different primary
amine can be used to complete this reaction. These
processes establish the pharmacophoric keto acid
moiety of the pyridine ring and allow for introduction
of a wide variety of N-substituents including those
possible with the Gould-Jacobs reaction, but now
aryl and cycloalkyl substituents that could not be
made in that manner are accessible as well. Ciprofloxacin is the most important of this group and bears
an N-cyclopropyl substituent. Temafloxacin and trovafloxacin have N-fluoroaromatic substituents, and
their preparation was also made possible by developing this reaction. This important reaction is now very
widely employed for the preparation of analogues,
including even N-tert-butyl-bearing substances.59
An adaptation of the Grohe-Heitzer synthesis to
synthesis on beads employing traceless linker combinatorial methods has been reported.60 Use of the
Groehe-Heitzer method in a solution-phase multiple
parallel synthesis has also been published.61
2.3. GersterHayakawa Syntheses

The Gerster-Hayakawa synthesis of the N-1 to C-8
bridged compounds is a variant of the Gould-Jacob
reaction (Figure 5). In the Gerster synthesis, a
carbocyclic analogue is produced by a Gould-Jacobs
reaction using a suitably substituted tetrahydroquinoline synthon.62 This synthesis was subsequently
modified by resolution of the tetrahydroquinolone
intermediate using an optically active amino acid
ester as a chiral auxiliary. The Gerster chemistry was
initially used to produce flumequine, the first fluoroquinolone, the first chiral member of this class, and
one of the first rigid analogues involving the N-1
substituent.62,63 Resolution demonstrated that the
absolute configuration of the N-1 substituent was
very important for maximal antimicrobial activity.
Despite the possession of so many trend-setting
characteristics, flumequine has not seen human use.
In the Hayakawa variant, bioisosteric C-1 oxo or
C-1 thio analogues became accessible by starting with
a nucleophilic aromatic displacement reaction on a
fluorinated nitrobenzene synthon (Figure 5).64
The rigidification resulting from these two methods
led to enhanced potency and greater activity against
Gram-positive analogues. These reactions are il-
Figure 5. Basic Gerster-Hayakawa syntheses: (a) original Gerster synthesis of the carbatricyclics, (b) Hayakawa
synthesis of the oxatricyclics.
ofloxacin.32
lustrated with the synthesis of

Resolution
of one of the intermediates led to its more active
enantiomer, levofloxacin, one of the market leaders.33
2.4. ChuMitscher Synthesis

Chu and Mitscher introduced an efficient chiral
synthesis of levofloxacin and its analogues (Figure
6).65 This method uses a chiral R-amino alcohol,
Figure 7. Chu-Li synthesis of C-cyclopropylpyridones.
analogues of the naphthyridine series (Figure 7). The

pyrimidine ring is assembled on a cyclopropylated
cyanoacetyl ester nucleus. Selective reduction to the
aldehyde allows formation of an alkylidine malonyl
ester. Heating then, analogous to the Gould-Jacobs
process, results in aroyl amide formation. Transformation of the hydroxyl moiety to a halogen subsequently enables the necessary nucleophilic aromatic
substitution reaction required to complete the synthesis.
Synthesis of 9-difluorophenylpyridopyrimidone analogues follows a closely similar path (Figure 8) but
starts with the aryl group in place at the outset.
Figure 6. Chu-Mitscher synthesis.
avoiding the necessity of a wasteful resolution step

at the end. Starting with optically active alanol and
employing a variant of the Grohe-Heitzer cycloarylation, chiral products were produced directly, and
the eutomeric configuration was established unambiguously to be S. The overall process is short and
very efficient. If somewhat forcing conditions are
employed in the tandem ring forming reaction, the
formation of the benzoxazine ring takes place also
and does not require a separate step.
2.5. ChuLi Syntheses

Synthesis of the investigational 9-cyclopropylpyrimidones required the development of chemistry
novel to the quinolone area.66 First are illustrated
Figure 8. Synthesis of C-arylpyridones.
Mitscher
Figure 10. quinolone chelates.
administration cannot be avoided, then a patient who

is a good complier with complicated administration
regimens can take the ion-rich drug or food 1 h before
or 2 h after taking the drug. This should minimize
the problem. Some toxic effects of quinolones are
exacerbated by this kind of interaction. There are
suggestions that photosensitivity is increased under
these conditions as is tendon erosion and even
mutation to resistance. The chemical nature of the
chelates is illustrated in Figure 10.
3.2. AcidBase Character

Figure 9. Synthesis of tricyclic pyridone derivatives.
Synthesis of tricyclic 3(S)-3-methyl-6-oxo-2,3-dihydro-6H-pyrano[2,3,4-ij]quinolizinone analogues requires initial replacement of the most reactive fluorine atom of pentafluoropyridine by an oxygen
substituent using tert-butoxide (Figure 9). This symmetrical product is alkylated and deprotected. Next,
reductive removal of the now surplus fluorine on the
other side of the pyridine nitrogen is carried out. A
nucleophilic aromatic substitution reaction completes
the second ring. A carbon version of the usual GouldJacobs reaction then follows to produce the tricyclic
ring system. Hydrolysis next is followed by a nucleophilic aromatic substitution reaction.
With one or another variation of these flexible
syntheses many thousands of pyridone analogues
have been prepared and evaluated.
Although the first generation of quinolones contains a number of monovalent, acidic examples of
largely hydrophobic character, the bulk of the quinolones of present clinical importance are amphoteric
substances possessing enhanced hydrophilicity. Consequently, the more recent compounds possess their
minimum aqueous solubility in the vicinity of neutral
tissue compartments. They are salts at pH extremes
and so have better solubility under these generally
nonphysiological conditions. Figure 11 illustrates this
3. Some Quinolone Chemical Reactions of

Significance to Their Medicinal Properties
The clinical behavior of the quinolone anti-infectives is strongly affected by a few of their chemical
properties.
3.1. Chelation
The carboxylic acid moieties of quinolones form
salts with metal ions, particularly in neutral to basic
solutions.67,68 The proximity of the carbonyl group at
C-4 leads to electron donation such that strong
chelate rings are formed. Chelation with metal ions
of higher valence, such as aluminum(III), magnesium(II), calcium(II), iron(II and III), copper(II), and
so on, often leads to water-insoluble complexes that
can interfere with blood levels following oral coadministration. Quinolones resemble the tetracyclines in this aspect. This is not only inconvenient
for formulation but leads to drug-food interactions
(especially with dairy products) and to drug-drug
interactions, leading to poor blood levels, particularly
with co-administration of certain antacids and with
hematinics. This problem is alleviated significantly
by administration in acidic media. If such co-
Figure 11. Protonation/deprotonation scheme for quinolones.
with ciprofloxacin and the proton equilibrium it

undergoes at various pH levels. At alkaline pH values
quinolones form carboxylate salts with reasonable
water solubility, and at acidic pH values they form
protonated amine salts likewise with reasonable
water solubility. At neutral pH values near their
isoelectric point they possess two forms in equilibrium. The zwitterion form is primarily responsible for
the degree of water solubility that they retain under
these conditions. On the other hand, the nonionized
form is also populated, and it is this form that is the
better absorbed. The particular relevance of these
features is that the quinolones are known to enter
most mammalian tissues and almost all bacterial
cells by a combination of passive uptake or porin
passage.69,70 Thus, a more accurate prediction of their

potential bioavailability following oral administration
is obtained by measuring their partition coefficient
under physiological conditions. Since they can form
insoluble salts with buffer components, particularly
when multivalent metal salts are present, the nature
of the buffer must be taken into consideration as
well.67,68
Since it is desired to administer quinolones by
injection and this is likely to suffer from the insolubility disadvantage at neutral pH values, it is common to buffer injectable preparations of quinolones
at acidic pH values where solubility is much improved, but it is therefore required to infuse them
comparatively slowly rather than by push to avoid
pain and blood vessel occlusion due to precipitation.
3.3. Photochemistry
Quinolone anti-infectives are often quite photoactive, especially under neutral or acid conditions.71,72
Product formations involve free radical intermediates, and the nature of the products depends on the
structure of the quinolone and the specific conditions
employed. Those quinolones substituted with a halogen at C-8 are particularly associated with these
reactions, whereas those with C-8 methoxy substitutents are less so.
Many quinolones, particularly those with halogen
substituents, absorb light in the 350-425 nm region
and are transformed thereby into singlet and triplet
states. The triplet state in particular is strongly
oxidizing, presumably leading in part to generation
of reactive oxygen species, and many of these agents
have nucleofugic groups (fluorine and chlorine atoms)
and so undergo facile nucleophilic aromatic substitution reactions. When a chlorine atom or a second
fluorine atom is present along with additional functional groups able to donate electrons to this
site, these tendencies are enhanced.72 Sparfloxacin,
lomefloxacin, and fleroxacin are examples.
Since patients undergoing quinolone therapy may
experience photosensitivity, it is believed that this
chemistry is relevant to this side effect. One notes
in particular that the 350-425 nm wavelength range
is that part of visible light associated with suntan
and sunburn.
At first glimpse the comparatively ready photochemical defluorination of quinolones is surprising
given the normally high stability and strength of the
aromatic C-F bond, but the quinolones are substituted in such a manner as to overcome these effects.
Once a quinolone radical is formed, the molecule
further reacts in one or more of several manners. The
N-alkyl substituent can be lost entirely or react with
the C-8 position to form a new ring, the C-3 carboxyl
group can be lost with or without further hydroxylation at either C-2 or C-3, a C-5 or C-6 fluorine can
be replaced by a phenolic hydroxyl or a hydrogen, the
C-7 aliphatic side chains can undergo a variety of ring
cleavages, and C-8 can lose its halogen with or
without reaction with the N-1 substituent. These
reactions are illustrated in Figures 12-18.
The early quinolones not possessing a C-6 fluorine
substituent or a piperazinyl moiety at C-7 underwent
Figure 12. Photochemistry of nalidixic acid, a nonfluorinated quinolone.
decarboxylation, oxidation, and dimerization under

photolysis conditions in water.73 This is illustrated
with nalidixic acid, which is transformed primarily
to its C-3 H, or to its C-2 keto analogues. These
products are antimicrobially inactive (Figure 12).
Decarboxylation to a C-3 H analogue can also take
place with ciprofloxacin, but this is a minor product.
The point of origin of the triggering radical is not very
obvious but can be posited to involve chelation with
suitable transition-state metals, such as ferrous iron.
With ofloxacin the more significant photolysis
product is one where the piperazinyl moiety at C-7
has been oxidatively cleaved.74 When an electrondonating group is attached to C-8, as with ofloxacin,
levofloxacin, and moxifloxacin, the products observed
often involve such bond cleavages in the C-7 sidechain moiety.75 This tendency is exaggerated in acidic
solutions. T-3761, however, has a rather different C-7
amino substituent, and on photolysis it produces a
C-3 OH analogue instead (Figure 13).76 The analogy
Figure 13. Photochemical reactions of T-3761 and ofloxacin.
between this carbon-linked molecule and nalidixic

acid is clear.
Radicals generated at C-8 by loss of halogen atoms
can interact with the N-1 substituent, resulting
either in its loss, as seen in Figure 14 in the case of
N-cyclopropyl-substituted sparfloxacin, or in formation of a carbocyclic ring as seen with N-1 ethylated
analogues.72 Alternately, they can be quenched by
hydrogen atoms, leading to replacement of F by H.
These reactions are also illustrated in Figure 14.
When a second halogen atom is added to C-5 or
C-8, photolysis is enhanced and replacement of the
halo group by H, OH, or Cl or interaction with the
alkyl group at N-1 is observed.77 Cleavage of a C-7amine-containing moiety is also observed. Quinolones
such as sparfloxacin, lomefloxacin, and fleroxacin also
undergo these reactions.
Photolytic loss of C-8 chlorine is more facile than
loss of C-8 fluorine, and C-8 fluorine is lost in
preference to C-5 fluorine as shown by the photochemistry of orbifloxacin (Figure 15), which possesses
molecular features similar to those of sparfloxacin
Mitscher
Figure 17. Novel quinolone anti-infectives apparently

relatively stable to photolysis.
Figure 14. Photolysis of sparfloxacin and selected Nethylated quinolones.
quench or prevent the formation of radicals at C-8,

and therefore, such quinolones promise to possess
milder phototoxicity. Such quenching groups are
1-aminodifluorophenyl and 1-isoxazoyl. These compounds possess at the same time quite satisfying
antimicrobial potency in vitro.71 Figure 17 presents
the structure of two of these.71 It will be interesting
to see whether exploitation of these findings produces
clinical benefits.
In summary, a large number of photodegradation
reactions are possessed by quinolones, and these
reactions are quite capable of generating reactive
oxygen species and may trigger photosensitivity reactions in patients. The recent finding that some members of the quinolone class retain very significant
antibacterial activity even though they do not have
a C-6 fluorine moiety has excited much interest.79-81
It is reasonable to suggest that they may have
reduced phototoxicity. None of these have yet reached
the marketplace, however.
4. In Vitro Antimicrobial Spectra
Figure 15. Photolytic reactions of orbifloxacin.
but with the addition of a fluorine moiety at C-5. In

addition to losses of N-1 cyclopropyl and C-8 fluoro,
it also undergoes C-7 piperazinyl cleavage and replacement of the C-5 fluorine by hydroxyl. One
perhaps can infer from this that halogen atoms at
C-5 and C-8 are more sensitive to photolysis than
halogens at C-6.72 Nonetheless, when there is only a
C-6 fluorine as with norfloxacin, enoxacin, and ciprofloxacin, photolytic replacement of their C-6 fluoro
atom by a hydroxyl group can take place as illustrated by a generic formula in Figure 16.78 Here,
Figure 16. Photochemical replacement of a C-6 fluoro

atom by OH.
as elsewhere, analogous chemistry takes place in the

quinolone and naphthyridinone ring systems when
they are similarly substituted.
It is interesting to note that the presence of certain
aromatic and heteroaromatic substituents at N-1 can
Overall, the quinolones possess antimicrobial spectra and potency attractive for clinical use. In particular, they are bactericidal in achievable oral doses
and possess significant postantibiotic effects. These
features are especially useful for treating infections
of immune-suppressed patients.
Microorganisms regarded as highly susceptible to
quinolones have minimum inhibitory concentration
values ranging from 0.01 to 0.2 g/mL. Examples
include E. coli, Klebsiella pneumoniae, Enterobacter,
Salmonella, Shigella, Vibrio, Hemophilus influenzae,
Neisseria, and Legionella. Less susceptible but still
sensitive microorganisms lie in the range of 0.25-2
g/mL. These organisms include those that become
resistant more easily. Especially notable are Pseudomonas aeruginosa and S. aureus (with particular
emphasis on MRSA). Organisms that are regarded
as insensitive have MIC values of 2 g/mL or higher.
Examples include Nocardia, Treponemia, and anaerobes.
Clinicians often classify quinolones as first-, second-, and third-generation agents on the basis of their
antimicrobial spectra. The classical first-generation
quinolones such as nalidixic acid and pipemidic acid
were of interest because of their activity against
Gram-negative microorganisms, with particular emphasis on the commonly encountered urinary tract
pathogen E. coli acquired in the community and
therefore less likely to be drug resistant. Their
specific potency is not very high, and resistance
development can occur even during the course of
therapy.
Table 1. Antimicrobial Sensitivities of Selected Quinolonesa

microorganism
Nor
Cipro
Enox
S. aureus
MRSA
Staphylococcus epidermidis
MRSE
Staphylococcus hemolyticus
Streptococcus pyogenes
Streptococcus viridans
Enterococcus faecalis
E. coli
Chlamydia trachomatis
Chlamydia pneumoniae
Enterobacter sp.
Gardnerella vaginalis
H. influenzae
K. pneumoniae
Legionella pneumoniae
Mycoplasma hominis
Mycoplasma pneumoniae
Neisseria gonorrhoeae
Proteus mirabilis
Proteus vulgaris
Providencia rettgeri
Providencia stuartii
P. aeruginosa
Salmonella typhi
Seratia marcescens
Shigella sp.
Ureaplasma urealyticum
Vibrio chloerae
+
+
+
+
+
+
+
+
+
+
+
Gati
Lome
Moxi
Spar
Trov
Alatro
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Gram-Positives
+
+
+
+
Levo
Oflox
Gram-Negatives and Special Microorganisms

+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Anaerobes
Bacteroides fragilis
Clostridium perfringes
a
+
+
+
+
+ means that the culture is normally regarded as sensitive to ordinary concentrations.
A comparison of the susceptibilities of a wide

number of pathogenic bacteria with representatives
of each generation of quinolones is presented in Table
1. If space allowed, the first generation of quinolones
would be represented by nalidixic acid. In that case,
only E. coli and Enterobacter sp. would be checked.
Other members of this generation include rosoxacin,
oxolinic acid, and cinnoxacin.
In the table the second generation of quinolones is
represented by ciprofloxacin, norfloxacin, enoxacin,82
lomefloxacin,83 and ofloxacin. The antimicrobial spectra of these agents, ciprofloxacin of which is the sales
leader, show a progressive broadening of the antimicrobial spectrum, retaining an emphasis on Gramnegatives and including some significant Grampositives. Activity against P. aeruginosa, S. aureus,
and Streptococcus pneumoniae is often observed, but
the effective dose is marginal, and breakthrough to
resistance is not uncommon. Anaerobes are only
occasionally inhibited. A number of microbes lacking
cell walls such as Legionella, Chlamydia, and Mycoplasma are also inhibited.
The third-generation quinolones in the table include levofloxacin, gatifloxacin, lomefloxacin, moxifloxacin, sparfloxacin, and trovafloxacin/alatrofloxacin. These agents are even broader in spectra,
retaining the overall spectrum of the second-generation agents but possessing in addition activity against
some strains marginally sensitive to the secondgeneration group and many resistant strains and
adding a number of anaerobes as well. They are,
however, somewhat more toxic. Of the newer quinolones, WCK-771 (the L-arginine salt of nadifloxacin)
has one of the most impressive in vitro anti-anaerobe
spectra.84
At present the search is on for quinolones that
retain the attractive features of the present members
but lack the constellation of side effects, most particularly the severe toxicities occasionally encountered with the most potent members.
Drlica and Hooper classify the microbes sensitive
to quinolones in a genetic/microbiological sense rather
than in a clinical sense.5 This is based upon the genes
encoding the target enzymes and the relative sensitivity they possess to inhibition by quinolones.
Type 1 includes a group of pathogens that are
comparatively insensitive to quinolones. These have
genes predominantly producing a DNA gyrase that
has a nonpolar alanine residue at a position normally
occupied by a polar serine or threonine residue in the
A subunits. This decreases the sensitivity of this
gyrase to quinolone action. This type also appears
not to have a significant content of topoisomerase IV
and so depends on its gyrase to decatenate, etc.
Microorganisms belonging to this class include Mycobacterium tuberculosis and related mycobacteria
Mitscher
Treponema pallidum and Helicobacter pylori.85,86

Types 2 and 3 are less precisely delineated. In type
2, both DNA gyrase and topoisomerase IV are present,
but the gyrase is more sensitive to quinolone inhibition. These organisms are mainly Gram-negatives.
Mutations of topoisomerase IV have comparatively
less effect on sensitivity unless accompanied by
mutations in gyrase.85,86 Type 3 microorganisms also
have both DNA gyrase and topoisomerase IV, but the
gyrase is less sensitive to quinolones so that resistance mutations in the genes for topoisomerase IV
are more significant for resistance. These microorganisms are mainly Gram-positives.85,86
5. StructureActivity Relationships
It is convenient to discuss quinolone structureactivity relationships position by position with the
caution that apparently distant constituents modulate each others properties because they are all
connected to the same electronic system and crosstalk is possible. The modulation normally involves
the intensity of specific properties rather than the
kind but should be kept in mind in making predictions and comparisons. Cross-talk is diminished
somewhat in the benzoxazines and quinolizines
where aliphatic atoms intervene.
With the structures that follow, those that have
names are those of particular prominence, especially
those that have been marketed. The marketed agents
are further identified by the year of their introduction. Each figure also contains a selection of microbes
and their sensitivities (g/mL) to the agent in question.
5.1. N-1 Ethyl Family

With the early, classic agents, small, nonpolar,
unbranched aliphatic agents proved best, with Nethyl being most useful.87 Polar substituents at N-1
have generally been disappointing. This relationship
has generally held up with later agents.
From Figure 18 it can be seen that the N-1 ethyl
substituent, sometimes halogenated, is still found in
many contemporarily important quinolone anti-infective agents. As time passed, there was a continual
enhancement in anti Gram-negative potency and
breadth of the spectrum when the various other
positions were explored. Norfloxacin was the breakthrough molecule as its potency and spectrum approximated those of the fermentation-derived antibiotics. Norfloxacin retains the classical N-ethyl
moiety. In recent years there has been an increased
emphasis on anti Gram-positive activity with these
agents, and this is reflected also in Figure 18. Activity
against anaerobes is generally lacking in this group.
5.2. N-1 Cyclopropyl Family

The N-1 cyclopropyl moiety first materialized with
ciprofloxacin.30 This substituent could not be introduced with the original Gould-Jacobs chemistry, but
the Groehe-Heitzer chemistry proved enabling for
it. As compared with norfloxacin, this change enhanced anti Gram-negative potency (Figure 19). This
moiety became standard for a long time thereafter
Figure 18. Structures and abbreviated antimicrobial

spectra of significant members of the N-1 ethyl family of
quinolones.
in quinolone analoguing, with potency being modulated by substituent changes at C-5, C-7, and C-8,
but ciprofloxacin has largely withstood these challenges and remains a market leader to this day. With
progressively later entries, anti Gram-negative activity intensified and anaerobic activity became more
frequent.
5.3. N-1 to C-8 Bridged (Tricyclic) Family

In this group a C-8 substituent is linked to an
N-ethyl moiety at N-1, resulting in the tricyclic family
of quinolones (Figure 20). This results in restricted
rotation of the ethyl group and introduces a chiral
atom. This feature was first present in flumequine,62,63 which incidentally also was the first fluoroquinoline anti-infective. Flumequine is used today
primarily for agricultural purposes, but its unmarketed S-analogue was shown to be the eutomer.63
This stereochemistry persists in all of the resolved
marketed analogues to this day.
market leaders. Rigidification of the N-1 substituent

in this way resulted in a significant enhancement in
anti Gram-positive activity although some anti Gramnegative activity was concomitantly lost. Resolution
of ofloxacin leads to a doubling of potency, and this
is apparently helpful against some otherwise marginally sensitive microbes such as S. pneumoniae.
The potency enhancement agrees with the finding
that the distomer has very little biopotency against
bacteria. The nature of the atom attached to C-8
seems comparatively unimportant as C, O, and S
bioisosteres possess similarly significant activity and
the products show the same chiral dependence.
5.4. N-1 Aryl Family

The introduction of an N-aryl substituent was
made possible by the Groehe-Heitzler methodology,
and this substituent change has proven to be quite
successful in specific instances (Figure 21). Difloxa-
Figure 19. Structure-activity relationships in the N-1

cyclopropyl family.
Figure 21. Structure-activity relationships in the N-aryl

family.
Figure 20. Structure-activity relationships of the tricyclic

quinolone family.
Racemic ofloxacin was the next member of this

group to be marketed, and it quickly became popular.
Subsequently, it was largely replaced by its resolved
S-analogue levofloxacin,64 which is one of the present
cin88 and sarafloxacin89 represent early entries that

did not see commercial use. The presence of a second
fluorine atom in the aromatic ring, as with tosufloxacin90 and trovafloxacin,91 however, further enhanced
both potency against Gram-positives and pharmacokinetics, and they were marketed. Unfortunately, as
will be noted in greater detail later, the addition of
the difluorobenzene substituent was accompanied by
uncommon but severe toxicities that would be difficult to detect in clinical trials unless a mammoth
number of patients were to be enlisted. The connection between this substituent and these unusual
toxicities is logical but remains to be firmly established. It seems that the connection will not be easy
to establish as the toxicities are rather different from
each other and so do not present a common pattern.
They appear to be class effects, but the specific
molecules in question possess them to an unusual
degree.
One supposes that a useful future role for genomics
might be in the detection of the genetic differences
between these unfortunates and the bulk of the
Mitscher
Figure 22. A comparison of the effect of important N-1

substituents on in vitro activity.
general population so those who should not receive

these agents could be reliably predicted. Not only
would this enhance the likelihood of success in
clinical evaluations but could also prevent devastating results encountered by particular patients and
their loved ones following general use. It would also
be welcomed by firms that must put forth very large
sums in research and development of quinolones at
the risk of suffering economic disaster at the latest
stages.
Of the various noncyclic N-1 substituents investigated, the N-tert-butyl analogues were found to be
unexpectedly potent in vitro, especially against Grampositives.59 This substituent is not very stable in
acidic solutions, and no example of this class ultimately reached the clinic.
Thus, it is clear that several important series of
antimicrobial agents are associated with particular
N-1 substituents. The properties of these agents are
clearly modulated by the nature of the C-7-aminecontaining substituent, and often the addition of a
suitable C-5 and C-8 substituent produces added
potency.
In each of these individual series the structures
depicted have emerged as best in show. Mostly, when
the other constituents are held constant, the potency
of N-1 analogues against Gram-positives follows the
order aryl > ethyl > 1,8-fused > c-propyl. Against
Gram-negatives the order is slightly different: ethyl
c-propyl > aryl > 1,8-fused (Figure 22). In Figure
22 the compounds chosen for comparison are identically substituted away from N-1.
In Figure 23 a selection of analogously substituted
earlier quinolones is presented, demonstrating that
ethyl is ordinarily the best of the aliphatic group.
Branching, increased chain length, and polar groups
are all usually detrimental. Vinyl and tert-butyl are
however close to ethyl in allowing potency. These
findings strongly influenced subsequent work.
5.5. Positions C-2, C-3, and C-4

Compared to the other positions, carbons 2, 3, and
4 have not been much represented in analoguing
studies, and no analogue modified at C-2 remains on
the market bearing other than a CH substituent.
This comparative lack of apparent exploration stems
from the early findings that alterations here are
usually unsatisfactory87 (Figure 24).92 It is possible
that the basis for this phenomenon is that the
carboxyl group at C-3 needs to be coplanar with the
C-4 carbonyl so that it can effectively hydrogen bond
Figure 23. An abbreviated list of quinolone substituent

potencies at N-1.
Figure 24. Some substituent changes explored involving

carbons 2-4.
with DNA bases made available by strand separation

catalyzed by DNA gyrase or topoisomerase IV. The
steric deficit that substitution at C-2 presumably
causes can be partly overcome by rigidifying a C-2
substituent in a ring connecting it with N-1, especially when the group attached to C-2 can accept a
hydrogen bond and form a virtual ring with the C-3
substituent.93 In support of this idea is the outstanding potency seen with formation of a thiazolone ring
fused to the carbonyl ring.94 When this is done, the
NH moiety is strongly acidified by resonance stemming from the aromaticity of the ring when enolization takes place. This produces a coplanar carboxyl
surrogate and leads to a striking enhancement of
antimicrobial potency. Unfortunately, this attractive
feature is negated by the propensity of these molecules to kill mammalian cells also by damaging their
DNA through intercalation. The flat fused three-ring
aromatic system is no doubt responsible. These
compounds might have a future as antitumor agents.
Aside from the thiazolone modification just noted,
replacement of the C-3 carboxyl by other moieties has
not led to economically valuable agents. Those that
retain significant acidity show some potency.
Some prodrugs have been made by esterification
of the C-3 carboxyl moiety or synthesis of the aldehyde (which is oxidized in the body to the acid).95-98
Figure 25. Substituent changes at the C-4a position: the

2-pyridones.
A commercially significant prodrug is alatrofloxacin,

an injectable form of trovafloxacin which is otherwise
too lipophilic for ordinary parenteral administration.99 The generally excellent absorption characteristics of the quinolones diminish motivation for
prodrugging.
The C-4 carbonyl is essential for bioactivity. This
finding provides one of the primary motivations for
alkylation of N-1. Unless this is done, the 4-pyridone
ring enolizes primarily to its 4-hydroxypyridine form,
and analogues of this type are inactive.
5.6. C-4a Substituted Analogues

The C-4a position in quinolones has no valences
left for substitution. It can, however, be exchanged
for the nitrogen atom normally found at N-1.66 This
allows for a redistribution of the electrons in the rings
so as to preserve aromaticity. These analogues have
been named 2-pyridones so as to differentiate them
clearly from the quinolones. Interestingly, these
double bioisosteric analogues possess substantially
increased in vitro antibacterial potency as shown by
a comparison of ABT 719 and the quinolone possessing the same substitution pattern at the other positions (Figure 25).100 Notably, potency against otherwise resistant microorganisms is significant among
the pyridones. A selected number of 2-pyridone
analogues have progressed into development, and at
least one has received significant clinical examination. Despite the passage of significant time, commercialization has not occurred yet. Although details
are sketchy, it is believed that they possess some
toxicities requiring caution. Interestingly, ABT 719
is also significantly lower melting and more watersoluble than the corresponding quinolone. X-ray
studies indicate that the C-methyl group twists the
ring system out of plane so that it does not stack as
compactly as the corresponding quinolone. This provides a probable rationale for these observations.

Substituents at C-5 are tolerated, especially if they
are small, and it helps potency if they are polar as
well. Groups that have been studied include N, COH,
CNH2, CNHMe, CNMe2, CNHAc, CCH3, CEt, CCl,
and CF. Sparfloxacin101 and grepafloxacin102 (illustrated under the N-1 cyclopropyl molecules) are
the most successful of these. As noted elsewhere in
this paper, the 5-halo substituents contribute significantly to phototoxicity, so there is a tradeoff
involved in using them. The other substituents have
been unsatisfying and have not progressed.
Figure 26. Comparison of the potencies of a C-6fluorinated and -desfluorinated quinolone.

Substituents at C-6 in the early investigations were
generally limited to the terminus of a methylenedioxy
bridge to C-7 (oxolinic acid,103 miloxacin,104 and
cinoxacin92), CH (nalidixic acid27 and rosoxacin105),
and N (pipemidic acid106 and piromidic acid107). Those
that were marketed have been largely supplanted
with analogues possessing a CF moiety after the
attractive properties of norfloxacin were revealed. It
is generally believed that the C-6 fluorine substituent
conveys enhanced DNA gyrase potency and enhances
cell penetration. These favorable features operate in
conjunction with compatible C-7 and C-8 substituents, and the need for a C-6 fluoro became dogma
until fairly recently.
It has now been shown in a number of cases that
the positive effect of a C-6 substituent is diminished
when the molecule contains other helpful substituents. A present vogue for investigation of nonfluorinated analogues has developed enhanced by the
suggestion that C-6 F may play a role in the potential
mammalian genotoxicity and central nervous system
side effects of quinolones.108,109 The examples in
Figure 26 reveal no difference in potency against E.
coli and S. aureus or the E. coli derived gyrase
whether the C-6 substituent is fluorine or hydrogen.
There is a slight advantage against topoisomerase
IV and human topoisomerase II, however.
Other C-6 substituents that have been investigated, with less satisfactory results, are Cl, Br,
methylketo, CN, nitro, methyl, and amino.

The most versatile position for substitution of
quinolones has been C-7 and its analogous position
in other ring systems. Many thousands of analogues
have been prepared employing various substituents
at this forgiving position, leading to the conclusion
that a cyclic system containing a secondary or
tertiary amino moiety is usually best. The beneficial
effects are usually believed to be enhanced potency
and favorable pharmacokinetics. The nature of the
C-7 substituent (along with C-8) also strongly affects
the target preferences (DNA gyrase and/or DNA
topoisomerase IV) of quinolones. Larger substituents
on the distal nitrogen generally decrease potency.
The earliest quinolones possessed a chlorine, a methyl (nalidixic acid), the terminus of a methylenedioxy
bridge to C-6 (oxolinic acid, miloxacin, or cinnoxacin),
a 4-pyridyl (rosoxacin), a pyrrolidinyl (piromidic acid),
or a piperazinyl moiety (pipemidic acid). On the basis
of these early findings, a consensus formed that

cyclized amino moieties were superior. One notes, on
the other hand, that a proximal electron-releasing
group at C-7 can stabilize a radical generated at C-8
and so could contribute to phototoxicity and genotoxicity.
After the discovery of norfloxacin28 the piperazinyl
moiety or its N-methyl analogue became a standard
feature among quinolones. Later it was found that a
pyrrolidinyl substituent with a pendant amine was
a suitable moiety also.24 The piperazinyl analogues
usually have enhanced potency, emphasizing Gramnegatives, whereas the pyrrolidinyl analogues have
enhanced activity against Gram-positives. Subsequently, bicyclic moieties of a variety of types were
also introduced. These had the virtue of retaining
potency but also diminishing metabolic liability. The
amino function that is proximal to the aromatic ring
appears primarily to be a synthetic convenience.61
Clearly, it is at the same time much less basic than
the distal nitrogen atom. The potency effect of the
stereochemistry of carbons near the distal nitrogen
is variable but often weak. This leads to the reasonable belief that this is not a point of close approach
to the edges of the binding pocket. On the other hand,
chiral substituents closer to the aromatic ring have
a more significant effect on potency.90 It is speculated
that this may influence the rotamer distribution of
the C-7 substituent.
The structures in the following few figures include
many of the groups that have been attached to C-7.
In these structures Q represents one of the quinolone, naphthyridine, or pyridone ring systems. The
looseness of the SAR associated with this position is
demonstrated by the wide variety of such moieties
that are associated with roughly similar potencies.
Mitscher
interested reader is drawn back to the properties of

lomefloxacin (Figure 18), balofloxacin (Figure 19),
danofloxacin, gatifloxacin, sparfloxacin, grepafloxacin, and nadifloxacin (Figure 20), and trovafloxacin
(Figure 21), important quinolones decorated with
these groups.
5.9.2. Pyrrolidinyl and Related Moieties

Figure 28 depicts a number of five-membered
heterocyclic rings that have been attached to C-7 of
various quinolones. Of these, isloxacin (Figure 18),
moxifloxacin, clinifloxacin, gemifloxacin, and olamufloxacin (Figure 19), and tosufloxacin (Figure 20)
stand out.
5.9.1. Piperazinyl and Related Moieties

Figure 27 depicts a selection of the many simple
and fused piperazines and piperidines that have been
attached to C-7 of quinolones. The attention of the
Figure 28. Some pyrrolidines and related moieties attached to quinolones.
5.9.3. Cyclobutylaminyl and Related Moieties

No prominent quinolones incorporate this moiety
as yet. Figure 29 shows some examples.
Figure 29. Some cyclobutylamines attached to quinolones.
5.9.4. Bicycloaminyl Moieties

Prominent quinolones with this structural feature
include moxifloxacin (Figure 19), danofloxacin, olamufloxacin, and trovafloxacin (Figure 21). Figure 30
depicts some additional examples.
5.9.5. Carbon-Linked Substituents

Figure 27. A selection of piperazinyl and related moieties
attached to quinolones.
There has only been a limited exploration of

substituents attached to C-7 through carbon in large
potency effect of a C-8 halogen atom is counterbalanced by an enhanced likelihood of phototoxicity and
mammalian clastogenicity. Prominent C-8-F-bearing
quinolones include lomefloxacin and sparfloxacin. An
O-methyl group, on the other hand, often increases
potency without increasing photoxic liability (see
gatifloxacin and moxifloxacin). Oxygen constrained
in a ring as in ofloxacin/levofloxacin, or sulfur as with
rufloxacin, seems to carry over these benefits. Bridging to N-1 with a carbon moiety such as with
flumequine and nadifloxacin may or may not also
convey these properties. The oxo and thio series can
be considered as rigid analogues of methoxyl- or
thiomethyl-bearing quinolones.
The first successful variant at C-8 was the bioisosteric replacement of CH by N (nalidixic acid). This
sort of replacement reoccurs regularly in the quinolone field and often leads to enhanced pharmacokinetic characteristics. Other groups at C-8 that have
been investigated with less salubrious effects include
O-ethyl, OH, OCH2F, OCHF2, OCF3, and SMe.
5.11. Resume of StructureActivity Relationships

of Quinolones
Figure 30. Bicycloamines attached to quinolones.
measure because they are more cumbersome to

synthesize. Most of these possess aromatic rings, and
a significant number possess enhanced activity against
Gram-positive microorganisms. They are often more
active against human topoisomerase II and so are
potentially toxic. A selection of these is shown in
Figure 31.
Figure 32 contains a pictorial summary of these

findings with reference to the parts of the molecule
that are presumed to be in contact with the enzyme,
with DNA, and with other quinolone moieties (thus
forming the ternary complex). At the left is a cartoon
viewed from the top. In this view, north and south
represent DNA-binding sites wherein the keto and
carboxyl groups hydrogen bond to single-stranded
segments made available to them by enzymic action.
West and east represent binding sites to DNA gyrase.
In the interior of the complex, the quinolone molecules are associated with each other through hy-
Figure 31. Some moieties attached to quinolones through

C-C bonds.
5.10. Substituents at C-8

C-8 substituents have an important effect on in
vivo efficacy and the antimicrobial spectrum of quinolones. In particular, the C-8 substituents appear
to play a significant role in determining the comparative affinity of quinolones for DNA gyrase or topoisomerase IV. This effect is modulated significantly
by the nature of the C-5 substituent, especially if it
is a fluorine atom. Unfortunately, the favorable
Figure 32. Proposed ternary complex drug-binding pocket.
drophobic contacts. Virtually every portion of the

quinolone molecule is employed in one or more of
these interactions. The requirement for the enzyme
to open a saturable drug-binding pocket is implicit.
The important role of having a basic moiety attached
to C-7 is also apparent. The need for the groups
attached to N-1 and C-8 to be small and hydrophobic
is also rationalized. The two molecules are aligned
head to tail with respect to each other by electrostatic interactions involving the carboxyl and the
amino substituents. A second pair of interacting
quinolone molecules can also be aligned in a vertical
stack by the - interactions and the same electrostatic interactions.
An end view with the N-1, C-8 edge closest to the
viewer is shown on the right side of the figure. The
significance of the chiral preference conveyed by the
methyl group of flumequine and levofloxacin and
their analogues is rationalized by invoking the presence of a lipophilic drug-binding site above and below
the pocket. The methyl portion of the N-ethyl moieties is illustrated as fitting into this pocket and
providing a favorable interaction. Aligned in this way,
the S-enantiomer of ofloxacin (levofloxacin) has its
methyl groups aligned in a Boston-San Diego
manner and, when axially oriented, fits this putative
pocket quite well. The R-enantiomer would have its
methyl groups aligned in a Seattle-Miami manner
and clearly would not benefit from this interaction.
This is discussed in greater detail in the next section.
Mitscher
Figure 33. 6-Methyl-6,7-dihydro-2H-benzo[a]quinolizin2-one-3-carboxylic acids.
the quinolones whose occupation results in significant

binding energy enhancing enzyme inhibitory power
and subsequent antibiosis. The methyl group of the
enantiomeric R-antipodes could not make this contribution (Figure 32).
The requirement for a small, unbranched, lipophilic
moiety attached to N-1, such as N-ethyl, is consistent
with this hypothesis. N-Cyclopropyl can be accommodated in a similar fashion although not as well.
The N-1 aryl series does not fit quite as well either,
but this substituent is not too bulky in the important
dimension to fit. One notes that the aromatic N-1
substituents are orthogonal to the major plane of the
molecule to the extent of at least 30. Most interestingly, the N-tert-butyl analogues surprising potency
can also be accommodated as it has three ways to
fill this putative cavity with each rotation. These
proposed interactions are illustrated in Figure 34.
5.12. Absolute Configuration

Chiral substitution of quinolones has largely been
associated with particular substituents bridging from
C-8 to N-1 and to groups attached to C-7. Of these,
by far the most successful have been the bridged
examples. The molecular rigidification of an N-1 ethyl
substituent brought about in this way produces a
chiral center. The first examples using this idea were
the enantiomers of flumequine and ibaquine.62,110
From this work it was found that a very substantial
preference was displayed for the S-enantiomer with
respect to antibacterial and enzyme inhibitory potency. The 1-R-antipodes were very weakly active.
Later the antipodes of commercially successful
racemic ofloxacin were prepared with analogous results. Levofloxacin, the S-antipode, is a market leader
among the quinolones today. Subsequently, analogous findings have been made repeatedly, and benofloxacin,111 nadifloxacin,112 pazufloxacin,113 S-12684,
GRB-23790, WIN-58161,114 and DV-7751a115 have all
shown the same enantiopreference. The single exception to this otherwise general rule is the assertion
in a meeting abstract, not subsequently published,
that with the 6-methyl-6,7-dihydro-2H-benzo[a]quinolizin-2-one-3-carboxylic acids, such as Ro-14-5319
and Ro-14-4299, the eutomer has the R absolute
stereochemistry (Figure 33).116
If it is fair to set aside this exception because it
has a different ring system, it is attractive to rationalize the absolute configurational findings with the
rest of the quinolones along with a number of other
observations by proposing the existence of a small
lipophilic cavity in DNA gyrase above the plane of
Figure 34. Schematic view of interactions of various

quinolone analogues with the putative DNA gyrase lipophilic binding pocket.
In contrast, when asymmetric substituents are

present in the cyclic amino moieties attached to C-7,
the biological consequences are by and large minimal
if the asymmetry is distal to the aromatic ring
(Figure 35). Ariens notes that exceptions to the
classical Pfeiffer rule correlating the intensity of
chiral recognition as a function of high potency
breaks down if the chirality is not near a point of
close approach of a ligand to a receptor.117 This would
suggest that the majority of quinolones only loosely
fill the sides of the binding pocket proposed for the
C-7 substituent. On the other hand, significant
eudesmic ratios are seen when asymmetry is proximal to the aromatic ring.90 This is likely to inhibit
free rotation selectively so as to produce a lateral but
not a linear directional bias. It is also possible that
the binding pocket is small in this region, resulting
in chiral contacts.
The work of Shen et al. demonstrates that the
quinolones serially associate (show cooperativity in
binding) rather like liquid crystals assemble in the
6. StructureToxicity Relationships
Figure 35. Enantiomers and diasteriomers at C-7.
drug-binding zone created by the interaction of DNA

gyrase and topoisomerase IV with DNA.118 In this
view, the first molecule enters the melt zone and
conditions it to accept the second more readily
possibly through induced conformational alterations
and water displacement. The third and fourth would
each bind more easily still. Morressey et al. have
extended this idea with a stacking model accommodating chirality.119 The lipophilic cavity idea illustrated in Figures 32 and 34 fits this proposal well
and extends it somewhat. There are various ways in
principle in which quinolones could assemble in the
pocket. Energetically stacking that involved acidic
moieties in proximity to basic moieties (i.e., head to
tail) would be favored. The oxazine ring can possess
two conformerssone in which the methyl group is
axial and the other in which it is equatorial. The two
conformers are approximately equal in energy, but
a conformer in which the methyl group is axial would
agree better with the prominent role that it plays in
interfering with the enzyme. A stack in which the
methyl groups are external would allow much closer
approach of the ring systems and therefore be energetically favored. If the methyl groups were internal
to a stack, then closeness of approach would be
difficult. Finally, the chirality would make the stack
itself chiral. That is, the stacked supermolecules
would be chiral themselves. A chiral stack such as
that illustrated would clearly rationalize enantiopreference and show how chiral association of quinolones induced by the enzyme would enable both
single-stranded segments of the DNA bubble and/or
both arms of the cut gate to interact with drug,
stopping further movement and thereby inhibiting
enzyme action. Other enzymes involved in DNA
replication and function, such as helicase, bumping
into the frozen bubble cannot function because of
this frozen blockade. This may well be the sort of
interaction that leads to apoptosis. In this view,
mammalian topoisomerase II would not have the
same lipophilic cavity, providing a molecular rationale for the difference in toxicity of marketed quinolones against DNA gyrase over mammalian topoisomerase II.120
Although a keenly important topic, an understanding of structure-toxicity relationships of quinolones

is not well developed yet. In general, certain substituents at C-5 and C-8 are sometimes associated with
genotoxicity and phototoxicity. Substituents at C-7
are associated with genetic toxicity, GABA binding,
and P-450-related drug-drug interactions.24,68 Particular substituents at N-1 are associated with genetic toxicity and P-450-related interactions. Since
there is cross-talk between substituents, some substituents modulate the effects of substituents at more
distant centers. Those quinolone molecules that have
had to be removed from the market or have been
restricted in use due to severe toxicities (temafloxacin, trovafloxacin, sparfloxacin, and grepafloxacin)
bear C-5 substituents or N-difluorobenzene moieties,
but this group of molecules is too small to support
convincing generalizations.
6.1. Chemotype Toxicities and Side Effects

There are a number of annoying but not lifethreatening toxicities that are regarded as chemotype
side effects and toxicities in that they are common
among quinolones but are not shared equally. They
are more or less prominent depending upon the
structure of the particular agent. They are listed in
all the package insert precautions.34 Since these
adverse effects are general, they contribute little to
an understanding of individual structure-toxicity
relationships among quinolones.
Gastrointestinal complaints, including diarrhea,
dyspepsia, and nausea, are comparatively common
among antibacterials and often are primarily due to
disturbances of the normal gut flora. Thus, these are
commonly mechanism-based problems rather than
structure-based problems. These often resolve in a
few days without requiring discontinuance of the
drug.
Neuropsychiatric complaints include headache,
dizziness, sleep disturbances, and occasionally even
seizures. The seizure tendency is often attributed to
GABAA receptor binding and is commonly potentiated
by co-administration of certain nonsteroidal antiinflammatory agents. Fenbufen is particularly singled
out as being involved as a potentiator of this effect.24,29 The problem is attributed to the biphenyl
acetic acid moiety formed from fenbufen upon mammalian metabolism.121
Allergic reactions to quinolones include rash, urticaria, skin rashes, serum sickness such as reactions,
anaphylactoid reactions, and photosensitivity. Photosensitivity has been discussed briefly in a previous
section and is associated particularly with the presence of halogens at C-5 and C-8 and a participating
amino function at C-7. The free radical generated is
quite capable of producing reactive oxygen species,
leading to inflammatory responses to doses of sunlight that would normally not lead to a significant
response. The relationship between this and other
allergic reactions is unclear.
On intravenous dosing, irritation at the site of
injection is not uncommon. This manifests itself as
erythrema and phlebitis. These effects are ameliorated when a suitable substituent is present at N-1
or C-8. Ofloxacin and levofloxacin must be given
slowly by drip rather than by push to prevent serious
side effects.122 The chelating capability of quinolones
discussed in an earlier section may contribute to
irritation on injection. Injectable quinolones are often
administered at acidic pH levels to help with this
problem as well as to enhance their water solubility.
Tendinitis up to and including tendon rupture has
been seen in a few cases.10,123-125 This side effect is
particularly prominent in juvenile beagle dogs but
appears to be uncommon in humans. It has been
suggested that this might be associated with inhibition of collaginase, leading to interference with
remodeling of collagen-dependent structures. Worries
about this have led to restrictions in the use of
quinolones in sexually active females and in adolescent children. Nonetheless, many courses of treatment have been given to children with severe infections with little apparent resulting tendon damage
so the import of this phenomenon is unclear.
There are no reports of teratogenicity in humans
taking quinolones even in the first trimester of
pregnancy.126 It is always advisable, however, to
administer drugs to pregnant females with caution,
and this caution certainly applies to quinolones. In
particular, high doses may lead to decreased weight
gain of fetuses, so caution is advised.127
6.2. Severe Toxicities Associated with Particular

Quinolones
Temafloxacin was withdrawn a few months after
marketing when severe hemolytic reactions, occasionally including clotting abnormalities and renal
failure, caused significant distress to a small number
of patients, several of whom died. These effects were
not seen in preclinical animal studies or in the
various phases of clinical study before marketing took
place.54
Trovafloxacin postmarketing surveillance revealed
a number of cases of hepatotoxicity, some of which
were severe enough to require liver transplantation,
and some patients died. Because of these problems,
trovafloxacin use is now restricted to those cases
where the potential benefits from its use outweigh
the potential risks.53 The high lipophilicity of trovafloxacin makes the liver one of the organs in which
drug might well accumulate, and this may contribute
to the observed problem.
Grepafloxacin was withdrawn following episodes
of postmarketing cardiotoxicity.128 Prolongation of the
QT interval is observed with most quinolones but
may well be more severe in particular cases. In this
regard, clinicians are cautious with sparfloxacin in
that a small percentage of patients demonstrate this
phenomenon, and a few cases of torsades-de-pointe
syndrome have been associated with its use.129 [Torsades-de-pointe is a French term used to describe a
specific ventricular rhythm/tachycardia most frequently seen in connection with a prolonged QT
interval. This phenomenon is associated mostly with
cardiovascular drugs but occurs with some quinolones in a few cases per million prescriptions.] Moxi-
Mitscher
floxacin use has also led to some reports of QT

interval prolongation, but these appear as yet to be
minor.130
Thus, the trend to enhanced breadth of the spectrum has been accompanied in some cases by severe
toxicities that limit or prevent the use of some of the
newer agents. These were rare enough and unusual
enough that they were not detected in detailed preclinical and clinical studies. Only administration to
a large number of patients following marketing revealed these problems. Several of these agents possess an N-1 2,4-difluorophenyl substituent. Whether
this is coincidental or ominous remains to be established. These instances were seen in the one-a-day
administration days, and it is not clear whether they
could have been avoided or minimized through use
of divided dosing schedules.
The underlying mechanisms leading to these severe toxicities have yet to be disclosed.
The rest of the quinolones have demonstrated
reasonable freedom from such severe side effects.
Millions of doses of quinolones have been administered over a period of several decades, demonstrating
a side effect incidence of 2-10%. Most of these are
discomforting but not severe enough to require discontinuation of administration. No commonality of
toxic response is seen among the idiosyncratically
severely toxic quinolones. Drug developers cannot
help but be nervous in the face of unexpected
problems of this type in a class of drugs previously
regarded as comparatively safe.
6.3. DrugDrug Interactions

Co-administration of quinolones with a number of
other classes of drugs can lead to interferences
complicating their best use.34
The interaction of quinolones with multivalent ionrich substances has been discussed above. Sucralfate
also interferes, and didanosine contains such ions in
its formulation and so can interfere with absorption.67,131
Co-administration of theophylline and drinking
significant amounts of caffeinated beverages (including some popular soft drinks) can lead to unusually
high blood levels of theophylline by competition for
P-450-based metabolic transformation.132,133 P-450catalyzed metabolic transformation is an important
means of metabolism for purines. Quinolones compete for the action of the same enzymes. Quinolones
that undergo significant metabolic transformation by
this means are thus more likely to cause negative
drug-drug interactions. Ciprofloxacin, trovafloxacin,
and enoxacin are such agents to watch.68
7. Molecular Mode of Action

The manner in which quinolones killed bacteria
was unknown when they were first discovered. Some
years later, it was discovered that DNA gyrase, a
previously unknown topoisomerase II essential for
altering the topology of DNA, was an important
target of these agents.39 Subsequent research has
shown that DNA gyrase is commonly the primary
target in most Gram-negative bacteria. Much later,
a second target enzyme was discovered to be DNA

topoisomerase IV.41 Topoisomerase IV is a primary
target for quinolones in certain Gram-positive bacteria, whereas in a number of bacterial species
activity is due to interference with a blend of both of
these enzymes.42 In mycobacteria, on the other hand,
DNA gyrase appears to be essentially the only
target.134 Thus, the question of the specific target
enzyme for a particular microbe is complex. Both
enzymes are vital for bacterial life and are very
widely distributed, readily rationalizing the broad
spectrum of the fluoroquinolones. Inhibition of either
enzyme by quinolones can be lethal to bacteria. The
one that is more sensitive is the primary determinant
of the minimum inhibitory concentration, but a
mutation in either enzyme can be influential. Thus,
a particular resistance mutation can be very or only
slightly important. Further, high-level resistance can
require the acquisition and consolidation of more
than one mutation.
Both of these enzymes are type II topoisomerases
since they temporarily cleave both strands of duplex
DNA and then reseal them following passage of an
uncut portion of the molecule through the temporary
gap. The net result is a change in topology of the DNA
molecule, hence their names. Since most of the
functional features of DNA are topologically dependent, these enzymes are essential for transcription,
translation, repair, and storage processes. The two
enzymes function in rather similar ways and possess
substantial amino acid sequence homology, so it is
believed appropriate to some degree to extrapolate
knowledge from one to the other. Less aptly, the
drug-enzyme-DNA complex bears similarities to
similar eukaryotic complexes of importance to anticancer chemotherapy as with, for example, adriamycin treatment.94 The quinolones of antibacterial
prominence are carefully crafted not to have significant inhibitory power against the human topoisomerase II. Some extrapolations of data can nonetheless be made to human topoisomerase II, but it
must be kept in mind that there are quite significant
drug structures involved.42 Other significant topoisomerases exist. These, however, belong to the
topoisomerse I type and operate rather differently in
that they produce transient single-strand cuts in
double-stranded DNA. Inhibitors of bacterial topoisomerases I and III have not become significant in
antimicrobial chemotherapy as yet. Although named
topoisomerase III, this is a class I topoisomerase.
By way of review, DNA can occur in circular form
in which the ends of the double helices are joined
covalently (Figure 36). For convenience this is more
often shown schematically as an untwisted ribbon or
tube (Figure 37). The natural situation is even more
complex in that the circular DNA molecules can be
twisted further around themselves, rather like one
does with a rubber band, to form a supercoiled
molecule. The further twisting can be done in either
a right-handed or a left-handed manner so as to
produce positively or negatively supercoiled molecules. The action of DNA gyrase is to produce
negative supercoils. Each time the molecules are
twisted over themselves, a node results. The linking
number is the sum of the positive and negative nodes
Figure 36. Circular DNA. Circular DNA is a double

helical molecule. It is most often illustrated, however, as a
ribbon as this is much easier to draw and to understand.
Reprinted with permission from ref 135, p 18, Figure 2.1
(originally in color). Copyright 1993 Oxford University
Press.
Figure 37. Supercoiling of relaxed DNA. The DNA is

depicted in ribbon form. Supercoiling can result in positive
or negative supercoils. DNA is intrinsically coiled in a righthand helix so positive supercoils result in increased torsional stress whereas negative supercoils unwind the
molecule. Each point of crossing over is called a node. If
the upper strand crosses in a clockwise manner, this is
assigned to be negative and vice versa. The linking number
(Lk) is the sum of the positive and negative nodes divided
by 2. Reprinted with permission from ref 136, p 934, Figure
24-17. Copyright 2005 W. H. Freeman and Co.
divided by 2 and is a measure of the degree of

twisting. Excellent treatments of this topic are available for the interested nonspecialist.135-137
Unless further processed, supercoiled DNA would
be tortionally unstable and spontaneously revert to
the relaxed circular state. Topoisomerases II stabilize
supercoiled DNA molecules by cutting both strands,
holding onto the cut ends, passing an uncut double
helical segment through this gap, and resealing the
cut ends. This process can take place once or several
times, thus producing a family of topoisomers differing from one another in the degree of twisting.
The resulting topoisomers differ in compactness
and so can be separated by gel electrophoresis.
Mitscher
Figure 39. Cartoon illustrating cleavage, strand passage,

and resealing of DNA.
Figure 38. Agarose gel electrophoretic examination of the

action of DNA gyrase to produce topoisomers (A, upper gel)
and strand breaks (B, lower gel) in DNA. (A) is reprinted
with permission from ref 135, p 32, Figure 2.9. Copyright
1993 Oxford University Press. (B) is reprinted with permission from ref 138, p 1285, Figure 5, p 1285. Copyright
2004 American Society for Microbiology.
Under normal circumstances isolated supercoiled

DNA of bacteria produces two kinds of electrophoretic
bands. The upper band in the left-hand track of
Figure 38, top, consists of open circular DNA, and
the lower band consists of negatively supercoiled
DNA. When acted upon by a topoisomerase, a series
of topoisomers are produced. These are visible as a
ladder in the right lane of the figure, top, where the
various degrees of supercoiling can be detected.
Visualization of the bands is greatly emphasized by
fluorescence following intercalation with the dye
ethidium bromide.135
When quinolones are present, the transitions are
inhibited in a concentration-dependent manner, providing the basis for an assay. The IC50 value is that
concentration of drug that inhibits supercoiling by
50%.138
The ability to relax supercoiled DNA is also crucial
in allowing transcription. Transcription requires
temporary separation of the duplex strands so that
each can be duplicated faithfully. Since the enzyme
that copies the base sequence moves only in one
direction, the bubble containing the single-strand
segments must move with it. As the bubble goes
forward it meets increasing resistance from supertwisting of the strands ahead and from the
undertwisting of the strands behind as a consequence
of the whole molecule being intertwined and stabilized by extensive hydrogen-bonding networks. Replication of small segments in this manner is much
more efficient than would be involved if each strand
of duplex DNA had to be completely separated into
single strands for transcription. This activity of DNA
is highly conserved, and interference with it serves
as a ready rationale for the broad-spectrum activity
of quinolones.136 Quinolones freeze the bubble, leading to rapid cell death.136 Assay of this effect must
be done in a different manner.
Another central feature of the activity cycle of
topoisomerases is to produce strand cuts in DNA. The
cut molecules can be released by detergent denaturation and protein digestion. The amount of cleavage
is a function of quinolone concentration. The CC50
value is the amount of drug that will trap half of the
maximal amount of linear DNA formed. This is
illustrated in the lower portion of Figure 38. In this
particular example, supercoiled circular DNA was
incubated with the DNA gyrase component proteins
(GyrA and GyrB) from M. tuberculosis without added
ATP but with increasing concentrations of levofloxacin (abbreviated as LVX) added. Without the enzyme,
the left lane shows nicked and supercoiled substrate
DNA. The next lane shows linearized DNA, and the
remaining lanes show increasing amounts of linear
DNA produced at the expense of supercoiled DNA
molecules.
It is believed that release of cut DNA strands
results in lethal consequences.139,140 Those particular
consequences are not well understood at present. It
appears that a lethal protein is biosynthesized when
cut ends are produced. The identity of this putative
protein is as yet unknown. The presumption that
such a cell poison is involved stems largely from the
observation that certain protein biosynthesis inhibitors, such as chloramphenicol, are partially antagonistic to the lethal action of quinolones with some
bacteria.141,142
The cleavage-passing process is illustrated in Figure 39. In the upper view (A) one views the process
from the top. A relaxed circular DNA molecule is
acted upon by DNA gyrase. First, the molecule is
distorted so that the segments overlap, producing a
positive and a negative node. Next, the phenolic OH
moiety of tyrosine 723 attacks the deoxyribose backbone of each of the sessile strands, producing fourbase-pair staggered cut ends with the ends covalently
attached to the phenolic oxygen of the enzyme. These
separate into two short single-stranded regions, and
then an uncut segment is passed through the gate.
The molecule is resealed behind, thereby changing

the linking number and producing two negative
nodes.
An alternate view of the process is presented in the
bottom part of the figure (B), illustrating the process
from the side. First, one sees an intact molecule with
one portion lying in the plane of the paper and
another segment of the same molecule coming out
directly at the reader and lying above the first. These
segments are attached to each other. Next, each
strand of the transverse segment is cut by the
enzyme to produce a gate. This opens with the
energy provided by ATP hydrolysis so that the
forward projecting segment of the molecule can pass
through to the bottom. Closing the gate and resealing
behind complete the catalytic cycle. This can repeat
a number of times, or the ADP can dissociate,
returning the enzyme to its ground state and releasing the DNA segment now possessing an altered
topology. The result is highly negatively twisted
(superhelical) but stable DNA.
Topoisomerase IV can also resolve the catenated
(knotted or intertwined) DNA that is a natural
consequence of replication of circular DNA. The
intertwined catenanes or knots can be converted into
individual closed circular or unknotted DNA molecules by double-strand cleavage, passage, and resealing. Topoisomerase IV therefore must possess
many similarities to DNA gyrase, but it also differs
in its substrate preferences and the outcome of its
action. It must differentiate between the two strands
in catenanes and knots and pass the segments in the
proper order and direction.
Structurally, DNA gyrase is a heterotetramer
composed of two copies each of an A and a B subunit
connected to each other by hinge strands. Unfortunately, there are no X-ray structures of the enzyme
either with or without substrate or inhibitor on board.
The present understanding of its structure is illustrated schematically in Figure 40, which is a
composite construction made from the amino-terminal ATP-binding domain of E. coli topoisomerase II
and the carboxyl-terminal fragment from yeast topoisomerase II.137 The smaller B subunits are on the
top, and the binding site for ATP/ADP resides there
as indicated. The A subunits are encoded by a gyrA
gene and the B subunits by a gyrB gene. The
combined A units are responsible for the breakage
and reunion functions and the B subunits for providing the ATPase activity that generates the energy
for the substantial cyclic movements that the enzyme
undergoes in functioning. Apparently, binding of ATP
to the B subunits is the signaling event that starts
the cycle of events resulting in topoisomeric change.
The negative supercoiling activity of DNA gyrase is
balanced by the relaxing action of topoisomerases I
and topoisomerases IV. In some bacteria DNA gyrase
apparently also performs some of the decatenations
that are normally the function of topoisomerases IV.
In mycobacteria, treponema, and helicobacter, for
example, topoisomerase IV is apparently entirely
lacking.86,134,143 The degree of dependence of decatenation on types of DNA gyrase is not yet entirely
clear but obviously is relevant to antimicrobial action.
Figure 40. A schematic view of DNA gyrase. Reprinted

with permission from ref 137, p 78, Figure 27.23. Copyright
2002 W. H. Freeman and Co.
The A subunits of DNA gyrase are 97 kDa molecules (in E. coli) containing the tyrosine moiety
whose phenolic OH group is the nucleophile that
cleaves the phosphodiester bonds of DNA and covalently holds the cut ends. Another region contains
those amino acids whose mutation is most influential in leading to quinolone resistance. The latter is
called the quinolone resistance-determining region
(QRDR).144 The N-terminal two-thirds of the A subunits is where the cleavage-resealing activity is found
whereas the C-terminal one-third is where the DNA
is wrapped around the enzyme.145
The B subunits are slightly smaller (90 kDa)
molecules that contain an ATP-binding site and
ATPase catalytic activity. The ATPase activity is
found in the N-terminal half of the B subunits,146
while the C-terminal portion is involved in DNA
binding and attachment to the A subunits.147-150
A variety of molecules bind to the B units and
interfere with the function of DNA gyrase. The B
region is sensitive to inhibition by the coumermycin
and cyclothialidine classes of antimicrobials following
binding to the ATP site.151,152 Novobiocin is best
known of the coumermycins and was once marketed
for antibiotic use. It interferes with the energy
transduction required for the substantial molecular
movements involved during the enzymatic action of
these key enzymes.153 Unfortunately, the coumermycins are not well tolerated in humans, and in addition, resistance develops readily. Cyclothialidine
Figure 41. A cartoon view of the catalytic cycle of DNA

gyrase cleavage of DNA. The DNA molecule (drawn as a
tube) wraps around DNA gyrase such that the crossing
segment of DNA enters the upper chamber consisting of
the two B segments of the enzyme. In the next step the
enzyme binds two molecules of ATP, closes the opening,
and cuts the transverse segment of DNA to produce a gate
through which the molecule can pass into the lower
chamber consisting of the two A segments of the enzyme.
The cut in the DNA molecule is resealed behind. Finally,
the molecule is ejected from the lower chamber, which
opens to allow this. The stage of the cycle that is sensitive
to quinolone binding is yet to be completely characterized.
Reprinted with permission from Corbett, K. D.; Schultzaberger, R. K.; Berger, J. M. The C-terminal domain of
DNA gyrase A adopts a DNA-bending beta-pinwheel fold.
Proc. Natl. Acad. Sci. U.S.A. 2004 101 (19), pp 7293-7298
Copyright 2004 National Academy of Sciences.
inhibits DNA gyrase efficiently in cell-free systems,

but the substance penetrates poorly into intact cells.
Creative analoguing has partially overcome this,
leading to molecules that possess a better hydrophilic/lipophilic balance.154 The clinical future of the
improved agents is not yet clear. Microcin B-17 is a
peptide-based natural product that binds to the B
region at a site apparently containing Trp751, and
this inhibits DNA gyrase.155 The microcin B-17 story
is in its infancy.155,156 Screening of a chemical library
for DNA gyrase inhibitors using an assay dependent
upon production of bacterial anucleate cells due to
failure of partitioning led to the discovery of pyrazole
analogues. Analoguing produced a substance with
significant inhibitory potency against DNA gyrase
and DNA topoisomerase IV that, at the same time,
did not significantly inhibit DNA topoisomerase
II.157-159
It will be interesting to see how many B-unitbinding agents appear in the future and whether any
of these will survive into clinical utilization. These
various agents are outside the scope of this review
and will not be treated further herein.
The various interactions postulated between DNA
and DNA gyrase in the catalytic cycle are illustrated
in Figure 41. The DNA molecule wraps around the
enzyme more or less at the AA/BB interface such that
the segment to be cut is in the plane of the paper
and the segment to be passed enters the BB chamber,
which opens to allow this and then closes to prevent
premature escape. Next, the portion of the DNA
molecule in the plane of the paper is cut to produce
a gate. The enzyme holds on to the cut ends to keep
the process under control. The enzyme then separates
the cut strands, and the uncut strand of the molecule
is passed through this gate into the lower chamber
(BB). The cut strand is resealed. Expulsion of the
uncut strand from the lower chamber and disassembly completes the catalytic cycle. Drug binding
resulting in freezing of the tertiary complex takes
Mitscher
place somewhere at or between the second or third

phases of the cycle, producing either a cleavable
complex or a cut segment in which the DNA becomes
trapped, neither able to progress nor to revert.118 The
ternary cleavable complex becomes or induces a
cellular poison so that the bacterial cell rapidly
dies.160 An important bit of evidence leading to this
conclusion is that the MIC values of quinolones are
frequently much lower than the IC50 or CC50 values.
It appears possible that the permanent gaps in the
DNA strands induce the biosynthesis of exonucleases
as repair enzymes, leading to poorly controlled repair
processes. Often this results in apoptosis. Interference with movement of replication bubbles has been
mentioned earlier. It is posited that an as yet
unidentified toxic protein is released or formed as
part of the interference by quinolones with these
processes. This rationalizes the finding that some
protein biosynthesis is essential for quinolone toxicity
as chloramphenicol, for example, can decrease sensitivity of bacteria to quinolones when employed
simultaneously.141 Much present speculation surrounds this issue, but the details are sketchy at best.
Although the primary sequence and X-ray structures are known for fragments of the gyrase A and
B subunits, no molecular level picture of the complete
enzyme or of the ternary complex is available.153,161-163
Indeed, given the various movements required for the
functioning of the enzyme and its substrate, it is not
clear that success in getting an X-ray picture will
clarify the precise mechanism of action. One suspects
that a single freeze-frame picture might well be
misleading and that a sequence of pictures might be
required. Thus, the present picture requires inferences from the structures of the agents and data from
resistance studies, and understanding lies still at the
cartoon level.
The quinolone-binding site on the A subunits is not
clearly identified, but amino acids 67-106 are referred to as the QRDR (E. coli numbering), with
major resistance mutations occurring at S83W, D87A,
D87G, D87H, D87N, and D87Y. Interestingly, the
serine at position 83 and the aspartate at position
87 are both polar, and the resistance mutations
mostly involve replacement by hydrophobic amino
acids. It is thought by many that the C-7 amino
substituents of quinolones interact with gyrase in the
ternary complex binding pocket. Replacement with
nonpolar side chains would lessen this interaction
considerably, and gyrases with hydrophobic amino
acids at this position are much less sensitive to
quinolones.
Other changes occur in the same general area but
are less significant from the standpoint of resistance
development. Interestingly, as might be expected
from susceptibility studies with intact bacteria, quinolones bearing halogens or alkoxy groups at C-8 are
less interfered with by these mutations.
Humans must also be able to alter the shape of
their DNA in a controlled manner, and the related
enzyme that carries out this function has considerable homology to the prokaryotic enzyme. This enzyme is called human topoisomerase II. Quinolones
bind much less avidly to the human counterpart
enzyme (100-1000 less) and so do not cause DNA

poisoning to humans at doses normally achievable.164
For example, ciprofloxacin inhibits DNA gyrase at
about 0.3 M, but it requires a concentration in
excess of 300 M to inhibit mammalian topoisomerase II.165,166 Etoposide inhibits mammalian
topoisomerase II at about 0.81 mM concentration, but
it takes more than 850 mmol to inhibit DNA gyrase.165,166 This explains not only the selective toxicity
of these agents but also how a drug that damages
DNA could escape being significantly genotoxic or
transforming to humans at normal doses. In this
regard it is important in passing to consider that
potent inhibition of mammalian topoisomerase II by
anthracyclines and etoposides, e.g., provides useful
antitumor chemotherapy.167 These drugs are too toxic
to be used as antibiotics.
Bacterial topoisomerase IV41,168 has functions that
partly duplicate functions of DNA gyrase and vice
versa. It is also composed of two sets of pairs of
subunits. parC genetically encodes the two ParC
subunits, and the parE gene encodes the two ParE
subunits. The C subunits of topoisomerase IV in E.
coli are about 36% homologous to the A units of DNA
gyrase. The E subunits are about 40% homologous
to the B units of DNA gyrase. As noted previously,
this enzyme catalyzes decatenation without wrapping
a segment around itself, so it produces relaxed DNA.
The intimate details of how this is done are still being
worked out.41
Mammalian topoisomerase I is sensitive to camptothecin analogues, and its inhibition plays a significant role in cancer chemotherapy.165,169 Mammalian
topoisomerase III is as yet not involved in chemotherapy.170
Antibiotic activity concentrations of a wide variety
of quinolones parallel those that inhibit the enzymes.42,119 Interestingly, quinolones have little affinity for purified DNA gyrase alone.171 They do have
affinity for DNA, particularly for single-stranded
DNA, but this affinity is not readily saturated at
meaningful concentrations.118 On the other hand,
when DNA gyrase and double-stranded DNA are
both present, a specific binding site for quinolones is
created that is saturable at antimicrobially relevant
concentrations.172 The binding is reversible and cooperative and saturates with up to four quinolone
molecules.118 It is also sensitive to the absolute
configuration of the quinolones.119 The cooperativity
is seen from the kinetics, which show the first
molecule to enter the active site fairly sluggishly, the
second more readily (perhaps as a consequence of a
compatible distortion of the active site, or perhaps
the first molecule provides additional binding interactions for the second), and a third and fourth
molecule to add sequentially to complete the saturation of the enzyme-induced binding pocket.118 Synthetic covalently bound model norfloxacin dimers
linked between the pyridine nitrogen atoms by methylene spacer groups bind well but fail to show an
anticipated entropic advantage, providing evidence
not only for the size and character of the binding
pocket but strongly in agreement with the cooperativity phenomena. Interestingly, a four-methylene
linker is optimal for inhibition of E. coli DNA gyrase

activity, whereas five are best for the enzyme from
Micrococcus luteus. This suggests that the quinolonebinding pocket is larger in Gram-positive microorganisms than in Gram-negatives. These tethered
probe molecules have fewer degrees of freedom than
the monomers have. They are also inactive in whole
cells presumably due to poor pharmacokinetic properties.118 More recently, a series of dimers of ciprofloxacin linked between the distal nitrogens of their
C-7 piperazinyl moieties also failed to show an
entropic advantage in potency over the monomers.
These compounds were, however, surprisingly active
in whole cell cultures of S. pneumoniae, approximating the MIC values for ciprofloxacin alone.173
The quinolones can also bind to supercoiled DNA
in the absence of the enzyme. This is believed to be
a consequence of the presence of small segments of
single-stranded regions created by the supercoiling.88
This probably is unimportant for their mode of action.
The possibility that enhanced drug binding was due
to the conversion of relaxed substrate DNA to the
supercoiled form was ruled out by the use of a
nonhydrolyzable triphosphate nucleotide in the binding mixture.174
The keto-carboxyl system at C-3 and C-4 is
proposed to hydrogen bond strongly to the singlestranded sides of the melt zone induced into the DNA
by the gyrase. This would dictate the primary orientation of the first drug molecule to enter the complex.
The bonding energy lost when the water associated
with the ketoacid moiety is displaced would be
regained upon hydrogen bonding to the DNA single
strands. The enzyme is also nearby, and genetic
studies with resistant mutants indicate that an
aspartate is important. This probably interacts electrostatically with the protonated distal nitrogen of
the piperazine ring or the distal nitrogen attached
as a side chain to a C-7 pendant pyrrolidinyl ring.
This would provide directionality to the drug binding.
The second quinolone molecule would align itself
such that the keto-carboxyl group interacts with the
other single strand of DNA in the melt bubble. The
carboxyl group and piperazinyl groups would line
themselves up with the first molecule acid to base
and base to acid. The interior regions of the two
quinolone molecules thus brought together would be
lipophilic and compact, providing a grease-grease
surface associating the two molecules. This requires
the remainder of the water present in the pocket to
be squeezed out as hydrogen bonding is much stronger than lipophilic bonding. This dimer is oriented
in reverse to the bases in double-stranded DNA in
that the hydrogen-bonding surfaces are on the outside and the lipophilic portions are on the inside. This
is perfectly suited to exploit the melt zone. Two more
molecules of quinolone can assemble themselves to
the original pair, forming a cube of four. This apparently saturates the binding pocket. This view would
rationalize satisfactorily the functional roles of the
known structure-activity features of the quinolones.
The experiments of Critchlow and Maxwell176 later
demonstrate a 2:1 ratio of quinolones in the complex.
They were cautious, however, to point out that their
methods of measurement were sufficiently vigorous

that they could have disrupted weaker interactions
in the complex so that a 4:1 ratio might still be
possible.
These and a number of other considerations have
been rationalized in a well-received model discussed
in the next section along with various other models.118,119
7.1. Models of the Ternary Complex

Obviously, the particular value of hypothetical
bioactivity models is their utility in rationalizing a
mass of observations in structural terms. Furthermore, the best models not only posit a credible role
for the functional contributions of individual portions
of quinolone molecules but also can provide the basis
for prediction of the potential contribution of new
chemical features in advance of synthesis.
The first and most successful of the models of the
ternary complex was advanced by Shen.118 It has
been modified over the years in light of further
information but remains substantially intact. Its
initial expression was based upon a series of important observations. Quinolones do not bind to DNA
gyrase at their inhibitory concentration although
they do bind to various types of DNA to various
degrees. They bind poorly to relaxed double-stranded
DNA. This binding is weak and nonsaturable. They
bind preferentially to single-stranded DNA but in a
nonspecific and a noncooperative manner. This form
of binding correlates with hydrogen-bonding capability. Indeed, homopolymer work showed that there
was a preference of quinolones for poly(dG), which
has more hydrogen-bonding functionality. Ring stacking could not be demonstrated and, indeed, appeared
unlikely to play a dominant role as the intensity of
quinolone binding did not correlate with the type of
DNA base present.174,175 They bind specifically to a
saturable site on supercoiled DNA in a highly cooperative manner at a concentration near their Ki
value. DNA gyrase enhances quinolone binding to
relaxed forms of DNA.171 The cooperative binding
observed saturated at about four molecules of quinolone per DNA site.174 These findings indicated a
significant role for the substrate DNA but did not
explicate the specific interactions with the enzyme.
It was, however, asserted that bound DNA gyrase
induces a binding site for the drug in the relaxed
DNA substrate and that this required the action of
ATP.118 Specificity apparently arose through the
specific action of this class of enzyme on its substrate.
This satisfactorily rationalized the general lack of
mutagenicity of these agents.
In its initial iteration, the model suggested that the
strong preference for single-stranded DNA regions
was satisfied when the enzyme cut the two DNA
strands to create short (four-residue) regions that
separated to allow passage of an uncut region and
passage behind. This model is illustrated in Figure
42. The upper portion of the figure illustrates in
cartoon form the DNA bubble attached to DNA
gyrase, saturation of the binding site by four quinolone molecules, and the putative cleavage sites.
The lower portion of the figure shows four head-to-
Mitscher
tail stacked norfloxacin molecules hydrogen bonding

to the GCA and T bases of a DNA segment. Note that
a preference for guanine would occur because of the
presence of two hydrogen-bond-donating NH groups
whereas CAT bases have only one such group. Selfassociation of the hydrogen-bonded quinolone molecules is envisioned as involving stacking in the
vertical dimension and hydrophobic interactions
involving the portions of the drugs opposite those
hydrogen-bonded to the edges of the DNA bubble.
Later work showed that inhibition of the catalytic
cycle by topoisomerases could occur without strand
cleavage.176 The most convincing demonstration of
this was made when the tyrosine residue in the active
site was exchanged for a phenylalanine residue. This,
of course, removed the essential nucleophilic phenolic
hydroxyl moiety from the enzyme, and yet quinolone
binding still could be measured. To accommodate
these data, the model was altered once again in that
the binding was attributed to a melt bubble being
formed. To visualize this, the reader can examine
Figure 42 again but imagine that the cut ends of the
DNA strands are still joined and therefore continuous. Localized separation of the strands by the
enzyme exposes single-stranded regions to which the
quinolones can bind in the manner previously postulated. Since two single-stranded regions are available in both the gate and the bubble forms, efficiency
and enhanced power would result from bonding to
each of them. The initial form of the model posited
four quinolone molecules binding to each binding site.
Later work suggested that two might be sufficient.
This disparity may be attributed to different experimental conditions being used. Association of two
molecules instead of four does not alter the model in
any fundamental way. In either case, the quinolone
molecules would have their polar edges available on
the outside of their aggregate and the internal
nonpolar regions would stabilize the complex by
lipophilic association. This is consistent with the
structure-activity relationships known to date, and
the original iteration of the model remains essentially
intact. The showing that cleavage was not a necessary precondition for formation of a ternary complex
does not rule out complexation with both the bubble
and the gate stages and does not indicate which of
these might be the most powerful or influential.
The binding cooperativity could be rationalized by
positing that the first quinolone molecule displaced
water present in the opening single-stranded region
and provided a template for the next quinolone
molecule as the gap produced by the enzyme widened. The enzyme-induced binding site would saturate with drug and freeze the complex so that it could
neither advance nor recede.
Later work suggested that the role of magnesium
ion had not been given sufficient prominence.177-179
It is not immediately clear whether the magnesium
ion influence is due to its well-known stabilizing
effect on DNA topology or to its equally well-known
ability to chelate with the keto and acid moieties of
quinolones or to both of these phenomena. If magnesium binding to quinolones is directly involved,
then instead of hydrogen bonding directly to DNA
Figure 42. Illustrations of the putative quinolone binding mode to DNA gyrase. The upper portion depicts the binding of
four molecules of quinolone (rectangles) to a cleaved bubble in DNA attached to DNA gyrase. The gyrase resembles a
Viking helmet. The lower portion depticts a computer-drawn cluster of four quinolone molecules hydrogen-bound to the
inner edges of a DNA bubble. Reprinted from ref 118. Copyright 1989 American Chemical Society.
bases, a chelated magnesium ion could serve as an

adapter. In Figure 42 this would place a magnesium
ion between each quinolone and its hydrogen-bonding
DNA base partner. This would require a bigger
bubble but is not precluded. The putative magnesium
ion effect is less attractive structurally as the binding
could alternatively be associated with linkage of
chelated quinolone to the phosphate backbone of
DNA instead of the bases. A preference for singlestranded regions is not readily accommodated by this
idea.
Despite its evolution as newer data become available, the model has served very well in rationalizing
structure-activity relationships in the quinolone
field. No other model has yet been brought forward
that has successfully supplanted it.
A particularly attractive feature of the modified

Shen model is its ability to rationalize the role of the
N-1 substituent (the C-3 position of the oxazine ring
of ofloxacin) and, most particularly, when it is chiral.
It is the only model that explicitly rationalizes this.
The Morrissey et al. modification of the original Shen
model has this as a central feature.119 The observation that (S)-ofloxacin (levofloxacin) binds more firmly
to the enzyme-induced pocket than its enantiomer
indicates that the methyl groups exploit the binding
site differently depending on their orientation into
three-dimensional space. Clearly, since racemic ofloxacin binds roughly half as well into the pocket as
levofloxacin, a methyl group with the wrong (R)
orientation fails to take complete advantage of the
pocket. This is confirmed by noting that (R)-ofloxacin
binds approximately 12-fold less intensely to the

enzyme-substrate complex. Furthermore, (R)-ofloxacin saturates the binding bubble when only two
molecules are added using conditions wherein (S)ofloxacin saturates it with four molecules. Thus, in
a competition, the S-enantiomer will clearly win out,
and the binding model proposed satisfactorily rationalizes these various observations. The relative rates
and final amounts of accumulation of these agents
into E. coli are closely similar, removing differential
uptake as an alternate explanation. Additional supportive data include the finding that des-C-methylofloxacin binds less well than ofloxacin itself and the
gem-dimethyl analogue also binds less well. When
the C-methyl group of ofloxacin is exchanged for an
olefinic linkage, once again binding decreases. These
data are most consistent with a preferred and axial
orientation of the methyl group in levofloxacin being
optimal for the enzyme. The interested reader should
refer back to Figure 32 for a structural view of this.
The Morrissey refinement of the Shen model is also
compatible with many other features of the antibacterial quinolones. For example, it successfully rationalizes the utility of the N-methyl compounds of the
first series. A methyl group would be too small to take
advantage of the positive binding features of the
methyl group of levofloxacin but could fit as well as
the less active des-C-methyl analogue. Possession of
an ethyl group at N-1 would allow, in one of the
rotamers, the distal methyl group to satisfy the needs
of the enzyme expressed in its preference for an axial,
, chiral methyl. It, not being rigid, however, would
not be as effective. The finding that bigger, branched
N-alkylquinolones are deleterious and polar substituents at N-1 are unsatisfactory would agree as the
pocket exploited by a suitable methyl could be posited
to be limited in size and to be hydrophobic in nature.
N-Cyclopropylquinolones are also compatible with
this hypothesis as are the N-vinyl- and N-arylsubstituted quinolones since X-ray studies indicate
that the aryl groups are about 30 orthogonal to the
ring system itself. The otherwise surprising antimicrobial activity of the N-tert-butylquinolones can also
be rationalized with this model. Rotation provides a
compatible methyl group every 120. This substance
could be viewed as being related to the gem-dimethyl
analogue of ofloxacin.180 The latter would be a
molecule possessing simultaneously the virtues of the
-methyl group of levofloxacin and the defects of the
R-methyl group of its enantiomer. These concepts are
captured in Figure 34.
One of the unusual features of these drugs is that
they bind to a pocket that is not present in the ground
state of the substrate DNA. Rather the binding site
is created by action of the enzyme. This satisfactorily
rationalizes their lack of significant genotoxicity
despite DNA being their target. They interfere with
the functions of DNA, not with the nature of its bases.
Another unusual feature is the number of drug
molecules required to complete inhibition of DNA
gyrase and the cooperative nature of this interference.118 The much more common arrangement, wellknown to medicinal chemists, is one molecule to a
single active site in the molecular target. This
satisfactorily rationalizes the puzzle that such a
Mitscher
small molecule (taken as a monomer) could have such

a profound effect on two such large targets.
Covalent tethering of two molecules that linked
quinolones at the N-1 position has provided significant support for at least the dimeric interaction. Four
methylenes proved optimal against the Gram-negative enzyme, and this coincided closely with the
relationship and orientation of neighboring nalidixic
acid molecules in the unit cell of its crystals.118
Tethering with a smaller or larger number of methylenes proved significantly less active. Interestingly,
five methylenes proved optimal against the Grampositive enzyme, providing evidence that homology
of the active site of different topoisomerases is not
absolute and suggesting that the difficulty in finding
quinolones with equal potency against Gram-positives and Gram-negatives may involve features additional to the comparative targeting of DNA gyrase
and topoisomerase IV.
The initial cartoons illustrating the complex were
silent about the specific role of the enzyme. This could
have been interpreted that the role of the enzyme was
to produce the binding site. Subsequently, it has come
to be believed that the C-7 amino function would
provide an orienting binding interaction also with the
enzyme.181-183 This would provide directionality to
the bound drug. Conveniently, this orientation is the
same as that chosen empirically for drawing the
structures on paper. The Morrissey modification of
the Shen model includes specific interactions between
quinolones and DNA gyrase as well as with DNA.
Thus, the particular value of the evolved version
of the Shen model is that it imputes not only a role
for each of the chemical features of the quinolones
but also for each of the components interacting in the
ternary complex. It is not only consistent with the
majority of experimental evidence but also predictive
about the possible role that novel quinolone structural features might play in the future.
In an alternative model advanced by Nakamura as
illustrated by Heddle and Maxwell (see Figure 43),
the interaction between the quinolones and the DNA
bases involves stacking and confirms that singlestranded regions are important.179,184,185 In this view,
the first step is intercalation in the bubble region
created by the enzyme. The bases in the bubble swing
outward next and are replaced by intercalating
quinolone molecules. The quinolone molecules would
be widely separated from each other and would not
self-associate. There is no literature support for
classical intercalation.171,177 If this occurs, it must be
transient. The model also pictures a binding site for
a single quinolone in a pocket near the juncture of
the GyrA and GyrB units of DNA gyrase. In this
model binding is enhanced by interactions with
certain amino acids of the enzyme and the DNA
phosphate backbone phosphates.182,185
Noble and Llorento advance a model that agrees
with many features of that of Nakamura.179,183 Significant features illustrated in Figure 44 include
interspersion of bound magnesium ion between a
quinolone molecule and DNA bases and phosphate
as well as intercalation/stacking and hydrogen bonding between serine 83 OH and the C-6 fluorine.
Figure 43. Nakamura model as illustrated by Heddle and

Maxwell. (A(i)) and (A(ii)) illustrate intercalation of quinolones in the bubble region of DNA followed by outward
rotation of the DNA bases, which are then replaced by the
quinolone molecules. (B) illustrates the putative quinolonebinding pocket at or near the junction of the GyrA and
GyrB interface. The incorrect double bond position in the
original of Figure 43B is trivial and has not been altered.
Reprinted with permission from ref 182, p 1813, Figure 8.
Copyright 2002 American Society for Microbiology.
Figure 44. Llorente model. Reprinted with permission

from ref 183. Copyright 1996 Elsevier.
Subsequent finding of intense activity possessed by

some nonfluorinated quinolones weakens the last
point.
Intercalation is also an important element of the

hypotheses of Hurleys group for the inhibition of
mammalian topoisomerase II by the quinobenzoxazines.186-188 The problem with the Hurley model
for this purpose is that the intercalating quinolones
involved in his work are cell poisons and not selective
antimicrobials. The antimicrobial quinolones that are
invoked as participating stack on the outside of the
DNA molecules. The Hurley model is more likely to
be applicable to the antitumor effects of certain fused
tricyclic quinolones than to the very different phenomenology associated with the antibacterial effects
of antibacterial chemotherapy.
Recently, Tuma proposed that quinolones form a
molecular cap, stabilizing the DNA duplex by forming
a covalent bond with the cut DNA C-5 segments.189
Unfortunately, for this view, there is no evidence for
a covalent bond between quinolones and DNA with
or without the intermediacy of DNA gyrase or the
presence of cut ends.
It is important to bear in mind that topoisomerase
IV has not been studied in the same detail and that
the Shen model may not apply as well to it. The
homology is only 40% between the two enzymes, and
they function somewhat differently. Also it has been
shown that the binding pocket of topoisomerase IV
is apparently larger than that of DNA gyrase.118
Furthermore, the lack of parallelism in structureactivity and potency between these enzymes and
human topoisomerase II makes it dubious whether
these considerations apply to the human enzyme
either. In particular, there is strong evidence that
intercalation is a significant feature of several drugs
inhibiting the human enzyme, and there is no evidence that intercalation is involved in inhibition of
the bacterial enzymes.
8. Assay Methods
To recapitulate briefly, two different sequillae are
measurable following tertiary complex formation as
illustrated in Figure 38 above. In the catalytic assay
(a, top), one measures the relative amounts of negatively supercoiled DNA and equilibrating relaxed
DNA topoisomers. In the absence of drug, bacterial
DNA normally equilibrates between closed negatively
supercoiled DNA and open nicked circular DNA.
Increasing concentrations of drug lead to progressive
conversion of negatively supercoiled DNA to topoisomers as seen by gel electrophoresis. The IC50
values are determined by measuring the concentration of drug required to inhibit DNA gyrase catalyzed
relaxation to topoisomers by 50%.138,190,191 A relaxation assay can also be used to measure inhibition
of topoisomerase IV.190 In the cleavage assay (b,
bottom), supercoiled DNA in the absence of ATP is
incubated in the presence of increasing amounts of
drug (in this example, levofloxacin). The mixture is
denatured with detergent and treated with a proteinase followed by electrophoresis to separate the
various states of DNA, and these are quantitated
following ethidium bromide staining. One notes the
concentration-dependent decrease in topoisomers and
the progressive appearance of linear DNA. The CC50
value is the concentration of drug that achieves 50%
of the maximal amount of cleavage as compared to

standard norfloxacin.192 For most quinolones, the IC50
and CC50 values correlate reasonably well.193 The
cleavage assay is easier to run and is less sensitive
to assay conditions than the catalytic assay.
9. Molecular Modes of Resistance

The optimistic early days of chemotherapy when
identification of a pathogen was often sufficient for
making a suitable choice of drug have lamentably
passed away. Widespread resistance now requires
much more diagnostic work if rational chemotherapy
is to be instituted. Compared to those of many other
classes of anti-infective agents, the resistance levels
to quinolones are as yet relatively low but are steadily
increasing. This phenomenon is attributed to a
variety of causes.194 Primary, of course, is the recognition that this is a natural evolutionary response
mediated by bacterial fecundity and genetic versatility. This war between man and microbes will be
punctuated by episodes in which one side or the other
gains a temporary advantage but neither will ever
achieve a total victory. Todays challenge is to find
means to keep the equilibrium point as much in
humanitys favor as is technically possible.
Resistance levels in developing countries are notably higher than those in advanced nations. One
wonders if this is significantly affected by a greater
dependence on older, cheaper, less potent quinolones
such as nalidixic acid, to which resistance develops
more easily. Certainly, even in advanced nations one
can point a finger toward widespread use of quinolones for agricultural purposes as contributing to the
resistance problem.195,196 To the extent that one buys
into any of these propositions, what is to be done?
One could ban the use of older, less effective quinolones that are more prone to resistance, but this is
probably not economically feasible. Easier to effect
would be withdrawal of quinolones from agricultural
uses that are not essential. Historically, this has not
proven easy to accomplish with other antimicrobials
either.
Significant clinical resistance was first encountered
in microorganisms that were not very sensitive to
quinolones and those at the same time requiring only
a single mutational step to engender high levels of
insensitivity. Such bacteria are S. aureus (especially
methicillin-resistant strains) and P. aeruginosa. For
most bacteria, levels of resistance that are troublesome in the clinic often involve more than one
mutational step. After a primary resistance step,
additional mutations occur that solidify and intensify
the level of resistance to quinolones.
Two main resistance mechanisms are clearly established. The most common and influential of these
is decreased cellular uptake and/or active expulsion
of quinolones, thereby denying them access to their
cellular targets. The second involves mutations in the
target enzymes, leading to decreased quinolone binding.197 A third and as yet minor resistance mode
involves a plasmid-mediated mechanism about which
comparatively little is known at present. Quinoloneresistant bacteria carrying plasmids are comparatively rare as yet but spreading rapidly. An interest-
Mitscher
ing feature of the bacteria carrying this plasmid is

that cells carrying it often also are resistant to
extended-spectrum -lactam antibiotics as well. This
makes this development potentially worrisome. The
gene product for this means of resistance is thought
to be a gyrase-protecting protein of unknown present
clinical relevance.198,199
The molecular mode of action of quinolones and
resistance to them are intimately interconnected, so
resistance has been touched upon briefly already in
several places in this review.
9.1. Uptake Inhibition

In wild Gram-negative strains access of quinolones
to bacterial cells is mediated through passive uptake
that can involve, in addition, comparatively nondiscriminating porin passage through the outer membrane.69 The classical hydrophobic quinolones such
as nalidixic acid are primarily taken up by passive
transport. The newer hydrophilic quinolones are
taken up partly through the water-lined porins and
partly by passive uptake through the membranes.
The peptidoglycan itself is thought to present almost
no passage difficulties to quinolones. Passive membrane passage is believed to involve the percentage
of a quinolone that is not ionized at physiological
conditions (see Figure 11). In some bacteria, mutations leading to a decrease in the number of porins
has been noted.200 In these cases, penetration of
hydrophilic quinolones decreases by about half, suggesting something of the degree to which these two
phenomena are involved in cellular uptake.
In other cases it has been demonstrated that some
of the ubiquitous ATP-coupled export proteins function to expel quinolones. At least 11 quinolone
expulsion pumps have been identified to date. These
are not quinolone specific but play other roles as
well.201 As yet, there is no definitive information
about the normal materials with which these exporters busy themselves.
Clearly reduced availability of porins coupled with
derepression of energy-requiring substrate pleotropic
efflux proteins can lead to significantly decreased
quantities of quinolones reaching the target enzymes
in the cytoplasm. Overexpression of these expulsion
proteins has been reported in a number of bacterial
strains.17 The efflux proteins span the inner membrane, periplasm, and outer membrane. When these
pumps are also modified by mutation, it can be seen
that multiple interlocking modes of resistance to
quinolones can occur.
From the standpoint of effecting chemical stratagems for overcoming this mechanism, it is possible
that providing a more avid substrate for the pumps
would keep them busy and allow the quinolones to
leak through into the cellular interior. This will be
complicated by the variety of such pumps and present
lack of knowledge of what the normal function of
these pumps and the identity of their substrates is.
Work along these lines has begun but has yet to reach
clinical significance.202 Off the shelf chemicals that
interfere with the nor A pump of Gram-positives
include reserpine199 and verapamil.144,203 These, of
course, are useless for practical purposes because
they would exert significant pharmacological effects

on the patient when given in the quantities required.
It is also important to recognize that for this approach to work at all the pharmacokinetic characteristics of the two companion drugs must closely
match. This is not easily accomplished.
9.2. Plasmid-Mediated Resistance

Recently, a plasmid has been isolated from a
resistant clinical strain that encodes for a gyraseprotecting protein.190,204,205 Very little is at present
known about the molecular details of its actions.
9.3. Enzymatic Alteration of Quinolone Structures

No enzymes have yet been identified that are
involved in quinolone structure alterations. This
mode of resistance is common among other antimicrobial classes but is not seen here.
9.4. Mutations of DNA Gyrase

Mutations that render the DNA gyrase and/or
topoisomerase IV less sensitive to their action are
common. Most commonly, chemists believe logically
that the more sensitive of the two enzymes controls
bacterial response, but there are suggestions that this
is not the whole story. One complicating feature, for
example, is that the two enzymes are not present in
equivalent amounts in all species.206 It is well established that the principal target of the quinolones in
Gram-negative microorganisms such as E. coli is
usually DNA gyrase, with topoisomerase IV playing
a lesser but enhancing role. Resistance involves
mutations in both genes. Mutations in the A unit
alone are sufficient to cause considerable resistance,
and further mutations in the B unit amplify this
effect.197 One notes, however, that there are mutations in other regions of the B subunit that are
independent of GyrA.181,207,208 Clearly, mutations in
either subunit that affect the supercoiling process or
alter the shape of the drug-binding pocket will
influence potency and could be involved in quinolone
resistance. The region in subunit A stretching from
amino acids 67-106 (E. coli numbering) constitutes
the QRDR, and serine 83 and aspartate 87 are the
most influential residues although a number of
amino acid exchanges occur elsewhere in clinical
strains.163,209-212 The QRDR lies comparatively near
the catalytic tyrosine residue 122.
A very useful summary of the specific resistanceassociated mutations involving amino acid changes
in quinolone target enzymes has been published
recently by Hooper.5
The bulk of the resistance-associated mutations
occurring in the GyrA subunit of DNA gyrase in
Gram-negatives occur between amino acids 51 and
119. Particularly prominent among these are mutations involving amino acids 83 (normally serine or
threonine) and 87 (normally aspartic acid). The
changes are many in that 83 becomes leucine, tryptophane, alanine, isoleucine, phenylalanine, or even
argenine, depending upon the species involved. Amino
acid 87 likewise undergoes many possible substitutions including exchange for asparagine, valine,
glycine, tyrosine, alanine, histidine, valine, lysine, or

histidine. By and large, but not always, these changes
involve substitution of a nonpolar residue for a polar
residue. Whether these substitutions affect more
prominently the shape of the binding site or its ability
to interact with quinolones is not yet clear. This
information would, of course, be invaluable in assisting future drug design. It is interesting to note that
the ability of these mutations to interfere with
quinolone binding is strongly influenced by quinolone
structure. This is encouraging from the standpoint
of drug design. It is obvious that alterations in amino
acid 122, the tyrosine residue that cleaves the
phosphate backbone to create an enzyme gate for
strand passage, is not involved in these resistance
mutations. If it were, it would simultaneously inactivate the enzyme and result in cell death.
An analogous region is found in the B subparticles.
Among Gram-negative microorganisms, mutations to
resistance in GyrB stretch between amino acids 406
and 495. These do not seem to be associated especially with specific amino acid residues but are singly
distributed in this region. This region is posited to
be near the quinolone-binding portion of the A
subunits and influential regarding the binding of
substrate DNA to the enzyme.179
Mutations in the B subunit of DNA gyrase are less
important unless the A unit has been previously
altered. Some structural inferences come from study
of the effect of drugs on the B subunit. In this regard,
the presence of a basic nitrogen in the form of a C-7
piperazine moiety is much more important than is
seen with drugs such as nalidixic acid that has a
neutral C-methyl moiety at C-7. It has been found
that the lys447glu mutation leads to enhanced binding of piperazinylquinolones but has little to no
influence on nalidixic acid.181 This is posited to
involve an electrostatic interaction between the
glutamate carboxylate and the protonated piperazinyl nitrogen in the B subunit. This interaction would
not take place with basic lysine. Being neutral (in
that region), nalidixic acid should be indifferent to
this change. Supporting this idea is the finding that
the asp426asn mutation decreases binding of piperazinylquinolones.213 Here the change from a putatively significant aspartoyl carboxylate to a neutral
asparagine residue would fit this developing picture.
This comparatively vague picture is all that is
available to guide synthesis at this moment.
These exchanges take place in the QRDR of the B
subunits. This region is distant from the analogous
QRDR region of the A subunits, so a mutual interaction would have to take place allosterically over a
considerable distance.144
9.5. Mutations of Topoisomerase IV

One recalls that the ParC subunits of topoisomerase IV are roughly equivalent to GyrA as is
ParE to GyrB. Furthermore, the QRDR is highly
conserved, so it seems safe to presume that there
should be significant parallels between quinolone
activity and the mechanisms of resistance to the two
enzymes. There are certainly some significant overlaps. For example, in E. coli mutations in oarC and
oarE genes convey significant quinolone resistance

only if gyrA mutations have taken place already. In
Gram-positives, significant resistance is seen following amino acid changes in ParC and ParE, with ParC
single amino acid changes being more common. Many
other mutational changes have also been observed.
In many Gram-positive pathogens the principal
quinolone-sensitive target is topoisomerase IV although mutations in the DNA gyrase further enhance
resistance. The sequence of resistance mutations here
is reversed compared with that in the gyrase occurring later. These gyrase mutations are, however,
quite similar to those occurring with Gram-negatives.
ParC, the subunit of topoisomerase IV comparable
to GyrA of DNA gyrase, undergoes resistance-associated mutational changes in Gram-negatives between
amino acids 78 and 116. Most commonly encountered
among these are changes in amino acids 80 and 84.
These are analogous in position to amino acids 83
and 87 of DNA gyrase, suggesting that they may
function similarly. Amino acid 80 is normally serine
and can become leucine, isoleucine, argenine, or
tryptophan depending upon the species involved.
Amino acid 84 is normally glutamic acid and is
exchanged for lysine, glycine, valine, or isoleucine.
The parallel with GyrA is striking.197
With some quinolones (notably sparfloxacin, nadifloxacin, and garenoxacin) DNA gyrase is the primary
target at least with S. aureus. The fairly substantial
homology between DNA gyrase and topoisomerase IV
suggests that similarities in the resistance pattern
might be found, so one is not surprised that this is
the case. In particular, the analogous QRDR segment
is highly conserved. Here mutations in ParC are more
influential than mutations in ParE, and there is
cooperativity between them in determining the overall level of resistance. The genetics of resistance have
been studied most thoroughly with S. aureus and S.
pneumoniae than with other Gram-positive microorganisms. In Gram-positives, the picture is less clear
and less data are yet available. Frequent mutational
changes in the GyrA subunit involve amino acids in
a narrower range stretching between 81 and 106.
Amino acid 83 is still among the frequently changed,
but amino acid 87 is replaced by amino acids at other
nearby positions. Amino acid 81, normally serine,
becomes phenylalanine, tyrosine, and cysteine. Amino
acid 83, still normally serine, becomes phenylalanine,
argenine, isoleucine, asparagine, or tyrosine. Amino
acid 84, normally serine, becomes leucine, alanine,
valine, lysine, tyrosine, or phenylalanine. Amino
acid 85, normally serine, becomes proline, lysine,
glutamine, or glycine.
The ParC mutational changes in Gram-positives
most frequently involve substitutions at amino acids
ranging from 23 to 176. Most often serines 79 and
80 and glutamic acid 84 are involved. Amino acid 79
correlates with amino acid 81 of GyrA and 80
correlates with amino acid 84, again suggesting some
parallel of effect. Serine 79 becomes phenylalanine,
tyrosine, isoleucine, leucine, or alanine. Serine 80
becomes phenylalanine, tyrosine, leucine, isoleucine,
or argenine. Glutamic acid 84 becomes lysine, leucine, valine, alanine, glycine, tyrosine, or asparagine.
Mitscher
Mutational changes in the GyrB and ParE subunits

of DNA gyrase and topoisomerase IV, respectively,
play a modulating role in quinolone resistance. It is
thought that changes in this subunit alter its topology and that this is transmitted to the A and C units.
In this way they are believed to degrade the fit of
quinolones to this part of the enzyme. As noted
elsewhere, mutations in both subunits are associated
with maximum resistance. Interestingly, mutations
in these subunits are significantly less frequently
associated with resistance than those in GyrA and
ParC.
Likewise in ParE they are singly distributed between residues 25 and 478.
9.6. Effect of Mutations on Microbial Vitality

It is significant to note that while mutations to
lesser sensitivity to quinolones have definite survival
value to pathogens, in a number of cases the microorganisms carrying these mutations are less vigorous
than wild strains. This suggests that drug rotations
or holidays might have a favorable effect on resistance levels.
9.7. Possible Approaches to Dealing with the

Resistance Problem
The prevailing belief is that quinolones bind to
GyrA, and possibly ParC, near amino acids 83 and
87. When these amino acids are modified, resistance
is common, especially when the mutations result in
less hydrophilic character. This is overcome to some
extent when C-8 of quinolones bears a chlorine,
bromine, or methoxy group.214,215 It can be speculated
that normally hydrogen-bond-donating and -receiving
enzyme side chains in this region are influential in
maintaining structure, and the structure is disordered when this effect is lost. Perhaps suitable
quinolone C-8 structural features compensate for
this. Alternatively, it is possible that the polar side
chains at positions 83 and 87 participate in hydrogen
bonding to the C-7 quinolone positions and that this
presents a steric hindrance to substrate processing.
In this view, loss of the hydrogen bond character
would make the drugs less effective as they would
no longer fit as well. In the mutants one would then
propose that the C-8 substituents provide an alternate binding capability, compensating for loss of
bonding due to C-7.
In partial support of these ideas, mutation to a
cysteine residue results in an enzyme that is less
tolerant of the size of a substituent at the distal
nitrogen of a C-7 moiety. This effect is significantly
lost when the substituent is attached to an adjacent
carbon compared to the nitrogen. The implication of
this is that the enzyme has only a little space
available in this region for the drug to fit into. If so,
this is not very helpful in terms of analogue production.
It is technically extremely difficult to reactivate a
defective enzyme through chemical means. Mostly
therapeutic manipulations affecting enzymes are
inhibitory. The best of current stratagems that might
Table 2. Officially Approved Clinical Indications for Common Quinolone Anti-Infectivesa

disease
Nal
Nor
Cipro
Ofl
Levo
urinary tract infection

sexually transmitted diseases
GI infections
upper respiratory tract
lower respiratory tract
skin and soft tissue
opthalmic
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Trova
Moxi
Gati
X
X
X
X
X
X
X
X
X
The data were assembled from the Physicians Desk Reference.34
Table 3. Approximate Pharmacokinetic Values for Commonly Used Quinolonesa

drug
nalidixic acid
norfloxacin
ciprofloxacin
ofloxacin
levofloxacin
trovafloxacin
moxifloxacin
gatifloxacin
grepafloxacin
UE
PPB
Cl
VD
T1/2
84
15
40
25
31
70
40
2.9
0.55
60
95
99
91
86
22
30
50
64
74
9
22
77
9.36
7.6
3.5
2.5
14
2.3
2.2
1.8
1.36
1.29
2.05
11.5
5
3.3
5.7
7
11.3
15.4
6.52
12.1
PT
PC
1.45
0.6
1.7
1.6
0.95
2.0
1.49
2.77
1.57
1.44
2.5
1.6
4.5
2.09
2.5
1.71
1.98
a Abbreviations: F ) bioavailability (%); UE ) percent active drug excreted in the urine; PPB ) plasma protein binding (%);
Cl ) clearance ((mL/min)/kg); VD ) volume of dilution (L/kg); T1/2 ) plasma half-life (h); PT ) time to peak concentration in
blood (h); PC ) peak blood concentration (g/mL).
ameliorate resistance due to enzyme mutations would

be to decrease the resistance development rate. The
most effective means of accomplishing this is to make
sure that sufficient drug is always present to kill the
organisms. Dead bacteria do not mutate. Alternatively, apparently under some circumstances in the
laboratory mutations causing DNA strand breaks in
the presence of reactive oxygen species can be protected against by use of certain antioxidant catechins.
These agents can also be shown to delay or prevent
resistance emergence at no other effect doses. This
suggests the possibility that antimutagenic agents
might be useful or, better yet, that quinolones possessing not only antimicrobial activity but also antimutagenic activity might be usefully developed.138,216
Resistance is often not an all or nothing phenomenon, so development of ever more potent quinolones
will undoubtedly be pursued. Pessimistically, one
notes that with widespread use resistance always
developsseither easily or after a delay.
10. Clinical Indications

The widespread utilization of the quinolones is a
consequence of their usefulness in treating a wide
variety of common infections, particularly those
encountered frequently in the community. For example, uncomplicated urinary tract, normal upper
respiratory tract, and skin and soft tissue infections
are quite common, and almost all of the quinolones
show utility for these conditions. The special frequency of prescription of ciprofloxacin, ofloxacin, and
levofloxacin is understandable from the table of
approved indications wherein a great many uses are
listed (Table 2).
11. Pharmacokinetics
The precise figures for the pharmacokinetic features of quinolone anti-infectives vary significantly
from source to source and are strongly dependent

upon the dosages given, but the following data are
representative of those published and are reasonably
near the consensus values. Figures are averaged, and
error bars, which are sometimes substantial, have
been omitted for clarity. Thus, the data in Table 3
are suitable for rough comparisons only.
From the data in the table it can be seen that the
various agents are significantly to outstandingly
bioavailable after oral administration, peak in the
blood soon thereafter, and vary widely in the percentage of administered drug appearing in the urine
in active form. Urinary excretion of quinolones
involves a blend of glomerular filtration and tubular
secretion. Tubular secretion is modestly enantioselective.52 Transintestinal and biliary excretion routes
are only significant for highly lipophilic quinolones
such as trovafloxacin.217 The more recently introduced agents have significantly longer half-lives that
often permit one-a-day dosing.
A variety of in vitro and in vivo investigations show
that many quinolones kill bacteria in a concentrationdependent manner.218 This contrasts with the classical pattern that is still followed in many other
classes of anti-infectives in which one strives to
maintain continuously reasonable multiples of the
minimum inhibitory or cidial concentrations in the
blood. In contrast, it is believed that with the quinolones efficacy is more closely related to the concentrations achievable than the dosage interval.
Animal infection models indicate that the area under
the curve to minimum inhibitory concentration ratio
is a significantly useful predictor of efficacy. The ratio
of the peak concentration achievable to the minimum
inhibitory concentration is believed more important
to prevent selection for resistance during the course
of therapy than to keep the concentration above a
given value. It is suggested that a 24 h AUC/MIC
ratio of 25-100, depending on the microorganism, is
satisfactory and is comparatively independent of the

dosage interval, the specific drug, the animal species,
and the site of infection.219 It is difficult, however, to
establish this proposition in patients. One also has
some concern that pushing peak drug levels ever
higher can lead to increased toxicity.
12. Future Prospects

The clinical impact of the quinolone anti-infectives
makes it a certainty that they will serve for many
years to come as mainstays in the unending struggle
of mankind against morbidity and mortality due to
infectious diseases. Whereas it is clear that future
analoguing studies will result in incremental improvements in their useful properties, recent findings
demonstrating potential impact in the areas of fungal220 and viral221 infections and in cancer chemotherapy94 suggest a role for carefully crafted analogues in these areas as well. On the other hand,
concerns about cross-resistance, particularly stemming from agricultural applications of quinolones,
suggest conservatism in application will be appropriate.
The interaction of quinolones with DNA-processing
enzymes is fascinating, and this story is still unfolding. Study of the quinolones has revealed much
interesting molecular biology, and the future will
reveal much more.
13. Conclusions
After nearly 40 years of intensive exploration,
approximately 20 quinolone anti-infectives have been
marketed, and two of these are present market
leaders. Utopiafloxacin remains elusive. Structureactivity relationships of the major present chemotypes are now reasonably clear although surprises
pop up from time to time. The molecular details of
their interaction with DNA gyrase, bacterial topoisomerase IV, and human topoisomerase II remain
largely conjectural and require more attention. Structure-toxicity relationships are not yet well understood although this is increasingly coming into focus.
Resistance emergence, as with every other family of
antibacterials, is increasing at a disturbing rate.
Whereas much detail is now available outlining the
alterations in bacteria that lead to resistance, practical means of minimizing this phenomenon have yet
to be found.
14. References
(1) Andriole, V. T. The Quinolones; Academic Press: San Diego,
2000.
(2) Beermann, D.; Kuhlmann, J.; Dalhoff, A.; Zeiler, H. J. Quinolone
antibacterials; Springer: Berlin, New York, 1998.
(3) Crumplin, G. C. The 4-Quinolones: Antibacterial Agents in vitro;
Springer-Verlag: New York, 1990.
(4) Hooper, D. C.; Wolfson, J. S. Quinolone antimicrobial agents,
2nd ed.; American Society for Microbiology: Washington, DC,
1993.
(5) Hooper, D. C.; Rubinstein, E. Quinolone antimicrobial agents,
3rd ed.; ASM Press: Washington, DC, 2003.
(6) Kuhlmann, J.; Dalhoff, A.; Zeiler, J.-J. Quinolone Antibacterials;
Springer-Verlag: New York, 1998.
(7) Ronald, A. R.; Low, D. E. Fluoroquinolone Antibiotics; Birkhauser: Boston, 2003.
(8) Siporin, C.; Heifetz, C. L.; Domagala, J. M. The New Generation
of Quinolones; Marcel Dekker: New York, 1990.
Mitscher
(9) Wolfson, J. S.; Hooper, D. C. Quinolone antimicrobial agents;
American Society for Microbiology: Washington, DC, 1989.
(10) Alghasham, A. A.; Nahata, M. C. Ann. Pharmacother. 2000, 34,
347, Quiz 413.
(11) Appelbaum, P. C.; Hunter, P. A. Int. J. Antimicrob. Agents 2000,
16, 5.
(12) Ball, P. J. Antimicrob. Chemother. 2000, 46 (Suppl. T1), 17.
(13) Bertino, J., Jr.; Fish, D. Clin. Ther. 2000, 22, 798-817, Discussion 797.
(14) Bush, K.; Goldschmidt, R. Curr. Opin. Invest. Drugs 2000, 1,
22.
(15) Casparian, J. M.; Luchi, M.; Moffat, R. E.; Hinthorn, D. South
Med. J. 2000, 93, 488.
(16) Gootz, T. D.; Brighty, K. E. Med. Res. Rev. 1996, 16, 433.
(17) Hooper, D. C. Clin. Infect. Dis. 2000, 31 (Suppl. 2), S24.
(18) ODonnell, J. A.; Gelone, S. P. Infect. Dis. Clin. North Am. 2000,
14, 489.
(19) Pickerill, K. E.; Paladino, J. A.; Schentag, J. J. Pharmacotherapy
2000, 20, 417.
(20) Varon, E.; Gutmann, L. Res. Microbiol. 2000, 151, 471.
(21) Maple, P. B.; Brumfitt, W.; Hamilton-Miller, J. M. J. Chemother.
1990, 2, 280.
(22) Walker, R. W.; Wright, A. J. Mayo Clin. Proc. 1991, 66, 1249.
(23) Gootz, T. D.; McGuirk, P. R. Expert Opin. Invest. Drugs 1994,
3, 93.
(24) Domagala, J. M. J. Antimicrob. Chemother. 1994, 33, 685.
(25) Chu, D. T. W.; Fernandes, P. B. Adv. Drug Res. 1991, 21, 39.
(26) Rosen, T. Prog. Med. Chem. 1990, 235.
(27) Lesher, G. Y.; Froelich, E. J.; Gruett, M. D.; Bailey, J. H.;
Brundage, R. P. J Med. Pharm. Chem. 1962, 91, 1063.
(28) Koga, H.; Ito, A.; Murayama, S.; Suzue, S.; Irikura, T J. Med.
Chem. 1980, 23, 1358.
(29) Jaber, L. A.; Bailey, E. M.; Rybak, M. J. Clin. Pharm. 1989, 8,
97.
(30) Wise, R.; Andrews, J. M.; Edwards, L. J. Antimicrob. Agents
Chemother. 1983, 23, 559.
(31) Sato, K.; Matsuura, Y.; Inoue, M.; Une, T.; Osada, Y.; Ogawa,
H.; Mitsuhashi, S. Antimicrob. Agents Chemother. 1982, 22, 548.
(32) Hayakawa, I.; Hiramitsu, T.; Tanaka, Y. Chem. Pharm. Bull.
1984, 32, 4907.
(33) Atarashi, S.; Yokahama, S.; Yamazaki, K.; Sakano, K.; Imamura,
M.; Hayakawa, I. Chem. Pharm. Bull. 1987, 35, 1896.
(34) Physicians Desk Reference; Thompson PDR: Montvale, NJ, 2004.
(35) Blondeau, J. M. Expert Opin. Invest. Drugs 2000, 9, 1877.
(36) Nightingale, C. H. Pharmacotherapy 2000, 20, 245.
(37) Ling, T. K.; Liu, E. Y.; Cheng, A. F. Chemotherapy 1999, 45, 22.
(38) Bron, N. J.; Dorr, M. B.; Mant, T. G.; Webb, C. L.; Vassos, A. B.
J. Antimicrob. Chemother. 1996, 38, 1023.
(39) Gellert, M.; Mizuuchi, K.; ODea, M. H.; Nash, H. A. Proc. Natl.
Acad. Sci. U.S.A. 1976, 73, 3872.
(40) Cozzarelli, N. R. Science 1980, 207, 953.
(41) Kato, J.; Nishimura, Y.; Imamura, R.; Niki, H.; Hiraga, S.;
Suzuki, H. Cell 1990, 63, 393.
(42) Takei, M.; Fukuda, H.; Kishii, R.; Hosaka, M. Antimicrob. Agents
Chemother. 2001, 45, 3544.
(43) Neuman, M. Int. J. Clin. Pharmacol. Res. 1987, 7, 173.
(44) Pan, X. S.; Fisher, L. M. Antimicrob. Agents Chemother. 1998,
42, 2810.
(45) Tanaka, M.; Onodera, Y.; Uchida, Y.; Sato, K.; Hayakawa, I.
(46) Hoshino, K.; Sato, K.; Une, T.; Osada, Y. Antimicrob. Agents
Chemother. 1989, 33, 1816.
(47) Scheld, W. M. Emerging Infect. Dis. 2003, 9, 1.
(48) Tanaka, M.; Wang, T.; Onodera, Y.; Uchida, Y.; Sato, K. J. Infect.
Chemother. 2000, 6, 131.
(49) Jalal, S.; Ciofu, O.; Hoiby, N.; Gotoh, N.; Wretlind, B. Antimicrob.
(50) Bazin, M.; Bosca, F.; Marin, M. L.; Miranda, M. A.; Patterson,
L. K.; Santus, R. Photochem. Photobiol. 2000, 72, 451.
(51) Ooie, T.; Terasaki, T.; Suzuki, H.; Sugiyama, Y. Drug Metab.
Dispos. 1997, 25, 784.
(52) Sorgel, F.; Kinzig, M. Am. J. Med. 1993, 94, 56S.
(53) Lazarczyk, D. A.; Goldstein, N. S.; Gordon, S. C. Dig. Dis. Sci.
2001, 46, 925.
(54) Blum, M. D.; Graham, D. J.; McCloskey, C. A. Clin. Infect. Dis.
1994, 18, 946.
(55) Satoh, Y.; Sugiyama, A.; Chiba, K.; Tamura, K.; Hashimoto, K.
J. Cardiovasc. Pharmacol. 2000, 36, 510.
(56) Jeffrey, A. M.; Shao, L.; Brendler-Schwaab, S. Y.; Schluter, G.;
Williams, G. M. Arch. Toxicol. 2000, 74, 555.
(57) Gould, R., Jr.; Jacobs, W. A. J. Am. Chem. Soc. 1939, 61, 2890.
(58) Grohe, K.; Heitzer, H. Liebigs Ann. Chem. 1987, 29.
(59) Bouzard, D.; Di Cesare, P.; Essiz, M.; Jacquet, J. P.; Remuzon,
P.; Weber, A.; Oki, T.; Masuyoshi, M. J. Med. Chem. 1989, 32,
537.
(60) MacDonald, A. A.; DeWitt, S. H.; Ramage, R. Chemia 1996, 50,
266.
(61) Frank, K. E.; Jung, M.; Mitscher, L. A. Comb. Chem. High
Throughput Screening 1998, 1, 89.

(62) Gerster, J. E.; Rohlfing; S. R.; Pecore, S. E.; Winandy, R. M.;
Stern, R. M.; Landmesser, J. E.; Olsen, R. A.; Gleason, W. B. J.
Med. Chem. 1987, 30, 839.
(63) Gerster, J. E.; Rohlfing, S. R.; Winandy, R. M. Proc. North Am.
Med. Chem. Symp., Toronto, Canada 1982, 153.
(64) Hayakawa, I.; Atarashi, S.; Yokohama, S.; Imamura, M.; Sakano,
K.; Furukawa, M. Antimicrob. Agents Chemother. 1986, 29, 163.
(65) Mitscher, L. A.; Sharma, P. N.; Chu, D. T. W.; Shen, L. L.;
Pernet, A. G. J. Med. Chem. 1987, 30, 2283.
(66) Li, Q. M.; Mitscher, L. A.; Shen, L. L. Med. Res. Rev. 2000, 20,
231.
(67) Lomaestro, B. M.; Bailie, G. R. Drug Saf. 1995, 12, 314.
(68) Polk, R. E. Am. J. Med. 1989, 87, 76S.
(69) Fresta, M.; Guccione, S.; Beccari, A. R.; Furneri, P. M.; Puglisi,
G. Bioorg. Med. Chem. 2002, 10, 3871.
(70) Fresta, M.; Spadaro, A.; Cerniglia, G.; Ropero, I. M.; Puglisi, G.;
Furneri, P. M. Antimicrob. Agents Chemother. 1995, 39, 1372.
(71) Hayashi, N.; Nakata, Y.; Yazaki, A. Antimicrob. Agents Chemother. 2004, 48, 799.
(72) Albini, A.; Monti, S. Chem. Soc. Rev. 2003, 32, 238.
(73) Vargas, F.; Rivas, C.; Machado, R. J. Photochem. Photobiol., B
1991, 11, 81.
(74) Morimura, T.; Ohno, T.; Matsukura, H.; Nobuhara, Y. Chem.
Pharm. Bull. (Tokyo) 1995, 43, 1000.
(75) Fasani, E.; Profumo, A.; Albini, A. Photochem. Photobiol. 1998,
68, 666.
(76) Shibata, T.; Nagasawa, M.; Iwai, N.; Miyazaki, M.; Kawamura,
Y.; Kodama, T. Jpn. J. Antibiot. 1995, 48, 861.
(77) Martinez, L. J.; Sik, R. H.; Chignell, C. F. Photochem. Photobiol.
1998, 67, 399.
(78) Monti, S.; Sortino, S.; Fasani, E.; Albini, A. Chem.sEur. J. 2001,
7, 2185.
(79) Howard, W.; Biedenbach, D. J.; Jones, R. N. Clin. Microbiol.
Infect. 2002, 8, 340.
(80) Miolo, G.; Viola, G.; Vedaldi, D.; DallAcqua, F.; Fravolini, A.;
Tabarrini, O.; Cecchetti, V. Toxicol. in Vitro 2002, 16, 683.
(81) Jones, R. N.; Biedenbach, D. J. Diagn. Microbiol. Infect. Dis.
2003, 45, 273.
(82) Matsumoto, J.; Miyamoto, T.; Minamida, A.; Nishimura, Y.;
Egawa, H.; Nishimura, H. J. Med. Chem. 1984, 27, 292.
(83) Koikeda, S.; Okabe, H.; Shibata, H.; Hase, T.; Nakanishi, N.;
Okezaki, E.; Kato, H.; Masamune, Y. Jpn. J. Antibiot. 1989, 42,
193.
(84) Peric, M.; Jacobs, M. R.; Appelbaum, P. C. Antimicrob. Agents
Chemother. 2004, 48, 3188.
(85) Fu, L. M.; Fu-Liu, C. S. Omics 2002, 6, 199.
(86) Tomb, J. F.; White, O.; Kerlavage, A. R.; Clayton, R. A.; Sutton,
G. G.; Fleischmann, R. D.; Ketchum, K. A.; Klenk, H. P.; Gill,
S.; Dougherty, B. A.; Nelson, K.; Quackenbush, J.; Zhou, L.;
Kirkness, E. F.; Peterson, S.; Loftus, B.; Richardson, D.; Dodson,
R.; Khalak, H. G.; Glodek, A.; McKenney, K.; Fitzegerald, L. M.;
Lee, N.; Adams, M. D.; Venter, J. C.; et al. Nature 1997, 388,
539.
(87) Albrecht, R. Prog. Drug Res. 1977, 21, 9.
(88) Chu, D. T.; Fernandes, P. B.; Claiborne, A. K.; Pihuleac, E.;
Nordeen, C. W.; Maleczka, R. E., Jr.; Pernet, A. G. J. Med. Chem.
1985, 28, 1558.
(89) Fernandes, P. B.; Chu, D. T.; Bower, R. R.; Jarvis, K. P.; Ramer,
N. R.; Shipkowitz, N. Antimicrob. Agents Chemother. 1986, 29,
201.
(90) Rosen, T.; Chu, D. T.; Lico, I. M.; Fernandes, P. B.; Shen, L.;
Borodkin, S.; Pernet, A. G. J. Med. Chem. 1988, 31, 1586.
(91) Alghasham, A. A.; Nahata, M. C. Ann. Pharmacother. 1999, 33,
48.
(92) Lumish, R. M.; Norden, C. W. Antimicrob. Agents Chemother.
1975, 7, 159.
(93) Segawa, J.; Kitano, M.; Kazuno, K.; Matsuoka, M.; Shirahase,
I.; Ozaki, M.; Matsuda, M.; Tomii, Y.; Kise, M. J. Med. Chem.
1992, 35, 4727.
(94) Chu, D. T.; Hallas, R.; Clement, J. J.; Alder, J.; McDonald, E.;
Plattner, J. J. Drugs Exp. Clin. Res. 1992, 18, 275.
(95) Kondo, H.; Sakamoto, F.; Inoue, Y.; Tsukamoto, G. J. Med. Chem.
1989, 32, 679.
(96) Kondo, H.; Sakamoto, F.; Kawakami, K.; Tsukamoto, G. J. Med.
Chem. 1988, 31, 221.
(97) Kondo, H.; Sakamoto, F.; Uno, T.; Kawahata, Y.; Tsukamoto,
G. J. Med. Chem. 1989, 32, 671.
(98) Li, Z. R.; Guo, H. Y.; Zhang, Z. P. Yao Xue Xue Bao 1991, 26,
111.
(99) Vincent, J.; Venitz, J.; Teng, R.; Baris, B. A.; Willavize, S. A.;
Polzer, R. J.; Friedman, H. L. J. Antimicrob. Chemother. 1997,
39 (Suppl. B), 75.
(100) Li, Q.; Chu, D. T.; Claiborne, A.; Cooper, C. S.; Lee, C. M.; Raye,
K.; Berst, K. B.; Donner, P.; Wang, W.; Hasvold, L.; Fung, A.;
Ma, Z.; Tufano, M.; Flamm, R.; Shen, L. L.; Baranowski, J.;
Nilius, A.; Alder, J.; Meulbroek, J.; Marsh, K.; Crowell, D.; Hui,
Y.; Seif, L.; Melcher, L. M.; Plattner, J. J.; et al. J. Med. Chem.
1996, 39, 3070.

(101) Miyamoto, T.; Matsumoto, J.; Chiba, K.; Egawa, H.; Shibamori,
K.; Minamida, A.; Nishimura, Y.; Okada, H.; Kataoka, M.; Fujita,
M.; et al. J. Med. Chem. 1990, 33, 1645.
(102) Imada, T.; Miyazaki, S.; Nishida, M.; Yamaguchi, K.; Goto, S.
(103) Kaminsky, D.; Meltzer, R. I. J. Med. Chem. 1968, 11, 160.
(104) Nagate, T.; Komatsu, T.; Izawa, A.; Ohmura, S.; Namiki, S.;
Mitsuhashi, S. Antimicrob. Agents Chemother. 1980, 17, 763.
(105) Braveny, I.; Machka, K. Arzneimittelforschung 1980, 30, 1476.
(106) Corelli, F.; Massa, S.; Stefancich, G.; Artico, M.; Panico, S.;
Simonetti, N. Farmaco, Ed. Sci. 1984, 39, 95.
(107) Matsumoto, J.; Minami, S. J. Med. Chem. 1975, 18, 74.
(108) Hayashi, K.; Takahata, M.; Kawamura, Y.; Todo, Y. Arzneimittelforschung 2002, 52, 903.
(109) Ledoussal, B.; Bouzard, D.; Coroneos, E. J. Med. Chem. 1992,
35, 198.
(110) Gerster, J. R.; Rohlfing, S. R.; Pecore, S. E.; Winandy, R. M. Proc.
North Am. Med. Chem. Symp., Toronto 1982, 153.
(111) Ishikawa, H.; Tabusa, F.; Miyamoto, H.; Kano, M.; Ueda, H.;
Tamaoka, H.; Nakagawa, K. Chem. Pharm. Bull. (Tokyo) 1989,
37, 2103.
(112) Takahashi, N.; Shiragiku, T.; Itoh, T.; Kaneko, E.; Shibahara,
T.; Kanbe, T.; Yamashita, S. Arzneimittelforschung 1994, 44,
1265.
(113) Fukuyama, M.; Oonaka, K.; Hara, M.; Imagawa, Y. Kansenshogaku Zasshi 1996, 70, 51.
(114) Coughlin, S. A.; Danz, D. W.; Robinson, R. G.; Klingbeil, K. M.;
Wentland, M. P.; Corbett, T. H.; Waud, W. R.; Zwelling, L. A.;
Altschuler, E.; Bales, E.; et al. Biochem. Pharmacol. 1995, 50,
111.
(115) Biedenbach, D. J.; Jones, R. N. Antimicrob. Agents Chemother.
1995, 39, 1636.
(116) Georgopapadakou, N. H.; Dix, B. A.; Angehrn, P.; Wick, A.;
Olson, G. L. Antimicrob. Agents Chemother. 1987, 31, 614.
(117) Ariens, E. J. Trends Pharmacol. Sci. 1993, 14, 68.
(118) Shen, L. L.; Mitscher, L. A.; Sharma, P. N.; ODonnell, T. J.;
Chu, D. W.; Cooper, C. S.; Rosen, T.; Pernet, A. G. Biochemistry
1989, 28, 3886.
(119) Morrissey, I.; Hoshino, K.; Sato, K.; Yoshida, A.; Hayakawa, I.;
Bures, M. G.; Shen, L. L. Antimicrob. Agents Chemother. 1996,
40, 1775.
(120) Hoshino, K.; Sato, K.; Akahane, K.; Yoshida, A.; Hayakawa, I.;
Sato, M.; Une, T.; Osada, Y. Antimicrob. Agents Chemother.
1991, 35, 309.
(121) Akahane, K.; Sekiguchi, M.; Une, T.; Osada, Y. Antimicrob.
(122) Vorbach, H.; Weigel, G.; Robibaro, B.; Schaumann, R.; Hlousek,
M.; Beil, W. J.; Griesmacher, A.; Georgopoulos, A.; Graninger,
W. Int. J. Clin. Pharmacol. Ther. 1997, 35, 235.
(123) Fleisch, F.; Hartmann, K.; Kuhn, M. Infection 2000, 28, 256.
(124) Shakibaei, M.; de Souza, P.; van Sickle, D.; Stahlmann, R. Arch.
Toxicol. 2001, 75, 369.
(125) Williams, R. J., III; Attia, E.; Wickiewicz, T. L.; Hannafin, J. A.
Am. J. Sports Med. 2000, 28, 364.
(126) Kim, J. C.; Yun, H. I.; Shin, H. C.; Han, S. S.; Chung, M. K.
Arch. Toxicol. 2000, 74, 120.
(127) Donders, G. G. Drugs 2000, 59, 477.
(128) Patmore, L.; Fraser, S.; Mair, D.; Templeton, A. Eur. J. Pharmacol. 2000, 406, 449.
(129) Kang, J.; Wang, L.; Chen, X. L.; Triggle, D. J.; Rampe, D. Mol.
Pharmacol. 2001, 59, 122.
(130) Chiba, K.; Sugiyama, A.; Satoh, Y.; Shiina, H.; Hashimoto, K.
Toxicol. Appl. Pharmacol. 2000, 169, 8.
(131) Allen, A.; Bygate, E.; Faessel, H.; Isaac, L.; Lewis, A. Int. J.
Antimicrob. Agents 2000, 15, 283.
(132) Bowles, S. K.; Popovski, Z.; Rybak, M. J.; Beckman, H. B.;
Edwards, D. J. Antimicrob. Agents Chemother. 1988, 32, 510.
(133) Randinitis, E. J.; Alvey, C. W.; Koup, J. R.; Rausch, G.; Abel,
R.; Bron, N. J.; Hounslow, N. J.; Vassos, A. B.; Sedman, A. J.
(134) Cole, S. T.; Brosch, R.; Parkhill, J.; Garnier, T.; Churcher, C.;
Harris, D.; Gordon, S. V.; Eiglmeier, K.; Gas, S.; Barry, C. E.,
III; Tekaia, F.; Badcock, K.; Basham, D.; Brown, D.; Chillingworth, T.; Connor, R.; Davies, R.; Devlin, K.; Feltwell, T.;
Gentles, S.; Hamlin, N.; Holroyd, S.; Hornsby, T.; Jagels, K.;
Barrell, B. G.; et al. Nature 1998, 393, 537.
(135) Bates, A. D.; Maxwell, A. DNA Topology; Oxford University
Press: New York, 1993.
(136) Nelson, D. L.; Cox, M. M. Lehninger Principles of Biochemistry,
4th ed.; W. H. Freeman and Co.: New York, 2005.
(137) Berg, J. M.; Tymoczko, J. L.; Stryer, L. Biochemistry, 5th ed.;
W. H. Freeman and Co.: New York, 2002.
(138) Aubry, A.; Pan, X.-S.; Fisher, L. M.; Jarlier, V.; Cambau, E.
(139) Chen, C. R.; Malik, M.; Snyder, M.; Drlica, K. J. Mol. Biol. 1996,
258, 627.
(140) Krasin, F.; Hutchinson, F. J. Mol. Biol. 1977, 116, 81.
(141) Mouton, R. P.; Koelman, A. Chemotherapy 1966, 11, 10.

(142) Gradelski, E.; Kolek, B.; Bonner, D. P.; Valera, L.; Minassian,
B.; Fung-Tomc, J. Int. J. Antimicrob. Agents 2001, 17, 103.
(143) Fraser, C. M.; Norris, S. J.; Weinstock, G. M.; White, O.; Sutton,
G. G.; Dodson, R.; Gwinn, M.; Hickey, E. K.; Clayton, R.;
Ketchum, K. A.; Sodergren, E.; Hardham, J. M.; McLeod, M. P.;
Salzberg, S.; Peterson, J.; Khalak, H.; Richardson, D.; Howell,
J. K.; Chidambaram, M.; Utterback, T.; McDonald, L.; Artiach,
P.; Bowman, C.; Cotton, M. D.; Venter, J. C.; et al. Science 1998,
281, 375.
(144) Gushchin, A. E.; Ladygina, V. G.; Govorun, V. M.; Taraskina,
A. M.; Savicheva, A. M. Mol. Gen. Mikrobiol. Virusol. 2000, 33.
(145) Reece, R. J.; Maxwell, A. J. Biol. Chem. 1991, 266, 3540.
(146) Ali, J. A.; Jackson, A. P.; Howells, A. J.; Maxwell, A. Biochemistry
1993, 32, 2717.
(147) Brown, P. O.; Peebles, C. L.; Cozzarelli, N. R. Proc. Natl. Acad.
Sci. U.S.A. 1979, 76, 6110.
(148) Tingey, A. P.; Maxwell, A. Nucleic Acids Res. 1996, 24, 4868.
(149) Chatterji, M.; Unniraman, S.; Maxwell, A.; Nagaraja, V. J. Biol.
Chem. 2000, 275, 22888.
(150) Bernard, P.; Kezdy, K. E.; Van Melderen, L.; Steyaert, J.; Wyns,
L.; Pato, M. L.; Higgins, P. N.; Couturier, M. J. Mol. Biol. 1993,
234, 534.
(151) Mizuuchi, K.; ODea, M. H.; Gellert, M. Proc. Natl. Acad. Sci.
U.S.A. 1978, 75, 5960.
(152) Nakada, N.; Gmunder, H.; Hirata, T.; Arisawa, M. Antimicrob.
(153) Tsai, F. T.; Singh, O. M.; Skarzynski, T.; Wonacott, A. J.; Weston,
S.; Tucker, A.; Pauptit, R. A.; Breeze, A. L.; Poyser, J. P.; OBrien,
R.; Ladbury, J. E.; Wigley, D. B. Proteins 1997, 28, 41.
(154) Angehern, P.; Buchmann, S.; Funk, C.; Goetschi, E.; Gmeunder,
H.; Hebeisen, P.; Kostrewa, D.; Link, H.; Luebbers, T.; Masciadri,
R.; Nielsen, J.; Reindl, P.; Ricklin, F.; Schmitt-Hoffmann, A.;
Theil, F.-P. J. Med. Chem. 2004, 47, 1487.
(155) Heddle, J. G.; Blance, S. J.; Zamble, D. B.; Hollfelder, F.; Miller,
D. A.; Wentzell, L. M.; Walsh, C. T.; Maxwell, A. J. Mol. Biol.
2001, 307, 1223.
(156) Pierrat, O. A.; Maxwell, A. J. Biol. Chem. 2003, 278, 35016.
(157) Tanitame, A.; Oyamada, Y.; Ofuji, K.; Fujimoto, M.; Iwai, N.;
Hiyama, Y.; Suzuki, K.; Ito, H.; Terauchi, H.; Kawasaki, M.;
Nagai, K.; Wachi, M.; Yamagishi, J. J. Med. Chem. 2004, 47,
3693.
(158) Tanitame, A.; Oyamada, Y.; Ofuji, K.; Kyoya, Y.; Suzuki, K.; Ito,
H.; Kawasaki, M.; Nagai, K.; Wachi, M.; Yamagishi, J. Bioorg.
Med. Chem. Lett. 2004, 14, 2857.
(159) Tanitame, A.; Oyamada, Y.; Ofuji, K.; Suzuki, K.; Ito, H.;
Kawasaki, M.; Wachi, M.; Yamagishi, J. Bioorg. Med. Chem. Lett.
2004, 14, 2863.
(160) Elsea, S. H.; Osheroff, N.; Nitiss, J. L. J. Biol. Chem. 1992, 267,
13150.
(161) Wigley, D. B.; Davies, G. J.; Dodson, E. J.; Maxwell, A.; Dodson,
G. Nature 1991, 351, 624.
(162) Prodromou, C.; Roe, S. M.; OBrien, R.; Ladbury, J. E.; Piper, P.
W.; Pearl, L. H. Cell 1997, 90, 65.
(163) Morais Cabral, J. H.; Jackson, A. P.; Smith, C. V.; Shikotra, N.;
Maxwell, A.; Liddington, R. C. Nature 1997, 388, 903.
(164) Albertini, S.; Chetelat, A. A.; Miller, B.; Muster, W.; Pujadas,
E.; Strobel, R.; Gocke, E. Mutagenesis 1995, 10, 343.
(165) Champoux, J. J. Annu. Rev. Biochem. 2001, 70, 369.
(166) Anderson, V. E.; Zaniewski, R. P.; Kaczmarek, F. S.; Gootz, T.
D.; Osheroff, N. Biochemistry 2000, 39, 2726.
(167) Zunino, F.; Capranico, G. Anticancer Drug Des. 1990, 5, 307.
(168) Okuda, J.; Okamoto, S.; Takahata, M.; Nishino, T. Antimicrob.
(169) Argaman, M.; Bendetz-Nezer, S.; Matlis, S.; Segal, S.; Priel, E.
Biochem. Biophys. Res. Commun. 2003, 301, 789.
(170) Gangloff, S.; de Massy, B.; Arthur, L.; Rothstein, R.; Fabre, F.
EMBO J. 1999, 18, 1701.
(171) Shen, L. L.; Pernet, A. G. Proc. Natl. Acad. Sci. U.S.A. 1985,
82, 307.
(172) Shen, L. L. Methods Mol. Biol. 2001, 95, 171.
(173) Gould, K. A.; Pan, X. S.; Kerns, R. J.; Fisher, L. M. Antimicrob.
(174) Shen, L. L.; Baranowski, J.; Pernet, A. G. Biochemistry 1989,
28, 3879.
(175) Tornaletti, S.; Pedrini, A. M. Biochem. Biophys. Acta 1988, 949,
279.
(176) Critchlow, S. E.; Maxwell, A. Biochemistry 1996, 35, 7387.
(177) Sissi, C.; Andreolli, M.; Cecchetti, V.; Fravolini, A.; Gatto, B.;
Palumbo, M. Bioorg. Med. Chem. 1998, 6, 1555.
(178) Palumbo, M.; Gatto, B.; Zagotto, G.; Palu, G. Trends Microbiol.
1993, 1, 232.
(179) Noble, C. G.; Barnard, F. M.; Maxwell, A. Antimicrob. Agents
Chemother. 2003, 47, 854.
(180) Wentland, M. P.; Perni, R. B.; Dorff, P. H.; Rake, J. B. J. Med.
Chem. 1988, 31, 1694.
Mitscher
(181) Yoshida, H.; Bogaki, M.; Nakamura, M.; Yamanaka, L. M.;
Nakamura, S. Antimicrob. Agents Chemother. 1991, 35, 1647.
(182) Heddle, J.; Maxwell, A. Antimicrob. Agents Chemother. 2002,
46, 1805.
(183) Llorente, B.; Leclerc, F.; Cedergren, R. Bioorg. Med. Chem. 1996,
4, 61.
(184) Gatto, B.; Capranico, G.; Palumbo, M. Curr. Pharm. Des. 1999,
5, 195.
(185) Heddle, J. G.; Barnard, F. M.; Wentzell, L. M.; Maxwell, A.
Nucleosides, Nucleotides, Nucleic Acids 2000, 19, 1249.
(186) Kwok, Y.; Zeng, Q.; Hurley, L. H. J. Biol. Chem. 1999, 274,
17226.
(187) Fan, J. Y.; Sun, D.; Yu, H.; Kerwin, S. M.; Hurley, L. H. J. Med.
Chem. 1995, 38, 408.
(188) Zeng, Q.; Kwok, Y.; Kerwin, S. M.; Mangold, G.; Hurley, L. H.
J. Med. Chem. 1998, 41, 4273.
(189) Tuma, J.; Connors, W. H.; Stitelman, D. H.; Richert, C. J. Am.
Chem. Soc. 2002, 124, 4236.
(190) Tran, J. H.; Jacoby, G. A. Proc. Natl. Acad. Sci. U.S.A. 2002,
99, 5638.
(191) Hooper, D. C.; Wolfson, J. S.; McHugh, G. L.; Winters, M. B.;
Swartz, M. N. Antimicrob. Agents Chemother. 1982, 22, 662.
(192) Saiki, A. Y.; Shen, L. L.; Chen, C. M.; Baranowski, J.; Lerner,
C. G. Antimicrob. Agents Chemother. 1999, 43, 1574.
(193) Barrett, J. F.; Bernstein, J. I.; Krause, H. M.; Hilliard, J. J.;
Ohemeng, K. A. Anal. Biochem. 1993, 214, 313.
(194) Hooper, D. C. Drug Resist Updates 1999, 2, 38.
(195) Moelbak, K.; Gerner-Smidt, P.; Wegener, H. Emerging Infect.
Dis. 2002, 8, 514.
(196) Anderson, A.; McClellan, J.; Rossiter, S.; Angulo, F. J. The
National Academies Press: National Academies of Sciences,
Washington, DC, 2003; Appendix A, p 231.
(197) Drlica, K.; Malik, M. Curr. Top. Med. Chem. 2003, 3, 249.
(198) Hooper, D. C. Emerging Infect. Dis. 2001, 7, 337.
(199) Zhanel, G. G.; Walkty, A.; Nichol, K.; Smith, H.; Noreddin, A.;
Hoban, D. J. Diagn. Microbiol. Infect. Dis. 2003, 45, 63.
(200) Hawkey, P. M. J. Antimicrob. Chemother. 2003, 51 Suppl 1, 29.
(201) Bhatt, K.; Banerjee, S. K.; Chakraborti, P. K. Eur. J. Biochem.
2000, 267, 4028.
(202) Lomovskaya, O.; Warren, M. S.; Lee, A.; Galazzo, J.; Fronko,
R.; Lee, M.; Blais, J.; Cho, D.; Chamberland, S.; Renau, T.; Leger,
R.; Hecker, S.; Watkins, W.; Hoshino, K.; Ishida, H.; Lee, V. J.
(203) Abramycheva, N.; Govorun, V. M. Antibiot. Khimioter. 2000, 45,
14.
(204) Chatterji, M.; Nagaraja, V. EMBO Rep. 2002, 3, 261.
(205) Wang, M.; Sahm, D. F.; Jacoby, G. A.; Hooper, D. C. Antimicrob.
(206) Pan, X. S.; Yague, G.; Fisher, L. M. Antimicrob. Agents Chemother. 2001, 45, 3140.
(207) Yamagishi, J.; Yoshida, H.; Yamayoshi, M.; Nakamura, S. Mol.
Gen. Genet. 1986, 204, 367.
(208) Nakamura, S.; Nakamura, M.; Kojima, T.; Yoshida, H. Antimicrob. Agents Chemother. 1989, 33, 254.
(209) Cullen, M. E.; Wyke, A. W.; Kuroda, R.; Fisher, L. M. Antimicrob.
(210) Bast, D. J.; Low, D. E.; Duncan, C. L.; Kilburn, L.; Mandell, L.
A.; Davidson, R. J.; de Azavedo, J. C. Antimicrob. Agents
Chemother. 2000, 44, 3049.
(211) Oram, M.; Howells, A. J.; Maxwell, A.; Pato, M. L. Mol. Microbiol.
2003, 50, 333.
(212) Sreedharan, S.; Oram, M.; Jensen, B.; Peterson, L. R.; Fisher,
L. M. J. Bacteriol. 1990, 172, 7260.
(213) Tanaka, M.; Onodera, Y.; Uchida, Y.; Sato, K. Antimicrob. Agents
Chemother. 1998, 42, 3044.
(214) Dong, Y.; Xu, C.; Zhao, X.; Domagala, J.; Drlica, K. Antimicrob.
(215) Sindelar, G.; Zhao, X.; Liew, A.; Dong, Y.; Lu, T.; Zhou, J.;
Domagala, J.; Drlica, K. Antimicrob. Agents Chemother. 2000,
44, 3337.
(216) Pillai, S.; Pillai, C.; Shankel, D. M.; Mitscher, L. A. Mutat. Res.
2001, 496, 61.
(217) Vincent, J.; Teng, R.; Dalvie, D. K.; Friedman, H. L. Am. J. Surg.
1998, 176, 8S.
(218) Zhanel, G. G.; Noreddin, A. M. Curr. Opin. Pharmacol. 2001, 1,
459.
(219) Andes, D.; Craig, W. A. Antimicrob. Agents Chemother. 2003,
47, 3935.
(220) Kunin, C. M.; Ellis, W. Y. Antimicrob. Agents Chemother. 2000,
44, 848.
(221) Kashiwase, H.; Momota, K.; Ohmine, T.; Komai, T.; Kimura, T.;
Katsube, T.; Nishigaki, T.; Kimura, S.; Shimada, K.; Furukawa,
H. Chemotherapy 1999, 45, 48.
CR030101Q
Chem. Rev. 2005, 105, 593620
593
DNA and RNA Synthesis: Antifolates

Ivan M. Kompis,* Khalid Islam, and Rudolf L. Then
ARPIDA Ltd, Dammstrasse 36, 4142 Munchenstein, Switzerland
Contents
1. Introduction
2. Folate Pathway Enzymes
2.1. Function and Biochemistry
2.2. Dihydrofolate Reductase
2.2.1. Mammalian DHFR
2.2.2. Bacterial DHFR
2.2.3. Other DHFRs
2.3. Dihydropteroate Synthase
2.4. Thymidylate Synthase
2.5. Bifunctional DHFRTS
2.6. GTP-Cyclohydrolase I
2.7. Dihydroneopterin Aldolase
2.8. 6-Hydroxymethyl-7,8-dihydropterin
Pyrophosphokinase
2.9. 7,8-Dihydroneopterin Triphosphate Epimerase
2.10. Serine Hydroxymethyltransferase
2.11. Multifunctional Folic Acid Synthesis Proteins
2.12. Recent Discoveries in the Folate Pathway
3. Impact of Bioinformatics
4. New Drugs and Drugs in Development
5. Inhibitors of Folate Pathway Enzymes
5.1. Screening and Methodology
5.2. Antibacterials
5.3. Antifungals
5.4. Inhibitors of Opportunistic Pathogens
5.4.1. Dihydrofolate Reductase Inhibitors
5.4.2. Dihydropteroate Synthase Inhibitors
5.4.3. Inhibitors of Thymidylate Synthase (TS)
and Multitargeted Antifolates (MTA)
5.5. Antimalarials and Other Antiprotozoal Agents
5.6. Anticancer Antifolates
5.6.1. Classical Inhibitors of DHFR
5.6.2. Inhibitors of Thymidylate Synthase
5.6.3. Nonlassical Inhibitors of Folate Enzymes
5.6.4. Inhibitors of Folylpolyglutamate
Synthetase
5.6.5. Inhibitors of Other Enzymes
6. Resistance to Antifolates
6.1. Bacteria
6.2. Protozoa
6.3. Fungi
6.4. Cancer Cells
7. Conclusion
593
594
594
595
595
595
596
596
596
596
597
597
597
597
598
598
598
599
599
601
601
601
602
603
603
609
609
610
612
612
613
614
614
615
615
615
615
616
616
617
Present Address: Actelion Percurex Ltd., Basel, Switzerland.
8. Acknowledgment
9. References
617
617
1. Introduction
Tetrahydrofolate cofactors are essential for the
synthesis of purines, certain amino acids, and thymidine. Most bacteria and plants produce these folate
cofactors by de novo biosynthesis, although some
bacteria and mammalian cells rely on the use of
preformed folates and have salvage pathways for
reduced folates, purines, and pyrimidines. Compounds that interfere with this pathway, antifolate
agents, have found use in the clinic as antibacterials,
antimalarials, and anticancer drugs. In the past
decade, an intensive search for drugs that could be
specifically used in a variety of opportunistic infections have been undertaken.
This review covers the major progress and developments in inhibitors of the enyzmes involved in folic
acid biosynthesis from January 1995 till mid-2004.
Outstanding comprehensive reviews on antifolates
prior to this period include those of Hitchings and
Smith,1 Sirotnak,2 Blakley,3 and Rosowsky.4 The
reader is also reffered to other important reviews that
emphasize particular aspects such as the selectivity
of antifolates,5 structure-activity relationships (SAR)
of inhibitors in this area,6,7 and general aspects of
dihydrofolate reductase (DHFR) inhibitors.8-12 Antifolates as antitumoral agents recently been reviewed by McGuire and Derouin.13,14
Starting from guanosine triphosphate, six enyzmes
are involved (see Scheme 1, not all enzymes are
shown) in the biosynthesis of tetrahydrofolic acid, and
the crystal structures of all but one have been
determined.15 In addition, inhibitors of thymidylate
synthase (TS)16 and dihydropteroate synthase17
(DHPS) have also been reported and will be dealt
with in this review. In a broad sense, inhibitors of
all these enzymes fall under the term antifolates.
However, because the overwhelming majority of
antifolates are inhibitors of DHFR, very often the
term antifolates is reduced to the inhibitors of this
enzyme and these constitute the major part of this
article.
Research efforts have concentrated on the discovery
of safer or more potent compounds or both when
compared with available antifolates in the corresponding therapeutic areas. In the past decade, these
efforts have been reflected in the publication of well

Ivan M. Kompis studied chemistry at the Slovak Technical University in

Bratislava (former Czechoslovakia) and got his Ph.D. degree from the
Slovak Academy of Sciences there in 1962. The years 19631964 were
spent as a postdoctoral fellow with Prof. Edward Leete at the University
of Minneapolis, MN, working on the biosynthesis of indole alkaloids. From
1968 to 1971, he was Assistant Professor at the Institute of Organic
Chemistry, University of Zurich, Switzerland, with main interest in structure
elucidation of natural products. In the years 19711995, he was working
in Pharma Research of F. Hoffmann-LaRoche Ltd., Basel, Switzerland,
and Nutley, NJ. In the last 10 years with Roche, he acted as the Head
of the Department of Infectious Diseases. He is co-founder and former
President of ARPIDA Ltd., a company devoted to research and
development of anti-infectives, a member of SAB of KOSAN Biosciences
Ltd. in Hayward, CA, and active as a consultant for medicinal chemistry.
Kompis et al.
Rudolf Then received a diploma in Biology at the University of Mainz in

1966 and later obtained his Ph.D. in microbiology 1969. Aside from
microbiology, he focused on biochemistry and pharmacology. After
spending a short period as scientific assistant at the University of Mainz,
he joined the antibiotic research group at F. Hoffmann-La Roche in Basel
in 1970. There he spent most of his scientific career in research, interrupted
by a short sabbatical in 1975 at the Roche Molecular Biology Institute in
Nutley, NJ. His main research activities were the discovery and
development of new antibacterial dihydrofolate reductase inhibitors, their
mode of action, and synergy with sulfonamides, new -lactam antibiotics,
and -lactamase inhibitors. He was always interested in the diverse
mechanisms of resistance to these agents. DNA-gyrase inhibitors and
screening assays for new antibiotics were other fields of his activities. He
has published numerous articles in these fields and some chapters in
text books. He is currently consultant for infectious diseases at Actelion
Pharmaceuticals in Basel.
Scheme 1
Khalid Islam obtained his Ph.D. in 1983 from Imperial College, University
of London. During his Ph.D. and postdoctoral studies, he has worked on
cellular mechanisms for the regulation of proteinprotein interactions, and
as an EMBO fellow (19851987), he worked on phosphorylation
dephosphorylation mechanisms. In 1987, he joined Marion Merrell-Dow
where he coordinated multidisciplinary teams of biologists (biochemists,
microbiologists, molecular and cell biologists) in drug discovery and
preclinical development. From 1996 to 1999, he worked in Hoechst Marion
Roussel (HMR), Paris, as a group leader in drug discovery and preclinical
development of antibacterials and antifungals. In July 1999, he moved to
Basel to join Arpida, where he is currently the President and Chief
Executive Officer. He referees for several international journals and is a
member of the editorial board of Current Drug Discovery Technologies.
He holds several patents and has published over 75 articles in leading
journals.
over 500 papers and the synthesis and evaluation of

hundreds of compounds as potential inhibitors of the
folate pathway enzymes. A number of new chemical
entities, which are or can potentially be used in
cancer chemotherapy or infectious diseases have been
identified. Indeed, at least four antifolates are under
development for the treatment of various cancers.
Moreover, with the advance of the HIV epidemic and
the concomitant emergence of opportunistic infections, in recent years a major effort has been dedicated to the search for inhibitors of folate pathway
enzymes of causative agents for these infections.
Despite the fact that several other enzymes of the
folate pathway have been characterized and their
structures solved,15 DHFR (E.C. 1.5.1.3), a key enzyme in folate utilization, and its inhibitors remain
the focus of research in this area.
2. Folate Pathway Enzymes

2.1. Function and Biochemistry
All living cells need tetrahydrofolate cofactors for
the synthesis of purines, some amino acids, and
especially thymidine. The biosynthesis of tetrahydrofolate from GTP is now well established,1,2,18 and
the antifolate target enzymes are outlined in Scheme
1. Although inhibitors of these target enzymes are
the primary subject of this review, we have also
included inhibitors of thymidylate synthase (TS) and
serine hydroxymethyl transferase (SHMT) enzymes
because these are intimately associated with folate

biosynthesis in the thymidylate cycle. The general
folate pathway can be accessed via the Internet at
http://ca.expasy.org/cgi-bin/show_image?A7 and http://
ca.expasy.org/cgi-bin/show_image?L2 or in the KEGG
database, http://www.genome.ad.jp/kegg/pathway/
map/map00790.html.
2.2. Dihydrofolate Reductase

DHFR is by far the most intensively studied
enzyme in the folate pathway. It is an essential and
almost (vide infra) ubiquitous enzyme. It generates
tetrahydrofolate (THF), various cofactors of which are
involved in the transfer reactions of the one carbon
unit used in the biosynthesis of nucleic and amino
acids, including methylation of dUMP to dTMP.
DHFR inhibitors act by halting synthesis of DNA,
RNA, and proteins, thereby arresting cell growth.
DHFR is an important target for drug development
against cancer and a variety of infectious diseases
caused by bacteria, protozoa, and fungi. The wealth
of the knowledge acquired on inhibitors of this
enzyme resulted relatively earlysin the 1950s and
1960ssin the well-known anticancer, antibacterial,
and antimalarial drugs, for example, methotrexate
(MTX), trimethoprim (TMP), and pyrimethamine
(PYR) (Chart 1). The recognition of the importance
of these enzymes and the routine availability of
protein crystallography led to an explosion of information of 3-D structures of DHFRs from human,
protozoal, fungal, and bacterial sources with numerous ligands and cofactors bound in their active
centers. Over 100 coordinates of 3-D structures of this
enzyme with substrate, cofactor, and a wide variety
of inhibitors are deposited in the Brookhaven Protein
Data Bank (http://www.rcsb.org/pdb/).
In discussing perspectives of enzyme catalysis, S.
Benkovic and S. Hammes-Schiffer have recently
reviewed the kinetics of the DHFR reaction.19 R.
Chart 1. Reference Compounds
Kisliuk has studied the synergistic interactions in

mammalian cells between anticancer antifolates,
particularly inhibitors of DHFR, TS, and glycinamide
ribonucleotide formyltransferase (GARFT)20 and observed significant synergistic cytotoxicity in many
human cell lines, dependent on folate levels and
polyglutamation.
The folding mechanisms of human DHFR, DHFR
from Escherichia coli, and DHFR from Lactobacillus
casei have been studied. Despite less than 30%
pairwise sequence identities, folding to the native
state occurs via parallel folding channels, and conservation of the fast-, intermediate-, and slow-folding
events provides convincing evidence for the hypothesis that evolutionarily related proteins achieve the
same fold via similar pathways.21
2.2.1. Mammalian DHFR

In contrast to microbial DHFRs, mammalian
DHFRs are highly conserved. A number of 3-D
structures of DHFR complexes have been described.22
The active site is somewhat larger than that of
bacterial DHFRs and provides poorer specificity for
contacts between TMP and surrounding residues.23
In human DHFR, the pteridine ring of MTX binds
in an inverse orientation as compared to the bound
folate found in other DHFRs. TMP, which is remarkably species-specific, binds in a different conformation
as compared to its orientation in the E. coli enzyme,
and the trimethoxyphenyl group occupies the upper
cleft of two hydrophobic pockets. The 4-amino group
makes one hydrogen-bond, in contrast to two hydrogen bonds in the bacterial enzymes.22
Expression levels of DHFR play a significant role
in MTX resistance in certain types of acute lymphoblastic leukemia (ALL) in children. Elevated levels
correlate with MTX-resistance in T-cell ALL.24 A
study of the expression of hamster DHFR minigenes
showed that expression in CHO cells was significantly higher in the presence of DHFR intron 1. The
protein encoded by the intronless construct was also
unstable, subject to lysosomal degradation, and had
a shorter half-life, suggesting that the DHFR intron
1 plays an important role.25
2.2.2. Bacterial DHFR

Close to 60 3-D structures of bacterial DHFRs,
either as apoenzymes or complexed with folate,
NADPH, or various other ligands or inhibitors, are
present in the protein data bank. The tight binding
of TMP to the active site results from close fitting
and optimal interatomic contacts between the drug
and the enzyme. The X-ray structure of enzyme from
the causative tuberculosis agent, Mycobacterium
tuberculosis, was solved recently by Li and coworkers.9 The binary complex with NADP and ternary complexes with NADP and different inhibitors
were determined at 1.7-2.0 resolution. Despite
only 26% sequence identity with human DHFR, the
overall fold is similar, but there are significant
differences that can be exploited for the design of
specific inhibitors.
2.2.3. Other DHFRs

In recent years, the DHFR from Pneumocystis
carinii has received significant attention because this
pathogen is often associated with AIDS and other
immunodeficiencies. The 3-D structure of this enzyme has been solved.26 The active site of P. carinii
DHFR is intermediate in size between those of E. coli
and vertrebrate DHFRs. TMP has sufficient van der
Waals interactions between the trimethoxybenzyl
group and the enzyme or cofactor to stabilize the
ligand in the bacterial mode of binding.23
The nomenclature for this organism has now been
revised because this pathogen is now known to be
species-specific. Pneumocystis jiroveci now refers to
the pathogen found in human hosts, and P. carinii
or P. carinii f. sp. ratti refers to the variants found
in rats.27 Most of the published studies on Pneumocystis DHFR, including the crystal structure, used the
enzyme derived from the variant found in rats. 27
Apart from Pneumocystis, where antifolates are
clinically used both as therapy and prophylaxis, no
antifolates are used as therapy against yeasts and
filamantous fungi. Nevertheless, DHFR is considered
a valid target for antifungal drug discovery and
DHFR from Candida albicans has been crystallized.28
Complexes with several 5-arylthioquinazolines, 3,
and NADPH were analyzed and two distinct modes
of binding were reported. The most selective compounds were found to bind in an unusual mode, displacing the dihydronicotinamide portion of NADPH
from its normal position within the enzyme active
site.
2.3. Dihydropteroate Synthase

Dihydropteroate synthase (DHPS, E.C. 2.5.1.15),
the target of sulfonamides and sulfones, has received
less attention than DHFR. Nonetheless, close to 170
nucleotide sequences of the DHPS gene (folP) from
several organisms have been determined, including
those from E. coli, Staphylococcus aureus, Streptococcus pneumoniae, M. tuberculosis, and P. carinii.
More than 200 partial or complete protein sequences
have been deposited, either for a monofunctional
DHPS or as part of a larger, folic acid synthesis
protein. The DHPS from S. aureus has been crystallized and the 3-D structure solved as the apoenzyme
and a binary complex with the substrate analogue
hydroxymethylpterin pyrophosphate at 2.2 and 2.4
, respectively.29 In common with other eukaryotic
DHPSs, the enzyme is a homodimer in solution and
only a single molecule of the substrate analogue
hydroxymethylpterin pyrophosphate is bound per
dimer. The 3-D structure of E. coli DHPS was also
solved in the same year.30 Moreover, a 1.7 resolution crystal structure of the DHPS from M. tuberculosis complexed with 6-hydroxymethylpterin monophosphate is also available.31 All three enzymes
belong to the TIM barrel proteins with eight
R-helices surrounding a central barrel composed of
eight parallel -strands. To date, about 10 3-D
structures of DHPS or its complexes with ligands
have been solved.
DHPS is usually expressed as a monomeric protein,
located in the cytoplasm, although in certain organ-
Kompis et al.
isms, for example, higher plants and fungi, it can be

part of a bifunctional or trifunctional folate biosynthesis enzyme.32
2.4. Thymidylate Synthase

Thymidylate synthase (E.C. 2.1.1.45) drives the
thymidylate cycle with the consumption of N5,N10methylene tetrahydrofolate. The dihydrofolate generated has to be reduced by DHFR (see Scheme 1). The
TS cycle is the sole de novo pathway for the synthesis
of dTMP. Complete blockade of TS ultimately leads
to thymineless death.33 As a key enzyme in general
metabolism and an important target for anticancer
agents, TS has been extensively studied. Reviews on
its structure, mechanism, and inhibition,34-39 its use
as a target for chemotherapy,40 or its function as a
translational regulator41 have been recently published. Close to 200 complete or partial sequences of
TS genes and more than 100 3-D structures are
currently deposited in the Protein Data Bank. The
native enzyme is a symmetrical dimer of structurally
similar subunits. Overall folding displays a series of
eight R-helices, 10 strands of -subunits, and several
segments of coils that connect the secondary structural elements.41 The TSs studied exhibit striking
structural homologies, and so far there are no appropriate selective inhibitors known for TS from
bacteria or protozoa. However, this situation could
alter based on the recent discovery of thymidylate
synthase complementing proteins in a number of
bacteria that exhibit a different kinetic and molecular
mechanism. These new proteins could be interesting
targets for selective inhibitors (see also section
2.12.).42,43
2.5. Bifunctional DHFRTS

Apicoplast parasites, such as Plasmodium falciparum (and other Plasmodium species such as Pl.
vivax, Pl. malariae, Pl. ovale), Toxoplasma gondii,
Leishmania major, Trypanosoma cruzi, or Cryptosporidium parvum, express a bifunctional DHFRTS enzyme coded by a single gene, in contrast to the
host, where both enzymes are separate gene products. The Pl. falciparum DHFR-TS, for example, is
a polypeptide of 608 amino acids, of which the first
231 residues constitute the DHFR domain. 44 A
junction region of 89 residues separates the DHFR
domain from the TS domain, which is composed of
288 residues. This junction is absent in the bifunctional DHFR-TS enzyme of Leishmania.
As with the plasmodia, Tr. cruzi, the parasite that
causes Chagas disease, harbors a bifunctional DHFRTS enzyme. Its crystal structure has not yet been
solved, but a homology model has been used for
inhibitor design.45
The DHFR-TS from Babesia bovis, an apicomplexan parasite of cattle, was recently sequenced and
cloned.46 It contains a moderately conserved 5-end
DHFR domain (190 aa), a nonconserved linker region
(33 aa), and a highly conserved 3-end TS domain
(288 aa).
Several theoretical models have been used for
inhibitor design.47-50 These include homology modeling of the enzymes from both Pl. falciparum and Pl.
vivax.46
Recently, crystal structures of the plasmodial

DHFR-TS have been solved for the wild-type and a
quadruple mutant complexed with NADPH, dUMP,
and either PYR or WR99210.51 Despite the presence
of truncated fragments resulting from proteolysis,
these studies provide a number of interesting insights
on the junction region linking the two domains, the
structure of the DHFR domain, and the possible roles
of the two Pl. falciparum-specific insert regions, as
well as on the binding and orientation of the inhibitors. The flexible WR99210 side chain adopts a
conformation that can still bind to the mutant active
site, in contrast to the more rigid cycloguanil or
related structures. All this information would be
useful in the design of new inhibitors able to overcome PYR resistance.43
The crystal structure of DHFR/TS from Cryptosporidium hominis (previously Cr. parvum) revealed
a novel architecture of the bifunctional enzyme.52 The
unique linker domain with an 11 residue R-helix
controls the relative orientation of the DHFR and TS
domains, which is different in the apicomplexans
(with Cr. hominis, Toxoplasma, Plasmodium) and the
kinetoplastids (with Leishmania and Trypanosoma).
The tertiary structure of the linker domain has
therefore been used in the classification of protozoa.51
2.6. GTP-Cyclohydrolase I
GTP-Cyclohydrolase I (GTP-CH-I) (E.C. 3.5.4.16)
catalyzes the first step in the synthesis of dihydroneopterin triphosphate and tetrahydrofolate from
GTP in bacteria, plants, or animals. The reaction
mechanism appears rather complex and poorly understood although a number of reaction mechanisms
have been proposed.53-55 The structures of the GTPCH-I from E. coli and humans have been solved.
These studies identified the key role of a zinc ion in
human and bacterial GTP-CH-I and provide a much
better understanding of the reaction mechanism.54
In the rat enzyme, zinc binds to the conserved Cys132, His-135, and Cys-203. In Pl. falciparum, the
gene for GTP-CH-I is located on chromosome 12 along
with the genes of other folate pathway enzymes.56
As the first enzyme in folate pathway, GTP-CH-I
would constitute an interesting target for selective
inhibitors. Although the amino acids involved in
substrate binding and catalysis and the role of zinc
seems to be identical in the E. coli and the human
enzymes,54 there are sufficient differences between
these enzymes that could be exploited for the design
of selective inhibitors. For example, the sequence
identity between human and E. coli enzyme is only
37%, and the human enzyme lacks the N-terminal
region. Similarly, in contrast to the bacterial enzyme,
the mammalian enzyme, which plays a key role in
the biosynthesis of tetrahydrobiopterin, is regulated
by feedback inhibition. A ternary complex between
GTP-CH-I, tetrahydrobiopterin, and an auxiliary
protein (GFRP, GTP-CH feedback regulatory protein)
is formed.54
2,4-Diamino-6-hydroxypyrimidine is a prototypic
inhibitor of GTP-CH-1 and exerts a dual mechanism
of inhibition.57 At low concentrations, it competes
with tetrahydrobiopterin and is part of the GFRP
system, while at higher concentrations, it directly

competes with the substrate GTP. It has been widely
employed as a tool in the study of tetrahydrobiopterin, which is an essential cofactor of nitric oxide
synthase and aromatic amino acid hydroxylases.54
This compound was recently shown to exert a positive
effect in rat postburn Staphylococcus aureus sepsis.58
At present, we are not aware of any current efforts
aimed at designing more potent and selective inhibitors of this enzyme for use as antimicrobial agents.
2.7. Dihydroneopterin Aldolase

Dihydroneopterin aldolase (DHNA) (E.C.4.1.2.25)
catalyzes the conversion of 7,8-dihydroneopterin to
6-hydroxymethyl-7,8-dihydropterin. The enzyme from
S. aureus has been purified after expression in E. coli
and crystallized.59 The X-ray structure at 1.65
resolution and the binding site of 6-hydroxymethyl7,8-dihydropterin have been determined. The protein
forms an octamer of 110 000 Da molecular weight.
The crystal structure of the homo-octameric protein
has been solved, and it has been shown that the
folding topology, quaternary structure, and amino
acid sequence is similar to that of the 7,8-dihydroneopterin triphosphate epimerase.60 The vibrational
structure of 7,8-dihydrobiopterin, an inhibitor of
DHNA, has been studied by Raman difference spectroscopy61 and the stereochemistry of the reaction by
using deuterated buffer.62 The gene for DHNA in Pl.
falciparum has not yet been identified.56
Although repeatedly discussed as a potentially
interesting target, only limited efforts have been
directed toward the design of new DHNA inhibitors.
A high-throughput screen carried out in the Roche
group in the 1990s yielded a number of hits with
moderate activity, but the potency of these could not
be substantially improved. Highly functionalized
6-substituted pteridines were recently prepared as
potential modulators of tetrahydrobiopterin activity
or DHNA inhibitors, but no biological results have
been reported.63
2.8. 6-Hydroxymethyl-7,8-dihydropterin
Pyrophosphokinase
6-Hydroxymethyl-7,8-dihydropterin pyrophosphokinase (E.C.2.7.6.3, HPPK) catalyzes the transfer of
pyrophosphate from ATP to 6-hydroxymethyl-7,8dihydropterin. Because this enzyme is absent in
mammalian cells, it is a potential target for selective
antimicrobial agents, but no useful inhibitors for this
enzyme are currently known. The mechanism of
HPPK-catalyzed pyrophosphoryltransfer and the crystal structure of the E. coli HPPK 3D-structure, in
complex with one and two product molecules have
been recently described.64
2.9. 7,8-Dihydroneopterin Triphosphate Epimerase

This epimerase is not shown in the Scheme 1
because no EC-number is available yet (A. Bacher,
personal communication). The enzyme catalyzes the
epimerization of carbon 2 in the triphosphates of
dihydroneopterin and dihydromonapterin. It can also
slowly cleave the side chain in the position 6 of
several pteridines. Since a deletion mutant of E. coli

exhibited normal growth properties, the physiological
role of the E. coli epimerase remains unknown.65 The
enzyme from E. coli has been crystallized.60
2.10. Serine Hydroxymethyltransferase

Serine hydroxymethyltransferase (SHMT) (E.C.
2.1.2.1) is a pyridoxal-5-phosphate-dependent enzyme.66 SHMT reversibly interconverts serine and
glycine with THF as the one-carbon carrier. SMHT
exists as a dimer in prokaryotic organisms but forms
a tetramer from obligate dimers in eukaryotic cells.12,67
There are about 10 entries in the protein data bank
for X-ray structures of SHMTs isolated from different
bacteria, animals, or man. The enzyme is complexed
either with glycine, serine, or 5-formylTHF. The
genome analysis of M. tuberculosis indicated two
putative SHMT enzymes, SHM1 and SHM2; recombinant proteins exist as homodimers under physiological conditions. In contrast to the usual stoichiometry of 2 mol of pyridoxal-5-phosphate (PLP) per
enzyme dimer, which applies to SHM2, only 1 mol
of PLP is bound per enzyme dimer in SHM1.68 The
role of proline residues in the folding of SHMT has
been studied with the E. coli enzyme.69 The structurefunction relationship in SHMTs was recently reviewed.12
Human SHMT is considered a target for anticaner
drugs,70 but there are currently no useful potent and
selective inhibitors known. The triazine antifolate
NSC 127755 has been found to inhibit also SHMT
from myeloma cells with an IC50 of 50 nM.71 Inhibition of SHMT, in addition to DHFR, may therefore
contribute to its cytotoxic effects on tumor cells.
2.11. Multifunctional Folic Acid Synthesis

Proteins
In some organisms, a number of enzymes in the
folate pathway form a multifunctional protein. For
example in P. carinii, the folic acid synthesis protein
(Fas) contains dihydroneopterin aldolase, 6-hydroxymethyl-7,8-dihydroneopterin pyrophosphokinase, and
dihydropteroate synthase.72,73 Single amino acid substitutions in the FasAB, such as FasAB-Met23, result
in a loss of DHNA activity and the ability to form
stable tetramers.74
In Pl. falciparum, the dihydrofolate synthetase and
folylpolyglutamate synthetase form a single protein,56
as do hydroxymethyl dihydropterin pyrophosphokinase and DHPS.75 The last two steps in purine
biosynthesis in man are catalyzed by the bifunctional
enzyme aminoimidazole-4-carboxamide ribonucleotide transformylase/IMP cyclohydrolase; the crystal
structure in complex with several inhibitors has been
solved76 (see Scheme 1).
2.12. Recent Discoveries in the Folate Pathway

The observation that certain bacteria, such as
Helicobacter pylori, did not contain a ubiquitous
enzyme like DHFR has for some time puzzled
biologists (see http://www.genome.ad.jp/dbget-bin/
get_pathway?org_name)hpj&mapno)00790). The explanation for this unusual finding has been provided
Kompis et al.
recently. Myllykallio and co-workers,77 in an elegant

piece of postgenomic work, demonstrated that a
number of pathogenic eubacteria, such as Helicobacter pylori, Campylobacter jejuni, or Treponema
pallidum, and some archaebacteria lacked DHFR.
These organisms also lack the classical thymidylate
synthase gene, thyA. Instead, they express a new
protein with thymidylate synthase function, named
ThyX. ThyX and ThyA use different reductive mechanisms, ThyX being dependent on reduced flavin
nucleotides, whereas ThyA uses tetrahydrofolate.
Occasionally, both thyA and thyX were found in the
same organism, for example in M. tuberculosis; the
functional consequences remain unclear. Since there
is no corresponding gene or protein in man, ThyX
could be an attractive target for new and selective
inhibitors for a number of important pathogens.42,77,78
Archaeabacteria usually perform C1-transfer reactions either by modified folates or in the absence of
folates. By contrast to many archaea, Haloferax
volcanii, an extremely halophilic archaeon, is significantly different in that it is sensitive to TMP. Search
for a dhfr gene actually revealed two distinct dfhr
genes, hdrA and hdrB. The hdrB gene is linked to
the gene for TS in a single transcription unit, as in
Bacillus subtilis. hDHFR-1 and hDHFR-2 share
about 38.8% amino acid sequence identity and 56%
identity at the nucleotide level. They differ considerably in stability and pH-optimum. The hdrB gene
alone can support the growth of H. volcanii in
minimal medium, whereas hdrA can support growth
only in the presence of thymidine.79 Another archaeon, Thermus thermophilus, also lacks a classical
DHFR (no dyrA gene was detected in its genome).
Instead, a dihydropteridine reductase (DHTt), related
to other short chain dehydrogenase/reductases (SDR)
is present. It is insensitive to inhibition by MTX and
TMP and displays considerable DHFR activity (at
20% of the DHPR activity detected with qPtH2). 80
It has been known for some time that folA-containing deletion mutants of E. coli can grow in the
presence of thymidine. This implied that another
enzyme could probably carry out the de novo synthesis of tetrahydrofolate. A candidate for this function was identified in E. coli recently. Once more, this
enzyme is a member of the SDR family, related to
the trypanosomatid pteridine reductases, and is able
to reduce dihydrobiopterin and dihydrofolate. This
enzyme is resistant to TMP but sensitive to inhibition
by MTX. The gene coding for this enzyme, ydgB, was
renamed folM, and the protein called FolM (Table
1).81 A BLAST search with the folM sequence as the
query identified a number of bacteria, showing that
this gene is widespread.
Trypanosomatid protozoans depend on exogenous
pteridines or folates for growth, and these essential
nutrients are accumulated by a specific folate and a
specific biopterin transporter.82 A broad spectrum
pteridine reductase, PTR1, was recently discovered,
which is able to reduce both pteridines and folate.83
The enzyme is essential and related to SDR and
forms tetramers from 30-kDa subunits. Compared to
DHFR-TS, it is less sensitive to inhibition by MTX.
The ability of different folate and pteridine substrates
Table 1. New Folate Pathway Enzymes

enzyme
E.C. number
host
FolM
E.C. 1.5.1.34
E. coli, other bacteria
PTR1
E.C. 1.5.1.33
Leishmania major, Crithidia,

other trypanosomatid
protozoans;
QDPR
E.C. 1.5.1.34
DHTt
E.C. 1.5.1.34
Leishmania major, T. brucei,

T. cruzi
Thermus thermophilus
hDHFR-2
E.C. 1.5.1.3
Haloferax volcanii
ThyX
E.C. 2.1.1.148
H. pylori, Cam. jejuni,

B. burgdorferi,
some archaea
to support growth of Leishmania correlated with

their substrate properties for PTR1.84 A separate
enzyme, a quinoid-dihydropteridine reductase (QDPR),
is found to maintain the H4-biopterin pool; it is
resistant to common antifolates.85
Gene expression of several folate pathway enzymes
in eukaryotes, such as DHFR, SHMT, and TS is
controlled by translational autoregulation. Besides
binding its cognate mRNA, TS is found to bind other
mRNAs.41 Regulation is different in DHFR from Pl.
falciparum and man, and this contributes to selectivity for inhibitors.86
function
dihydropteridine reductase/dihydrofolate
reductase; insensitive to TMP,
sensitive to MTX
broad spectrum
pteridine reductase
essential in pteridine salvage,
reduces folates and pteridines;
resistant to MTX
regeneration of H4-biopterin
dihydropteridine reductase/dihydrofolate
reductase; insensitive to TMP and MTX
2nd DHFR besides hDHFR-1;
can maintain THF-pool
but not recycle thymidine
alternative thymidylate synthase,
using reduced flavin as reductant
ref
81
83, 84
85
80
79
77
Chart 2. Marketed Antifolates
3. Impact of Bioinformatics
The number of sequenced genomes from bacteria,
parasitic protozoa, and other human pathogens is
constantly growing at a rapid pace. There are currently around 170 microbial genomes accessible in
public databases, such as EMBL-EBI, GenomeNet,
KEGG, or the NCBI Entrez Genome Data Base. This
information allows easy comparison of the degree of
conservation, the differences between taxa, and differences between host and parasite enzyme and thus
greatly improves the drug discovery process. Detection of new enzymes and their function, as listed
under 2.11 and 2.12, was greatly aided by applying
these tools.
Most important information for drug discovery,
however, is deduced from three-dimensional structures of the targeted proteins. They reveal information on inhibitor binding, conformational changes,
enzyme-inhibitor-cofactor complexes, and exploitable differences between the parasite and host enzyme.
4. New Drugs and Drugs in Development

In recent years, DHFR as drug target for new
antimicrobial agents has received little attention in
big pharmaceutical companies. A number of academic
institutions, however, actively pursue antifolate
projects.87 In addition to the established antifolate
drugs (Chart 1), new antifolates have reached the
market recently (Chart 2). Several new investigational drugs are at different stages of development
(Chart 3). The development status of several new

drugs against malaria has been reviewed in 2003.88
Antifolates in clinical development for treatment of
cancer until 1997 have been reviewed by Takimoto.89
More recent reviews on new antifolates in development include those by Purcell and Ettinger 90 and
McGuire.13
Pemetrexed. Pemetrexed disodium (Alimta,
LY231514) is a novel antifolate for use in oncology.
Several tumor types were found to respond to
pemetrexed in clinical trials, such as mesothelioma,
non-small cell lung cancer, and colon, pancreatic, and
breast cancers. Its properties have recently been
reviewed.91-94 It is called a multitargeted antifolate
(MTA) because it inhibits de novo pyrimidine and
purine pathways by TS, DHFR, glycinamide ribonucleotide formyltransferase, and aminoimidazole
carboxamide ribonucleotide formyltransferase. It becomes fully active after polyglutamation, the polyglutamate being 60-fold more active against TS than

Chart 3. Investigational Drugs
the monoglutamate, whereas polyglutamation has no

effect on its activity against DHFR. The compound
was developed by Eli Lilly and approved in February
2004 by the FDA as the first treatment of malignant
pleural mesothelioma, a condition usually associated
with asbestos exposure. It is used in combination
with cisplatin.95
Raltitrexed. Raltitrexed (Tomudex, ZD1694) is a
selective TS inhibitor developed by AstraZeneca. It
is transported into the cells by the reduced folate
carrier where it is extensively polyglutamated. The
compound has successfully completed clinical trials
and was first launched in the U.K. in 1996 and is
available in a number of countries for the intravenous
treatment of colorectal cancer. Raltitrexed extends
the range of tumors that are responsive to antifolates
as MTX is ineffective in colon cancer. A. L. Jackman
et al. have recently reviewed the development of
Tomudex (ZD1694).96
Trimetrexate. Trimetrexate (NeuTrexin) has been
approved for the treatment of P. carinii pneumonia
in 1993. It is used as the glucuronate salt for
intravenous application. Due to its inherent toxicity co-administration of leucovorin is mandatory
(NeuTrexin product information).97
Piritrexim. Since 1998, piritrexim isethionate has
been granted an orphan drug status for the treatment
of infections caused by P. carinii, T. gondii, and
Mycobacterium avium-intracellulare. Clinical trials
Kompis et al.
in several cancers, such as MTX-resistant tumors or

bladder carcinomas, are currently ongoing.98,99
ZD-9331. ZD-9331 is a nonpolyglutamable TS
inhibitor in development by AstraZeneca for treatment of solid tumors. It is a potent inhibitor of
various cancer cell lines at submicromolar concentrations.100 In contrast to raltitrexed, it freely effluxes
across the plasma membrane. It may have advantages over raltitrexed in ovarian cell lines expressing
low folylpolyglutamate synthetase (FPGS) mRNA.
Both intravenous and oral formulations are in clinical
trials. The NDA submission is foreseen in 2004.
PT 523. A side chain modified analogue of aminopterin, NR-(4-amino-4-deoxy-pteroyl-N-hemiphthaloyl-L-ornithine (PT 523) has been synthesized in
Rosowskys group. It is 10-100-fold more potent than
MTX against a large number of human cancer cell
lines in culture. Its affinity for the transporter
responsible for uptake of folates (RFC) is about
10-fold higher than that of MTX. Because PT 523
lacks a glutamate side chain, it is not a substrate for
FPGS and cannot be polyglutamated once it enters
the cell. The compound is in Phase I/Phase II clinical
trials.101,102
Lometrexol. Lometrexol (DDATHF) is an antipurine antifolate that selectively inhibits GARFT. It
is a good substrate for FPGS. It is in clinical studies
as an anticancer agent, for example, against melanomas, renal cell carcinomas, and other cancers. Due
to high accumulation after polyglutamation, it exhibits considerable systemic toxicity, which could be
reduced by oral coadministration of folic acid or
folinic acid.103 This and other similar compounds have
been reviewed by Purcell and Ettinger.90
Brodimoprim. This close analogue of trimethoprim was developed for single therapy for respiratory
tract infections.8,11 In contrast to TMP, it has a long
elimination half-life of about 30 h, allowing once daily
treatment. The compound has been commercialized
in 1993 but did not find wide acceptance. In 2000, it
was withdrawn from the market.
Iclaprim. Iclaprim (ICL, AR-100), a novel diaminopyrimidine, resulted from a program aimed at new
broad-spectrum DHFR inhibitors with increased
potency against Gram-positive bacteria at F. Hoffmann-LaRoche in Basel. The compound has been
licensed and is under development by ARPIDA Ltd.
ICL exhibits excellent activity against staphylococci,
including most of the TMP-resistant variants.104 The
antimicrobial properties and mode of action of ICL91
have been first presented at the Interscience Conference on Antimicrobial Agents and Chemotherapy in
2002 and 2003 and later published.105 The agent is
targeted as monotherapy for serious hospital infections, particularly those by methicillin-resistant staphylococci. ICL also demonstrates good activity against
respiratory tract pathogens, for example, pneumococci, including penicillin-resistant strains. In 2003,
it successfully completed Phase II clinical trials for
complicated skin and skin structure infections in
hospitalized patients.
Epiroprim. A diaminopyrimidine with attractive
properties for the treatment of Gram-positive infections or infections by several opportunistic proto-
zoa,106,107 epiroprim is one of the most active DHFR

inhibitors against Mycobacterium ulcerans and exerted synergism when combined with dapsone.108
Epiroprim is also synergistic with dapsone against
Mycobacterium leprae in vivo.109 To the best of our
knowledge, there is no development work ongoing
with this compound.
Pyrimethamine/Dapsone (Maloprim). This combination is used together with chloroquine for prophylaxis against malaria in certain areas with a high
risk of chloroquine resistance.110
Chlorproguanil/Dapsone (Lapdap). This synergistic combination of a DHFR inhibitor, chlorcycloguanil, and a DHPS inhibitor, dapsone, has undergone clinical trials in Africa and has been approved
in the U.K. in 2003. The development of a fixed triple
combination, chlorproguanil/dapsone/artesunate, is
under way.88
WR 99210. The DHFR inhibitor WR 99210 has
been developed as a promising antimalarial agent as
far back as 1973. Clinical evaluation of this drug has
been hampered by its gastrointestinal intolerance.111
DHFR for determining the selectivity.112 Similar

approaches have been used by other groups, for
example, A. Rosowsky, et al., using DHFRs of T.
gondii, P. carinii, and M. avium. Simultaneous
testing of new inhibitors was greatly improved with
an elegant genetic assay in which the DHFR from
Saccharomyces cerevisiae has been replaced by the
DHFRs from Pl. falciparum, T gondii, P. cariniii,
Cryptosporidium parvum, or humans.113,114
DHFR from M. tuberculosis is also considered an
important target for drug design.115 Since M. tuberculosis is a slowly growing microorganism and dangerous to handle, a simple in vitro screen with Sac.
cerevisiae expressing the M. tuberculosis DHFR has
been used for the selection of new DHFR inhibitors.
Recombinant wild-type and mutant Pl. falciparum
DHFR, derived from a synthetic gene of the DHFR
domain of the bifunctional pfDHFR-TS, has been
used for the screening of new inhibitors with activity
against Pl. falciparum.116
5. Inhibitors of Folate Pathway Enzymes
In recent years, only limited efforts have been

directed at DHFR inhibitors for common bacteria.
Kuyper et al. have synthesized conformationally
restricted analogues of trimethoprim, designed to
mimic the conformation of the drug observed in its
complex with bacterial DHFR. The restriction has
been achieved by linking the 4-amino group with the
methylene of TMP by one- and two-carbon bridges.
Of the three analogues prepared, 2-amino-4-methyl5-(3,4,5-trimethoxyphenyl)-7H-pyrrolo[2,3-d]pyrimidine shows good activity and inhibits four bacterial
DHFRs with a potency similar to TMP. However, it
is significantly less selective than the reference
compound.117 Efforts aimed at identification of selective and potent DHFR inhibitors for Gram-positive
bacteria, particularly S. aureus, yielded iclaprim,
which is much more active than TMP against both
wild-type and TMP-resistant S. aureus strains (see
Chart 3, Investigational Drugs). The development of
even more potent and selective inhibitors failed,
largely because of poor solubility and high protein
binding of these inhibitors.87
The design of antifolates with activity against
protozoal pathogens, often results in interesting
activities against the Mycobacterium avium complex
(MAC).118 M. avium as a target organism for new
DHFR inhibitors has also been pursued in a novel
approach to use pharmacophores in a series of 2,4diamino-5-deazapteridines.119
Mycobacterium tuberculosis and particularly the
multiresistant strains pose emerging threats to public health in many countries, and new drugs are
urgently needed. No DHFR inhibitor is currently
used for tuberculosis treatment. Epiroprim has recently been evaluated against various M. tuberculosis
strains and been found to exhibit weak activity. It
exhibits synergy with isoniazid (INH) in INH/
rifampicin-sensitive strains and prevents development of INH resistance. More potent inhibitors of the
M. tuberculosis DHFR would be needed to become
useful chemotherapeutic agents. The triazine DHFR
inhibitor WR99210 has been shown to exhibit rea-
All DHFR inhibitors exhibiting IC50s in the micromolar range or less contain the 2,4-diamino-1,3-diaza
pharmacophore 1. Therefore, a primary classification
of antifolates according to their structures appears

impractical. The authors have therefore opted to
organize their efforts primarily on the search for
inhibitors of a potential target, medical indication,
or pathogenic organisms and only then in terms of
the structure of the compounds. Therefore throughout the current review, after the description of the
general aspects of screening and methodology, the
new antifolates will be presented according to their
targeted medical utility.
In the literature, a distinction is frequently made
between classical and nonclassical antifolates. Classical antifolates are structural analogues of the
substrate folic acid and thus bear an acidic moiety
in the distal part of the molecule. They are, as a rule,
substrates for folylpolyglutamate synthetase (FPGS).
Nonclassical antifolates are lipophilic analogues of
the substrate molecule, lacking the glutamate portion, or its surrogates, of the folate molecule. They
are, therefore, not substrates for FPGS.
5.1. Screening and Methodology

It seems to be an economic and logical procedure
during design and synthesis of new antimicrobial
antifolates to test them against multiple enzymes
rather than a single enzyme or organism. This
approach has been used by the Roche group, evaluating potential new antifolates against the wild-type
and resistant E. coli and S. aureus DHFR, the P.
carinii DHFR, a pneumococcal DHFR, and human
5.2. Antibacterials
sonable in vitro activity against M. tuberculosis, as

well as against other mycobacteria. This structural
class has been further pursued using a genetically
modified Sac. cerevisiae strain for screening. 115 The
selectivity of WR99210, however, is not sufficient for
use as an antimicrobial agent because it shows poor
gastrointestinal tolerance. A series of analogues have
been synthesized and tested and show activity that
is comparable to the lead molecule.
The three-dimensional structure of the M. tuberculosis DHFR complexed with several inhibitors has
been solved, and the enzyme has been characterized.9,120 Although the general fold of the protein is
essentially the same as that of the human enzyme,
a number of differences were detected, which could
be exploited for inhibitor design.
A diaminopyrimidine covalently linked to dapsone,
named K-130, has been shown to potently inhibit the
DHFR from Mycobacterium lufu, a model organism,

and to exhibit good in vitro and in vivo activity
against M. lufu, and M. leprae.121
A series of 2,4-diamino-5-deazapteridine derivatives122,123 exhibit significant activity against recombinant M. avium DHFR with IC50-values in the
nanomolar range and correlating in vitro activities
with minimum inhibitory concentrations (MICs) below 0.13 g/mL. Many of these possess selectivity
ratios of >100 against the human DHFR. However,
these compounds are significantly less potent against
M. tuberculosis in vitro. Although these data are
encouraging, an improved selectivity ratio would be
required for therapeutic applications.
A new approach for the discovery of inhibitors of
DHNA as potential antibacterial agents has been
undertaken by scientists at Abbott.124 The S. aureus
DHNA has been used to screen a library of 10 000
compounds by directly diffusing the compounds into
the crystals. In this way, several hits, both purines
and pyrimidines, have been detected and inhibited
DHNA at concentrations of 28-80 M. Structurebased design has been used to further improve the
potency of these hits. The most active compound, a
dichlorobenzyl-substituted azapurine derivative, 2,
Kompis et al.
Table 2. Inhibition of DHFR and in Vitro Antifungal
Activity of Reference Compounds127
compd
TMP
PYR
TMX
PTX
3a
3b
3c
3d
3e
3f
3g
DHFR IC50 (M)

C. albicans
human
50
5.0
0.04
0.004
0.05
0.13
0.03
0.05
0.06
0.057
0.008
490
2.6
<0.001
0.002
>10
70
3.1
2.0
0.32
0.82
2.0
selectivity
ratio
C. albicans
MIC (g/mL)
10
0.5
<0.03
0.005
>200
540
100
40
5.0
14
250
>50
>50
>50
>10
1.0
>50
0.25
0.25
2.5
1.0
0.10
Haemophilus influenzae. It has been suggested that

these compounds were not potent enough to compete
successfully with the intracellular substrate levels,
but the reasons for this lack of in vitro activity
remain to be explained.
In contrast to earlier findings recent studies on the
mechanism of action of sulfonamides in yeast indicate
that sulfa-analogues formed via DHPS are active and
compete with dihydrofolate. The precise in vivo target
of sulfa-dihydropteroate remains to be determined,
but this finding may lead to novel antifolate inhibitors.125,126
5.3. Antifungals
So far, no effective antifungal agent based on
inhibition of DHFR is used in therapy. The known
inhibitors are neither sufficiently potent nor sufficiently selective.
The search for DHFR inhibitors as antifungal
agents has been undertaken in the former Wellcome
Research Laboratories.127 Several compounds belonging to the class of 5-(arylthio)-2,4-diamino-quinazolines have been identified as potent inhibitors of C.
albicans DHFR. The compounds are up to 540-times
less active against human DHFR, and most of the
selected compounds are also good inhibitors of C.
albicans cell growth (Table 2). The most selective
inhibitor, compound 3b, though showing a good
selectivity index of 540 shows a poor MIC of >50 g/

mL against C. albicans.
A series of 7,8-dialkylpyrrolo[3,2-f] quinazolines, 4,
have been evaluated as inhibitors of C. albicans and
exhibited an IC50 of 68 nM against the enzyme. None

of the compounds, however, inhibit common bacteria
in vitro, including hypersusceptible strains of E. coli
and Acr- efflux pump mutants of E. coli and
Table 3. Biological Data for 7,8-Dialkylpyrrolo[3,2-f]

Quinazolines117
compd
C. albicans
DHFR Ki (nM)
human DHFR
Ki (pM)
C. albicans
MIC (g/mL)
4a
4b
4c
4d
4e
4f
0.16
0.12
0.22
0.22
0.0071
0.030
10
<2.0
6.4
4.5
0.4
0.3
0.05
0.05
0.025
0.001
0.025
0.025
human DHFR and for growth inhibition of fungal and

human cells.117 Several compounds displayed exceptional, albeit nonselective, affinity for C. albicans
DHFR and cell growth inhibition. The values for six
compounds, 4a-4f, out of 21 tested are presented in
Table 3. The level of growth inhibition did not
correlate with the inhibition of DHFR. Compound 4f
is also a potent inhibitor of P. carinii and T. gondii
DHFR and has also been tested in vivo against P388
leukemic cells and fungal and P. carinii infection in
mice.
In an attempt to exploit the synergism observed
when both DHFR and DHPS are inhibited, new 2,4diaminopyrimidines and 4-substituted 4-aminodiphenyl sulfones have been prepared and tested
against enzyme systems from C. albicans.127,128 A
representative compound, HH-136, inhibits C. albi-
cans DHFR with an IC50 of 0.031 M. For at least

some of the derivatives, the differences in the activities against enzyme and whole cells are in part due
to active efflux systems.
5.4. Inhibitors of Opportunistic Pathogens

5.4.1. Dihydrofolate Reductase Inhibitors
Since the outbreak of the AIDS epidemic and the
recognition that opportunistic infections are the
premier cause of morbidity and mortality, major
efforts have been undertaken from the early 1990s
onward to find novel, potent, and selective inhibitors
of the DHFRs of the causative organisms. Two
groups, namely, those of Andre Rosowsky and Aleem
Gangjee, a selected list of their publications in this

area can be found in refs 128 and 129, made major
contributions to the developments in this area.
The frequently targeted pathogens are the fungus
Pneumocystis carinii the protozoa Cryptosporidium
parvum, Leishmania ssp., Trypanosoma cruzi, and
Toxoplasma gondii. Some inhibitors of DHFR from
the bacterial pathogens Mycobacterium avium and
M. tuberculosis have also been reported.
Low doses of co-trimoxazole are used for the treatment and prophylaxis of Pneumocystis pneumonia,
the most common opportunistic infection.130 One of
the active ingredients, TMP, is a selective but rather
weak inhibitor of P. carinii DHFR. It is, therefore,
used in a combination with sulfamethoxazole. An
alternative treatment with piritrexim, trimetrexate,
or both,129 both potent inhibitors of the target enzyme128 but lacking selectivity for the mammmalian
DHFR, requires concomitant and expensive rescue
therapy with leucovorin. Thus, the goal of the research efforts in this area is the design and synthesis
of antifolates for use against one or more of the abovementioned organisms that are potent enough not to
require co-administration of a sulfa drug, as is the
case with TMP, and selective enough not to require
leucovorin rescue.
The compounds synthesized have been, as a rule,
tested for their inhibitory activities against P. carinii
and T. gondii and rat liver DHFRs and less frequently against DHFRs from other organisms mentioned above. Because the procedures for the determination of inhibitory activities are standardized, the
IC50 values, although extracted from different sources
as summarized in the tables, are comparable. In the
following part, representative structures of the inhibitors are presented. Within these groups, the
values of the selected compounds with the best
potency, selectivity, or both are shown in the tables.
The IC50 values and selectivity indices of the
reference compounds used throughout this section
are presented in Table 4. The values shown in Table
4 reflect the previously mentioned problem of potency
versus selectivity. In addition, MTX and other inhibitors with an acidic side chain fail to enter the
bacterial cell.124,127,128
Studies performed in a rat model of dual infection
with P. carinii and T. gondii in which TMP combined
with sulfamethoxazole and PYR in vitro and in vivo
have shown that the biguanide PS-15 and epiroprim
were effective against both P. carinii and T. gondii.133
Table 4. Inhibition of P. carinii, T. gondii, M. avium, Rat, and Human DHFRs by the Reference Compounds
IC50 (M)
compd
P. carinii
PTXb
TMXb
TMPb
ICLd
MTXe
epiroprimf
PYRg
pemetrexedh
a
0.031
0.042
12
2.4
0.011
2.6
3.65
c
T. gondii
selectivity
rat liver
M. avium
0.004
0.016
2.7
0.001
0.003
133
0.011c
0.022
0.47
0.39
0.0002
0.006
170
2.30
0.01c
0.3d
human
rl/pc
rl/tg
rl/ma
>300
>300
0.022
0.05
0.07
14
>125a
20a
12.8
0.63
0.1
0.3
65
5.4c
5.3
610
0.041
1
70.6
5.9
4100
0.002
d
Human. Reference 202. Reference 158. Reference 105. Reference 164. Data on file, F. Hoffmann-La Roche AG. g Reference
138. h Reference 16.
Kompis et al.
Table 5. Inhibition of P. carinii, T. gondii, and Rat

DHFRs with Compounds 6
IC50 (M)
selectivity ratio
compd P. carinii T. gondii rat liver

PYR
6a
6b
6c
6d
3.65
20.4
12.9
3.0
0.84
0.39
1.2
16.4
1.4
0.41
2.3
6.1
11.4
3.1
1.2
rl/pc
rl/tg
ref
0.63
0.3
0.9
1.0
1.4
5.9
5.1
0.7
0.5
2.9
138
155
155
155
155
Several reviews dealing with partial aspects of these

efforts have been published.84,134,135
Substituted Monocyclic 2,4-Diaminopyrimidines and 4,6-Diamino-1,2-dimethyl-1-(X-phenyl)-s-triazines. In a quantitative structure-activity
relationship (QSAR) study, the inhibitory activities
of 60 2,4-diaminopyrimidines and 4,6-diamino-1,2dimethyl-1-(X-phenyl)-s-triazines, 5, versus purified,
pcDHFR in the low micromolar range (Table 6).

Compound 7a (TAB) is one of the most selective
inhibitors of the P. carinii enzyme.138 The crystal
structure and modeling studies on pcDHFR-cofactor
complexes with TAB have been reported.139 The
overall structure of the ternary complex is similar to
that observed for other antifolates. The most notable
feature of the binding orientation of TAB in this
complex is that its binding is disordered in such a
way that there are two alternative positions for the
binding of the benzyl and acetyloxy groups.
In an effort to discover more active and selective
antifolates, a series of 2,4-diamino-5-(2-methoxy-5substituted)-benzylpyrimidines containing a carboxyl
group at the distal end of the 5-substituent have
been synthesized and tested. Based on the analysis
of the structure of pcDHFR, compounds 8a-8h have
recombinant pcDHFR were analyzed, and their activities were compared with inhibition of DHFRs
from different sources, including recombinant human
DHFR.136 Although it is difficult to draw definitive
conclusions from this and a similar type of work,137
the study indicates that such comprehensive analysis
could be helpful in identifying potent and potentially
selective antifolates as therapeutic agents.
Ten previously untested 1-aryl-4,6-diamino-1,2dihydro-s-triazines, 6, were assayed. Both their activ-
ity and selectivity were poor (Table 5).

SAR data of a series of triazenyl-pyrimethamine
derivatives, 7a-7d, exhibited IC 50 values against
been designed such that their -carboxyl group could

interact with the basic residues of Arg-75 and Lys37 in the active site of the enzyme.128,140,141 These
compounds are about 1 order of magnitude less
Table 6. Inhibition of P. carinii, T. gondii, M. avium, and Rat DHFR by Substituted 2,4-Diaminopyrimidines
IC50 (M)
selectivity ratio
compd
P. carinii
T. gondii
M. avium
rat liver
rl/pc
rl/tg
PTX
7a (TAB)
7b
7c
7d
8a
8b
8c
8d
8e
8f
8g
8h
9a
9b
9c
9d
9e
0.013
0.17
0.68
0.26
0.053
0.054
0.15
0.13
0.028
0.87
1.20
0.001
1.2
0.27
0.27
0.53
0.58
0.10
0.004
0.69
0.69
2.01
0.001
0.003
19.4
7.35
7
0.28
4.6
4.1
4.0
2.2
25
21
5
88
0.029
0.038
0.008
0.060
0.013
0.26
114
10.87
27
5.36
85
27
31
79
28
17
5000
73
0.01
0.01
0.01
0.01
0.13
0.76
28
3.66
0.11
0.0084
0.097
0.032
0.072
2.0
0.034
0.042
0.0018
0.0031
0.037
0.0043
0.0062
3.9
0.058
0.016
0.004
0.008
0.041
0.060
0.002
0.082
1.4
42
490
41
69
340
11
150
210
1.61
1.23
2.03
1.4
2.1
rl/ma
ref
5.4
128
138
138
138
138
128
128
128
128
128
128
141
141
142
142
142
142
142
2.01
79
260
910
280
590
340
2100
11000
Table 7. Inhibition of P. carinii, T. gondii, M. avium, Rat Liver, and Human DHFRs by 2,4-Diaminopyrimidines
with Fused Five-Membered Rings
IC50 (M)
selectivity ratio
compd
P. carinii
T. gondii
M. avium
rat liver
PTX
10a
10b
10c
10d
10e
10f
11a
11b
11c
12a
12b
12c
12d
12e
12f
12g
12h
13
0.013
0.6
7.7
0.90
0.035
>4.0
8.3
1.8
1.3
7.5
45.7
35.3
0.038
0.044
11.1
29.0
72
0.77
19.5
0.0043
11.6
45.4
0.70
19.8
>4.0
>3.9
0.14
0.14
26
1.70
1.4
0.21
0.15
2.60
3.3
14
0.037
6.7
0.0006
0.003
12.3
137
1.3
0.4
>37
25.6
0.51
0.22
10
156
14.4
0.044
0.06
16.7
9.6
52
14
67
0.067
human
0.45
0.22
>26
>25
0.26
252
potent against pcDHFR when compared with PTX

but exhibit higher selectivity (Table 6). Against
tgDHFR several compounds show both higher potency and selectivity than the reference compound.
Analogues of TMP and brodimoprim, designed with
the same rationale,131,132 have been shown to be
inactive in vitro and in vivo against E. coli and L.
casei, respectively. For recently synthesized compounds, it it remains to be seen whether they are
taken up into intact cells.
Several derivatives of 2,4-diamino-5-[4-(substituted)3-nitrophenyl]-6-ethyl-pyrimidines, 9a-9e, have been
synthesized and evaluated as inhibitors of P. carinii

and T. gondii DHFR.142 The compounds exhibit
potent inhibitory activity against the latter enzyme,
whereas all are relatively weak inhibitors of the
former one (Table 6). These compounds exhibit little
or no selectivity based on their activity on the rat
liver enzyme.
Bicyclic 2,4-Diaminopyrimidines with Fused
Five- and Six-Membered Rings. Antifolates 10a10f containing a furo[2,3-d] pyrimidine ring system
have been synthesized as potential dual inhibitors
rl/pc
rl/tg
rl/ma
ref
0.026
18.9
17.8
1.44
11.4
0.077
1.1
3.02
1.85
0.02
0.55
146
144
144
143
143
143
143
145
145
152
148
148
148
147
147
146
146
146
149
0.28
0.17
1.4
3.4
0.4
1.2
1.36
1.50
0.3
0.72
5.4
13
3.64
1.57
0.39
92
10.3
0.21
0.40
6.42
0.3
0.72
2.60
38
0.69
0.78
35
of DHFR and TS.143 The compounds have been tested

against P. carinii and T. gondii DHFR, as well as
against recombinant human and L. casei DHFR
(Table 7). The classical analogues 10c and 10d have
also been evaluated as inhibitors of TS, glycinamide
ribonucleotide formyltransferase, and 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase. They are inactive against these enzymes.
Nonclassical 2,4-diamino-5-(substituted)furo[2,3-d]
pyrimidine antifolates with a variation X in the
bridge region and an appended substituted aromatic
ring have been designed such that the aromatic ring
could specifically interact with Phe-69 of pcDHFR.
Compounds 10a and 10b show significant potency
and selectivity for the target enzyme. The X-ray
structure of 10a with pcDHFR has further confirmed
the design rationale.144
The synthesis and inhibitory properties of (R,S)2,4-diamino-5-[(3,4,5-trimethoxyphenyl)alkyl]-6,7-dihydro-5H-cyclopenta[d]pyrimidines 11a and 11b as
anologues of TMP have been reported. Unlike TMP,

both compounds were better inhibitors of rat liver
enzyme than of microbial enzymes145 (Table 7).
Classical and nonclassical 2,4-diamino-5-(substituted)methyl)pyrrolo[2,3-d] pyrimidines 12a and 12e
Kompis et al.
Table 8. Inhibition of P. carinii, T. gondii, M. avium, and Rat Liver DHFRs by 2,4-Diaminopteridines and
2,4-Diaminoquinazolines
IC50 (M)
compd
P. carinii
T. gondii
14a
14b
14c
14d
14e
14f
14g
14h
14i
14j
15a
15b
15c
16a
16b
16c
16d
16e
16f
16g
0.082
0.2
1.1
2.6
0.41
0.28
1.6
12
5.3
0.14
0.092
0.12
0.15
4.6
0.095
0.30
0.114
0.502
0.171
0.10
0.028
0.033
1.0
0.46
0.013
0.0064
0.017
0.054
0.007
0.015
0.017
0.01
0.022
0.023
selectivity ratio
M. avium
0.09
0.007
with various substitution patterns of the phenyl ring

have been prepared to better understand the effect
of the bridge variations, in particular of N-methylation on one hand and the effect of introducing of 4L-glutamate substitution on the inhibitory potency
on the other hand. These studies show that Nmethylation does not greatly influence the parameters measured but the introduction of a glutamate
moiety increases the potency by several orders of
magnitudes (Table 7). The substances with amine
and sulfide groups in the bridge region are almost
equipotent.146-148
Compound 13 is the most active compound of a the
series of 22 2,6-diamino-8-substituted purines designed as TMP analogues in which rotation around

the two flexible bonds of TMP, linking the pyrimidine
ring and the side chain phenyl ring, was restricted
by incorporation into a purine ring system. It is as
potent as TMP with a selectivity ratio of 13 for P.
carinii and 38 for T. gondii DHFR.149 (Table 7)
Structural data for human and pcDHFR inhibitor
complexes have been corroborated in V. Codys group
for a number of antifolates including TMP, MTX, and
folate,150 as well as for a number of novel classical
and nonclassical furopyrimidine antifolates144,151 and
TAB.139 Analysis of quinazoline and pyrido[2,3-d]
pyrimidine N9-C10 reversed bridge antifolates in
complex with NADPH+ and pcDHFR152 and with PT
653153 has also been performed. Taken together, these
data show that despite a number of residue changes
compared with human DHFR, the key structural
features in the active site of pcDHFR are well
conserved. The volume in the active site is slightly
rat liver
rl/pc
rl/tg
0.32
1.1
2.7
27
5
2.3
5.9
19
7.9
0.8
0.028
0.012
0.042
0.29
0.038
0.26
0.071
0.011
0.067
0.047
3.9
5.5
2.5
10
12
8.2
3.7
1.6
1.5
5.7
0.3
1
0.28
0.06
0.40
0.9
0.62
0.22
0.39
0.50
11.4
33.3
2.7
59
2.2
1.9
2.5
5.4
5.43
17.3
4.20
11.0
3.05
2.04
rl/ma
3.1
6
ref
154
154
154
154
155
155
155
155
155
155
146
155
146
156
156
156
156
156
156
156
smaller in pcDHFR than in the human enzyme.

These small changes can enhance binding affinity
and, consequently, the selectivity for one enzyme over
the other.139
Several novel substituted diaminopteridines have
been identified and tested against P. carinii in
vitro.154 Two of them, 14f (GR 92754) and 14g,
antagonize the uptake of a folate precursor, paminobenzoic acid, and are at least 10-100 times
more active than TMP in this assay. The inhibition
and selectivity data are shown in Table 8.
Twenty-eight 2,4-diamino quinazolines 14 substituted with alkyl, halogen, or alkoxy groups, eight 2,4diaminopteridines, nine 4,6-diamino-1,2-dihydro-striazines, and five 1,3-diamino-7,8,9,10-tetrahydropyrimido[4,5-c]isoquinolines have been evaluated as
inhibitors of P. carinii and T. gondii DHFR enzymes.
Generally speaking, these compounds, as well as
compounds 15, exhibit a modest selectivity, for
example, compound 14d (Table 8).146,155
Twenty 6-substituted 2,4-diaminotetrahydroquinazolines 16 have been designed, synthesized as their
A series of 2,4-diamino-6-(substituted)pyrido[2,3d]pyrimidines 18 with variations in the X bridge
racemic mixtures, and biologically evaluated. NSubstitution was conducive to potency. As shown in
Table 8, the compounds have been significantly more
potent and selective against T. gondii DHFR, and
compound 16c shows also an exceptionally high
inhibitory activity, IC50 ) 5.4 10-8 M, against the
growth of T. gondii cells in culture. Selected analogues have been evaluated as inhibitors of tumor
cells in culture.156
2,4-Diamino-6-(benzylamino)pyrido[2,3-d]pyrimidine antifolates 17, lacking a 5-methyl substitution,
which has been shown to be important for increased

human DHFR activity, have been synthesized and
tested. They contain a reversal of the C9-N10 bridge
present in folates and most antifolates. C9-Methylated compounds 17a and 17c-17f are the most
potent inhibitors in this series (Table 9). Compound
Table 9. Inhibition of P. carinii, T. gondii, Human,
and Rat Liver DHFRs by 2,4-Diamino-6(methylamino-substituted)pyrido[2,3-d]pyrimidines157
IC50 (M)
selectivity ratio
compd P. carinii T. gondii human rat liver rl/pc rl/tg

17a
17b
17c
17d
17e
17f
17i
0.068
14.1
0.061
0.079
0.076
0.084
3.8
0.032
0.35
0.014
0.026
0.031
0.0028
0.31
0.7
76
0.53
2.9
8.5
0.014
3.3
0.033
0.030
0.072
0.057
0.35
2.1
0.23
0.5
0.4
0.9
0.7
0.09
h/tg
4.4
20
9.4 192
2.4
1.2
20.4
2.3 100
9.0 304
1.1
17f with 2,5-dimethoxy-substitution of the phenyl

ring shows a selectivity of 9.0 when compared to
rlDHFR and a selectivity of 304, when compared to
recombinant human DHFR. Compound 17f when
tested in vivo for the inhibition of T. gondii trophozoites in mice demonstrated a distinct prolongation
of survival of infected animals.157 Compounds 17g
and 17h have also been evaluated as antitumor
agents.152 Three crystal structures of a 5-methyl-6N-methylanilino pyridopyrimidine antifolate complex with hDHFR have been determined and analyzed.158
connecting the pyridopyrimidine part of the compound with the distal phenyl moiety (18a-18t) have
been prepared.144,146,158-161 Compounds 18o and 18u18z shown in the Table 10 represent examples with
the highest specificity ratio for hDHFR versus MAC
DHFR. The N-Me analogue 18j exhibits the best
balance of potency and selectivity against both
tgDHFR and pcDHFR, whereas the ethylene bridge
has a detrimental effect on the potency (Table 11).
The 10-deaza analogues are generally less potent and
selective than compounds with an amino group in
this position. The activity and selectivity of the
5,6,7,8-tetrahydro derivative of 18l (data not shown)
against pcDHFR was markedly decreased.159
The highest potency and selectivity, especially
against T. gondii, is found in the group of 2,4diamino-6-(substituted)pyrido[2,3-d]pyrimidines and
is lower for the other heterocyclic types.
The compound 18s is over a 1000-fold more selective than PTX and inhibits T. gondii cell growth with
IC50 ) 0.1 M. Several other compounds (e.g., 18r
Table 10. Inhibition of Mycobacterium Avium
Complex and Human DHFR by 2,4-Diamino5-methyl-6-(substituted)pyrido[2,3-d]pyrimidines123
IC50 (nM)
compd
MAC
human
selectivity ratio
18o
18u
18v
18w
18x
18y
18z
1.1
1.0
1.9
0.84
1.4
1.5
4.5
1100
7300
710
2300
1000
990
1200
909
7300
374
2738
714
660
293
Kompis et al.
Table 11. Inhibition of P. carinii, T. gondii, M. avium, and Rat DHFR by

2,4-Diamino-6-(substituted)pyrido[2,3-d]pyrimidines
IC50 (M)
compd
P. carinii
T. gondii
18a
18b
18c
18d
18e
18f
18g
18h
18i
18j
18k
18l
18m
18n
18o
18p
18r
18s
18t
0.013
0.210
0.013
0.084
0.44
0.35
1.30
0.17
0.038
0.042
1.5
0.24
0.19
5.0
0.011
0.044
0.03
0.34
0.29
0.001
0.036
0.0009
0.006
0.3
0.98
0.47
0.09
0.3
0.009
0.049
0.2
0.25
1.4
0.014
0.022
0.016
0.0079
0.03
selectivity ratio
M. avium
0.04

DHFR by 2,4-Diamino-6-substituted-pteridines160
selectivity
ratio
IC50 (M)
compd
P. carinii
T. gondii
rat liver
rl/pc
rl/tg
19a
19b
19c
19d
19e
9.5
6.2
3.9
21
7
0.77
6.9
0.21
10.6
1
246
22.9
0.47
21
1.9
25.9
3.7
0.1
1
0.27
319
3.3
2.2
2.2
2
with IC50 ) 0.01 M) demonstrate even lower values

in the same test (Table 11).
A large group of 2,4-diaminopteridines with a two
atom bridge and an aryl group attached to the
6-position of the heterocyclic moiety 19 (Table 12),
37 2,4-diamino-6-(substituted)pyrido[2,3-d]pyrimidines, and four 2,4-diaminoquinazolines have been

synthesized and evaluated.160,161
Seventy-seven 2,4-diamino-5-methyl-6-(substituted)pyrido[2,3-d]pyrimidines 18 have been evaluated
in vitro as potential drugs against Mycobacterium
avium complex (MAC).123 The activities of the compounds were assessed against recombinant MAC
DHFR and human DHFR. Most of these derivatives
show good activity against three strains of MAC with
MICs ranging from 0.13 to 1.3 g/mL.
Another series of 2,4-diamino-5-methyl-6-substituted-pyrido[2,3-d]pyrimidines 20 have been de-
rat liver
rl/pc
rl/tg
0.0076
0.37
0.008
0.057
6.9
4.6
1.9
0.022
1.9
0.28
0.12
1.14
0.26
12.9
0.01
0.02
0.12
0.77
0.55
0.6
1.8
0.6
0.7
15.7
13
1.46
1.29
1.3
1.2
0.63
0.23
0.9
2.58
0.94
0.5
4.0
2.3
0.03
0.85
10.3
8.9
9
23
4.7
4.04
2.44
6.3
31
3.2
5.7
1.0
9.2
0.73
1
7.5
97.5
18.3
rl/ma
ref
158
146
144
144
144
144
161
161
159
159
159
159
159
159
160
160
160
160
160

DHFR by Compounds 20 and 21
IC50 (M)
selectivity ratio
compd P. carinii T. gondii rat liver

20a
20b
20c
20d
21a
21b
21c
21d
21e
21f
21g
21h
21i
0.29
0.25
0.57
0.35
0.086
0.023
0.56
0.12
2.00
0.36
1.40
0.13
0.02
0.048
0.057
0.077
0.033
0.019
0.001
0.063
0.044
0.13
0.048
0.1
0.047
0.98
0.015
0.17
0.47
0.23
0.002
0.0004
0.52
0.052
0.52
0.086
0.43
0.026
0.32
rl/pc
rl/tg
ref
0.52
0.68
0.83
0.7
0.21
0.17
0.93
0.43
0.26
0.23
0.31
0.20
1.5
3.1
3.0
6.1
7.0
0.95
0.45
8.25
1.18
4
1.8
4.3
5.5
0.33
161
161
161
161
260
260
261
261
262
204
263
263
263
signed and synthesized as conformationally restricted

analogues of TMX so that the side chain nitrogen is
incorporated in an indoline or indole system. Compound 20d and its congeners have been designed to
investigate the role of the pyrrolo-substitution of the
phenyl ring, a group present also in epiroprim.
Conformational restriction in the form of an indoline
or an indole ring did not result in analogues with
better potency or selectivity when compared with the
previously synthesized open chain analogues (Table
13). Among these compounds, 20c also inhibits the
growth of T. gondii cells in culture.161
A novel easy access to 2,4-diaminopyrido[2,3-d]
pyrimidines 21 has been developed.146 In this study,
13 previously untested 2,4-diamino-6-(substituted
benzyl)quinazolines have also been evaluated as
inhibitors of DHFR isolated from major opportunistic

pathogens.260-263 All the compounds tested are less
active against P. carinii, T. gondii, and M. avium
DHFR than the reference compounds PTX and MTX
(Table 13). The modest gain in selectivity was achieved
at the cost of decreased potency.146
5.4.2. Dihydropteroate Synthase Inhibitors

Sulfonamides are well-known inhibitors of the
folate pathway enzyme DHPS(EC 2.5.1.15) and have
been used clinically for over 60 years. Today, the best
clinical utility as antibacterial agents is their combination with TMP or PYR. The combination of TMP
with sulfamethoxazole (SMZ) has also found use
against nonbacterial pathogens such as P. carinii, T.
gondii, and Pl. falciparum. The sulfonamides used
in these combinations have, however, very modest
potency against the target enzyme. Recently, Chio
et al.17 reported the discovery of a class of sulfonamides 22, which are active in culture models against
P. carinii and T. gondii albeit not better than

standard agents against the DHPS enzyme (Table
14).
Table 14. Inhibition of Dihydropteroate Synthases
from T. gondii, P. carinii, M. avium, and E. coli17
IC50 (M)
compd
T. gondiia
M. aviuma
P. cariniib
E. colia
SMZ
sulfadiazine
22a
22b
22c
22d
7.30
2.83
32.9
111
33.4
66.5
1.8
0.23
0.62
1.95
0.58
0.99
4.05
5.8
1.2
1.2
0.82
2
5.6
a
Substrate (PABA) c ) 11 M. b Substrate (PABA) c )2.2
M.
Pemetrexed is on the market as an antitumor agent.

The 2,4-diaminopyrimidine moiety is considered mandatory for potent DHFR inhibitors. The activity of
pemetrexed, which contains a 2-amino-4-oxopyrimidine ring, is unusual. One possible explanation is that
one considers an alternate model of DHFR, in which
the pyrrole NH of this and similar compounds mimics
the 4-amino group of the 2,4-diaminopyrimidine ring
system. Indeed, Gangjee et al. have demonstrated
that 4-methyl analogue 23c does bind in an alternate
mode to DHFR.162 Among the synthesized classical
analogues, the replacement of a CH2 group by a NH
did not result in an improved activity against either
TS or DHFR. Within the group of 10 nonclassical
inhibitors with the same structural change as described above, the 2,4-dichloro-substituted compound
is the most potent of all analogues tested.16
Tricyclic 2,4-Diaminopyrimidines with Fused
Five- or Six-Membered Rings. Twenty-one conformationally restricted PYR analogues 24 with
substituents at different positions of the phenyl ring

have been synthesized and tested as DHFR inhibitors. Heterocyclic systems studied include indeno[2,1d]pyrimidines, benzo[f]quinazolines, and benzo[3,4]cyclohepta[2,1-d]pyrimidines. Neither the potency
nor the selectivity of these compounds is substantially better than that of the reference compound
(Table 15). Computer-simulated docking into the
active site pocket of P. carinii and human DHFR
suggests that the rotationally restricted tricyclic
structures are at a disadvantage relative to PYR in
that the torsional relief between the chlorine atoms
and the critical serine and threonine residues in the
active site is prevented by the bridge.163
Novel classical and nonclassical, partially restricted, linear tricyclic 5-deaza antifolates represented by the compounds 25a-25d have been syn-
5.4.3. Inhibitors of Thymidylate Synthase (TS) and

Multitargeted Antifolates (MTA)
Classical and nonclassical analogues 23 of the
multitargeted antifolate pemetrexed have also been
tested as inhibitors of TS and DHFR as potential agents against opportunistic infections.162
thesized and tested against DHFR from different

sources and for antitumor activity.164 The nonclassical analogues show moderate, but better than that
of PTX, selectivity against DHFR from pathogenic
microbes compared to recombinant human DHFR,
which supports the idea that the removal of 5-methyl
group of PTX, along with the restriction of the side
chain, can translate into selectivity for DHFR from
pathogens (Table 15).
Seven novel tricyclic pyrimido[4,5-c][2,7]naphtyridones and the corresponding naphtyridines 26
have been synthesized as conformationally restricted
analogues of the inhibitors of DHFR as antitumor or
Kompis et al.
Table 15. Inhibition of P. carinii, T. gondii, M. avium, Rat Liver, and Human DHFR by Tricyclic Compounds
IC50 (M)
compd
P. carinii
28a
28b
28c
28d
27a
27b
27c
27d
27e
25a
25b
25c
25d
24a
24b
24c
24d
>30
1.0
>30
6.8
>9.0
22.6
24.1
40.5
10.9
16
14
15
2
7.2
0.59
0.12
1.3
T. gondii
1.4
13.1
22.3
31.7
21.5
1.4
2.9
2.7
0.9
4.7
0.038
0.011
0.006
selectivity ratio
M. avium
rat
>15
>50
>20
>81.2
0.97
2.8
0.062
0.18
0.45
15
50.9
20.6
81.2
85.8
human
32
2.3
7.35
133
19
0.036
0.016
0.005
rl/pc
rl/tg
rl/ma
ref
88
155
155
155
155
166
166
166
166
166
164
164
164
164
163
163
163
163
<0.1
<0.1
<0.1
<0.1
2.3
0.9
2.0
7.9
2a
1.48a
0.32a
3a
2.6
0.06
0.13
<0.01
>10
3.9
2.5
4.0
25.6b
2.7b
2.1b
6b
4
0.95
1.5
0.83
Ratio rh/pc. b Ratio rh/tg.
antiinfectious agents or both. The tricyclic compounds

are about 2 orders of magnitude less potent inhibitors
than their bicyclic analogues when tested against
P. carinii and T. gondii DHFR. The activity and lack
of selectivity of these compounds lends further credence to the idea that the inappropriate orientation
of the substituted phenyl ring in the compounds 26a
and 26b is responsible for their inhibitory properties.165
Seven nonclassical and one classical antifolate have
been designed as conformationally restricted analogues of TMX. They show moderate inhibitory activity against the pathogen DHFRs (Table 15). Compound 27e was 88-fold more potent against M. avium
In an attempt to summarize the results of the

different substance types, it can be stated that
substituted monocyclic 2,4-diaminopyrimidines in
general exhibit high selectivity but rather modest
inhibitory potency. The group of 2,4-diaminopyrimidines with fused five- or six-membered rings contains the most potent compounds, but unfortunately,
they lack selectivity; they are almost equally active
against mammalian and target pathogen organisms.
Conformationally restricted tricyclic analogues of the
above-mentioned types are practically devoid of
inhibitory activities.
Several 3-D structures of DHFRs from the targeted
organisms mentioned above with different ligands
are available and have been studied intensively.
Hundreds of compounds designed using this knowledge have been synthesized and tested. Despite these
efforts, today we are still not in a position to
understand the subtle differences in the active centers of the enzyme to the extent that would enable
us to rationally design a compound that would be
both a potent and a selective inhibitor of the enzyme
of one or more targeted opportunistic pathogens.
5.5. Antimalarials and Other Antiprotozoal Agents
DHFR than against rat liver DHFR.166

Examples of angular tricyclic 1,3-diamino-7,8,9,10tetrahydro-pyrimido[4,5-c]isoquinolines 28 have been
prepared and tested as inhibitors of rat liver and

pcDHFR. Neither analogue exhibits the desired
selectivity155 (Table 15).
Malaria each year kills about 2 million children

and debilitates over 500 million individuals worldwide, and its incidence continues to increase.167
Although the need for antimalarial agents is acute,
in recent years very few new antifolates have been
designed, synthesized, and tested against Plasmodium spp., the causative agent of malaria. The
research in this field has been concentrated almost
exclusively in academia, becuase most, if not all, big
pharmaceutical companies terminated their engagement in tropical diseases in the past decade.
The medical need for new antimalarial drugs,
resistance, and drug development efforts have been
reviewed by Ridley in 2002.168 Debaert summarized
the recent results in this field,169 and structure-based
approaches have been discussed by Brady and Cameron.170 The research efforts have concentrated on the
mechanisms of resistance from malarial DHFR-
Table 16. Inhibition of Trypanosomal, Leishmanial,

and Human DHFRsa
Ki (nM)
compd
Le. major
Tr. cruzii
Tr. brucei
human
ref
TMP
PYR
MTX
29a
29b
29c
29d
29e
30a
30b
30c
120 (12)
250 (0.49)
1000 (1.3)
98 (1.2)
0.038 (4.7)
1130 (2.1)
710 (0.57)
220 (4.2)
23 (60)
200 (4.9)
0.107 (18.4)
0.131 (11.1)
0.183 (8.9)
10 (134)
11 (11)
1380
120
0.179
2400
400
930
1400
1000
1.97
1.45
1.63
178
178
180
178
178
178
178
180
180
180
180
97 (25)
48 (83)
150 (6.2)
160 (8.5)
65 (15)
24 (100)
19 (21)
3.6 (257)
8.8 (156)
11 (88)
The selectivity is shown in parentheses.
TS,44,48,171-176 the biology of the parasite, and epidemiology of its resistance.177

I. H. Gilberts group, in collaboration with others,
explored the SAR of 2,4-diamino-5-benzylpyrimidines
as inhibitors of trypanosomal and leishmanial DHFR,
starting with a lead compound 29a, which was
reported to have a 100-fold selectivity for the leishmanial enzyme (Table 16). The best compounds show
some selectivity for parasite over human DHFR. They
are surprisingly more active against Tr. brucei, given
the high structural similarity between the Tr. cruzi
and Tr. brucei enzymes. The compounds have also
been tested against the clinically relevant forms of
the intact parasite. Among others, compound 29a is
also active in vivo against Tr. brucei.178
Knighton and co-workers created a homology model
of Tr. cruzi DHFR using the published structure of
Leishmania major DHFR.179 Based on the differences
in the active site of these enzymes and the binding
mode of MTX, compounds 30a-30c have been de-
signed as inhibitors of Tr. cruzi DHFR. None of the

compounds are either significantly selective for the
parasite enzyme (Table 16) or active in vitro against
amastigote stage of the parasite.180
Computational screening of commercially available
compounds has been performed using a 3-D structural model of the DHFR domain of the bifunctional
DHFR-TS of Pl. falciparum.181 Twenty-one compounds identified in this screen have been also
assayed for their inhibitory activities. Two of these,
31 and 32, inhibit the recombinant Pl. falciparum
domain with Ki values of 8.7 and 0.54 M, respectively. These results support the validity of the model
and the docking experiments. However, compound 31

has previously been reported to be mutagenic in
bacteria and carcinogenic in animals.
Another molecular docking strategy aimed at discovery of compounds that are active against Pl.
falciparum DHFR is described by Rastelli et al.50
Twelve compounds, N-hydroxyamidines, pyrimidines,
triazines, urea, and thiourea-derivatives, unrelated
to known antifolates, were identified as micromolar
inhibitors of the wild-type and resistant mutant
pfDHFR harboring the widespread single, double,
triple, and quadruple mutations of this enzyme. In
agreement with the design, they bind with similar
affinity to the wild-type and mutated DHFRs. Insights into how these inhibitors bind to their targets
is presented.
An inspiration and a fresh start for the design and
discovery of novel folate inhibitors was provided by
the publication of the 3-D structure of Pl. falciparum
DHFR-TS, the target for the clinically established
antimalarial drugs pyrimethamine and cycloguanil.
The structure reveals insights into the nature of
inhibitor WR 99210 binding in complex with adenine
dinucleotide phosphate and 2-deoxyuridylate in drug
resistance and autologous gene repression, all of
which influence species-specific drug sensitivity.44
The historical background and importance of this
work for further developments in the area are
outlined in the paper of Rathod and Phillips.182
PS-15 is a prodrug form of WR 99210. It is
metabolized in vivo to the active form by microsomal

mixed-function oxidases. Both compounds are active
against Pl. falciparum and P. carinii infections in
vivo and are active against M. avium complex in
vitro.183
Recently twenty-two novel analogues of PYR and
twenty-four of cycloguanil (CYC) represented in 33
and 34 have been synthesized and tested as inhibi-
tors of Pl. falciparum DHFR carrying triple and

quadruple mutations responsible for antifolate re-
sistance.184 The inhibitors were designed to avoid

steric clashes in the active site of the mutant enzymes. Several compounds show inhibition constants
at a low nanomolar level against the mutant enzymes. A number of these inhibitors have also been
shown to have good antiplasmodial activity against
resistant strains of Pl. falciparum in vitro with IC50
values at low micromolar level and relatively low
toxicities. The properties of the compounds demonstrate the feasibility of developing antifolates against
mutated targets in Pl. falciparum.
5.6. Anticancer Antifolates

Although the number of antifolates synthesized as
potential anticancer agents in the last 10 years has
been smaller than those synthesized for antimicrobial
targets, the success rate of these efforts is impressive.
Several new targetssTS, FPGS, GarFTaseshave
been exploited, and novel chemical entities have
progressed to clinical development or reached the
market (section 4). The current state of the research
and development of anticancer antifolates has been
reviewed recently.13
5.6.1. Classical Inhibitors of DHFR

MTX and TMX are highly potent but nonselective
DHFR inhibitors in use. MTX is a mainstay in single
or combination chemotherapy of lymphoblastic leukemias and other cancers. It is also considered a gold
standard in the treatment of rheumatic arthritis and
is used to treat psoriasis.185-190
DL-4,4-Difluoroglutamic acid and DL-,-difluoromethotrexate have been synthesized. Their evaluation revealed that the former compound is a poor
alternate substrate for FPGS and the latter is neither
a substrate nor an inhibitor of human FPGS.191
The stereospecific synthesis of methotrexate analogues containing L-threo-(2S,4S)-4-fluoroglutamic
acid and DL-3,3-difluoroglutamic acid has been reported. The compounds do not act as substrates for
FPGS and inhibit human DHFR at a similar level
as MTX.192
In recent years, MTX has also been used as a
therapeutic agent in the treatment of patients with
Crohns disease, a chronic inflammatory bowel disease. Classical DHFR inhibitors resembling the MTX
structure were synthesized as potential drugs for this
indication. They contain ester bridges in the central
parts of the MTX molecule, and the pteridine ring
system has been substituted by a quinazoline. The
compounds inhibit rat DHFR in the low nanomolar
range. Compound 35 is also active in vivo in the
corresponding disease model. However, the mechanism of action is not well understood, and the results
cannot be fully explained by DHFR inhibition or by
inhibition of lymphocyte cell proliferation.193
Kompis et al.
The synthesis of two thiophene analogues of 5chloro-5,8-dideazafolic acid has been reported. Compounds 36a and 36b were tested as inhibitors of
tumor cell growth in culture, and their IC50s against

CCRF-CEM human leukemic lymphoblasts are 1.8
and 2.1 M, respectively.194
Compound 37, in which the bicyclo[2.2.2]octane
ring system replaces the phenyl ring of the paminobenzoate moiety of aminopterin has been synthesized and tested for antifolate activity. It is
ineffective against L1210 DHFR and three tumor cell
lines. In contrast to many classical DHFR inhibitors
bearing appropriate aromatic ring systems in the side
chain, the compound negates the stoichiometric binding to the target enzyme.195
Gangjee et al. have studied the effect of C9-methyl
substitution and C8-C9 conformational restriction
on the antifolate and antitumor activity of classical
5-substituted 4-diaminofuro[2,3-d]pyrimidines.196 Compound 10i with a 9-methyl group shows increased
inhibitory potency against recombinant human DHFR,
as well as against the growth of CCRF-CEM tumor
cells in culture. Conformationally restricted analogues are significantly less active. The analogues
with the C-C bridge are also good substrates for
human FPGS, indicating that FPGS is relatively
tolerant to conformations in the bridge region.
With the design and synthesis of 10j, the effect of
homologation of a C9-methyl to an ethyl on DHFR
inhibition and antitumor activity was investigated.
The extension doubles the inhibitory potency against
hDHFR (IC50 ) 0.21 M) when compared with its
lower homologue and is 4-fold more potent than the
C9-H analogue 10h. It also demonstrates increased
growth inhibitory potency against several human
tumor cell lines in culture with GI50 values of <1.0
10-8 M and is a weak inhibitor of rhTS. Compounds
10i and 10j are efficient substrates of human FPGS.
Further evaluation of the cytotoxicity of the latter
compound in MTX-resistant CCRF-CEM cell lines
and metabolite protection studies implicated DHFR
as the primary intracellular target. The authors
conclude that alkylation of the C9 position in the C8C9 bridge of the classical 5-substituted 2,4-diaminofuro-[2,3-d]pyrimidine is highly conducive to DHFR
and tumor inhibitory activity, as well as FPGS
substrate efficiency.197
Two new analogues of a nonpolyglutamable antifolate PT 523, which is currently in clinical development, have been synthesized.198 Compounds 39a and
Table 17. Cell Growth Inhibition and DHFR Binding

by Compounds 39 and 40
compd
cell growth
IC50 (nM)
DHFR binding
Ki (pM)
PT 523
39a
39b
39c
39d
40a
40b
40c
40d
40e
1.5 ( 0.39
0.69 ( 0.04
1.3 ( 0.35
0.64 ( 0.04
0.53 ( 0.07
0.63 ( 0.08
1.2 ( 0.25
54 ( 4.9
1.2 ( 0.22
4.4 ( 1.1
0.33 ( 0.04
0.21 ( 0.005
0.60 ( 0.02
0.014 ( 0.005
0.35 ( 0.06
ref
101, 102, 198
198
198
199
199
201
201
201
201
201
39b were tested in a 72 h growth inhibition assay

against cultures of CCRF-CEM human leukemic
lymphoblasts (Table 17). The activities are comparable to those of PT 523 and the previously studied
analogues 39c and 39d.199 However, they are more
active than aminopterin despite the fact that they
cannot form -polyglutamated metabolites as classical antifolates with a glutamate side chain.
Replacement of ornithine in PT 523 by L-2,4diaminobutanoic acid or lysine affects the binding to

hDHFR but results in a loss of activity against some
carcinoma cells in culture. 3,5-Dichlorosubstitution
in the p-aminobenzoic moiety decreases neither
DHFR binding nor cytotoxicity.202
Compounds 41 have been designed as antitumor
agents acting as dual inhibitors of TS and DHFR.

Compared to pemetrexed, inhibitory potency against
human DHFR of compounds 41b and 41c is 1 and 2
orders of magnitude lower that that of the reference
compound, respectively.203 Both 41b and 41c are
more potent than pemetrexed against E. coli TS.
Against human TS, 41b is 7-fold less potent than
pemetrexed and 41c shows similar inhibitory activity
as pemetrexed. In contrast to 41b, which is an
efficient substrate of human FPGS, 41c is substantially less active. Compound 41b shows GI50 values
in the nanomolar range against more than 18 human
tumor cell lines in the standard NCI preclinical in
vitro screen.
Many compounds, designed as inhibitors of DHFR
of opportunistic pathogens, such as 10c, 10d,143 15,156
or 18,204 have also been tested against a variety of
tumor cell lines in culture. Others have been evaluated in the in vitro screening program of the National
Cancer Institute (e.g., 17 and 18).
5-Deazafolate analogues with a rotationally restricted glutamate or ornithine side chain, 42, have
The consequences of changes in the ring B of PT

523 on the inhibition of hDHFR and growth inhibition of a large panel of tumor cell lines, performed
also at NCI, have been reported and results with the
39e are discussed in detail. 200
Synthesis of a series of analogues of PT 523 with
modifications in the side chain, p-aminobenzoyl
moiety, or C9-C10 bridge has been reported.198,201
The growth inhibition values of the selected compounds 40a-40e are shown in the Table 17. The shift
been synthesized and tested as substrates for FPGS

and as inhibitors of the growth of CCRF-CEM cells.205
Whereas compounds 42b and 42d are potent inhibitors of rhFPGS, compounds 42a and 42c are exceptionally efficient FPGS substrates. All four compounds are inactive in the CCRF-CEM cell growth
assay.
5.6.2. Inhibitors of Thymidylate Synthase
of the terminal ortho-carboxyl group to the meta or

para position has a detrimental effect on the activity.
The potential of TS inhibitors in cancer therapy has

been recently reviewed by McGuire et al.,206 N. L.
Lehman,207 and Ackland et al.208 The inhibitory
concentrations of compound considered as references
for inhibition of TS are given in the Table 18.
Kompis et al.
Table 18. Inhibitory Concentrations (IC50, M) of the

Reference Compounds against Isolated TS204
compd
rec
rpc
rh
pemetrexed
raltitrexed
CB 3717
MTX
1.1 10-4
8.0 10-6
5.8 10-8
1.8 10-4
5.7 10-5
5.7 10-5
1.0 10-6
1.5 10-7
3.6 10-5
5.0 10-8
Compounds 43 have been designed as dual inhibitors of TS and DHFR and as antitumor agents.162 The
triaminopyrimidines, for which DHFR is the major

target, 45c and 45d are more potent than MTX in
inhibiting the growth of H23/0.3 cell line.
5.6.4. Inhibitors of Folylpolyglutamate Synthetase
X-ray crystal structure of the ternary complex of 43a,

DHFR, and NADPH shows that the inhibitor binds
in a 2,4-diamino mode, where the pyrrolo-nitrogen
mimics the 4-amino moiety of 2,4-diaminopyrimidines. The compounds have been evaluated as
inhibitors of L. casei, E. coli, rat, and human TS,
along with reference compounds pemetrexed, ZD
1694, and PDDF. Compound 43a, an excellent substrate for FPGS, is similar in potency to ZD 1694 and
4.4-fold more active than pemetrexed. The activities
against DHFR from the above-mentioned sources, as
well as against CCRF-CEM human leukemia and
FaDu squamous cell carcinoma have also been reported.162
Compounds 44 are in the bridge C8-C9 isosteric
with MTA pemetrexed. Both classical and nonclassical analogues have been prepared and tested as
antitumor agents and agents against opportunistic
infections. The compounds are poor inhibitors of P.
carinii DHFR and possess similar potency as TMP
against T. gondii DHFR. The nonclassical analogues
are also inactive against TS. Compound 44a marginally (IC50 ) 46 M) inhibits human TS, but it is a
potent inhibitor of several cell carcinoma lines.204
5.6.3. Nonlassical Inhibitors of Folate Enzymes

With TMX and PTX in clinical development, the
search for nonclassical antifolates as anticancer
agents slowed considerably. Compounds synthesized
for other target indications have often been evaluated
in basic screens either at NCI or at the particular
institution for their anticancer activities.
A series of 5-(N-phenylpyrrolidin-3-yl)-2,4,6-triaminopyrimidines 45a and 2,4-diamino-(N-phenylpyrrolidin-3-yl)-6(5H)-oxopyrimidines 45b have been
synthesized and evaluated for their in vitro cytotoxicity.209 The studies revealed that the former compounds are more cytotoxic than their 2,4-diamino6(5H)-oxopyrimidine counterparts. Among 2,4,6-
Folylpolyglutamate synthetase (FPGS) (E.C.6.3.2.17)

plays a critical role in endogenous folate metabolism,
as well as in the cellular pharmacology of classical
antifolates and, more specifically, in the therapeutic
selectivity of these drugs and the development of
resistance. It is responsible for the conversion of
naturally occurring folates (and antifolates) to their
poly(-glutamate) derivatives, forms required for
intracellular retention of folates and the preferred
substrates for most folate-dependent enzymes. Because of its role, several studies have been performed
to identify the structural requirements for binding
of folates and folate antagonists to the active site of
the enzyme. The developments in the area till 1999
and the definitions of polyglutamable and nonpolyglutamable inhibitors of types A and B have been
reviewed by A. Rosowsky.210 Recent advances in the
chemistry and biology of folyl-poly(-glutamate) synthetase substrates and inhibitors have been reviewed
by Gangjee et al. in 2002.211
Surprisingly, suramine, a substance outside of the
class of known FGPS inhibitors, is reported to be a
potent inhibitor of human FPGS.212 The effects of
suramine on growth of CCRF-CEM cells and a MTXresistant subline, expressing low levels of FPGS
activity, suggest that inhibition of folate metabolism
could be involved in the mechanism of action of
suramine.
Folate and MTX analogues 38a-38c, with L-
histidine in place of L-glutamate, have been synthesized as potential inhibitors of FPGS. No significant
inhibition of the target enzyme by these compounds
is observed.213
Partially restricted tricyclic antifolates 25d and
27e are reasonable substrates for FPGS but virtually
inactive against CCRF-CEM human leukemia cells.
The compounds and their congeners have also been
evaluated in the NCI preclinical antitumor screening
program.164,166
Compound 46 and its congeners have been designed as mechanism-based inhibitors of FPGS where
a phosphonate moiety mimics the tetrahedral intermediate in the ligation reaction. They do not act as
substrates but are potent and competitive inhibitors

of this enzyme.214
5.6.5 Inhibitors of Other Enzymes

DHFR and TS are established targets for anticancer agents. As mentioned in section 4, a number
of newer inhibitors are known to inhibit more than
one enzyme in the folate pathway or related reactions. Thus, pemetrexed is an inhibitor of DHFR and
TS, as well as glycinamide ribonucleotide formyltransferase. The latter is a crucial enzyme in purine
biosynthesis. The finding that 5,10-dideazatetrahydrofolate (lometrexol, see Chart 3) is an effective
inhibitor of GARFT and an efficacious antitumor
agent establishes purine biosynthesis as a viable
target for anticancer agents. Specific inhibitors of
essential enzymes in purine synthesis, such as GARFT
or AICARFT, for example, conformationally restricted 10-formyl-tetrahydrofolate analogues have
been synthesized.215,216 This topic is not further
addressed in the present review.
6. Resistance to Antifolates
Resistance to DHFR inhibitors or inhibitors of
DHPS in bacteria, protozoa, fungi, or cancer cells can
be caused by a variety of mechanisms. Point mutations in the target enzyme that alter the binding of
the inhibitor thereby leading to resistance are frequently found in Gram-positive bacteria, in protozoa,
and in other parasites. Resistant plasmid-borne
bypass enzymes are the main cause of resistance in
Gram-negative bacteria. Almost 20 different TMPresistant DHFR bypass enzymes have been found in
Gram-negative bacteria and a detailed discussion of
their characteristics, their genetic location, and their
epidemiology is beyond the scope of this article.
Mechanisms of TMP resistance in bacteria have been
studied intensively for years, and there is a wealth
of information available.217-220 Considerable new
information has been acquired, however, in recent
years on antifolate resistance in protozoa, particularly in malaria parasites, aided by the application
of modern biochemical and genomic tools.221 We
therefore focus on the latter.
6.1. Bacteria
Streptococcus pneumoniae is a major human pathogen, causing upper and lower respiratory tract infections. Penicillin-resistant strains are now prevalent,
and many of these strains are co-resistant to TMP
and sulfonamides. A number of amino acid changes
in the Str. pneumoniae DHFR have been reported,
but only a single point mutation, isoleucine 100 to
leucine (Ile100Leu), can lead to high TMP resistance.222,223 Similarly, a single amino acid substitution,
that is, phenylalanine 98 to tyrosine (Phe98Tyr), was
found to be responsible for TMP resistance in S.
aureus.218 X-ray crystallography with the ternary

complex of the Phe98Tyr DHFR with folate-NADPH
showed that the mutation resulted in a loss of a
hydrogen bond between the 4-amino group of TMP
and the carbonyl oxygen of Leu-5. This mechanism
is predominant in both transferable plasmid-encoded
and nontransferable chromosomally encoded resistance.224
The effect of mutations on the interactions between
dimers of the R67 plasmid-encoded DHFR from E.
coli have been investigated. The native enzyme is a
tetramer.225
Numerous point mutations in DHPS have been
described in many species, conferring resistance to
sulfonamides and sulfones.217,226 For high level, transferable sulfonamide resistance, mainly in Gramnegative bacteria, two genes, sul1 and sul2, have
been found.
6.2. Protozoa
Pl. falciparum, the most important causative agent
of malaria, has acquired resistance to many of the
established agents. The consequences of parasite
resistance for the dynamics of malaria spread and
public health have been comprehensively reviewed
recently.227-229 The mechanism of antifolate resistance was reviewed by Warhurst.230 Mutations in the
DHFR domain of the bifunctional DHFR-TS enzyme
have been associated with antifolate resistance.
Several recent reviews describe single, double, or
multiple mutations44,228,231-235 in the gene. The prevalent single mutation is Ser108AsN, which confers a
moderate level of resistance to PYR and CYC. Higher
resistance levels are observed with double mutations, such as Cys59Arg and Ser108Asn, triple mutations, Asn51Ile, Cys59Arg, and Ser108AsN or
Cys59Arg, Ser108Asn, and Ile164Leu, and the quadruple mutant, Asn51Ile, Cys59Arg, Ser108Asn, and
Ile164Leu. The latter is highly resistant to both
agents. One double mutant, Ala16Val and Ser108Asn,
confers CYC resistance only. The Ser108Asn mutation could be specifically identified in field isolates
with PCR-based fluorescent probes.236
A single Asp54Glu mutation in the pfDHFR domain greatly decreases the catalytic activity of the
enzyme and affects both Km values for substrate and
Ki values for PYR, CYC, and WR99210237 (Table 19).
Resistance to CYC was found to increase in several
African countries, as monitored by the Ser108Asn
mutation, from 19.8% in 1995 to 43.6% in 1997.238
Fixed trimethoprim-sulfamethoxazole (5:1) combination (co-trimoxazole) is widely used in Africa for
prophylaxis against opportunistic infections in HIVinfected individuals. Pyrimethamine-sulfadoxine and
co-trimoxazole select for antifolate resistance in Pl.
falciparum, and there is cross-resistance between
these two agents.239
Drug resistance alleles for both mutated dhfr and
dhps genes are frequently found in Pl. falciparum
isolated from many parts of the world where resistance is common, that is, in African countries, Java,
Indonesia, or South America.240,241
Resistance to pyrimethamine-sulfadoxine in Africa is mostly due to point mutations at codons 108,
Kompis et al.
Table 19. Inhibition Kinetics and in Vitro Sensitivity of PfDHFR/TS and Parasites44
antiplasmodial activity
(IC50, M)
inhibition constant
(Ki, nM)
DHFR type
PYR
CYC
WR99210
PYR
CYC
WR00210
wild-type
Cys59Arg/Ser108Asn
Asn51Ile/Cys59Arg/
Ser108Asn/Ile164Leu
0.2
9.8
283
0.3
6.2
254
0.011
0.02
0.037
0.08
30.9
>100
0.037
2.4
>100
0.00057
0.0023
0.018
51, and 59 of dhfr and codon 437, 540, or both of dhps.

In contrast to South East Asia and South America,
the Ile164Leu mutation in dhfr was not detected
until recently.242
Point mutations in the gene for dhps of Pl. falciparum are responsible for resistance to sulfadoxine,
dapsone, and other sulfonamides. Mutations at codons
436, 437, 540, 581, and 613 of dhps were observed in
isolates from Kenya, but the poor correlation between
genotype and in vitro resistance suggests that additional factors contribute to resistance; these include
the folate content of the medium and utilization of
exogenous folates.243
Expression of mutated pfDHPS in E. coli unequivocally demonstrates that these enzymes confer resistance to sulfonamides and sulfones.244
DHFR and DHPS genotypes, analyzed in 70 Pl.
falciparum isolates, correlate with resistance to PYR,
TMP, sulfadoxine, and sulfamethoxazole.245
Genotyping of pyrimethamine-sulfadoxine-resistant Pl. falciparum has been recently performed by
matrix-assisted laser desorption/ionization time-offlight mass spectrometry (MALDI-TOF MS). This
technique conveniently identified single nucleotide
polymorphisms (SNPs) occurring at position 16, 51,
59, and 108 of the pfdhfr gene, which are associated
with PYR resistance.246
Similarly, as in Plasmodium species, mutations in
the DHFR-TS protein have been found to be responsible for resistance to pyrimethamine of Toxoplasma
gondii, Arg-59 and Asn-108 (Pl. falciparum numbering system). Arg-36 and Ser-83 in T. gondii do not
exhibit significant fitness defects in vitro but exhibit
a 1.8% fitness defect per generation in mice. The
high-level PYR resistant mutant Arg-59 and Ser-223
exhibits a >2.8% fitness defect both in vitro and in
vivo. This high cost of mutation is assumed to be
responsible for the fact that this mutation is not
observed in the field.247
There is very little information about the prevalence and mechanism of antifolate resistance in
Toxoplasma. Several mutations in the T. gondii
DHFR-TS generated in vitro have been shown
to confer resistance to PYR, namely, Trp25Arg,
Leu98Ser, and Leu134His.248
6.3. Fungi
Sulfonamides are more active than TMP in the
treatment of Pneumocystis carinii pneumonia (PCP),
and TMP is a moderately active inhibitor of the P.
carinii DHFR with an IC50 of 43 M.86 A considerable
body of information on the mechanism of resistance
to DHFR inhibitors and sulfonamides has accumulated in the past few years. This has been made
possible through the use of modern genetic tools to

study this difficult to grow pathogen.
Mutations in the DHFR of patient isolates of P.
carinii have been detected repeatedly. Amino acid
substitutions Ala67Val and Cys166Tyr have been
found in two patients in Japan.249
Genotyping of DHPS from P. carinii from AIDS
patients revealed several mutations, the most frequent being Thr55Ala and Pro57Ser, which are
located in the sulfa-binding site and may occur singly
or as a double mutation in the same isolate. Although
these mutations often lead to failures of prophylaxis
with co-trimoxazole, therapy is often successful.250,251
These mutants are still infrequent but are increasing
and linked to prior sulfonamide prophylaxis.252
Two case reports published in 1994 and 2002 show
failure of co-trimoxazole prophylaxis or therapy with
concomitant administration of leucovorin (5-formyltetrahydrofolate), which is sometimes given to reduce
the incidence of neutropenia.253,254 This suggests that
P. carinii might be able to use exogenous folates.
6.4. Cancer Cells

Various mechanisms of resistance to antifolates in
cancer cells have been described and recently reviewed.255,256 Alterations in transport, efflux, polyglutamation, and hydrolase activities are the major
determinants for MTX resistance. Increased dhfr
gene copy number, mutations in DHFR, and changes
in transcriptional regulation are additional resistance
mechanisms. These mechanisms are responsible for
a considerable failure rate in the treatment of acute
lymphoblastic leukemia in pediatric patients. Similarly, increased levels of TS, decreased uridine monophosphate kinase (UMPK), or changes in thymidine
phosphorylase or dihydropyrimidine dehydrogenase
may be responsible for resistance to 5-fluorouracil or
its derivatives.
Alterations in membrane transport are an important mechanism of resistance that affects a number
of antifolates.257
For pemetrexed, mutations in the reduced folate
carrier leading to a decrease in the activity of FPGS
and an increase in the activity of -glutamylhydrolase
or of TS have been described as resistance mechanisms.92 Loss of FPGS activity is a dominant mechanism of resistance to polyglutamation-dependent
novel antifolates in human leukemia cell sublines.258
A number of multiple mutants of murine DHFR
have been constructed and analyzed for stability and
resistance to MTX and trimetrexate. The Ki values
of the Phe31Ala/Phe34Ala mutant are >10 000-fold
higher for MTX than wild-type values, but only 13.5fold higher for trimetrexate.259
7. Conclusion
Inhibition of the folate pathway enzymes in the
past decade continued to be an area of intensive
efforts, both in academia and in industry. The
ubiquitous nature of these enzymes provided the
basis to target several indication areas. Thus, in
addition to antifolates aimed at combating bacterial
pathogens, especiallly those involved in opportunistic
infections, fungi, protozoa, and, in particular, cancer
cells remain an area of high interest. Antifolate
research has become an exercise field for scientists
from different disciplines to demonstrate the power
of modern methodologies to contribute to better
understanding of the basic processes in the folate
biosynthesis, in the utilization of folates, and in
resistance mechanisms to antifolates. The combination of X-ray crystallography of numerous enzymes
from different biological sources, molecular modeling,
and skilled synthetic work resulted in design and
synthesis of many hundreds of antifolates and to the
identification of almost a dozen of new investigational
drugs. A number of these drugs have reached the
market. New discoveries in the folate pathway,
greatly aided by the application of genomic and
proteomic tools, not only improved our general understanding of this key pathway in all living cells,
its conservation, and its modifications but also offer
new possibilities for drug discovery.
8. Acknowledgment
The authors are grateful to Dr. Johannes Hoffner,
Arpida, Ltd., for the valuable technical assistance.
9. References
(1) Hitchings, G. H.; Smith, R. L. Adv. Enzyme Regul. 1980, 18, 349.
(2) Sirotnak, F. M. Folate Antagonists as Therapeutic Agents; John
Wiley & Sons: New York, 1984.
(3) Blakley, R. L. Adv. Enzymol. Relat. Areas Mol. Biol. 1995, 70,
23.
(4) Rosowsky, A. Progress in Medicinal Chemistry; Elsevier Science
Publishers: Amsterdam, 1989; pp 1-252.
(5) Baccanari, D. P.; Kuyper, L. F. J. Chemother. 1993, 5, 393.
(6) Bowden, K.; Harris, N. V.; Watson, C. A. J. Chemother. 1993, 5,
377.
(7) Blaney, J. M.; Hansch, C.; Silipo, C.; Vittoria, A. Chem. Rev.
1984, 84, 333.
(8) Periti, P. J. Antimicrob. Chemother. 1995, 36, 887.
(9) Li, R.; Siriwaraporn, R.; Chitnumsub, P.; Siriwaraporn, W.;
Wooden, J.; Athappilli, F.; Turley, S.; Hol, W. G. J. J. Mol. Biol.
2000, 295, 307.
(10) Berman, E. M.; Werbel, L. M. J. Med. Chem. 1991, 34, 479.
(11) Hartman, P. G. J. Chemother. 1993, 5, 369.
(12) Appaji Rao, N.; Ambili, M.; Jala, V. R.; Subramanya, H. S.;
Savithri, H. S. Biochim. Biophys. Acta 2003, 1647, 24.
(13) McGuire, J. J. Curr. Pharm. Des. 2003, 9, 2593.
(14) Deroiun, F. Curr. Opin. Invest. Drugs 2001, 10, 1368.
(15) Costi, M. P.; Ferrari, S. Curr. Drug Targets 2001, 2, 135.
(16) Gangjee, A.; Vidwans, A. P.; Elzein, E.; McGuire, J. J.; Queener,
S. F.; Kisliuk, R. L. J. Med. Chem. 2001, 44, 1993.
(17) Chio, L. C.; Bolyard, L. A.; Nasr, M.; Queener, S. F. Antimicrob.
(18) Blakley, R. L.; Benkovic, S. J. Chemistry and Biochemistry of
Folates; John Wiley & Sons: New York, 1984.
(19) Benkovic, S. J.; Hammes-Schiffer, S. Science 2003, 301, 1196.
(20) Kisliuk, R. L. Pharmacol. Ther. 2000, 85, 183.
(21) Wallace, L. A.; Matthews, C. R. Biophys. Chem. 2002, 101, 113.
(22) Oefner, C.; DArcy, A.; Winkler, F. Eur. J. Biochem. 1988, 174,
377.
(23) Champness, J. N.; Achari, A.; Ballantine, S. P.; Bryant, P. K.;
Delves, C. J.; Stammers, D. K. Structure 1994, 2, 915.
(24) Matherly, R. L.; Taub, J. W.; Wong, S. C.; Ekizian, R.; Buck, S.;
Williamson, M.; Amylon, M.; Pullen, J.; Camitta, B.; Ravindranath, Y. Blood 1997, 90, 578.

(25) Noe, V.; MacKenzie, S.; Ciudad, C. J. J. Biol. Chem. 2003, 278,
38292.
(26) Cody, V.; Galitsky, N.; Luft, J. R.; Pangborn, W.; Rosowsky, A.;
Queener, S. F. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2002,
58, 946.
(27) Stringer, J. R.; Beard, C. B.; Miller, R. F.; Wakefield, A. E.
Emerging Infect. Dis. 2002, 8, 891.
(28) Whitlow, M.; Howard, A. J.; Stewart, D.; Hardman, K. D.; Chan,
J. H.; Baccanari, D. P.; Tansik, R. L.; Hong, J. S.; Kuyper, L. F.
J. Med. Chem. 2001, 44, 2928.
(29) Hampele, I. C.; DArcy, A.; Dale, G.; Kostrewa, D.; Nielsen, J.;
Oefner, C.; Page, M. G. P.; Schonfeld, H. J.; Then, R. L.; Stuber,
D. J. Mol. Biol. 1997, 268, 21.
(30) Achari, A.; Somers, D. O.; Bryant, P. K.; Rosemond, J.; Stammers, D. K. Nat. Struct. Biol. 1997, 4, 490.
(31) Baca, A. M.; Sirawaraporn, R.; Turley, S.; Sirawaraporn, W.; Hol,
W. G. J. Mol. Biol 2000, 302, 1193.
(32) Lane, B. R.; Ast, J. C.; Hossler, P. A.; Mindell, D. P.; Bartlett,
M. S.; Smith, J. W.; Meshnik, S. R. J. Infect. Dis. 1997, 175,
482.
(33) Ahmad, S. I.; Kirk, S. H.; Eisenstark, A. Annu. Rev. Microbiol.
1998, 52, 591.
(34) Carreras, C. W.; Santi, D. V. Annu. Rev. Biochem. 1995, 64, 721.
(35) Montfort, W. R.; Weichsel, A. Pharmacol. Ther. 1997, 76, 29.
(36) Danenberg, P. V.; Makki, H.; Swenson, S. Semin. Oncol. 1999,
26, 621.
(37) Costi, M. P.; Tondi, D.; Rinaldi, M.; Barlocco, D.; Peconari, P.;
Soragni, F.; Venturelli, A.; Stroud, R. M. Biochim. Biophys. Acta
2002, 1587, 206.
(38) Stroud, R. M.; Finer-Moore, J. S. Biochemistry 2003, 42, 239.
(39) Finer-Moore, J. S.; Santi, D. V.; Stroud, R. M. Biochemistry 2003,
42, 248.
(40) VanTriest, B.; Peters, G. J. Oncology 1999, 57, 179.
(41) Liu, J.; Schmitz, J. C.; Lin, X.; Tai, N.; Yan, W.; Farrell, M.;
Bailly, M.; Chen, T.; Chu E. Biochim. Biophys. Acta 2002, 1587,
175.
(42) Mathews, I. I.; Deacon, A. M.; Canaves, J. M.; McMullan, D.;
Lesley, S. A.; Agarwalla, S.; Kuhn, P. Structure 2003, 11, 677.
(43) Agrawal, N.; Lesley, S. A.; Kuhn, P.; Kohen, A. Biochemistry
2004, 43, 10295.
(44) Yuvaniyama, J.; Chitnumsub, P.; Kamchonwongpaisan, S.;
Vanichtanankul, J.; Sirawaraporn, W.; Taylor, P.; Walkinshaw,
M. D.; Yuthavong, Y. Nat. Struct. Biol. 2003, 10, 357.
(45) Zuccotto, F.; Zvelebil, M.; Brun, R.; Chowdhury, S. F.; DiLucrezia, R. I.; Maes, L.; Ruiz-Perez, L. M.; Pacanowska, D. G.;
Gilbert, I. H. Eur. J. Med. Chem. 2001, 36, 395.
(46) Gaffar, F. R.; Wilschut, K.; Franssen, F. F. J.; de Vries, E. Mol.
Biochem. Parasitol. 2004, 133, 209.
(47) Delfino, R. T.; Santos-Filho, O. A.; Figueroa-Villar, J. D. Biophys.
Chem. 2002, 98, 287.
(48) Santos-Filho, O. A.; Hopfinger, A. J. J. Comput.-Aided Mol. Des.
2001, 15, 1.
(49) Lemcke, T.; Christensen, I. T.; Jrgensen, F. S. Bioorg. Med.
Chem. 1999, 7, 1003.
(50) Rastelli, G.; Pacchioni, S.; Sirawaraporn, W.; Sirawaraporn, R.;
Parenti, M. D.; Ferrari, A. M. Bioorg. Med. Chem. 2000, 8, 1117.
(51) Chitnumsub, P.; Yavaniyama, J.; Vanichtanankul, J.; Kamchonwongpaisan, S.; Walkinshaw, M. D.; Yuthavong, Y. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2004, 60, 780.
(52) ONeil, R. H.; Lilien, R. H.; Donald, B. R.; Stroud, R. M.;
Anderson, A. C. J. Biol. Chem. 2003, 278, 52980.
(53) Nar, H.; Huber, R.; Meining, W.; Schmid, C.; Weinkauf, S.;
Bacher, A. Structure 1995, 3, 459.
(54) Bracher, A.; Fischer, M.; Eisenreich, W.; Ritz, H.; Schramek, N.;
Boyle, P.; Gentili, P.; Huber, R.; Nar, H. J. Biol. Chem. 1999,
274, 16727.
(55) Auerbach, G.; Herrmann, A.; Bracher, A.; Bader, G.; Guttlich,
M.; Fischer, M.; Neukamm, M.; Garrido-Franco, M.; Richardson,
J.; Nar, H.; Huber, R. Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
13567.
(56) Lee, C. S.; Salcedo, E.; Wang, Q.; Sims, P. F.; Hyde, J. E.
Parasitology 2001, 122, 1.
(57) Xie, L.; Smith, J. A.; Gross, S. S. J. J. Biol. Chem. 1998, 273,
21091.
(58) Li, H. Y.; Yao, Y. M.; Shi; Z. G.; Dong, N.; Yu, Y.; Lu, I. R.; Sheng,
Z. Y. Crit. Care Med. 2002, 30, 2520.
(59) Hennig, M.; DArcy, A.; Hampele, I. C.; Page, M. G.; Oefner, C.;
Dale, G. E. Nat. Struct. Biol. 1998, 5, 357.
(60) Ploom, T.; Haussmann, C.; Hof, P.; Steibacher, S.; Bacher, A.;
Richardson, J.; Huber, R. Structure 1999, 7, 509.
(61) Deng, H.; Callender, R.; Dale, G. E. J. Biol. Chem. 2000, 275,
30139.
(62) Illarionova, V.; Eisenreich, W.; Fischer, M.; Haussmann, C.;
Romisch, W.; Richter, G.; Bacher, A. J. Biol. Chem. 2002, 277,
28841.
(63) Guiney, D.; Gibson, C. L.; Suckling, C. J. J. Org. Biomol. Chem.
2003, 1, 664.
(64) Blaszcyk, J.; Shi, G.; Li, Y.; Yan, H.; Ji, X. Structure 2004, 12,
467.

(65) Haussmann, C.; Rohdich, F.; Schmidt, E.; Bacher, A.; Richter,
G. J. Biol. Chem. 1998, 273, 17418.
(66) Agrawal, S.; Kumar, A.; Srivastava, V.; Mishra, B. N. J. Mol.
(67) Zanetti, K. A.; Stover, P. J. J. Biol. Chem. 2003, 278, 10142.
(68) Chaturvedy, S.; Bhakuni, V. J. Biol. Chem. 2003, 278, 40793.
(69) Fu, T. F.; Boja, E. S.; Safo, M. F.; Schirch, V. J. Biol. Chem.
2003, 278, 31088.
(70) Matthews, R. G.; Drummond, J. T.; Webb, H. K. Adv. Enzyme
Regul. 1998, 38, 377.
(71) Snell, K.; Riches, D. Cancer Lett. 1989, 44, 217.
(72) Ballantine, S.; Volpe, F.; Delves, C. J. Protein Expression Purif.
1994, 5, 371.
(73) Volpe, F.; Ballantine, S.; Delves, C. J. Gene 1995, 160, 41.
(74) Thomas, M. C.; Ballantine, S. P.; Bethell, S. S.; Bains, S.;
Kerllam, P.; Delves, C. J. Biochemistry 1998, 37, 11629.
(75) Warhurst, D. C. Sci. Prog. 2002, 85 (Part 1), 89.
(76) Cheong, C. G.; Wolan, D. V.; Greasley, S. E.; Horton, P. A.;
Beardsley, G. P.; Wilson, I. A. J. Biol. Chem. 2004, 279, 18034.
(77) Myllykallio, H.; Lipowski, G.; Leduc, D.; Filee, J.; Forterre, P.;
Liebl, U. Science 2002, 297, 105.
(78) Murzin, A. G. Science 2002, 297, 61.
(79) Ortenberg, R.; Rozenblatt-Rosen, O.; Mevarech, M. Mol. Microbiol. 2000, 35, 1493.
(80) Wilquet, V.; Van de Casteele, M.; Gigot, D.; Legrain, C.;
Glansdorff, N. J. Bacteriol. 2004, 186, 351.
(81) Giladi, M.; Altman-Price, N.; Levin, I.; Levy, L. J. Bacteriol.
2003, 185, 7015.
(82) Cunningham, M. L.; Beverley, S. M. Mol. Biochem. Parasitol.
2001, 113, 199.
(83) Bello, A.; Nare, B.; Freedman, D.; Hardy, L.; Beverley, S. M.
(84) Nare, B.; Luba, J.; Hardy, L. W.; Beverley, S. Parasitology 1997,
114, 101.
(85) Lye, L. F.; Cunningham, M. L.; Beverley, S. M. J. Biol. Chem.
2002, 277, 38245.
(86) Zhang, K.; Rathod, P. K. Science 2002, 296, 545.
(87) Then, R. L. J. Chemother. 2004, 16, 3.
(88) Olliaros, P. L.; Taylor, W. R. J. J. Exp. Biol. 2003, 206, 3753.
(89) Takimoto, C. H. Semin. Oncol. 1997, 24, 18.
(90) Purcell, W. T.; Ettinger, D. S. Curr. Oncol. Rep. 2003, 5, 114.
(91) Norman, P. Curr. Opin. Invest. Drugs 2001, 2, 1611.
(92) Exinger, D.; Exinger, F.; Mennecier, B.; Limacher, J. M.; Dufour,
P.; Kurtz, J. E. Cancer Ther. 2003, 1, 1315.
(93) Adjei, A. A. Expert Rev. Anticancer Ther. 2003, 3, 145.
(94) Calvert, H. Semin. Oncol. 2003, 30, 2.
(95) OShaughnessy, J. A. Semin. Oncol. 2002, 29, 57.
(96) Jackman, A. L.; Boyle, F. T.; Harrap, K. R. Invest. New Drugs
1996, 14, 305.
(97) Durand, R.; Savel, J. Expert Opin. Ther. Pat. 2001, 11, 1285.
(98) Roth, B. J.; Manola, J.; Dreicer, R.; Graham, D.; Wilding, G.
Invest. New Drugs 2002, 20, 425.
(99) Liu, G.; Bailey, H. H.; Arzoomanian, R. Z.; Alberti, D.; Binger,
K.; Feierabend, C.; Marnocha, R.; Wilding, G.; Thomas, J. P. Am.
J. Clin. Oncol. 2003, 26, 280.
(100) Jackman, A. L.; Melin, C. J.; Kimbell, R.; Brunton, L.; Aherne,
G. W.; Theti, D. S.; Walton, M. Biochim. Biophys. Acta 2002,
1587, 215.
(101) Rosowsky, A.; Wright, J. E.; Vaidya, C. M.; Forsch, R. A.
Pharmacol. Ther. 2000, 85, 191.
(102) Rosowsky, A. Antifolate Drugs in Cancer Therapy; Humana
Press: Totowa, New Jersey, 1999; p 59.
(103) Matherly, L. H.; Angeles, S. M.; McGuire, J. J. Biochem.
Pharmacol. 1993, 46, 2185.
(104) Boggs, A. F.; Hecker, S. J. Expert Opin. Ther. Pat. 2002, 12, 1159.
(105) Schneider, P.; Hawser, S.; Islam, K. Bioorg. Med. Chem. Lett
2003, 13, 4217.
(106) Then, R. L.; Hartman, P. G.; Kompis, I.; Stephan-Guldner, M.;
Stockel, K. Drugs Future 1994, 19, 446.
(107) Locher, H. H.; Schlunegger, H.; Hartman, P. G.; Angehrn, P.;
Then, R. L. J. Antimicrob. Agents Chemother. 1996, 40, 1376.
(108) Dhople, A. M. J. Antimicrob. Agents Chemother. 2001, 47, 93.
(109) Dhople, A. M. Int. J. Antimicrob. Agents 2002, 19, 71.
(110) McKeage, K.; Scott, L. Drugs 2003, 63, 597.
(111) Kinyanjui, S. M.; Mberu, E. K.; Winstanley, P. A.; Jacobus, D.
P.; Watkins, W. M. A. J. Trop. Med. Hyg. 1999, 60, 943.
(112) Wyss, P. C.; Gerber, P.; Hartman, P. G.; Hubschwerlen, C.;
Locher, H.; Marty, H. P.; Stahl, M. J. Med. Chem. 2003, 46, 2304.
(113) Lau, H.; Ferlan, J. T.; Brophy, V. H.; Rosowsky, A.; HopkinsSibley, C. Antimicrob. Agents Chemother. 2001, 45, 187.
(114) Brophy, V. H.; Vasquez, J.; Nelson, R. G.; Forney, J. R.;
Rosowsky, A.; Hokins-Sibley, C. Antimicrob. Agents Chemother.
2002, 44, 1019.
(115) Gerum, A. B.; Ulmer, J. E.; Jacobus, D. P.; Jensen, N. P.;
Sherman, D. R.; Sibley, C. H. Antimicrob. Agents Chemother.
2002, 46, 3362.
(116) Brobey, R. K.; Sano, G.; Itoh, F.; Aso, K.; Kimura, M.; Mitamura,
T.; Horii, T. Mol. Biochem. Parasitol. 1996, 81, 225.
Kompis et al.
(117) Kuyper, L. F.; Baccanari, D. P.; Jones, M.; Hunter, R.; Tansik,
R. L.; Joyner, S. S.; Boytos, C. M.; Rudolph, S. K.; Knick, V.;
Wilson, H. R.; Cadell, J. M.; Friedman, H. S.; Comley, J. C.;
Stables, J. N. J. Med. Chem. 1996, 39, 892.
(118) Donkor, I. O.; Li, H.; Queener, S. F. Eur. J. Med. Chem. 2003,
46, 2304.
(119) Debnath, A. K. J. Med. Chem. 2002, 45, 41.
(120) White, E. L.; Ross, L. J.; Cunningham, A.; Escuyer, V. FEMS
(121) Dhople, A. M. Arzneim.-Forsch. 1999, 49, 267.
(122) Suling, W. J.; Maddry, J. A. J. Antimicrob. Chemother. 2001,
47, 51.
(123) Suling, W. J.; Seitz, L. E.; Pathak, V.; Westbrook, L.; Barrow,
E. W.; Zywno-Van-Ginkel, S.; Piper, J. R.; Barrow, W. W.
(124) Sanders, W. J.; Nienaber, V. L.; Lerner, C. G.; McCall, J. O.;
Merrick, S. M.; Swanson, S. J.; Harlan, J. E.; Stoll, V. S.;
Stamper, G. F.; Betz, S. F.; Condroski, K. R.; Meadows, R. P.;
Severin, J. M.; Walter, K. A.; Magdalinos, P.; Jakob, C. G.;
Wagner, R.; Beutel, B. A. J. Med. Chem. 2004, 47, 1709.
(125) Patel, O.; Satchell, J.; Baell, J.; Fernley, R.; Coloe, P.; Macreadie,
I. Microbiol. Drug Resist. 2003, 9, 139.
(126) Patel, O. G.; Mberu, E. K.; Nzila, A.; Macreadie, I. G. Trends
Parasitol. 2004, 20, 1.
(127) Chan, J. H.; Hong, J. S.; Kuyper, L. F.; Baccanari, D. P.; Joyner,
S. S.; Tansik, R. L.; Boytos, C. M.; Rudolph, S. K. J. Med. Chem.
1995, 38, 3608.
(128) Rosowsky, A.; Forsch, R. A.; Queener, S. F. J. Med. Chem. 2003,
46, 1726.
(129) Gangjee, A.; Adair, O. O.; Queener, S. F. J. Med. Chem. 2003,
46, 5074.
(130) Kovacs, J. A.; Gill, V. J.; Meshnick, S.; Masur, H. J. Am. Med.
Assoc. 2001, 286, 2450.
(131) Kompis, I.; Then, R. L. Eur. J. Med. Chem. 1984, 19, 529.
(132) Kuyper, L. F.; Roth, B.; Baccanari, D. P.; Ferone, R.; Bedell, C.
R.; Champness, J. N.; Stammers, D. K.; Dann, J. G.; Norrington,
F. E.; Baker, D. J.; Goodford, P. J. J. Med. Chem. 1985, 28, 303.
(133) Brun-Pascuad, M.; Rajagopalan-Levasseur, P.; Chau, F.; Bertrand, G.; Garry, L.; Derouin, F.; Girard, P. M. Antimicrob.
(134) Deroiun, F. Curr. Opin. Invest. Drugs 2001, 10, 1368.
(135) Gilbert, I. H. Biochim. Biophys. Acta 2002, 18, 1587.
(136) Marlowe, C. K.; Selassie, C. D.; Santi, D. V. J. Med. Chem. 1995,
38, 967.
(137) Gelhahaar, D. K.; Moerder, K. E.; Zichi, D.; Sherman, C. J.;
Ogden, R. C.; Freer, S. T. J. Med. Chem. 1995, 38, 466.
(138) Chan, D. C.; Laughton, C. A.; Queener, S. F.; Stevens, M. F. J.
Med. Chem. 2001, 44, 2555.
(139) Cody, V.; Chan, D.; Galitsky, N.; Rak, D.; Luft, J. R.; Pangborn,
W.; Queener, S. F.; Laughton, C. A.; Stevens, M. Biochemistry
2000, 39, 3556.
45, 233.
(141) Forsch, R. A.; Queener, S. F.; Rosowsky, A. Bioorg. Med. Chem.
2004, 14, 1811.
(142) Robson, C.; Meek, M. A.; Grunwaldt, J. D.; Lambert, P. A.;
Queener, S. F.; Schmidt, D.; Griffin, R. J. J. Med. Chem. 1997,
40, 3040.
(143) Gangjee, A.; Devraj, R.; McGuire, J. J.; Kisliuk, R. L.; Queener,
S. F.; Barrows, L. R. J. Med. Chem. 1994, 37, 1169.
(144) Gangjee, A.; Guo, X.; Queener, S. F.; Cody, V.; Galitsky, N.; Luft,
J. R.; Pangborn, W. J. Med. Chem. 1998, 41, 1263.
(145) Rosowsky, A.; Papoulis, A. T.; Queener, S. F. J. Med. Chem.
1998, 41, 913.
(146) Rosowsky, A.; Chen, H.; Fu, H.; Queener, S. F. Bioorg. Med.
Chem. 2003, 11, 59.
(147) Gangjee, A.; Mavandadi, F.; Queener, S. F. J. Med. Chem. 1997,
40, 1173.
(148) Gangjee, A.; Mavandadi, F.; Queener, S. F.; McGuire, J. J. J.
Med. Chem. 1995, 38, 2158.
(149) Gangjee, A.; Vasudevan, A.; Queener, S. F. J. Med. Chem. 1997,
40, 3032.
(150) Cody, V.; Galitsky, N.; Rak, D.; Luft, J. R.; Pangborn, W.;
Queener, S. F. Biochemistry 1999, 38, 4303.
(151) Cody, V.; Galitsky, N.; Luft, J. R.; Pangborn, W.; Galitsky, N.;
Queener, S. F.; Blakley, R. L. Acta Crystallogr., Sect D: Biol.
Crystallogr. 1997, 53, 638.
(152) Cody, V.; Galitsky, N.; Luft, J. R.; Pangborn, W.; Queener, S.
F.; Gangjee, A. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2002,
58, 1393.
(153) Cody, V.; Galitsky, N.; Luft, J. R.; Pangborn, W.; Rosowsky, A.;
Queener, S. F. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2002,
58, 946.
(154) Jackson, H. C.; Biggadike, K.; McKilligin, E.; Kinsman, O. S.;
Queener, S. F.; Lane, A.; Smith, J. E. Antimicrob. Agents
Chemother. 1996, 40, 1371.
(155) Rosowsky, A.; Hynes, J. B.; Queener, S. F. Antimicrob. Agents
Chemother. 1995, 39, 79.

(156) Gangjee, A.; Zaveri, N.; Kothare, M.; Queener, S. F. J. Med.
Chem. 1995, 38, 3660.
(157) Gangjee, A.; Vasudevan, A.; Queener, S. F.; Kisliuk, R. L. J. Med.
Chem. 1996, 39, 1438.
(158) Cody, V.; Luft, J. R.; Pangborn, W.; Gangjee, A. Acta Crystallogr.,
Sect. D: Biol. Crystallogr. 2003, 59, 1603.
(159) Gangjee, A.; Devraj, R.; Queener, S. F. J. Med. Chem. 1997, 40,
470.
(160) Piper, J. R.; Johnson, C. A.; Krauth, C. A.; Carter, R. L.; Hosmer,
C. A.; Queener, S. F.; Borotz, S. E.; Pfefferkorn, E. R. J. Med.
Chem. 1996, 39, 1271.
(161) Gangjee, A.; Vasudevan, A.; Queener, S. F. J. Med. Chem. 1997,
40, 479.
(162) Gangjee, A.; Yu, J.; McGuire, J. J.; Cody, V.; Galitsky, N.;
Kisliuk, R. L.; Queener, S. F. J. Med. Chem. 2000, 43, 3837.
(163) Rosowsky, A.; Queener, S. F.; Cody, V. Drug Des. Discovery 1999,
16, 25.
(164) Gangjee, A.; Zeng, Y.; McGuire, J. J.; Kisliuk, R. L. J. Med. Chem.
2002, 45, 5137.
(165) Gangjee, A.; Shi, J.; Queener, S. F. J. Med. Chem. 1997, 40, 1930.
(166) Gangjee, A.; Elzein, E.; Queener, S. F.; McGuire, J. J. J. Med.
Chem. 1998, 41, 1409.
(167) Miller, L. H.; Baruch, D. I.; Marsh, K.; Duombo, O. K. Nature
2002, 415, 498.
(168) Ridley, R. G. Nature 2002, 415, 686.
(169) Debaert, M. Ann. Pharm. Fr. 2001, 59, 75.
(170) Brady, R. L.; Cameron, A. Curr. Drug Targets 2004, 5, 137.
(171) Yuthavong, Y. Microbes Infect. 2002, 4, 175.
(172) Hankins, E. G.; Warhurst, D. C.; Sibley, C. H. Mol. Biochem.
Parasitol. 2001, 117, 91.
(173) Ferlan, J. T.; Mookherjee, S.; Okezie, I.; Fulgence, L.; Sibley, C.
H. Mol. Biochem. Parasitol. 2001, 113, 139.
(174) Sirawaporn, W.; Yongkiettrakul, S.; Sirawaraporn, R.; Yuthawong,
Y.; Santi, D. V. Exp. Parasitol. 1997, 87, 245.
(175) van Dijk, M. R.; McConkey, G. A.; Vinkenoog, R.; Waters, A. P.;
Janse, C. J. Mol. Biochem. Parasitol. 1994, 68, 167.
(176) Leartsakulpanich, U.; Imwong, M.; Pukrittayakamee, S.; White,
N. J.; Snounou, G.; Sirawaraporn, W.; Yuthavong, Y. Mol.
Biochem. Parasitol. 2002, 11, 63.
(177) Basco, L. K.; leBras, J. Bull. Soc. Pathol. Exot. 1997, 90, 90.
(178) Chowdhury, S. F.; Villamor, V. B.; Guerrero, R. H.; Leal, I.; Brun,
R.; Croft, S. L.; Goodman, J. M.; Maes, L.; Ruiz-Perez, L. M.;
Pacanowska, D. G.; Gilbert, I. H. J. Med. Chem. 1999, 42, 4300.
(179) Knighton, D. R.; Kann, C. C.; Howland, E.; Janson, C. A.;
Hostowska, Z.; Welsh, K. M.; Matthews, D. A. Nat. Struct. Biol.
1994, 1, 186.
(180) Zuccotto, F.; Brun, R.; Pacanowska, D. G.; Ruiz Perez, L. M.;
Gilbert, I. H. Bioorg. Med. Chem. Lett. 1999, 9, 1463.
(181) Toyoda, T.; Brobey, R. K.; Sano, G.; Horii, T.; Tomioka, N.; Itai,
A. Biochem. Biophys. Res. Commun. 1997, 27, 515.
(182) Rathod, P. K.; Phillips, M. A. Nat. Struct. Biol. 2003, 10, 316.
(183) Meyer, C. C.; Majumder, S. K.; Cynamon, M. H. Antimicrob.
(184) Kamchonwongpaisan, S.; Quarrell, R.; Charoensetakul, N.;
Ponsinet, R.; Vilaivan, T.; Vanichtanankul, J.; Tarnchompoo, B.;
Sirawaraporn, W.; Lowe, G.; Yuthavong, Y. J. Med. Chem. 2004,
47, 673.
(185) Schroeder, F.; Schroeder, J. O. Dtsch. Med. Wochenschr. 2000,
125, 1435.
(186) Kopytek, S. J.; Dyer, J. C.; Knapp, G. S.; Hu, J. C. Antimicrob.
(187) Andersson, S. E.; Johansson, H. L.; Lexmuller, K.; Ekstrom, G.
M. Eur. J. Pharm. Sci. 2000, 9, 333.
(188) Goodsell, D. S. Stem Cells 1999, 17, 314.
(189) Schilsky, R. L. Oncologist 1996, 1, 244.
(190) Blakley, R. L. Hum. Mutat. 1998, 11, 259.
(191) Tsukamoto, T.; Kitazume, T.; McGuire, J. J.; Coward, J. K. J.
Med. Chem. 1996, 39, 66.
(192) Hart, B. P.; Haile, W. H.; Licato, N. J.; Bolanowska, W. E.;
McGuire, J. J.; Coward, J. K. J. Med. Chem. 1996, 39, 56.
(193) Graffner-Nordberg, M.; Fyfe, M.; Brattsand, R.; Mellgard, B.;
Hallberg, A. J. Med. Chem. 2003, 46, 3455.
(194) Forsch, R. A.; Wright, J. E.; Rosowsky, A. Bioorg. Med. Chem.
2002, 10, 2067.
(195) Reynolds, R. C.; Johnson, C. A.; Piper, J. R.; Sirotnak, F. M.
Eur. J. Med. Chem. 2001, 36, 237.
2000, 43, 3125.
2002, 45, 1942.
(198) Vaidya, C. M.; Wright, J. E.; Rosowsky, A. J. Med. Chem. 2002,
45, 1690.
(199) Wright, J. E.; Vaidya, C. M.; Chen, Y.; Rosowsky, A. Biochem.
Pharmacol. 2000, 60, 41.
(200) Rosowsky, A.; Wright, J. E.; Vaidya, C. M.; Bader, G.; Forsch,
R. A.; Mota, C. E.; Pardo, J.; Chen, C. S.; Chen, Y. N. J. Med.
Chem. 1998, 41, 5310.
(201) Rosowsky, A.; Wright, J. E.; Vaidya, C. M.; Forsch, R. A.; Bader,
H. J. Med. Chem. 2000, 43, 1620.

(202) Rosowsky, A.; Bader, H.; Wright, J. E.; Keyomarsi, K.; Matherly,
L. H. J. Med. Chem. 1994, 37, 2167.
(203) Gangjee, A.; Yu, J.; Kisliuk, R. L.; Haile, W. H.; Sobrero, G.;
McGuire, J. J. J. Med. Chem. 2003, 46, 591.
(204) Gangjee, A.; Adair, O. O.; Queener, S. F. J. Med. Chem. 2003,
46, 5074.
(205) Rosowsky, A.; Forsch, R. A.; Null, A.; Moran, R. G. J. Med. Chem.
1999, 42, 3510.
(206) McGuire, J. J.; Magee, K. J.; Russell, C. A.; Canstrari, J. M.
Oncol. Res. 1997, 9, 139.
(207) Lehman, N. L. Expert Opin. Invest. Drugs 2002, 11, 1175.
(208) Ackland, S. P.; Clarke, S. J.; Beale, P.; Peters, G. J. Cancer
Chemother. Biol. Response Modif. 2002, 20, 1.
(209) Huang, Y.; Lin, C. F.; Lee, Y. J.; Li, W. W.; Chao, T. C.;
Bacherikov, V. A.; Chen, K. T.; Chen, C. M.; Su, T. L. Bioorg.
Med. Chem. 2003, 11, 145.
(210) Rosowsky, A. Curr. Med. Chem. 1999, 6, 329.
(211) Gangjee, A.; Dubash, N. P.; Zeng, Y.; McGuire, J. J. Curr. Med.
Chem. Anticancer Agents 2002, 2, 331.
(212) McGuire, J. J.; Haile, W. H. Arch. Biochem. Biophys. 1996, 335,
139.
(213) Mao, Z.; Pan, J.; Kalman, T. I. J. Med. Chem. 1996, 39, 4340.
(214) Tsukamoto, T.; Haile, W. H.; McGuire, J. J.; Coward, J. K. Arch.
Biochem. Biophys. 1998, 355, 109.
(215) Kisliuk, R. L. Curr. Pharm. Des. 2003, 9, 2615.
(216) Boger, D. L.; Labroli, M. A.; Marsilje, T. H.; Jin, Q.; Hedrick,
M. P.; Baker, S. J.; Shim, J. H.; Benkovic, S. J. Bioorg. Med.
Chem. 2000, 8, 1075.
(217) Huovinen, P.; Sundstrom, L.; Swedberg, G.; Skold, O. Antimicrob. Agents Chemother. 1995, 39, 279.
(218) Huovinen, P. Clin. Infect. Dis. 1997, 24 (Suppl. 1), 63.
(219) Huovinen, P. Clin. Infect. Dis. 2001, 32, 1608.
(220) Skold, O. Vet. Res. 2001, 32, 261.
(221) Wernsdorfer, W. H.; Noedl, H. Curr. Opin. Infect. Dis. 2003, 16,
553.
(222) Adrian, P. V.; Klugman, K. P. Antimicrob. Agents Chemother.
1997, 41, 2406.
(223) Maskell, J. P.; Sefton, A. M.; Hall, L. M. C. Antimicrob. Agents
Chemother. 2001, 45, 1104.
(224) Dale, G. E.; Broger, C.; DArcy, A.; Hartman, P. G.; DeHoogt,
R.; Jolidon, S.; Kompis, I.; Labhardt, A. M.; Langen, H.; Locher,
H.; Page, M. G.; Stuber, D.; Then, R. L.; Oefner, C. J. Mol. Biol.
1997, 266, 23.
(225) Dam, J.; Rose, T.; Goldberg, M. E.; Blondel, A. J. Mol. Biol. 2000,
302, 235.
(226) Swedberg, G.; Ringertz, S.; Skold, O. Antimicrob. Agents Chemother. 1998, 42, 1062.
(227) May, J.; Meyer, C. G. Trends Parasitol. 2003, 19, 432.
(228) Talisuna, A. O.; Bloland, P.; DAllesandro, U. Clin. Microbiol.
Rev. 2004, 17, 235.
(229) Hayton, K.; Su, X. Z. Curr. Drug Targets Infect. Disord. 2004,
4, 1.
(230) Warhurst, D. C. Sci. Prog. 2002, 85, 89.
(231) Basco, L. K.; de Pecoulas, P. E.; Wilson, C. M.; Le Bras, J.;
Mazabraud, A. Mol. Biochem. Parasitol. 1995, 69, 135.
(232) Yuthavong, Y. Microbes Infect. 2002, 4, 175.
(233) Wang, P.; Lee, C. S.; Bayuomi, R.; Dijmde, A.; Doumbo, O.;
Svedberg, G.; Dao, L. D.; Mshinda, H.; Tanner, M.; Watkins,
W. M.; Sims, P.; Hyde, J. E. Mol. Biochem. Parasitol. 1997, 89,
161.
(234) Plowe, C. V.; Kublin, J.; Duombo, O. K. Drug Resist. Updates
1998, 1, 389.
(235) Cowman, A. F. Malaria: Parasite Biology, Pathogenesis, and
Protection; ASM Press: Washington, DC, 1998; Chapter 22, p
317.
(236) Durand, R.; Eslahpazire, J.; Jafari, S.; Delabre, J. F.; MarmoratKhuong, A.; DiPiazza, J.-P.; leBras, J. Antimicrob. Agents
Chemother. 2000, 44, 3461.
(237) Sirawaraporn, W.; Sirawaraporn, R.; Yongkiettrakul, S.; Anuwatwora, A.; Rastelli, G.; Kamchonwongpaisan, S.; Yuthavong,
Y. Mol. Biochem. Parasitol. 2002, 121, 185.
(238) Durand, R.; DiPiazza, J. P.; Longuet, C.; Secardin, Y.; Clain, J.;
leBras, J. Ann. Trop. Med. Parasitol. 1999, 93, 25.
(239) Iyer, J. K.; Milhous, W. K.; Cortese, J. F.; Kublin, J. G.; Plowe,
C. V. Lancet 2001, 358, 1066.
(240) Alifrangis, M.; Enosse, S.; Khalil, I. F.; Tarimo, D. S.; Lemnge,
M. M.; Thompson, R.; Bygbjerg, I. C.; Ronn, A. M. Am. J. Trop.
Med. Hyg. 2003, 69, 601.
(241) Syafruddin, D.; Asih, B. P.; Aggarwal, S. L.; Shankar, A. H. Am.
J. Trop. Med. Hyg. 2003, 69, 614.
(242) Ochong, E.; Nzila, A. M.; Kimani, S.; Kokwaro, G.; Mutabingwa,
T.; Watkins, W.; Marsh, K. Malar. J. 2003, 2, 46.
(243) Mberu, E. K.; Nzila, A. M.; Nduati, E.; Ross, A.; Monks, S. M.;
Kokwaro, G. O.; Watkins, W. M.; Sibley, C. H. Exp. Parasitol.
2002, 101, 90.
(244) Berglez, J.; Iliades, P.; Sirawaraporn, W.; Coloe, P.; Macreadie,
I. Int. J. Parasitol. 2004, 34, 95.
(245) Khalil, I.; Ronn, A.; Alifrangis, M.; Gabar, H. A.; Satti, G. M.;
Bygbjerg, I. C. Am. J. Trop. Med. Hyg. 2003, 68, 586.

(246) Marks, F.; Meyer, C. G.; Sievertsen, J.; Timmann, C.; Evans,
J.; Horstmann, R. D.; May, J. Antimicrob. Agents Chemother.
2004, 48, 466.
(247) Fohl, L. M.; Roos, D. S. Mol. Microbiol. 2003, 50, 1319.
(248) Reynolds, M. G.; Oh, J.; Roos, D. S. Antimicrob. Agents Chemother. 2001, 45, 1271.
(249) Takahashi, T.; Endo, T.; Nakamura, T.; Sakashitat, H.; Kimurat,
K.; Ohnishit, K.; Kitamura, Y.; Iwamoto, A. J. Med. Microbiol.
2002, 51, 510.
(250) Nahimana, A.; Rabodonirina, M.; Helweg-Larsen, J.; Meneau,
I.; Francioli, P.; Bille, J.; Hauser, P. M. Emerging Inf. Dis. 2003,
9, 864.
(251) Nahimana, A.; Rabodonirina, M.; Francioli, P.; Bille, J.; Hauser,
P. M. Eukaryotic Microbiol. 2003, 50 (Suppl.), 656.
(252) Ma, L.; Kovacs, J. A.; Cargnel, A.; Valerio, A.; Fantoni, G.; Atzori,
C. J. J. Infect. Dis. 2002, 185, 1530.
(253) Safrin, S.; Lee, B. L.; Sande, M. A. J. Infect. Dis. 1994, 170, 912.
(254) Razavi, B.; Lund, B.; Allen, B. L.; Schlessinger, L. Infection 2002,
30, 41.
(255) Jansen, G.; Barr, H.; Kathmann, I.; Bunni, M. A.; Priest, D. G.;
Noordhuis, P.; Peters, G. J.; Assaraf, Y. G. Mol. Pharmacol.
1999, 55, 761.
Kompis et al.
(256) Banerjee, D.; Mayer-Kuckuk, P.; Capiaux, G.; Budak-Alpdogan,
T.; Gorlick, R.; Bertino, J. R. Biochim. Biophys. Acta 2002, 1587,
164.
(257) Brzezinska, A.; Wilska, P.; Baliska, M. Acta Biochim. Pol. 2000,
47, 735.
(258) Liani, E.; Rothem, L.; Bunni, M. A.; Smith, C. A.; Jansen, G.;
Assaraf, Y. G. Int. J. Cancer 2003, 103, 587.
(259) Kanter, P. P.; McIvor, R. S.; Benkovic, S. J.; Rosowsky, A.;
Wagner, C. R. J. Med. Chem. 2003, 46, 2816.
(260) Gangjee, A.; Zhu, Y.; Queener, S. F. J. Med. Chem. 1998, 41,
4533.
(261) Gangjee, A.; Zhu, Y.; Queener, S. F.; Francom, P.; Broom, A. D.
J. Med. Chem. 1996, 39, 1836.
(262) Gangjee, A.; Adair, O. O.; Queener, S. F. Biorg. Med. Chem. 2001,
11, 2929.
38, 2615.
CR0301144
Chem. Rev. 2005, 105, 621632
621
RifamycinsMode of Action, Resistance, and Biosynthesis

Heinz G. Floss* and Tin-Wein Yu
Department of Chemistry, Box 351700, University of Washington, Seattle, Washington 98195-1700
Contents
1. Introduction
2. Mechanism of Action
3. Rifamycin Resistance
3. 1. Resistance Due to Modification of rpoB
3. 2. Other Resistance Mechanisms
4. Rifamycin Biosynthesis
5. Autoresistance of Amycolatopsis Mediterranei
6. Concluding Remarks
7. Acknowledgment
8. References
621
621
623
623
624
626
629
630
631
631
1. Introduction
The rifamycins1 belong to the family of ansamycin
antibiotics,3,4 so named because of their basket-like
molecular architecture comprising an aromatic moiety bridged at nonadjacent positions by an aliphatic
chain (Latin: ansa ) handle).5 The aromatic moiety
can either be a naphthalene or a naphthoquinone
ring system, as in the naphthalenic ansamycins
rifamycin and naphthomycin6 or the streptovaricins,7
or it can be a benzene or benzoquinone ring, as in
the benzenic ansamycins geldanamycin8 or ansamitocin9 (Figure 1). The rifamycins were first isolated
by Sensi and co-workers at Lepetit SA in Milan in
1959 as a complex mixture of several congeners.10
The producing organism, an Actinomycete, was originally classified as Streptomyces mediterranei,11 then
reclassified as Nocardia mediterranea,12 and finally
assigned to a newly defined genus as Amycolatopsis
mediterranei.13 Fermentation in the presence of
added diethylbarbituric acid led to the production of
predominantly rifamycin B,14 the structure of which
was determined by chemical means and X-ray
crystallography.15-18 Subsequently, it was possible by
mutagenesis of the producing organism to eliminate
the requirement for addition of diethylbarbituric acid
to the fermentation.19 Since then, numerous other
rifamycins have been isolated from the fermentation
of A. mediterranei or its mutants and their structures
have been determined.19 Very closely related compounds have been isolated from other Actinomycetes,
for example, tolypomycin (together with rifamycins
B and O) from Amycolatopsis tolypomycina20,138 and
the halomicins from Micromonospora halophytica
(Figure 2).21
The rifamycins display a broad spectrum of antibiotic activity against Gram-positive and, to a lesser
extent, Gram-negative bacteria.22 Rifamycin B, the
product of the commercial fermentation, has only

very modest activity, but it can be converted chemically, enzymatically, or by biotransformation into
rifamycin SV (Figure 3) (cf. ref 19),23-25 which has
much more potent activity and was the first rifamycin used clinically.26 Rifamycin SV is a biosynthetic
precursor of rifamycin B,27 and mutagenesis of the
producing organism has succeeded in blocking the
terminal conversion step, resulting in the accumulation of rifamycin SV.28 However, strain optimization
of this mutant has not been as successful as that of
its rifamycin B-producing parent strain, and most
commercial fermentations appear to produce rifamycin B, which is then converted into rifamycin SV.19
Following the clinical introduction of rifamycin SV,
extensive programs of semisynthesis, primarily at the
Lepetit group and at Ciba-Geigy, led to the preparation and evaluation of large numbers of analogues
of rifamycin.2,3,29,30 From these, rifampicin (Figure 3)
was selected as the next generation clinical candidate.23 Rifampicin shows more pronounced activity
against Gram-positive bacteria, particularly mycobacteria, better activity against Gram-negative bacteria, and importantly, excellent oral bioavailability.31
It has become one of the mainstay agents in the
treatment of tuberculosis, leprosy, and AIDS-associated mycobacterial infections.32 Since resistance to
rifampicin develops rather rapidly,33 the drug is
typically used in combination with other antimycobacterial agents, particularly isoniazid.34 Other semisynthetic rifamycin derivatives, such as rifabutin35
and rifapentine36 (Figure 3), were subsequently introduced for clinical use. Rifabutine in particular is
active against a number of rifampicin-resistant clinical pathogens.35
2. Mechanism of Action
The antibacterial action of rifampicin results from
its inhibition of DNA-dependent RNA synthesis.37
This inhibition is due not to interaction with the
template, but to strong binding to the DNA-dependent RNA polymerase of prokaryotes.38 Binding constants for prokaryotic RNA polymerases are in the
range of 10-8 M; eukaryotic enzymes are at least 102
to 104 times less sensitive to inhibition by rifampicin.39 The inhibition of DNA-dependent RNA polymerase seems to be the common mechanism for all
antibacterially active rifamycins;38 the many structural modifications made in these molecules primarily alter the pharmacokinetic properties of the
molecules4 and their affinity for eukaryotic DNA-
10.1021/cr030112j CCC: $53.50 2005 American Chemical Society

A native of Berlin, Germany, Professor Heinz G. Floss studied chemistry

at the Technical Universities of Berlin and Munich, obtaining his Ph.D. in
1961. Following a postdoctorate in biochemistry at UC Davis, he joined
the faculty of the Department of Medicinal Chemistry and Pharmacognosy
at Purdue University in 1966, where he rose through the ranks to the
position of Lilly Distinguished Professor and Head of Department. After
serving as Professor of Chemistry and Department Chair at The Ohio
State University from 19821987, he moved to the Department of
Chemistry, University of Washington, Seattle. At the end of 2000 he was
awarded Emeritus status, but he continues to pursue his research interests
in the biosynthesis of natural products. He published about 400 papers,
and his honors include the NIH Research Career Development Award,
the Research Achievement Award in Natural Products of the Academy of
Pharmaceutical Sciences, the Volwiler Award of the American Association
of Colleges of Pharmacy, the Humboldt US Senior Scientist Award,
Honorary Doctor of Science degrees from Purdue University and from
the University of Bonn, Germany, the Research Award of the American
Society of Pharmacognosy, the Kitasato Award in Microbial Chemistry,
and Honorary Membership of the Kitasato Institute, Tokyo.
Tin-Wein Yu received a Ph.D. in Biology under Professor Sir David A.

Hopwood with an emphasis on microbial genetics, from the University of
East Anglia, England, in 1995. He did postdoctoral research with Professor
Heinz G. Floss at the University of Washington and is currently an assistant
Professor at the Department of Biological Sciences in Louisiana State
University. His research interests include bioinformatics-based natural
product discovery, bioactive molecules, and biosynthetic networks in living
organism.
dependent RNA polymerases,40 but do not change the

principal mechanism of action.
The interaction of rifampicin with DNA-dependent
RNA polymerase has been studied in considerable
detail since its initial discovery, culminating recently
in the solution of the crystal structure of the complex
of rifampicin with the enzyme from Thermus aquaticus and the development of a detailed model for
rifampicins mechanism of inhibition of RNA synthesis by this enzyme.41 Early work had revealed that
Floss and Yu
the antibiotic interferes with the initiation phase of

RNA synthesis;42,43 once RNA synthesis has progressed beyond an early stage and the enzyme carries
a longer oligoribonucleotide chain, the process is no
longer sensitive to rifampicin.44 DNA-dependent RNA
polymerase is a complex enzyme with an R2
subunit structure. The binding site for rifampicin has
been located at the subunit encoded by the rpoB
gene. This follows from the construction of chimeric
RNA polymerases containing subunits from rifampicin-sensitive and rifampicin-resistant strains.45,46 An
R2 complex, the core enzyme, binds rifampicin,
that is, the subunit is not required,47 but the
subunit is necessary for rifampicin binding.48 A
seminal observation was made by McClure and
Cech,49 who reported that the rifampicin-inhibited
RNA polymerase released dinucleotides when the
reaction was initiated with a nucleoside triphosphate,
whereas the reaction was terminated after the second
phosphodiester bond formation when it was initiated
with nucleoside di- or monophosphates. This led them
to propose that the binding of rifampicin to the RNA
polymerase-DNA complex sterically blocks the extension of the nascent RNA chain after the first or
second condensation step. Although other mechanisms have been proposed, for example, allosteric
effects on protein conformation,50 the crystal structure of the core RNA polymerase-rifampicin complex
bears out the mechanism proposed by McClure and
Cech.41
The 3.3 crystal structure of the complex of T.
aquaticus core DNA-dependent RNA polymerase
with rifamicin was solved and interpreted following
the earlier determination of the structure of the
enzyme alone by the same group51 and the development of a functional model for the enzyme based on
the structural data.52 According to the structure of
the complex,41 rifampicin binds to the subunit deep
within the main DNA/RNA channel, about 12 away
from the active site Mg2+ ion, consistent with biochemical data.53,54 Binding to the protein involves
hydrogen bonding interactions between the four
hydroxyl groups at C-1, C-8, C-21, and C-23, which
are essential for biological activity of rifamycins,55 as
well as the carbonyl oxygen of the C-25 acetoxy
group, and amino acid residues R409, S411, Q393,
H406, D396, and F394. In addition, hydrophobic
interactions with E445, I452, G414, L413, L391, and
Q390 contribute to the binding of the antibiotic
(Figure 4). The position of the bound antibiotic is such
that it physically gets in the way of the growing
oligonucleotide chain after the first or second chain
elongation step. The model clearly indicates that
rifamycin does not inhibit the initiation step or the
translocation step, and it explains why the enzyme
after a number of elongation steps, when it carries a
longer RNA chain, is no longer sensitive to inhibition
by rifamycin. It also neatly explains the release of
di- or trinucleotides from the inhibited enzyme.
Comparative studies with DNA-dependent RNA
polymerases from Gram-negative and Gram-positive
bacteria have shown that they have comparably high
sensitivities to rifampicin. Thus, the lower sensitivity
of Gram-negative bacteria to rifamycins must be due
Figure 1. Structures of representative ansamycins.
almost exclusively in drug combinations, most commonly with isoniazid,63 and why its use, at least in
the United States, is restricted to the treatment of
tuberculosis or clinical emergencies.cf 41 By far the
predominant mechanism of resistance to rifamycins
is modification of the drug target, rpoB, by mutation.
Resistance by modification of the antibiotic (inactivation) has also been described, but its clinical significance, at least in M. tuberculosis, does not seem to
be as high.
3.1. Resistance Due to Modification of rpoB

Figure 2. Structures of tolypomycin Y and halomicin B.
to poorer penetration of the antibiotic through the

cell membrane.33 These transport problems can be
partially overcome by structural modifications, which
is apparently the reason that rifampicin has better
activity against Gram-negative bacteria than the
original rifamycins, such as rifamycn SV.19
Rifamycins were also investigated for their potential as antitumor agents and, based on potential
inhibition of reverse transcriptase,56 as antiviral
agents.57 However, these inhibitory effects were not
potent and/or selective enough to lead to clinical
candidates.19 Rifamycins also interact with some
other cellular targets. For example, rifampicin displays some immunosuppressive activity in addition
to toxic side effects on the liver.58 The former effect
was traced to binding to and activation of the human
glucocorticoid receptor by rifampicin.59 There are also
reports that rifampicin inhibits multidrug resistance
and enhances anticancer drug accumulation in multidrug-resistant cells60 due to down-modulation of
P-glycoproteins.61
3. Rifamycin Resistance
Pathogens develop resistance to rifampicin at a
high rate, 10-8 to 10-9 per bacterium per cell division.33,62,63 This is the reason the antibiotic is used
The vast majority of mutations to rifampicin resistance map to the rpoB gene in E. coli46,64,65 as well
as in M. tuberculosis66-68 and other microorganisms68,69 examined (Figure 5). Following the primary
structure determination of E. coli rpoB by Ovchinnikov and co-workers,70 several laboratories analyzed
RifR mutants of E. coli for the nature of the mutations.71-76 It was found that 95% of these mapped to
four small regions in the N-terminal half of the
encoded protein, the vast majority to region I spanning amino acids 505-537 (E. c. numbering) (Figure
5).68 Most of these mutations are point mutations
resulting in single amino acid substitutions, with a
few deletions or insertions. The rif I region of rpoB
is rather highly conserved among prokaryotic organisms, but not between prokaryotes and eukaryotes,
such as yeast, Drosophila melanogaster, and humans.41 The different mutations of prokaryotic rpoB
genes lead to different levels of rifamycin resistance;
that is, insusceptibility of rpoB to rifamycins is not
an all or nothing phenomenon.33 Different mutants
also display different degrees of fitness, that is,
normal or impaired growth patterns.63 RifR mutations
in other microorganisms similarly mapped to equivalent regions in their respective rpoB genes.77-82 In
the rpoB gene of M. tuberculosis, all but one mutation
mapped to the rif I region spanning amino acids
419-451 (M. t. numbering) (Figure 5), with 41% of
Floss and Yu
Figure 3. Conversion of rifamycin B into rifamycin SV and clinically used derivatives.
Figure 4. Interaction of rifampicin with proximal amino

acids of Thermus aquaticus DNA-dependent RNA polymerase in the antibiotic-enzyme complex. Amino acid
numbers refer to the Thermus aquaticus RpoB, except for
the three shown in parentheses, which represent the
changes in the rifamycin-resistant Amycolatopsis mediterranei RpoB. (Modified with permission from ref 41. Copyright 2001 Elsevier.)
the resistant clinical isolates carrying a mutation of

S455, 36% of H440, and 9% of D430.67 Although some
other mutations to rifampicin resistance are induced
at a high rate, they do not manifest themselves in
clinical isolates, presumably due to reduced fitness
in a competitive environment.63
The structural work on rpoB from T. aquaticus
shows that, of the twelve amino acids involved in
hydrogen bonding or van der Waals interactions with
the bound rifampicin, all but one (E445) are susceptible to mutation to rifampicin resistance.41 It may
be assumed that mutation of E445 impairs the
function of the enzyme sufficiently to make this a
lethal mutation. The three amino acids most frequently mutated in resistant clinical isolates of
M. tuberculosis, corresponding to H406, S411, and
D396 (T. a. numbering), are involved in hydrogen
bonding interactions with the oxygens at C-8 and
C-21. The remaining 12 of the 23 sites known to
be susceptible to mutation to rifampicin resistance
do not make direct contact with the bound antibiotic
but are located in a second sphere and are likely
to affect rifamycin binding through subtle changes
in the structure of the mutated protein.41 The mutations of rpoB to rifampicin resistance result in a
decreased affinity of the enzyme to the antibiotic,
which binds to the wild-type protein in a very
tight one-to-one complex. This decreased affinity
between antibiotic and target correlates with the
decreased susceptibility of the organism to inhibition
by rifampicin.33
3.2. Other Resistance Mechanisms

Different prokaryotic organisms show different
degrees of susceptibility to inhibition by rifamycins.
In some instances, this is due to decreased sensitivity
of the DNA-dependent RNA polymerase to inhibition
by rifamycin. For example, the enzyme from T.
aquaticus is intrinsically less sensitive than that from
M. tuberculosis.41 In other cases, including M. smegmatis83 and Pseudomonas fluorescens, however, different mechanisms of resistance seem to be operating, including impeded cellular uptake of the antibiotic.84-86 The about 200-fold higher sensitivity of
E. coli to the rifamycin derivative CGP 4832,87
Figure 5. Regions of the rpoB genes from E. coli, Thermus aquaticus, Amycolatopsis mediterranei, and Mycobacterium
tuberculosis carrying mutations which confer rifampicin resistance upon the enzyme. The three amino acids highlighted
in the A. mediterranei RpoB, N447, D438, and Q432 are responsible for the rifampicin resistance of this enzyme. (Modified
with permission from ref 41. Copyright 2001 Elsevier.)
Figure 6. Inactivation products of rifampicin generated by different rifampicin-resistant bacteria.
compared to rifampicin, is due to its active cellular

uptake via the FhuA-TonB transport system.88-90
There is, however, little evidence for a role of permeability barriers in acquired high-level rifamycin
resistance in M. tuberculosis and M. leprae.66,77
Another mechanism, antibiotic modification, has
been demonstrated in a number of microorganisms.
Dabbs first reported rifampicin inactivation by a
Rhodococcus species and by Mycobacterium smegmatis through an inducible mechanism requiring de
novo protein synthesis.91 A gene responsible for this
activity was subsequently cloned from nocardioform
DNA92 and later from M. smegmatis (arr gene).93 The
products of this modification were shown to be 23-
O-R-D-ribosylrifampicin and 3-formyl-23-O-R-D-ribosylrifamycin SV (Figure 6), both antibacterially inactive compounds.94,95 Ribosylation of rifampicin
contributes significantly to the natural low susceptibility of M. smegmatis to rifampicin; inactivation
of the arr gene changed the MIC for rifampicin from
20 to 1.5 g/mL.93 Homologues of the arr gene, arr2, have also been isolated from a multiply resistant
strain of P. aeruginosa from a patient in Thailand,96
from a clinical isolate of Klebsiella pneumoniae,97 and
from some Enterobacteriaceae.98,99 In most of these
cases, the gene was located on and could be transferred by a plasmid. Two other modes of chemical
inactivation of rifampicin have been reported in
Floss and Yu
Figure 7. rif biosynthetic gene cluster and flanking genes encoding ribosomal proteins and the and subunits of
DNA-dependent RNA polymerase.
pathogenic Nocardia species which are naturally

rifamycin-resistant. Nocardia brasiliensis converts
rifampicin into the 23-O--D-glucosyl derivative and
into 3-formyl-23-O--D-glucosylrifamycin SV,94,100 and
N. otidiiscaviarum metabolizes the compound to
21-O-phosphorylrifampicin and 3-formyl-21-O-phosphorylrifamycin SV (Figure 6).94,101 Consistent with
the essential role of the 21 and 23 OH groups, all
four compounds lack antibacterial activity. In a
subsequent survey, the distribution of the three modification mechanisms among various Nocardia and
Mycobacterium species and related taxa was examined.102 The inactivation mechanisms were found to
be rather species specific. It is clear that antibiotic
modification plays a role in clinical resistance of nonmycobacterial species and perhaps also in M. avium,
whereas it seems much less prominent in the clinical
resistance of M. tuberculosis and M. leprae.
4. Rifamycin Biosynthesis
Feeding experiments with isotopically labeled precursors as well as mutagenesis and complementation
experiments have demonstrated that rifamycin is a
polyketide assembled from an aromatic starter unit,
3-amino-5-hydroxybenzoic acid (AHBA), through chain
extension by two acetate and eight propionate units
(cf. ref 19). The origin of the AHBA starter unit is
related to the shikimate pathway, but shikimate or
earlier intermediates of the pathway were not incorporated.103 Rather, their amino analogues, 3,4-dideoxy4-amino-D-arabino-heptulosonic acid 7-phosphate (aminoDAHP), 5-deoxy-5-amino-3-dehydroquinic acid
(aminoDHQ), and 5-deoxy-5-amino-3-dehydroshikimic
acid (aminoDHS) were efficiently converted into
AHBA in cell-free extracts of A. mediterranei.104 On
the basis of this information, the enzyme AHBA
synthase converting aminoSA into AHBA was purified to homogeneity and the gene encoding it was
cloned from A. mediterranei genomic DNA by reverse
genetics.105 This gene was then used as a probe to
isolate cosmids from a cosmid library of A. mediterranei DNA which carried this gene and other rifamycin biosynthetic genes. Further chromosome walking allowed the sequencing and analysis of about 96
kb of DNA contiguous with the AHBA synthase gene,
which accounted for most of the genes considered

necessary for rifamycin biosynthesis.106 The rif PKS
was also cloned independently by Schupp et al.,107
and the cluster was subsequently expanded to include additional genes located upstream of rifA (see
no. 5)108 (Figure 7, Table 1).
The rif cluster contains a set of genes, rifG through
rifN, which were shown by inactivation and heterologous expression to be involved in the biosynthesis
of AHBA.109 The role of three of these genes, rifL,
rifM, and rifN, was obscure for some time, but recent
work indicates that the formation of aminoDAHP is
much more complicated than originally anticipated103
and invokes participation of these three genes, as
well as rifK, in a second role as a transaminase, in
the formation of the AHBA precursor, kanosamine
(Figure 8).110-112 Five large open reading frames, rifA
through rifE, encode a type I modular polyketide
synthase (PKS)106,107,113 with a loading module114
which was identified as a non-ribosomal peptide
synthase (NRPS) adenylation/thiolation didomain;115,116
that is, the rif PKS is actually a hybrid NRPS/PKS.
rifA-E are responsible for the assembly of a linear
undecaketide and are followed by rifF encoding an
amide synthase which catalyzes the release of this
undecaketide and its cyclization to a macrolactam.
The Rif F protein was heterologously expressed and
purified, and its structure was modeled by the group
of Sim.117 The function of the amide synthase as the
terminating enzyme was demonstrated by inactivation of the rifF gene, which, surprisingly, led to the
accumulation of a series of linear ketides ranging
from a tetraketide to the undekaketide.118,119 It was
subsequently found that traces of the same ketides
are present in fermentations of the wild-type organism and that a type II thiosterase (encoded by rifR)
is not responsible for the premature shedding of these
assembly intermediates from the NRPS/PKS.120 The
structures of the accumulated ketides revealed that
the ring closure of the aromatic moiety from a
benzenoid to a naphthalenic structure must occur on
the PKS between the third and the fourth chain
extension steps. The tetraketide has a benzenoid
Table 1. Homologies and Putative Functions of rif Genes from Amycolatopsis mediterranei S699
ORF
RplJ
(191 aa)
RplL
(127 aa)
Orf21
(390 aa)
Orf22
(250 aa)
Orf23
(278 aa)
Orf24
(441 aa)
Orf25
(342 aa)
Orf26
(328 aa)
Orf27
(394 aa)
Orf28
(390 aa)
Orf29
(421 aa)
Orf30
(191 aa)
Orf31
(346 aa)
Orf32c
(340aa)
RifS
(322 aa)
RifT
(255aa)
Orf35
(75 aa)
Orf0
(396 aa)
RifA
(4735 aa)
RifB
(5060 aa)
RifC
(1763 aa)
RifD
(1728 aa)
RifE
(3413 aa)
RifF
(260 aa)
Orf1
(62 aa)
RifG
(351 aa)
RifH
(441 aa)
RifI
(263 aa)
RifK
(388 aa)
RifL
(358 aa)
RifM
(232 aa)
RifN
(235 aa)
properties or content
ribosomal protein L10
ribosomal protein L7/L12
possible ABC transport ATP-binding protein
putative ABC transporter integral membrane protein
putative ABC transporter permease protein
(Rv0168, 289 aa/50%; Rv1965, 271 aa/48%;
Rv3500c, 280 aa/51%; Rv0588, 295 aa/54%)
putative secreted protein: virulence factor mce family
protein (mce1, 454aa/32%; mce2, 404 aa/29%; mce3,
425 aa/31%; mce4, 400 aa/30%)
putative lipoprotein: virulence factor mce family protein
(Rv0170, 346 aa/38%; Rv1967, 342 aa/37%; Rv3498c,
350 aa/36%; Rv0590, 275 aa/40%)
putative lipoprotein: virulence factor mce family protein
(Rv1968, 410 aa/37%; Rv0171, 515 aa/33%; Rv3497c,
357 aa/32%; Rv0591, 481 aa/30%)
putative secreted protein: virulence factor mce family
protein (Rv1969, 423 aa/35%; Rv0172, 530 aa/31%;
Rv3496c, 451 aa/36%; Rv0592, 508 aa/33%)
putative secreted protein: virulence factor mce
family protein (lprM, 377 aa/36%; lprN, 384 aa/37%;
lprK, 390 aa/35%; lprL, 402 aa/31%)
putative secreted protein: virulence factor mce
family protein (Rv1971, 437 aa/33%; Rv3494c,
564 aa/31%; Rv0174, 515 aa/31%; Rv0594, 516 aa/31%)
putative membrane protein (RNA polymerase
sigma-54 factor, RpoN)
putative integral membrane protein: similar to zinc
finger type transcription factor MZF-3
conserved hypothetical protein: a member of the
lipocalin superfamily
putative NADH-dependent dehydrogenase
putative NADH-dependent dehydrogenase
hypothetical protein
cytochrome-P450-like protein
rifamycin polyketide synthase protein (Loading
domain: AD, ACP. Module 1: KS, AT, DH*,
KR, ACP. Module 2: KS, ATm, ACP. Module 3:
KS, AT, KR*, ACP)
rifamycin polyketide synthase protein (Module 4:
KS, AT, DH, KR, ACP. Module 5: KS, AT, DH*,
KR, ACP. Module 6: KS, AT, DH, KR, ACP)
rifamycin polyketide synthase protein
(Module 7: KS, AT, DH, KR, ACP)
rifamycin polyketide synthase protein
(Module 8: KS, AT, DH, KR, ACP)
rifamycin polyketide synthase protein (Module 9:
KS, ATm, DH, KR, ACP. Module 10:
KS, AT, DH, KR, ACP)
amide synthase (N-acyl transferase)
similarity
RplJ (Z92772: 66%) Mycobacterium
tuberculosis (strain H37RV)
RplL (Z92772: 70%) M. tuberculosis
(strain H37RV)
(Z95972: 73%) hypothetical protein
Rv0655sM. tuberculosis (strain H37RV)
(SCC42.02C: 76%) putative ABC transporter
integral membrane proteinsS. coelicolor A3(2)]
(AL022073: 48%) hypothetical protein Rv1965s
M. tuberculosis (strain H37RV)
(SC8A2.07C: 43%) putative secreted proteins
S. coelicolor A3(2)
S. coelicolor A3(2)
S. coelicolor A3(2)
S. coelicolor A3(2)
S. coelicolor A3(2)
S. coelicolor A3(2)
(SC4A7.39C: 31%) putative membrane proteins
S. coelicolor A3(2)
(SC4A7.38C: 36%) putative integral membrane
proteinsS. coelicolor A3(2)
(AL096743: 29%) conserved hypothetical proteins
S. coelicolor A3(2)
(AP004604: 28%) NADH-dependent dehydrogenase
[Oceanobacillus iheyensis]
(AP004604: 23%) NADH-dependent dehydrogenase
[Oceanobacillus iheyensis]
(AL138668: 32%) hypothetical protein SC4A9.08s
S. coelicolor A3(2)
(M31939: 41%) cytochrome-P450-like protein (choP)
[Streptomyces sp.]
(AF453501: 36%) amide synthasesActinosynnema

pretiosum subsp. auranticum
aminodehydroquinate synthase
aminoDAHP sythase
aminoquinate dehydrogenase
AHBA synthase
oxidoreductase
phosphatase
kanosamine kinase
(AF131877: 73%) aminodehydroquinate synthases

Streptomyces collinus
(AF131877: 61%) amino-deoxyarabinoheptulosonate7-phosphate synthasesStreptomyces collinus
(AF131877: 70%) shikimate/quinate dehydrogenases
(AF131879: 71%) aminohydroxybenzoic acid synthases
(AE013146: 30%) predicted dehydrogenases and
related proteinssThermoanaerobacter tengcongensis
(AE007074: 33%) hydrolase, haloacid dehalogenaselike familysMycobacterium tuberculosis CDC1551
(AF131877: 59%) NapI kinasesStreptomyces collinus
Floss and Yu
Table 1 (Continued)
ORF
RifO
(255 aa)
Orf2
(310 aa)
RifP
(522 aa)
RifQ
(242 aa)
Orf3c
(166 aa)
Orf4c
(403 aa)
Orf5c
(421 aa)
Orf6c
(435 aa)
Orf7
(381 aa)
Orf8
(214 aa)
Orf9c
(430 aa)
Orf10c
(330 aa)
Orf11
(321 aa)
properties or content
putative regulatory protein
similarity
putative esterase
(AL450350: 34%) uncharacterized lmbE-like proteins

S. coelicolor A3(2)
(U70619: 50%) heroin esterasesRhodococcus sp.
efflux transporter protein
(AB019519: 73%) VarSsStreptomyces virginiae
putative tetR-like transcription regulatory

protein
hypothetical protein (ATP-binding protein)
(AB046994: 64%) VarRsStreptomyces virginiae
putative cytochrome P450 oxidoreductase

cytochrome P450 monooxygenase
dNTP-hexose dehydratase
dNTP-hexose glycosyl transferase
dNTP-hexose 3,5-epimerase
aminotransferase
probable dNDP-hexose-3-ketoreductase
flavin-dependent oxidoreductase
(AL392178: 44%) conserved hypothetical proteins

S. coelicolor A3(2)
(AF072709: 51%) putative cytochrome P450
oxidoreductasesStreptomyces lividans
(M54983: 36%) EryF: 6-deoxyerythronolide B
hydroxylase (6-DEB hydroxylase, erythomycinA
biosynthesis hydrolase) (cytochrome
P450 107A1) (CYPCVIIA1) (P450eryF)s
Saccharopolyspora erythraea
(AF269227: 71%) NDP-hexose 3,4-dehydratase UrdQs
Streptomyces fradiae
(AF164960: 43%) glycosyl transferases
Streptomyces fradiae
(AJ006985: 49%) StrM: dTDP-4-keto-6-deoxyglucose
3,5-epimerasesStreptomyces glaucescens
(AB005901: 58%) deduced aminotransferases
Streptomyces kasugaensis
(AF080235: 60%) oxidoreductase homologues
Streptomyces cyanogenus
(U67594: 28%) N5,N10-methylenetetrahydromethanopterin reductase (mer)s
Methanococcus jannaschii
(X58791: 27%) luciferase R subunitsVibrio harveyi
Orf17
(356 aa)
Orf18
(473 aa)
Orf19c
(501 aa)
alkanal monooxygenase R-chain
Orf20c
(403 aa)
Orf12c
(RifR)
(259 aa)
Orf13c
(422 aa)
25-O-acetyltransferase
(AF237895: 59%) dTDP-4-keto-6-deoxyglucose

2,3-dehydratasesStreptomyces antibioticus
(AP005277: 40%) 2-polyprenyl-6-methoxyphenol
hydroxylase and related FAD-dependent
oxidoreductasessCorynebacterium glutamicum
ATCC 13032
(Z83857: 31%) papA5sMycobacterium tuberculosis H37Rv
thioesterase
(AB070940: 55%) thioesterasesStreptomyces avermitilis
(M54983: 36%) EryF: 6-deoxyerythronolide B

hydroxylase (6-DEB hydroxylase, erythomycin A
biosynthesis hydrolase) (cytochrome
P450 107A1) (CYPCVIIA1) (P450eryF)s
Saccharopolyspora erythraea
(AB090952: 50%) putative D-glucose O-methyltransferases
Lechevalieria aeroclonigenes
(Z29635: 55%) orf3sRhodococcus fascians
Orf14
(272 aa)
Orf15
(533 aa)
Orf16c
(389 aa)
RifJ
(163 aa)
Orf36
(404 aa)
Orf37
(161 aa)
RpoB
(1167 aa)
-105,774
RpoC
putative 2, 3-dehydratase
3-(3-hydroxyphenyl)propionate hydroxylase
27-O-methyltransferase
transketolase
aminoDHQ dehydratase
putative regulatory protein
DNA-dependent RNA polymerase -subunit
(AF127374: 44%) cytochrome P450 hydroxylase ORF4s

Streptomyces lavendulae
(AF127374: 74%) MmcFsStreptomyces lavendulae
(AF534707: 34%) putative transcriptional activator RebRs
Lechevalieria aerocolonigenes
(NZ_AAAC01000306: 32%) hypothetical proteins
Burkholderia fungorum
(AE006964: 80%) DNA-directed RNA polymerase,
subunitsMycobacterium tuberculosis CDC1551
DNA-dependent RNA polymerase -subunit
structure, but the penta- to undecaketides carry a

naphthoquinone ring.118,119 Interestingly, module 4 of
the NRPS/PKS does not process a ketide in which
the naphthalene ring closure for some reason has not
occurred, but releases the corresponding tetraketide.
This compound, called P8/1-OG, was first isolated
from a mutant of A. mediterranei blocked in rifamycin biosynthesis,121 and the corresponding analogues
of P8/1-OG were obtained when a rifK(-) mutant was

complemented with AHBA analogues.114 Heterologous expression of rifA in E. coli has recently been
reported, leading to the production of P8/1-OG, albeit
in very low yield.122 A gene in the rif cluster, rif orf19,
has been tentatively identified as being involved in
the ring closure reaction; its inactivation leads to the
accumulation of P8/1-OG.123
Figure 8. Biosynthetic pathway for the rifamycin polyketide starter unit, 3-amino-5-hydroxybenzoic acid (AHBA).
The nature of the first cyclic product released from

the rif NRPS/PKS has been a matter of controversy.
On the basis of mutagenesis experiments, it had been
proposed that protorifamycin I, a naphthoquinone
derivative lacking the 8-hydroxyl group, is an intermediate in the biosynthesis of rifamycin B.124 Although seemingly supported by the 8-deoxynaphthoquinone structures of the penta- to undecaketides
accumulated in the rifF mutants,118,119 this proposal
is incompatible with the finding that one atom of 18O
from the C18O2H group of AHBA is retained in
rifamycin B.118,125 On the basis of this result and of
the isolation from the rifF(-) mutant of a pentaketide
with a 7,8-dihydro-8-hydroxynaphthoquinone structure, we proposed the structure of protorifamycin X
for the first cyclic PKS product (Figure 9), suggesting
that the 8-deoxy compounds are shunt metabolites
resulting from spontaneous dehydration.118 Recent
work by Stratmann et al. has indeed shown that the
8-deoxy compounds are intermediates on a shunt
pathway to 8-deoxyrifamycins, rather than rifamycin
B precursors.126
The rif cluster also contains several potential
regulatory genes and a set of genes predicted to
encode the formation of a sugar nucleotide which,
however, appear to be silent. Also present are a
substantial number of genes apparently responsible
for the modification of the original polyketide during
or after its assembly. Their functional analysis by
gene inactivation and heterologous expression is in
progress.123 The gene rif orf14 encodes a methyltransferase which has been shown to use 27-Odemethylrifamycin SV, not its quinone or its 25-Odesacetyl derivative, as substrate, shedding some
light on the late stages of rifamycin formation.127 The
biosynthetic pathway to rifamycin B and suggested
or proven assignments of genes to individual transformation steps are shown in Figure 9.
5. Autoresistance of Amycolatopsis Mediterranei

Notably, the rif cluster does not contain any obvious candidates for genes conferring resistance on A.
mediterranei to its own antibiotic,106 a feature of most
antibiotic biosynthesis gene clusters.128 Experiments
with whole cells and with the partially purified DNAdependent RNA polymerase of A. mediterranei have
shown that rifamycin resistance is expressed throughout the entire culture period independent of the time
of antibiotic production.129,130 In the process of defining the boundaries of the rifamycin biosynthetic gene
cluster in A. mediterranei, we found the rpoB gene
to be located on a 3.9 kb DNA fragment on the righthand side of the rif gene cluster.108 Sequencing
revealed an 1168 amino acid open reading frame with
81% identity to the M. tuberculosis rpoB, followed by
the 5 end of the rpoC gene. Southern hybridization
revealed that these represented the only copies of
these genes in the A. mediterranei genome and must
thus represent the respective housekeeping genes.
Cloning of the A. mediterranei rpoB gene into rifampicin-sensitive M. smegmatis conferred rifampicin resistance upon the organism. The same was observed
when the rif I region of the M. tuberculosis rpoB was
replaced with that from the A. mediterranei rpoB and
introduced into M. smegmatis.108 In the rif I region
of A. mediterranei rpoB, five amino acids, Q432,
T434, I437, D438, and N447, differ from their counterparts in the rpoB from wild-type M. tuberculosis
(Figure 5). Three of these, Q432, D438, and N447,
were sufficient to confer rifampicin resistance upon
A. mediterranei; a triple mutation, Q432D, D438S,
and N447S, in the rpoB gene resulted in a high level
Floss and Yu
Figure 9. Biosynthetic pathway to rifamycin B and the established or proposed role of individual rif biosynthetic genes.
of rifampicin sensitivity (MIC 0.01 g/mL) in a

rifamycin nonproducing mutant of A. mediterranei.
Site-specific mutagenesis of the corresponding positions in the M. tuberculosis rpoB gene and expression
in M. smegmatis showed that in fact each one of these
three amino acid substitutions, D430Q, S436D, and
S445N (M. t. numbering), alone was sufficient to
confer resistance upon that organism. Autoresistance
of the producing organism is thus predominantly, if
not exclusively, due to a rifamycin-insensitive DNAdependent RNA polymerase.108
The arrangement of the rpoB/C genes and the
genes rplJ and rplL, encoding ribosomal proteins, is
highly conserved in archaebacteria and eubacteria.131-137 It was therefore a question where on the
A. mediterranei genome the rplJ and rplL genes are
located. They were found 16.6 kb upstream of the rifA
gene. The region between them and rifA contains 12
genes encoding transporter-related lipoproteins, which
are likely to be involved in antibiotic efflux, and
several post-PKS processing genes (Figure 7).108
Thus, in A. mediterranei the entire rifamycin biosynthetic gene cluster is inserted between some of
the genes encoding the cellular machinery targeted
by the antibiotic. Several other bacterial strains
producing rifamycin-related antibiotics were analyzed for the arrangement of the rplL-rpoB genes.
In four non-Amycolatopsis strains, Micromonospora
lacustris, Micromonospora nigra, and two Strepto-
myces species producing streptovaricins and awamycin, respectively, the rplL and rpoB genes were found
to be closely linked. However, in Amycolatopsis
tolypomycina and A. vancoremycina, no such linkage
was detectable, suggesting that their antibiotic biosynthesis gene clusters are also located in the intergenic region between rplL and rpoB, as in A. mediterranei. All six organisms showed pronounced
rifamycin resistance and carried amino acid substitutions in the rif I region consistent with a rifamycininsensitive rpoB.108
Interestingly, the presence or absence of rifamycin
production and resistance in A. mediterranei has
pronounced effects on growth, susceptibility to phage
infection, and spore production. Under laboratory
culture conditions, spore production in a rifamycin
nonproducing mutant is delayed by a moderate
supplement of rifamycin. The rifamycin nonproducing mutant also revealed a higher sensitivity to phage
infection, particularly in the rifamycin-sensitive strains
that carry a mutated rpoB allele. These results could
suggest a mediator role for rifamycins.108
6. Concluding Remarks
Like most colonial organisms, A. mediterranei
exploits elaborate systems of intra- and intercellular
communication to facilitate the adaptation to changeable environmental conditions. The messages by
which bacteria communicate take the form of chemical signals released from the cells which can elicit
profound physiological changes. In the transcription
process, the cellular RNA polymerase operates as a
complex molecular machine with extensive interactions with the template DNA, the product RNA, and
regulatory molecules. It seems plausible that many
distinct sites exist where the binding of a mediator
molecule, such as rifamycin, could switch critical
features of the functional mechanism. Indeed, rifamycin is active against a large variety of organisms,
including many bacteria, eukaryotes, and viruses. It
is, therefore, possible that rifamycin represents a
widely recognized ancient signaling molecule and
regulates diverse behaviors across distant genera.
The discovery that the rifamycin biosynthetic gene
cluster is closely linked to the housekeeping genes
encoding the ribosomal proteins and the RNA polymerase subunits could provide an alternative view
of the natural role of this broadly used antimicrobial
agent.
7. Acknowledgment
Work from the authors laboratory was supported
by NIH Grant AI20264. We thank Mrs. Kay B.
Kampsen for her help in the preparation of this
manuscript.
8. References
(1) The name rifamycin (originally rifomycin) is derived from the
title of a French movie, Rififi, popular at the time of the
antibiotics discovery.2
(2) Sensi, P. Rev. Infect. Dis. 1983, 5 (Suppl. 3), S402.
(3) Rinehart, K. L., Jr.; Shields, L. S. Fortschr. Chem. Org. Naturst.
1976, 33, 231.
(4) Wehrli, W. Top. Curr. Chem. 1977, 72, 21.
(5) Prelog, V.; Oppolzer, W. Helv. Chim. Acta 1973, 56, 2279.
(6) Balerna, M.; Keller-Schierlein, W.; Martius, C.; Wolf, H.; Zahner,
H. Arch. Mikrobiol. 1969, 65, 303.
(7) Rinehart, K. L., Jr.; Maheshwari, M. L.; Antosz, F. J.; Mathur,
H. H.; Sasaki, K.; Schacht, R. J. J. Am. Chem. Soc. 1971, 93,
6273.
(8) Deboer, C.; Meulman, P. A.; Wnuk, R. J.; Peterson, D. H. J.
Antibiot. 1970, 23, 442.
(9) Higashide, E.; Asai, M.; Ootsu, K.; Tanida, S.; Kozai, Y.;
Hasegawa, T.; Kishi, T.; Sugino, Y.; Yoneda, M. Nature 1977,
270, 721.
(10) Sensi, P.; Margalith, P.; Timbal, M. T. Farmaco, Ed. Sci. 1959,
14, 146.
(11) Margalith, P.; Beretta, G. Mycopathol. Mycol. Appl. 1960, 13,
321.
(12) Thiemann, J. E.; Zucco, G.; Pelizza, G. Arch. Mikrobiol. 1969,
67, 147.
(13) Lechevalier, M. P.; Prauser, H.; Labeda, D. P.; Ruan, J.-S. Int.
J. Syst. Bacteriol. 1986, 36, 29.
(14) Margalith, P.; Pagani, H. Appl. Microbiol. 1961, 9, 320, 325.
(15) Oppolzer, W.; Prelog, V.; Sensi, P. Experientia 1964, 20, 336.
(16) Brufani, M.; Fideli, W.; Giacomello, G.; Vaciago, A. Experientia
1964, 20, 339.
(17) Leitich, J.; Oppolzer, W.; Prelog, V. Experientia 1964, 20, 343.
(18) Oppolzer, W.; Prelog, W. Helv. Chim. Acta 1973, 56, 2287.
(19) Lancini, G.; Cavalleri, B. In Biotechnology of Antibiotics; Strohl,
W. R., Ed.; Marcel Dekker: New York, 1997; p 521.
(20) Kishi, T.; Yamana, H.; Muroi, M.; Harada, S.; Asai, M.; Hasegawa, T.; Mizuno, K. J. Antibiot. 1972, 25, 11.
(21) Ganguly, A. K.; Szmulewicz, S.; Sarre, O. Z.; Greeves, D.; Morton,
J.; McGlotten, J. J. Chem. Soc., Chem. Commun. 1974, 395.
(22) Wehrli, W.; Staehelin, M. Bacteriol. Rev. 1971, 35, 290.
(23) Sensi, P.; Ballotta, R.; Greco, A. M.; Gallo, G. G. Farmaco, Ed.
Sci. 1961, 16, 165.
(24) Maggi, N.; Pasqualucci, C. R.; Ballotta, R.; Sensi, P. Chemotherapia 1966, 11, 285.
(25) Sensi, P.; Thiemann, J. E. Prog. Ind. Microbiol. 1967, 6, 21.
(26) Bergamini, N.; Fowst, G. Arzneim.-Forsch. 1965, 15, 951.
(27) Roos, R.; Ghisalba, O. Experientia 1980, 36, 486.
(28) Lancini, G. C.; Hengeller, C. J. Antibiot. 1969, 22, 637.

(29) Lancini, G.; Zanichelli, W. In Structure-activity relationships
among the semisynthetic antibiotics; Perlman, D., Ed.; Academic
Press: New York 1977; p 531.
(30) Traxler, P.; Vischer V. A.; Zak, O. Drugs Future 1988, 13, 845.
(31) Binda, G.; Domenichini, E.; Gottardi, A.; Orlandi, B.; Ortelli,
E.; Pacini, B.; Fowst, G. Arzneim.-Forsch. 1971, 21, 1907.
(32) Shinnick, T., Ed. Current Topics in Microbiology and Immunology; Academic Press: New York, 1996.
(33) Wehrli, W. Rev. Infect. Dis. 1983, 5 (Suppl. 3), S407.
(34) Hobby, G. L.; Lenert, T. F. Am. Rev. Respir. Dis. 1970, 102, 462.
(35) Brogden, R. N.; Fitton, A. Drugs 1994, 47, 983.
(36) Mealy, N. E. Drugs Future 1979, 4, 255.
(37) Calvori, C.; Frontali, L.; Leoni, L.; Tecce, G. Nature 1965, 207,
417.
(38) Hartmann, G.; Honikel, K. O.; Knusel, F.; Nuesch, J. Biochim.
Biophys. Acta 1967, 145, 843.
(39) Hartmann, G.; Behr, W.; Beissner, K.-A.; Honikel, K.; Sippel,
A. Angew. Chem., Int. Ed. Engl. 1968, 7, 693.
(40) Meilhac, M.; Tysper, Z.; Chambon, P. Eur. J. Biochem. 1972,
28, 291.
(41) Campbell, E. A.; Khorzheva, N.; Mustaev, A.; Murakami, K.;
Nair, S.; Goldfarb, A.; Darst, S. A. Cell 2001, 104, 901.
(42) Sippel, A.; Hartmann, G. Biochim. Biophys. Acta 1968, 157, 218.
(43) Umezawa, H.; Mizuno, S.; Yamasaki, H.; Nitta, K. J. Antibiot.
1968, 21, 234.
(44) Sippel, A.; Hartmann, G. Eur. J. Biochem. 1970, 16, 152.
(45) Rabussay, D.; Zillig, W. FEBS Lett. 1969, 5, 104.
(46) Heil, A.; Zillig, W. FEBS Lett. 1970, 11, 165.
(47) Wehrli, W.; Handschin, J.; Wunderli, W. In RNA polymerase;
Losick, R., Chamberlin, M., Eds.; Cold Spring Harbor Laboratory: Cold Spring Harbor, NY, 1976; p 397.
(48) Naryshkina, T.; Mustaev, A.; Darst, S. A.; Severinov, K. J. Biol.
Chem. 2001, 276, 13308.
(49) McClure, W. R.; Cech, C. L. J. Biol. Chem. 1978, 253, 8949.
(50) Schulz, W.; Zillig, W. Nucleic Acids Res. 1981, 9, 6889.
(51) Zhang, G.; Campbell, E. A.; Minakhin, L.; Richter, C.; Severinov,
K.; Darst, S. A. Cell 1999, 98, 811.
(52) Korzheva, N.; Mustaev, A.; Kozlov, M.; Malhotra, A.; Nikiforov,
V.; Goldfarb, A.; Darst, S. A. Science 2000, 289, 619.
(53) Mustaev, A.; Zaychikov, E.; Severinov, K.; Kashlev, M.; Polyakov,
A.; Nikiforov, V.; Goldfarb, A. Proc. Natl. Acad. Sci. U.S.A. 1994,
91, 12036.
(54) Severinov, K.; Mustaev, A.; Severinova, E.; Kozlov, M.; Darst,
S. A.; Goldfarb, A. J. Biol. Chem. 1995, 270, 29428.
(55) Brufani, M.; Cerrini, S.; Fedeli, W.; Vaciago, S. J. Mol. Biol. 1974,
87, 409.
(56) Bartolucci, C.; Cellai, L.; Di Filippo, P.; Segre, A.; Brufani, M.;
Filocamo, L.; Bianco, A. D.; Guiso, M.; Brizzi, V.; Benedetito, A.
D.; Di Caro, A.; Elia, G. Il Farmaco 1992, 47, 1367.
(57) Lal, R.; Lal, S. Bioessays 1994, 16, 211.
(58) Grosset, J.; Leventis, S. Rev. Infect. Dis. 1983, 5 (Suppl. 3), S440.
(59) Calleja, C.; Pascussi, J. M.; Mani, J. C.; Maurel, P.; Vilarem, M.
J. Nature Medicine 1998, 4, 92.
(60) Fardel, O.; Lecureur, V.; Loyer, P.; Guillouzo, A. Biochem.
Pharmacol. 1995, 49, 1255.
(61) Courtois, A.; Payen, L.; Vernhet, L.; deVries, E. G. E.; Guillouzo,
A.; Fardel, O. Cancer Lett. 1999, 139, 97.
(62) David, H. L. Appl. Microbiol. 1970, 20, 810.
(63) Gillespie, S. H. Antimicrob. Agents Chemother. 2002, 46, 267.
(64) Ezekiel, D. H.; Hutchins, J. E. Nature 1968, 220, 276.
(65) Wehrli, W.; Nuesch, J.; Knusel, F.; Staehelin, M. Biochem.
Biophys. Res. Commun. 1968, 32, 284.
(66) Telenti, A.; Imboden, P.; Marchesi, F.; Lowrie, D.; Cole, S.;
Colston, M. J.; Matter, L.; Schopfer, K.; Bodmer, T. Lancet 1993,
341, 647.
(67) Ramaswamy, S.; Musser, J. M. Tuberc. Lung Dis. 1998, 79, 3.
(68) Heep, M.; Rieger, U.; Beck, D.; Lehn, N. Antimicrob. Agents
Chemother. 2000, 44, 1075.
(69) Heep, M.; Beck, D.; Bayerdorffer, E.; Lehn, N. Antimicrob. Agents
Chemother. 1999, 43, 1497.
(70) Ovchinnikov, Y. A.; Monastyrskaya, G. S.; Gubanov, V. V.;
Guryev, S. O.; Chertov, O. Y.; Modyanov, N. N.; Grinkevich, V.
A.; Makarova, I. A.; Marchenko, T. V.; Polovnikova, I. N.; Lipkin,
V. M.; Sverdlov, E. D. Eur. J. Biochem. 1981, 116, 621.
(71) Ovchinnikov, Y. A.; Monastyrskaya, G. S.; Guriev, S. O.;
Kalinina, N. F.; Sverdlov, E. D.; Gragerov, A. I.; Bass, I. A.;
Kiver, I. F.; Moiseyeva, E. P.; Igumnov, V. N.; Mindlin, S. Z.;
Nikiforov, V. G.; Khesin, R. B. Mol. Gen. Genet. 1983, 190, 344.
(72) Lisitsyn, N. A.; Guriev, S. O.; Sverdlov, E. D.; Moiseeva, E. P.;
Nikiforov, V. G. Bioorg. Khim. 1984, 10, 127.
(73) Lisitsyn, N. A.; Sverdlov, E. D.; Moiseyeva, E. P.; Danilevskaya,
O. N.; Nikiforov, V. G. Mol. Gen. Genet. 1984, 196, 173.
(74) Jin, J. D.; Gross, C. A. J. Mol. Biol. 1988, 202, 45.
(75) Severinov, K.; Soushko, M.; Goldfarb, A.; Nikiforov, V. J. Biol.
Chem. 1993, 268, 14820.
(76) Severinov, K.; Soushko, M.; Goldfarb, A.; Nikiforov, V. Mol. Gen.
Genet. 1994, 244, 120.
(77) Honore, N.; Cole, S. T. Antimicrob. Agents Chemother. 1993, 37,
414.

(78) Carter, P. E.; Abadia, F. J. R.; Yakubu, D. E.; Pennington, T. H.
(79) Yee, Y. C.; Kisslinger, B.; Yu, V. L.; Jin, D. J. J. Antimicrob.
Chemother. 1996, 38, 133.
(80) Alekshun, M.; Kashlev, M.; Schwartz, I. Gene 1997, 186, 227.
(81) Drancourt, M.; Raoult, D. Antimicrob. Agents Chemother. 1999,
43, 2400.
(82) Heep, M.; Odenbreit, S.; Beck, D.; Decker, J.; Prohaska, E.;
Rieger, U.; Lehn, N. Antimicrob. Agents Chemother. 2000, 44,
1713.
(83) Hetherington, S. V.; Watson, A. S.; Patrick, C. C. Antimicrob.
(84) Hui, J.; Gordon, N.; Kajoika, R. Antimicrob. Agents Chemother.
1977, 11, 773.
(85) Abadi, F. J.; Carter, P. E.; Cash, P.; Pennington, T. H. Antimicrob. Agents Chemother. 1996, 40, 645.
(86) Chandrasekaran, S.; Lalithakumari, D. J. Med. Microbiol. 1998,
47, 197.
(87) Wehrli, W.; Zimmermann, W.; Kump, W.; Tosch, W.; Vischer,
W.; Zak, O. J. Antibiot. 1987, 40, 1733.
(88) Pugsley, A. P.; Zimmermann, W.; Wehrli, W. J. Gen. Microbiol.
1987, 133, 3505.
(89) Braun, V. Drug Resist. Updates 1999, 2, 363.
(90) Ferguson, A. D.; Kodding, J.; Walker, G.; Bos, C.; Coulton, J.
W.; Diederichs, K.; Braun, V.; Welte, W. Structure 2001, 9, 707.
(91) Dabbs, E. R. FEMS Microbiol. Lett. 1987, 44, 395.
(92) Andersen, S. J.; Dabbs, E. R. FEMS Microbiol. Lett. 1991, 79,
247.
(93) Quan, S.; Venter, H.; Dabbs, E. R. Antimicrob. Agents Chemother. 1997, 41, 2456.
(94) Morisaki, N.; Iwasaki, S.; Yazawa, K.; Mikami, Y.; Maeda, A. J.
Antibiot. 1993, 46, 1605.
(95) Dabbs, E. R.; Yazawa, K.; Mikami, Y.; Miyaji, M.; Morisaki, N.;
Iwasaki, S.; Furihata, K. Antimicrob. Agents Chemother. 1995,
39, 1007.
(96) Tribuddharat, C.; Fennewald, M. Antimicrob. Agents Chemother.
1999, 43, 960.
(97) Arlet, G.; Nadjar, D.; Herrmann, J.-L.; Donay, J.-L.; Rouveau,
M.; Lagrange, P. H.; Philippon, A. Antimicrob. Agents Chemother. 2001, 45, 2971.
(98) Girlich, D.; Poirel, L.; Leelaporn, A.; Karim, A.; Tribuddharat,
C.; Fennewald, M.; Nordmann, P. J. Clin. Microbiol. 2001, 39,
175.
(99) Naas, T.; Mikami, Y.; Imai, T.; Poirel, L.; Nordmann, P. J.
Bacteriol. 2001, 183, 235.
(100) Yazawa, K.; Mikami, Y.; Maeda, A.; Akao, M.; Morisaki, N.;
Iwasaki, S. Antimicrob. Agents Chemother. 1993, 37, 1313.
(101) Yazawa, K.; Mikami, Y.; Maeda, A.; Morisaki, N.; Iwasaki, S. J.
(102) Tanaka, Y.; Yazawa, K.; Dabbs, E. R.; Nishikawa, K.; Komaki,
H.; Mikami, Y.; Miyaji, M.; Morisaki, N.; Iwasaki, S. Microbiol.
Immunol. 1996, 40, 1.
(103) Floss, H. G. Nat. Prod. Rep. 1997, 14, 433.
(104) Kim, C.-G.; Kirschning, A.; Bergon, P.; Zhou, P.; Su, E.; Sauerbrei, B.; Ning, S.; Ahn, Y.; Breuer, M.; Leistner, E.; Floss, H. G.
J. Am. Chem. Soc. 1996, 118, 7486.
(105) Kim, C.-G.; Yu, T.-W.; Fryhle, C. B.; Handa, S.; Floss, H. G. J.
Biol. Chem. 1998, 273, 6030.
(106) August, P. R.; Tang, L.; Yoon, Y. J.; Ning, S.; Muller, R.; Yu,
T.-W.; Taylor, M.; Hoffmann, D.; Kim, C.-G.; Zhang, X.; Hutchinson, C. R.; Floss, H. G. Chem. Biol. 1998, 5, 69.
Floss and Yu
(107) Schupp, T.; Toupet, C.; Engel, N.; Goff, S. FEMS Microbiol. Lett.
1998, 159, 201.
(108) Yu, T.-W.; Pogosova-Agadjanyan, E.; Kuan, L.-Y.; Bai, L.; Tin,
A. M.; Floss, H. G. Manuscript in preparation.
(109) Yu, T.-W.; Muller, R.; Muller, M.; Zhang, H.; Draeger, G.; Kim,
C.-G.; Leistner, E.; Floss, H. G. J. Biol. Chem. 2001, 276, 12546.
(110) Guo, J.; Frost, J. W. J. Am. Chem. Soc. 2002, 124, 528.
(111) Guo, J.; Frost, J. W. J. Am. Chem. Soc. 2002, 124, 10642.
(112) Arakawa, K.; Muller, R.; Mahmud, T.; Yu, T.-W.; Floss, H. G. J.
Am. Chem. Soc. 2002, 124, 10644.
(113) Tang, L.; Joon, Y. J.; Choi, C. H.; Hutchinson, C. R. Gene 1998,
216, 255.
(114) Hunziker, D.; Yu, T.-W.; Hutchinson, C. R.; Floss, H. G.; Khosla,
C. J. Am. Chem. Soc. 1998, 120, 1092.
(115) Admiraal, S. J.; Walsh, C. T.; Khosla, C. Biochemistry 2001, 40,
6116.
(116) Admiraal, S. J.; Khosla, C.; Walsh, C. T. J. Am. Chem. Soc. 2003,
125, 13664.
(117) Pompeo, F.; Mustaq, A.; Sim, E. Protein Expression Purif. 2002,
24, 138.
(118) Yu, T.-W.; Shen, Y.; Doi-Katayama, Y.; Tang, L.; Park, C.; Moore,
B. S.; Hutchinson, C. R.; Floss, G. H. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 9051.
(119) Stratmann, A.; Toupet, C.; Schilling, W.; Traber, R.; Oberer, L.;
Schupp, T. Microbiology 1999, 145, 3365.
(120) Doi-Katayama, Y.; Yoon, Y. J.; Choi, C.-Y.; Yu, T.-W.; Floss, H.
G.; Hutchinson, C. R. J. Antibiot. 2000, 53, 484.
(121) Ghisalba, O.; Fuhrer, H.; Richter, W. J.; Moss, S. J. Antibiot.
1981, 34, 58.
(122) Watanabe, K.; Rude, M. A.; Walsh, C. T.; Khosla, C. Proc. Natl.
Acad. Sci. U.S.A. 2003, 100, 9774.
(123) Xu, J.; Floss, H. G.; Mahmud, T. Manuscript in preparation.
(124) Ghisalba, O.; Traxler, P.; Nuesch, J. J. Antibiot. 1978, 31, 1124.
(125) Anderson, M. G.; Monypenny, D.; Rickards, R. W.; Rothschild,
J. M. J. Chem. Soc., Chem. Commun. 1989, 311.
(126) Stratmann, A.; Schupp, T.; Toupet, C.; Schilling, W.; Oberer, L.;
Traber, R. J. Antibiot. 2002, 55, 396.
(127) Xu, J.; Mahmud, T.; Floss, H. G. Arch. Biochem. Biophys. 2003,
411, 277.
(128) Maplestone, R. A.; Stone, J. S.; Williams, D. H. Gene 1992, 115,
151.
(129) Watanabe, S.; Tanaka, K. Biochem. Biophys. Res. Commun.
1976, 72, 522.
(130) Blanco, M. G.; Hardisson, C.; Salas, J. A. J. Gen. Microbiol. 1984,
130, 2883.
(131) Lindahl, L.; Jaskunas, S. R.; Dennis, P. P.; Nomura, M. Proc.
Natl. Acad. Sci. U.S.A. 1975, 72, 2743.
(132) Dabbs, E. R. J. Bacteriol. 1984, 159, 770.
(133) Mekhedov, S. L.; Bass, I. A. Mol. Biol. (Moscow) 1986, 20, 92.
(134) Liao, D.; Dennis, P. P. J. Biol. Chem. 1992, 267, 22787.
(135) Clark, M. A.; Baumann, L.; Baumann, P. Curr. Microbiol. 1992,
25, 283.
(136) Aboshkiwa, M.; al-Ani, B.; Coleman, G.; Rowland, G. J. Gen.
Microbiol. 1992, 138, 1875.
(137) Kuster, C.; Piepersberg, W.; Distler, J. Mol. Gen. Genet. 1998,
257, 219.
(138) Wink, J. M.; Kroppenstedt, R. M.; Ganguli, B. M.; Nadkarni, S.
R.; Schumann, P.; Seibert, G.; Stackebrandt, E. Syst. Appl.
Microbiol. 2003, 26, 38.
CR030112J
Chem. Rev. 2005, 105, 633683
633
Biosynthesis and Mode of Action of Lantibiotics

Champak Chatterjee, Moushumi Paul, Lili Xie, and Wilfred A. van der Donk*
Department of Chemistry, University of Illinois at Urbana-Champaign, 600 South Mathews Avenue, Urbana, Illinois
Contents
1. Introduction
2. Gene Organization
3. Structures of Lantibiotics
3.1. Type AI Lantibiotics: Nisin Group
3.1.1. Primary Structure
3.1.2. Three-Dimensional Structure
3.2. Type AI Lantibiotics: Epidermin Group
3.3. Type AI Lantibiotics: Pep5 Group
3.4. Type AII Lantibiotics: Lacticin 481 Group
3.5. Type B Lantibiotics: Mersacidin Group
3.6. Type B Lantibiotics: Cinnamycin Group
3.7. Two-Component Lantibiotics: Lacticin 3147
4. Biosynthesis of Lantibiotics
4.1. Lantibiotic Precursor Peptides
4.2. The LanB Dehydratases
4.3. The LanC Cyclases
4.4. The LanM Bifunctional Enzymes
4.5. The LanD Enzymes
4.6. Other Posttranslational Modifications
4.7. Proteases and Transporters
5. Regulation of Lantibiotic Production
6. Self-Immunity of the Producing Strains
7. Lantibiotic Engineering
7.1. In Vivo Protein Engineering
7.2. In Vitro Protein Engineering
8. Mode of Action
8.1. Pore Formation in Model Membranes
8.2. Highjacking of Lipid II for Pore Formation
8.3. Inhibition of Spore Outgrowth
8.4. Other Biological Activities
9. Resistance Against Nisin
9.1. Gram-Negative Bacteria
633
635
638
638
638
639
640
640
641
641
641
642
642
642
643
644
644
644
644
644
645
646
646
647
647
647
650
650
654
655
656
657
659
661
664
664
666
668
668
669
673
673
673
673

vddonk@uiuc.edu.
10.
11.
12.
13.
14.
9.2. Gram-Positive Bacteria

Summary and Outlook
Abbreviations
Acknowledgments
Note Added in Proof
References
674
676
677
678
678
678
1. Introduction
Lantibiotics are peptide-derived antimicrobial agents
that are ribosomally synthesized and posttranslationally modified to their biologically active forms.
The name lantibiotics was introduced in 1988 as an
abbreviation for lanthionine-containing antibiotic
peptides.1 The unusual amino acid lanthionine consists of two alanine residues cross-linked via a
thioether linkage that connects their -carbons (S(alaninyl-3-yl)-cysteine) (Figure 1). These residues
are the unifying structural motif present in all
lantibiotics. Horn and co-workers reported the first
isolation of this thioether-cross-linked amino acid
from the treatment of wool with sodium carbonate
and introduced the name lanthionine (Latin, lana )
wool).2 In all natural lantibiotics, the lanthionines are
believed to have the meso-stereochemistry (Lan,
Figure 1),3 although this has only been rigorously
established for a subset of known lantibiotics includingnisin,4 subtilin,5 epidermin,6 Pep5,7,8 cinnamycin,9-11
ancovenin,12,13 actagardine,14,15 and mersacidin,16 and
the meso stereochemistry is generally assumed for
other family members.
Lantibiotics are produced by a large number of
Gram-positive bacteria and have their lanthionines
imbedded within cyclic peptides. They usually also
contain a methyl-substituted lanthionine derivative,
(2S,3S,6R)-3-methyllanthionine17 (MeLan, Figure 1)
and typically (but not always)18 contain the unsaturated amino acids 2,3-didehydroalanine (Dha) and (Z)2,3-didehydrobutyrine (Dhb).19 Less frequently encountered posttranslationally introduced structures
are lysinoalanine, -hydroxy-aspartate, D-alanine,
2-oxobutyrate, 2-oxopropionate (pyruvate), 2-hydroxypropionate (lactate), S-aminovinyl-D-cysteine, and
S-aminovinyl D-methylcysteine (Figure 1), and it is
possible that other modifications remain to be discovered.
The widespread application of the prototype lantibiotic nisin (Figure 2) as a safe alternative for
chemical reagents in food preservation (>80 countries
for over 40 years)20-22 spurred a rapid expansion of
research activities directed at understanding lantibiotic biogenesis. Early investigations showed that
their production by Gram-positive bacteria was sus-
10.1021/cr030105v CCC: $53.50 2005 American Chemical Society

Chatterjee et al.
Champak Chatterjee was born in Calcutta, India. He received a B.S. in

Chemistry in 1997 from the University of Bombay. After receiving an M.S.
from I. I. T. Bombay in 1999 working with Professor Sambasivarao Kotha,
he moved to the University of Illinois at Urbana-Champaign and joined
the laboratories of Professor Wilfred A. van der Donk. He has been
engaged in studying the mechanism and substrate specificity of the lacticin
481 synthetase LctM, and the protease involved in lacticin 481 maturation.
Lili Xie was born in China in 1976. She graduated from the University of
Science and Technology of China (USTC) with a B.S. degree in Chemical
Physics in 1998. She received her Ph.D. degree from the University of
Illinois at Urbana-Champaign in 2003 under the supervision of Professor
Wilfred van der Donk, where she successfully developed the first in vitro
reconstitution of lantibiotic biosynthesis. Currently, she is carrying out her
postdoctoral research with Professor Jon Clardy at Harvard Medical School.
Moushumi Paul was born in Mississippi and raised in Philadelphia, PA.

She earned a B.A. in Chemistry from Barnard College in 1999 and is
currently in the final stages of completing her Ph.D. in chemistry under
the guidance of Prof. Wilfred van der Donk at the University of Illinois,
Urbana-Champaign. Her doctoral work has focused on understanding the
enzymes involved in lantibiotic biosynthesis.
Wilfred van der Donk received his B.S. and M.S. from Leiden University,
The Netherlands, under the guidance of Prof. Jan Reedijk and his Ph.D.
from Rice University in the laboratories of Prof. Kevin Burgess. After
postdoctoral work with Prof. JoAnne Stubbe at MIT, he joined the faculty
at the University of Illinois at Urbana-Champaign in 1997. The research
in his laboratory focuses on understanding of the molecular mechanisms
of enzyme catalysis and the use of enzymes in organic chemistry.
ceptible to inhibition by compounds that disrupt

protein biosynthesis,22 suggesting that lantibiotics
are ribosomally synthesized as precursor peptides
(prepeptides), which subsequently undergo posttranslational modifications.23,24 As such, they are
considered members of the bacteriocins.25 To distinguish lantibiotics from other bacteriocins that are not
posttranslationally modified, they have been designated as a separate subgroup, the class I bacteriocins
(Table 1). The biosynthesis of lantibiotics also distinguishes them from classical nongene encoded
peptide antibiotics such as gramicidin, which are
produced by modular nonribosomal peptide synthetases (NRPS).26-30
Nisin, the most studied lantibiotic, is produced by
Lactococcus lactis and has been used extensively as
a food preservative without substantial development
of bacterial resistance.20 Discovered in 1928,31,32 one
year prior to penicillin,33 the compound is one of the
oldest known antibacterial agents but its structure
was not determined until elegant landmark studies
by Gross and Morell in 1971 (Figure 2).4 Around the
same time as the structure elucidation studies,

Ingram proposed that the dehydro amino acids in
lantibiotics are the result of dehydration of serine and
threonine residues to produce Dha and Dhb structures, respectively, and that the lanthionine and
methyllanthionine rings are generated by intramolecular conjugate additions of cysteines to these R,unsaturated amino acids.35,36 This hypothesis was
confirmed when the first gene clusters responsible
for the biosynthesis of a number of lantibiotics were
sequenced in the late 1980s.1,37-39 In these studies,
the genes encoding the precursor peptides for epidermin,1 subtilin,37,38 and nisin39 were shown to
contain codons for Ser, Thr, and Cys residues at the
sites of posttranslational modifications. Only very
recently has the biosynthesis of a lantibiotic (lacticin
481) been reconstituted in vitro,40 which has provided
further support for Ingrams hypothesis.
Nisin is active at low concentrations (MICs low nanomolar) against many strains of Gram-positive bacteria,20 including drug resistant strains41 and the
Figure 2. The structure of nisin A.
enzyme by ancovenin.58 Moreover, mersacidin exhibits comparable antimicrobial activity against methicillin-resistant Staphylococcus aureus (MRSA) as
vancomycin without showing any cross-resistance,59-61
and mutacin B-Ny266 displays activities comparable
to vancomycin and oxacillin against many strains and
remains active against vancomycin-resistant strains.62
We will not cover in this review the extensive studies
dealing with practical applications of lantibiotics and
refer the reader to various excellent recent reviews.20,55,63,64 Instead, this work will focus on the
available information regarding the mechanisms of
biosynthesis and mode of action of this intriguing
class of compounds.
2. Gene Organization
Figure 1. Structural motifs that are introduced into

lantibiotics by posttranslational modifications. A shorthand notation for these modification that will be used
elsewhere in this review is presented underneath several
structures.
food-borne pathogens Clostridium botulinum and Listeria monocytogenes.42-45 With an estimated 76 million cases of food-related illness in the United States
each year,46 translating into a cost of between $6.5
and 34.9 billion in 1997,47 research into the modes
of action and biosynthesis of nisin has increased
dramatically in the past decade culminating in the
demonstration that the cell wall biosynthetic intermediate lipid II constitutes its specific target.41,48-54
Other lantibiotics show different interesting biological activities. These include high potency of epidermin against Propionibacterium acnes,55 which may
be exploited for topical treatment of acne, inhibition
of phospholipase A2 by cinnamycin and duramycin,56,57 and inhibition of angiotensin converting
Similar to most biosynthetic pathways in bacteria,

the genes for lantibiotic biosynthesis are clustered
and have been designated the generic locus symbol
lan, with a more specific genotypic designation for
each lantibiotic member (e.g., nis for nisin, gdm for
gallidermin, cin for cinnamycin). They may be found
on conjugative transposable elements (e.g., nisin), on
the chromosome of the host (e.g., subtilin), or on
plasmids (e.g., epidermin, lacticin 481). Many of the
lan genes have been sequenced in the past 15
years.23,65,66 These studies have demonstrated a high
level of similarity in the gene organization for production of the various compounds. The gene clusters
for the biosynthesis of a select number of lantibiotics
are depicted in Figure 3.67 Although the gene order,
complexity, and transcriptional organization of the
various clusters differ, three genes have been identified that are involved in the biosynthesis of all type
AII and type B lantibiotics (lanAMT), and four genes
are present in all type AI lantibiotics gene clusters
(lanABCT) (blue-colored genes, Figure 3. For a
description of the type A and B classification, see
section 3). These essential genes obviously include
the structural genes encoding the precursor peptides
for posttranslational maturation (prepeptides), which
have been designated lanA, except for subtilin whose
structural gene historically is named spaS. The lanA
genes produce prepeptides that have an extension of
23-59 amino acids at their N-terminus compared to
the mature lantibiotic product, which provided the
first indication that an N-terminal leader peptide is
Chatterjee et al.
Figure 3. Representative biosynthetic gene clusters of the lantibiotics nisin,214,259,305 subtilin,68,73,211 epidermin,76 Pep5,75
lacticin 481,148 lactocin S,155 cinnamycin,78 mersacidin,262 and lacticin 3147.227 In blue are those genes that are present in
all known members with the LanB and LanC genes substituted by one LanM gene in some cases. Promoters for the
transcriptional units in these clusters (where known) are indicated by red wedges.23,75,76,155,211,260,268,272,289,294,295,329,467,468
Table 1. Classification of Bacterial Antimicrobial Peptides (Bacteriocins)25,34
characteristics
size
Class I
class
posttranslationally modified
peptides containing
(methyl)lanthionines
(lantibiotics)
< 5 kDa
Class II
heat-stable peptides
of 37-58 amino acids;
leader peptide removed
during maturation
<10 kDa
Class III
heat labile, large proteins
>30 kDa
important in lantibiotic biosynthesis. As mentioned

in the introduction, sequencing of the lanA genes also
indicated that Ser and Thr residues are the precursors to Dha and Dhb structures found in the final
products, and that Ser+Cys and Thr+Cys residues
are the precursors to the formation of the characteristic Lan and MeLan structures, respectively.
For the type AI lantibiotics, two genes, lanB and
lanC code for proteins that have no similarity with
any other entries in the databases.68-70 They are
believed to be required for the dehydration of Ser and
Thr to Dha and Dhb, respectively (section 4.2), and
the conjugate addition of Cys residues to these
dehydro amino acids (section 4.3) to form Lan and
subclasses
example
Type A: elongated shape
nisin
Type B: globular shape

Type IIa: N-terminal consensus
YGNGVXC, Listeria-active,
contain 1-2 disulfides
mersacidin
leucocin A
Type IIb: two-peptide systems
lactococcin G
helveticin J
MeLan (e.g., Figure 4 for nisin). In the type AII and

type B lantibiotics, the lanBC genes are not present
and instead a single gene (lanM) producing a protein
with sequence homology at its C-terminus with the
LanC proteins is found. No homology can be detected
between the LanM and LanB proteins, indicating the
lanM genes do not originate from a gene fusion
event.66 It was initially postulated, and recently
confirmed both in vivo and in vitro,40,71 that the lanM
products carry out both the dehydration and cyclization reactions (section 4.4). It is very interesting that
the operons for type AII lantibiotics as well as the
sequences of their leader peptides have very strong
similarity to those involved in the biosynthesis of
Figure 4. Representative example of the posttranslational maturation process of lantibiotics. The prepeptide NisA is
ribosomally synthesized, followed by NisB catalyzed dehydration of underlined Ser and Thr residues in the propeptide
region of NisA. NisC catalyzes the conjugate addition of Cys residues in a regio- and stereospecific manner to five of the
Dha (green) and Dhb (purple) residues to generate five cyclic thioethers: one lanthionine (red) and four methyllanthionines
(blue). It should be emphasized that although it is generally assumed that LanB proteins complete their dehydration of
the targeted Ser and Thr residues before LanC proteins catalyze the cyclizations, at present it cannot be ruled out that
the two proteins pass the substrate between them such that LanB dehydrates one Ser/Thr followed by LanC catalyzed
ring formation before LanB dehydrates the next Ser/Thr. After dehydration/cyclization is complete, the leader peptide is
proteolytically removed by the protease NisP. Sequence of the leader peptide: MSTKDFNLDLVSVSKKDSGASPR.
nonlanthionine-containing bacteriocins (class II, Table

1) except for the addition of the lanM gene. This may
indicate that the recruitment of this single gene
resulted in the transformation of class II bacteriocin
producing bacteria to lantibiotic producing organisms.
An ATP-binding cassette (ABC) transport system
(lanT) is found in all lantibiotic gene clusters except
for epicidin 280.72 The LanT proteins are responsible
for secretion of either the final mature product or the
posttranslationally modified product still attached to
its leader sequence (section 4.7). Gene disruption and
heterologous expression studies have shown that
these transport systems are absolutely required in
some cases (e.g., subtilin73 and nisin74), but that
alternative transport systems can substitute during
biosynthesis of other lantibiotics (e.g., Pep575 and
epidermin69,76).
Although all lantibiotics require proteolytic removal of the leader peptide, genes encoding the
proteases are not always located in the biosynthetic
gene clusters, suggesting that other cellular proteases
can fulfill this role (e.g., subtilin77 and cinnamycin).78
When genes for dedicated proteases are found in the
biosynthetic operons, they have been designated lanP
and code for subtilisin-type serine proteases (section

4.7). Some clusters lacking a lanP gene have lanT
genes with an N-terminal protease domain fused to
the ABC-type transporter, similar to the transport
systems of class II bacteriocins.79 These protease
domains appear to be cysteine proteases by sequence
comparison with LagD, a transport-proteolysis system involved in the biosynthesis of the nonlantibiotic
class IIa bacteriocin lactococcin G.79 In addition to
the transport system that excretes the lantibiotic,
several gene clusters contain a second transport
system comprised of three genes (lanEFG) that has
been implicated in self-immunity in lantibioticproducing strains that contain them (section 6). In
addition to these transport proteins, a protein encoded by lanI is also believed to be involved in selfprotection for some family members. Finally, two
often found regulatory genes (lanKR) are important
for regulation of lantibiotic production and comprise
a two-component sensory system (section 5). A schematic representation of the overall process of lantibiotic biosynthesis is depicted for subtilin as a
representative example in Figure 5.
As depicted in Figure 3, other genes are found in
certain gene clusters, and they are in many cases
Figure 5. Schematic representation of the process of

subtilin biosynthesis. Subtilin serves as the ligand for the
receptor kinase SpaK,272 which upon sensing subtilin first
autophosphorylates a His residue and subsequently transfers the phosphate to an Asp residue on SpaR.261 After
phosphorylation, this transcription factor regulates the
expression of three transcriptional units (spaS, spaBTC,
and spaIFEG) involved in biosynthesis of and self-immunity against subtilin.272 The SpaS substrate is acted
upon by a membrane-associated multi-enzyme complex.216
The biosynthetic enzymes SpaB and SpaC introduce the
dehydro amino acids and (Me)Lan residues, respectively,
and the modified peptide is secreted in an ATP-dependent
manner by SpaT. Outside of the cell proteases remove the
leader peptide generating mature subtilin.77 SpaEFG
constitute another ABC-type transporter believed to be
important for immunity and SpaI is also involved in selfprotection.319
involved in additional, less frequently encountered

posttranslational modifications discussed in sections
4.5 and 4.6. A dramatic example was recently reported for cinnamycin with no less than 21 likely
open reading frames (orfs) in the vicinity of the cinA
structural gene (Figure 3).78 This lantibiotic contains
the highly unusual lysinoalanine residue as well as
the unique -hydroxy aspartate (Figure 6), and some
of these unassigned orfs may be involved in their
biogenesis as they have no similarities to known
proteins.
3. Structures of Lantibiotics
At present about 40 different lantibiotics are
known with varying structure, size, and mode of
action. A representative collection is depicted in
Figure 6 illustrating the high level of posttranslational modifications that typically amount to structural changes to about one-third of all amino acids
in the peptide. The lantibiotics were classified by
Jung as type A or B, based on the topology of their
ring structures and their biological activities (Table
2).80 The type A lantibiotics, with nisin as the
prototype, exist as elongated amphipathic screwshaped structures in solution, varying in length from
20 to 34 residues and bearing a net positive charge.
Initially, their bactericidal action was believed to
predominantly involve the formation of short-lived
pores in cell membranes. More recently, a growing
number of lantibiotics have been shown to interfere
with peptidoglycan biosynthesis by binding to lipid
II, but this activity is not confined to the type A
Chatterjee et al.
lantibiotics (section 8). Type B lantibiotics such as

cinnamycin and mersacidin are more globular and
compact in structure (Figure 6), and they generally
have no net charge or a negative charge at pH 7. A
further subdivision within the type A lantibiotics is
based upon the modification enzymes involved in
their biosynthesis. The type A lantibiotics in which
the Lan and MeLan residues are formed by the action
of two distinct enzymes (LanB and LanC) are classified as type AI, whereas those that are formed by
a single enzyme (LanM) are termed type AII (Table
2).81 The structures and unique modifications present
in representative examples from seven subgroups,
each of which is likely derived from a common
ancestor, will be discussed in this section (for an
alternative classification scheme based on genetic
data, see section 4.1).
3.1. Type AI Lantibiotics: Nisin Group

Nisin is produced by L. lactis,38 and as mentioned
previously, has been used as a preservative in the
food industry for over 40 years without the appearance of significant bacterial resistance. The efforts
to understand the molecular basis of its action have
rendered it the most extensively studied lantibiotic.82
The two common forms of nisin are nisin A and Z,
which differ by a single amino acid at position 27,
which is His in nisin A and Asn in nisin Z.83,84
Recently, another natural variant, nisin Q, has been
isolated from L. lactis 61-14 that differs at four
positions (Val15, Leu21, Asn27, and Val30) from
nisin A (Ala15, Met21, His27, and Ile30).85 The name
nisin is derived from Lancefield Group N inhibitory
substance, the initial classification of the compound.86
The structure of nisin A was worked out by Gross in
19714 and later confirmed by genetic analysis of the
prepeptide38 and the landmark total synthesis by
Shiba and co-workers.87 Nisin contains three dehydrated amino acids and five thioether rings, which
are not amenable to amino acid analysis by Edman
sequencing. Gross was able to overcome this impediment by reductive desulfurization of the rings with
Raney nickel to yield D- and L-Ala in the case of mesolanthionine, and D-R-aminobutyric acid and L-Ala in
the case of MeLan, which were identified by sequencing before or after proteolysis of the peptide. The
positions of Lan and MeLan were identified by
performic acid oxidation of the thioether linkages to
the corresponding sulfones, which upon prolonged
heating in sodium bisulfite underwent -elimination
to generate the R,-didehydro amino acids. Subsequent addition of bisulfite generated sulfonic acid
derivatives that could be identified by sequencing.121
Dha and Dhb residues also interfere with Edman
sequencing (and amino acid analysis) by the formation of a pyruvyl group under hydrolytic conditions.122
Hence, these groups were hydrogenated or treated
with thiol reagents prior to sequencing. The Sconfiguration at the -position of MeLan was established by comparison of retention times during cation
exchange chromatography with authentic samples.17
Interestingly, in the 30 odd years since the initial
determination of their stereochemistry, no reports
Figure 6. Representative structures of some lantibiotics using the shorthand notation and color coding defined in Figure
1. The ring numbering is shown for nisin and cinnamycin and is typically alphabetical from the N- to C-terminus.
have suggested the occurrence of other diastereomers

of Lan/MeLan in lantibiotic isolates. Although this
may be pending a more thorough characterization of
various more recent isolates, a strong intrinsic preference for the natural diastereomers has been shown
in the nonenzymatic biomimetic cyclization of the
B-rings of subtilin and nisin (section 4.3).123-126
Subtilin is produced by Bacillus subtilis ATCC
663337 and is structurally closely related to nisin,
showing 63% sequence identity including one Lan
and four MeLan rings of identical size and position
along the peptide chain (Figure 6). In addition, both
contain Dha residues at position 5 and as their
penultimate amino acid. The structure of subtilin was
determined by Gross5 and confirmed by two-dimensional NMR methods by Roberts and co-workers.127
A natural variant of subtilin that is succinylated on
the N-terminus, [NR-succinyl-Trp1]-subtilin, has been
identified in the culture broth of B. subtilis ATCC
6633 and found to have reduced antibacterial activity
compared to subtilin (MICs against Micrococcus
luteus are 0.05 and 0.34 g/mL, respectively).128
Ericin A and S88 whose precursor peptides have

high identity with the subtilin precursor (75 and 92%,
respectively) have been classified as type AI lantibiotics. A putative thioether bridging pattern has been
proposed based on the results of Edman sequencing,
digestion with peptidases, mass spectral analysis,
and the similarity in the placement of Ser, Thr, and
Cys residues with those of subtilin. Streptin isolated
from Streptococcus pyogenes129 is another type AI
lantibiotic, the structure of which has been proposed
based on its similarity to nisin.89

The solution structure of nisin has been determined
using high-resolution NMR spectroscopy both in
aqueous solution and in the presence of dodecylphosphocholine (DPC) and sodium dodecyl sulfate (SDS)
micelles that mimic the cellular membrane.130-132
These studies revealed an overall extended conformation and the presence of two amphipathic screwshaped domains consisting of the N-terminal A-, B-,
and C-rings, and the C-terminal fused rings D and
Chatterjee et al.
Table 2. Lantibiotics Isolated as of June 2004

lantibiotic
Nisin Aa
Nisin Za
Subtilina
Ericin Sb
Ericin Ab
Streptinb
Epidermina
[Val1-Leu6]-epiderminb
Gallidermina
Mutacin 1140a
Mutacin B-Ny266b
Mutacin IIIb
Mutacin Ib
Pep5a
Epilancin K7a
Epicidin 280b
Lacticin 481a
Variacinb
Mutacin IIa
StreptococcinA-FF22a
Salivaricin Ab
[Lys2,Phe7]-salivaricin Ab
Lactocin Sb
Cypemycina
Plantaricin Ca
Sublancin 168b
Butyrivibriocin OR79Ab
producer strain
ref
Type A (I)
Lactococcus lactis ATCC 11454
4
Lactococcus lactis N8, NIZO22186
83
Bacillus subtilis ATCC6633
5
Bacillus subtilis A1/3
88
Bacillus subtilis A1/3
88
Streptococcus pyogenes BL-T, M25
89
Staphylococcus epidermis Tu3298
6
Staphylococcus epidermis
90
BN-V1, BN-V301
Staphylococcus gallinarium Tu3928
91
Streptococcus mutans JH1140
92
Streptococcus mutans Ny266
93
Streptococcus mutans UA787
94
Streptococcus mutans CH43
95
Staphylococcus epidermis 5
7
Staphylococcus epidermis K7
96
Staphylococcus epidermis BN280
72
Type A(II)
Lactococcus lactis CNRZ 481
97
Micrococcus varians MCV8
98
Streptococcus mutans T8
99
Streptococcus pyogenes FF22
100
Streptococcus salivarius 20P3
18
Streptococcus pyogenes T11
101
(M type 4)
Lactobacillus sakei L45
102
Streptomyces OH-4156
103
Lactobacillus plantarum LL441
104
Bacillus subtilis 168
105
Butyrivibrio fibriosolvens
106
Lan
MeLan
Dha
Dhb
amino
acids
1
1
1
1
1
2
2
2
4
4
4
4
4
1
1
1
2
2
2
2
1
0
0
0
1
1
1
1
0
3
1
1
34
34
32
32
29
23
22
22
2
2
2
2
3
2
2
1
1
1
1
1
0
1
1
2
0
1
1
1
2
0
2
0
1
1
1
1
0
2
2
1
22
22
22
22
24
34
31
30
2
2
2
1
1
1
1
1
1
2
2
2
0
0
0
0
0
0
1
1
1
1
0
0
27
25
27
26
22
22
2
0
1
0
1
0
0
3
1
2
0
0
1
1
0
1
4
0
0
1
37
22
27
37
25
1
1
2
2
0
0
0
0
19
19
1
1
1
0
1
1
2
2
2
3
2
2
0
0
1
1
0
0
0
0
0
0
0
0
19
19
19
20
19
20
2
2
4
3
1
2
2
1
0
0
Dha/Dhb
Dha/Dhb
0
1
Dha/Dhb
Dha/Dhb
2
2
3
2
0
1
4
4
30
29
30
28
29
32
38
21
Duramycin Ba
Duramycin Ca
Ancovenina
Mersacidina
Actagardinea
Ala(0)-actagardineb
Type B
Streptomyces cinnamoneus
10,107
Streptoverticillium hachijoense
108
DSM 40114
Streptoverticillium R 2075
109,110
Streptomyces griseoluteus R 2107
109,110
Streptomyces sp. A647P-2
13
Bacillus sp. strain HIL Y-85,54728
16
Actinoplanes linguriae ATCC 31048
14
Actinoplanes linguriae ATCC 31048
111
Lacticin 3147A1a
Lacticin 3147A2a
Staphylococcin C55Rb
Staphylococcin C55b
Plantaricin WRb
Plantaricin Wb
Cytolysin LLb
Cytolysin LSb
Two-Component Lantibiotics
Lactococcus lactis DPC3147
112
2
Lactococcus lactis DPC3147
1
Staphylococcus aureus C55
113
Lan/MeLan
Staphylococcus aureus C55
Lan/MeLan
Lactobacillus plantarum LMG 2379
114
2
Lactobacillus plantarum LMG 2379
1
115
Lan/MeLan
Lan/MeLan
Cinnamycina
Duramycina
Ruminococcin A
Carnocin UI 49
Macedocin
Bovicin HJ50
Nukacin ISK-1
SapB morphogen
a
Structures Not Yet Determined

Ruminococcus gnavus E1
116
Carnobacterium piscicola UI49
117
Streptococcus macedonicus
118
ACA-DC 198
Streptococcus bovis HJ50
119
Staphylococcus warneri ISK-1
120
Streptomyces coelicolor
474
Structure established independently. b Proposed structure.
E that are joined by a flexible hinge region (residues

20-22, Asn-Met-Lys) as depicted in Figure 7. The
secondary structure of subtilin was found to be
similar to nisin.127 The presence of a Dhb residue at
position 18 in subtilin compared to Gly in nisin A did
not significantly affect the conformational flexibility
of the C-ring. In both compounds, the four-amino acid
containing thioether rings are enforced to adopt a
-turn131,133 (type I in rings B and C and type II and
type II in rings D and E).54 Although some regions
of rigidity are present within the individual lanthionine rings, both nisin and subtilin were found to be
overall flexible molecules.127,130
3.2. Type AI Lantibiotics: Epidermin Group

The structure of the 22-residue lantibiotic epidermin was elucidated by Jung and co-workers in 1985.6
Epidermin contains one Dhb, one MeLan, and two
Lan residues besides the unusual ring structure
containing S-[(Z)-2-aminovinyl]-D-cysteine (AviCys)
(Figures 1 and 6). Epidermins structural elucidation
was carried out by chemical and enzymatic fragmentation coupled with Raney-Ni catalyzed desulfurization, Edman sequencing, and FAB-MS analysis. The
AviCys residue was characterized by its conversion
III and mutacin 1140 were initially isolated, characterized, and named by different research groups
but have identical structures that differ by two amino
acids from mutacin B-Ny266, and share 77% identity
with epidermin and 62.5% identity with mutacin I.
All members of the epidermin group have the characteristic Lan ring between positions 3 and 7; however, only the mutacins have a Dha at position 5,
which is also found in the nisin subgroup and has
been implicated in its biological activity (section
8.3).136-138
Figure 7. (A) One of the NMR structures of nisin in the

presence of DPC micelles.131 The molecule adopts a more
or less extended conformation with the N- and C-termini
somewhat curling back toward each other. The four amino
acid rings B, D, and E all have enforced -turn conformations, which is also adopted in a noncovalent fashion by
residues 21-24. (B) Molecule shown from the same viewpoints as in panel A with the A-ring in blue, the B-ring in
yellow, the C-ring in cyan, and the fused D- and E-rings
in orange. Figures were generated using the program
RASMOL.469
to S-[2-aminoethyl]-D-cysteine (thialysine) by reduction of epidermin with Pd/C and subsequent hydrolysis, and also by the formation of L-alanine-Nethylamide upon Raney-Ni desulfurization of the
C-terminal tryptic fragment. The absolute configuration of the Lan and MeLan residues was confirmed
to be identical to that found in nisin by gas chromatography with a chiral stationary phase.
The natural epidermin variant gallidermin, produced by Staphylococcus gallinarum, differs from
epidermin by a single amino acid, Leu6, which is Ile
in epidermin.91 The structural elucidation of the
polypeptide revealed the presence of four thioether
bridges identical to epidermin. Another natural variant of epidermin, initially named staphylococcin T,
has been isolated from Staphylococcus cohnii T.134
Amino acid analysis and DNA sequencing indicated
that it is identical to gallidermin.
Mutacin 1140,92,135 mutacin B-Ny266,93 mutacin I,95
and mutacin III94 are other lantibiotics in the epidermin group that are all isolated from various
strains of Streptococcus mutans. They bear close
homology to each other and to epidermin. Mutacin
In the presence of the structure inducing solvent

trifluoroethanol (TFE) gallidermin adopts an extended amphiphilic screw-shape with a lipophilic
C-terminus and positively charged hydrophilic Nterminus.139,140 Flexibility in the fairly rigid peptide
backbone is due to a hinge region from residues 12
to 15. The calculated average length of 30 and
average diameter of 8-10 would allow the molecule
to span the cell membrane once, which may be
relevant for its pore forming activity. Obviously,
multiple molecules would have to come together to
generate the pore. The recent solution structure of
mutacin 1140 as determined by Edison and coworkers in acetonitrile-water (80:20) retains the
rigidity within the lanthionine rings that are seen
in nisin and gallidermin as well as the flexibility of
the hinge region.141 The structure differs from that
of gallidermin in being bent in the hinge region,
giving it a horseshoe-like appearance.
3.3. Type AI Lantibiotics: Pep5 Group

The representative lantibiotic Pep5 isolated from
S. epidermis 5142 is large (34 residues, 3488 Da) and
basic in nature (pI > 10.5) due to the presence of
eight positively charged amino acids. The structure
of Pep5 was determined by Jung and co-workers7 and
contains two Dhb residues, two Lan, and one MeLan.
The stereochemistry of the Lan and MeLan residues
was identical to that in nisin and subtilin. Fragmentation of Pep5 by enzymatic cleavage with chymotrypsin and endoproteinase Arg-C gave rise to four
fragments that were partially identified by Edman
sequencing and FAB-MS. The N-terminus of Pep5 is
blocked to sequencing, and 13C NMR spectroscopy
suggested the presence of a 2-oxobutyryl residue at
this position. This was confirmed with the synthesis
of an N-terminally 2-oxobutyrylated heptapeptide
corresponding to the N-terminal proteolytic fragment
of Pep5. The deduced structure was also consistent
with the subsequently determined pre-Pep5 gene
sequence that revealed a codon for Thr at position
1.143 Presumably, this Thr at the N-terminus is
converted to a Dhb during the maturation process,
which spontaneously hydrolyzes to the observed
2-oxobutyryl structure. To eliminate the difficulty in
Edman sequencing due to the presence of this residue
and the other unusual amino acids in Pep5, Meyer
and co-workers have outlined a noteworthy method
of chemical derivatization.144 They achieved the
Chatterjee et al.
residues.72 It bears 75% identity in amino acid

sequence to Pep5, which has led to the suggestion of
a similar pattern of thioether bridging. Like epilancin
K7, its N-terminus is blocked by the 2-hydroxypropionyl group.

The solution structure of Pep5 has been investigated in water and water-TFE mixtures by circular
dichroism (CD) and NMR spectroscopy.146 In CD
studies, Pep5 presented a disordered, random coil
type structure in water alone, while the addition of
TFE was found to induce helicity. The helical nature
of Pep5 was also seen in the presence of sodium
dodecyl sulfate micelles that mimic the membrane
environment. Thus, it was concluded that Pep5
remains in a disordered state in aqueous solution and
adopts a (pore forming) helical shape in a lipophilic
environment. NMR experiments in water echoed the
mostly disordered state observed by CD. Similar to
gallidermin, some rigidity was observed in the backbone of the thioether rings and residues immediately
next to the Dhb. In the presence of TFE, the inflexibility observed in water was further extended as
reflected in a higher number of backbone NH-NH
cross-peaks. On the basis of these findings, some
degree of helicity was inferred in the segment spanning residues 14-23, suggesting a partly rod-shaped
structure in a membrane-like dielectric.
3.4. Type AII Lantibiotics: Lacticin 481 Group

Figure 8. Modification of Dha, Dhb, Lan, and MeLan with
ethanethiol to generate products that are amenable to
Edman degradation.144
almost complete sequencing of Pep5 by the initial

treatment with an alkaline mixture of ethanethiol
that led to the opening of thioether rings and formation of cysteine and S-ethylcysteine from Lan, and
-methyl-S-ethylcysteine together with cysteine or
-methylcysteine and S-ethylcysteine from MeLan
(Figure 8). This procedure was followed by an oxidative step with trifluoroperacetic acid that released
propionic acid, CO2, and freed the N-terminal amine
of Pep5. This method was also shown to derivatize
the AviCys residue in epidermin and gallidermin and
allowed for the complete sequencing of the latter.
Epilancin K7, isolated from S. epidermis K7, is
three residues shorter than Pep5 and contains two
Dha residues that are absent in Pep5.96 The structure
of epilancin K7 was determined by extensive twodimensional 1H-homonuclear and 1H,13C-heteronuclear NMR spectroscopy as well as three-dimensional
1
H-homonuclear NMR spectroscopy. An unprecedented 2-hydroxypropionyl (lactate) group was identified at the N-terminus and corresponds to a Ser
residue in the ElkA prepeptide as determined by
sequencing of the structural gene.145 Like the oxobutyryl moiety in Pep5, this structure is probably the
result of dehydration of the Ser residue followed by
spontaneous hydrolysis of the N-terminal Dha residue. Unlike Pep5, the resulting ketone is subsequently reduced to the alcohol in a process for which
currently the stereochemistry is not known. Epicidin
280 contains one Dhb, one Lan, and two MeLan

Lacticin 481147 (also isolated as lactococcin DR)148
was purified and partially sequenced by Piard and
co-workers from L. lactis CNRZ481.149 Two probable
structures of lacticin 481 were proposed based on
Edman degradation, amino acid analysis, NMR spectroscopy, and comparison with the predicted translational product of the lctA gene.150 Lacticin 481 (ESIMS 2901 Da) is 27 residues long with a high Gly
content (11%), a high proportion of hydrophobic
residues (75%), and no net charge at pH 7. It contains
two Lan, one MeLan, and one Dhb residue. The
precise location of the thioether bridges was determined by van de Hooven et al.97 after cyanogen
bromide digestion of lacticin 481, and analysis of the
fragments by FAB-MS and NMR spectroscopy (Figure 9). The compact C-terminus, arising from overlapping rings, and the low homology of the prelacticin
leader peptide with the prenisin and presubtilin
leader sequences (section 4.1), led to a distinct
classification of lacticin 481 and its analogues as type
AII lantibiotics (Table 2). Mutacin II (3245 Da), a
close relative of lacticin 481 (Figure 9), was isolated
from Streptococcus mutans T8 by Caufield and coworkers.151 It contains 27 amino acids, including one
MeLan, two Lan, and one Dhb residue.151,152 The
complete structure, including the bridging topology,
was established by a combination of cyanogen bromide digestion, mass spectrometry, site-directed mutagenesis, and NMR spectroscopy (Figure 9).99 Its
primary structure agrees with the product of the
mutA gene sequence.153 The compound bears high
Figure 9. Structures of lacticin 481,97 mutacin II,99 sublancin 168,105 actagardine,14 and cypemycin.103 The natural derivative
Ala(0)-actagardine is shown as a dotted circle.
identity with lacticin 481 (59%), including an identically clustered ring structure and invariant position
of a Dhb.
Lactocin S produced by Lactobacillus sakei L45154
is a unique lantibiotic in that its structure contains
three Ala of D-configuration.102,155 The compound
(3764 Da) consists of 37 residues, including two Lan
and one Dhb residue. The N-terminus of lactocin S
is blocked to Edman sequencing due to the presence
of a 2-oxopropionyl group.102 Its structure has been
proposed but as yet has not been firmly established.
StreptococcinA-FF22 (SA-FF22),156 salivaricin A,18
[Lys2,Phe7]-salivaricin A (salivaricin A1),101,157 variacin,98 plantaricin C,158 and butyrivibriocin OR79A106
are the other members of this family that share high
homology, and a similar pattern of ring formation to
that found in lacticin 481. At present, the ring
structure has only been determined unequivocally for
SA-FF22100 and plantaricin C.104 Salivaricin is an
interesting case in that salivaricin A is active against
strains of S. pyogenes, but use of salA from Streptococcus salivarius 20P3 as a DNA hybridization probe
showed that no less than 63 out of 65 S. pyogenes
strains tested contained a salA homologue.101 Some
of these strains were further investigated revealing
that those strains in which the production of a
salivaricin analogue was disrupted by deletions in the
biosynthetic genes were sensitive to salivaricin A,

whereas those that actually produced an analogue
were not. One of the latter strains (S. pyogenes T11)
was shown to produce the lantibiotic [Lys2,Phe7]salivaricin A (salivaricin A1).157
Sublancin 168, produced by B. subtilis 168,105
differs significantly from the other lantibiotics of this
group due to the presence of two disulfide linkages,
besides a MeLan and a Dha residue. The pattern of
rings is also different from that of lacticin 481 (Figure
9), and it contains an as yet unknown modification
of +164 Da.105 Sublancins inclusion with the lacticin
subgroup results from the similarity of its leader
peptide with other members of this group (section
4.1). Cypemycin159 is an extraordinary lantibiotic
containing four Dhb residues, one allo-Ile, one AviCys, and an N-terminal N,N-dimethylalanine residue
(Figure 9).103 This molecule illustrates the range of
posttranslational modifications that may take place
during lantibiotic maturation.

The type AII lantibiotics are made up of a characteristic linear N-terminus and globular C-terminus.
On the basis of NMR studies, the structure of
mutacin II has been proposed to consist of an Nterminal amphipathic R-helix from residues 1-8,
with a hinge region around Pro9 separating it from

the C-terminal Lan/MeLan rings.160 The importance
of this hinge region has been demonstrated by the
mutation P9A, which leads to loss of antimicrobial
activity.161 CD studies conducted with SA-FF22 in
water-TFE mixtures, 1% aqueous SDS, and in the
presence of vesicles showed a change in secondary
structure and suggested an ordered conformation
different from that of the type AI lantibiotics Pep5
and gallidermin.100 The solution structure of plantaricin C has been determined by NMR spectroscopy.
It shows two distinct regions comprised of a positively
charged, flexible N-terminus (residues 1-6) and a
C-terminal rigid globular domain.104
3.5. Type B Lantibiotics: Mersacidin Group

The representative lantibiotic from this group,
mersacidin, was isolated from Bacillus HIL Y-85,54728.16,60 Its primary sequence was investigated by
two-dimensional NMR spectroscopy, GC-MS analysis
of the acid hydrolysate, and MS/MS studies of its
desulfurized analogue.162 Mersacidin is one of the
smaller lantibiotics, 20 residues long (1825 Da), and
contains three MeLan rings, one Dha, and the
residue S-[(Z)-2-aminovinyl]-(3S)-3-methyl-D-cysteine,
AviMeCys (Figure 1). The AviMeCys residue is
presumably formed by the same oxidative decarboxylation mechanism as seen for the formation of AviCys
in epidermin (section 4.5). The primary amino acid
sequence of mersacidin was confirmed by cloning of
the mrsA gene from the producing strain by Bierbaum et al.61 In contrast to the type AI lantibiotics,
mersacidin has no net charge.
Actagardine is a tetracyclic lantibiotic, 19 residues
in length (1890 Da), and bears a single negative
charge in neutral solution (Figure 9).23,163 It has also
been isolated under the name gardimycin.164-167 The
primary structure of this lantibiotic was first published by Kettenring and co-workers in 1990,168 but
Zimmermann and Jung proposed a revised structure
in 1995 that is currently accepted.14 Edman degradation and amino acid sequencing, before and after
modification of actagardine with -mercaptoethanol
as per the protocol of Meyer et al. (i.e., Figure 8),144
along with ESI-MS and multidimensional NMR
experiments, established the presence of one Lan and
three MeLan residues. Interestingly, the authors
observed a major portion of actagardine to possess
an unprecedented oxidized C-terminal MeLan residue (sulfoxide), while only a small amount corresponded to the non-oxidized actagardine.14 A natural variant of actagardine, Ala(0)-actagardine, was
isolated from Actinoplanes linguriae ATCC 31048
(Figure 9).111 NMR spectroscopy, ESI-MS, Edman
sequencing, and amino acid analysis indicated a
primary sequence identical to actagardine with an
additional N-terminal Ala residue. A synthetic Ala(0)-actagardine, prepared by coupling previously
purified actagardine with Boc-protected O-N-hydroxysuccinimidyl alanine and subsequent deprotection of
the N-terminus, had identical retention times during
Chatterjee et al.
HPLC purification as the isolated wild-type compound.111

The activity of mersacidin59 against the virulent
strain S. aureus (see also Note Added in Proof) has
produced efforts to uncover its key structural features
in solution169 and crystalline state.170 The solution
NMR structure in methanol showed three distinct
structural domains formed by the thioether rings
spanning residues 1-3, 4-12, and 13-20 (Figure
10).169,171 The overall structure is globular with
mostly neutral side-chains and only a few charged
groups (Glu17 and the N-terminus) exposed to solvent. The glycine-rich sequence in domain II (residues 4-12) confers some conformational flexibility.
Intramolecular hydrogen bonds were observed between domains I and III and within domain III that
promote backfolding and rigidify the structure. The
structure of mersacidin has been solved by X-ray
crystallography using crystals obtained from a saturated solution of benzene and methanol.170 This was
the first and only example of an X-ray structure of a
lantibiotic and showed good correspondence with the
solution structure proposed by Griesinger and coworkers (vide supra) except for the glycine rich region
in domain II.
Zimmerman and Jung determined the solution
structure of actagardine in 1997 by two- and threedimensional NMR spectroscopy in a water-acetonitrile mixture (3:7).15 Actagardine has a compact
globular structure comprised of two domains joined
between residues 6 and 7 (Figure 11). The N-terminal
domain consists of a single Lan ring, while the
C-terminal domain is composed of three intertwisted
MeLan rings. Residues 7-8, 9-12, and 17-19 form
a small, three-stranded -sheet with one antiparallel
and two parallel strands that is not common to
lantibiotic structures, and provides rigidity along
with the thioether bridges. The fixed torsional angle
between residues 6 and 7 and van der Waals interactions lead to an L-shaped planar arrangement of the
two domains. Two putative binding pockets are
present in actagardine. A hydrophilic pocket is made
by Glu11 and Ser2 and the thioether bridges between
residues 1-6 and 7-12. This resembles the sequence
between residues 9 and 18 in mersacidin, and the
similar position of the Glu residue suggests a role in
the mode of action in both mersacidin and actagardine. The second pocket is formed by the backbone
amide loop of residues 12-17 and the thioether
bridge from residue 14-19 and may be involved in
binding a hydrogen bond acceptor.
3.6. Type B Lantibiotics: Cinnamycin Group

This group consists of the lantibiotics cinnamycin,
duramycin, duramycin B and C, and ancovenin, all
of which are very similar in their ring structure
and amino acid composition (Figure 12). Cinnamycin56,78,107,172 isolated from Streptomyces cinnamoneus
was also purified as Ro 09-0918173-175 from Streptoverticillium griseoverticillatum and as lanthiopep-
Figure 10. (A) Stereoview of one of the NMR structures of mersacidin in water-methanol.171 (PDB accession number
1MQX). (B) View of mersacidin in the same orientation as in panel A in which the rings are highlighted and the sidechains are omitted. Shown in blue is the A-ring, in red the B-ring, in green the C-ring, and in magenta the D-ring. Figures
were generated using the program RASMOL.469
tin by Shiba and co-workers from Streptoverticillium

L337-2.176 Structural determinations by various groups
have shown that these compounds are identical.56,174,176
The structure of cinnamycin was first suggested by
Gross,177 with later corrections to the amino acid
sequence by Kessler174 and Fredenhagen et al.56,109
The revised structure is in agreement with the
prepeptide gene sequences from S. griseoverticillatum172 and S. cinnamoneus.78 Cinnamycin, the duramycins, and ancovenin are all 19 residues long and
contain one Lan and two MeLan rings in conserved
positions. Except for ancovenin they all possess the
unusual lysinoalanine ring that joins Lys19 to Dha6.172
In ancovenin, Ser6 undergoes dehydration but no
cyclization takes place onto Dha6.13 Ancovenin is also
the only family member in which Asp15 is not
-hydroxylated (section 4.6). To further distinguish
them from the type A lantibiotics, the cinnamycin
group contains two MeLan residues where the nucleophilic Cys is positioned N-terminally to the Dhb
(Figure 12). This reversed direction of ring formation
is also seen in the type-B lantibiotics mersacidin, the
A1 peptide of the two-component lantibiotic lacticin
3147, and actagardine, but is not found in type A
lantibiotics.
The current structure of duramycin was proposed
by Hayashi108 and Fredenhagen.56 Duramycin (also
isolated as leucopeptin by Kondo et al.178) differs from
cinnamycin at a single residue at position 2, while
the duramycins B and C differ by one and six amino
acids, respectively (Figure 12). Ancovenin isolated
from Streptomyces sp. A647P-2 contains two MeLan
rings, one Lan ring, and one Dha residue. Wakamiya
et al.12,13 confirmed the presence of Dha at position 6
by reduction of ancovenin with Pd/H2, which resulted
in Ala at residue 6, as well as by nucleophilic addition
of the methyl ester of mercaptoacetic acid (HSCH2COOCH3). The position and stereochemistry of the
sulfide bridges were assigned by comparison with
synthetic samples and chiral GC analysis.12,13

The solution structure of cinnamycin has been
determined in DMSO and a water-acetic acid mixture (9:1) by NMR spectroscopic techniques.179 Its
ring structure can be viewed as four structural
domains. Residues 1-4 and 14-18 along with the
thioether bridges spanning Ala1-Abu18 and Ala4Ala14 together constitute the A-ring (Figure 6). This
ring exists in an antiparallel -sheet structure supported by backbone H-bonds between the Arg2 carbonyl and amide of Abu18, the carbonyl of Ala4 and
amide of Gly16, and the carbonyl of the side-chain
of Asn17 and the amide of Ala4. The C-ring formed
by residues 5-11 showed a high degree of flexibility.
The D-ring comprising Lys19 to Ala6 and the lysinoalanine is placed above the plane occupied by the
other rings resulting in an amphipathic structure.
The conformation of cinnamycin undergoes a strong
change in the presence of SDS micelles.180 NOE
measurements conducted in water and SDS micelles
indicated that the change takes place predominantly
in a hinge region that connects the lipophilic and
lipophobic portions of the molecule. Moreover, studies
with cinnamycin in SDS bilayers180 and in the presence of 1-dodecanoyl-sn-glycerophosphoethanolamine
(C12-LPE)181 indicated a conformational change in
the lipophilic portion of the molecule due to interactions with hydrophobic segments of the lipids. Zimmerman et al.110 have determined the solution structures of duramycins B and C and also observed a
hinged amphiphilic structure. The authors noted that
Chatterjee et al.
Figure 11. (A) Stereoview of one of the NMR structures of actagardine reported by Jung and Zimmermann.15 (PDB
accession number 1AJ1) (B) View of actagardine in the same orientation as in panel A in which the rings are highlighted
and the side-chains are omitted. Shown in blue is the A-ring that makes up domain A and in magenta domain B formed
by ring B (green), ring C (red), and ring D (purple). Figures were generated using the program RASMOL.469
overall structure but led to a decreased amphiphilicity in these compounds.
3.7. Two-Component Lantibiotics: Lacticin 3147

A separate subgroup is formed by the two-component lantibiotics that consist of two posttranslationally modified peptides that individually have weak
activity but synergistically display strong antibacterial action. Two-component systems had been previously characterized in the class II bacteriocins, and
only a few such systems have been found so far in
lantibiotics.182 At present, only the two polypeptides
of lacticin 3147 have been structurally characterized.

Figure 12. Sequence comparison of compounds from the
cinnamycin group. The residues involved in ring formation
are depicted in three letter code, whereas one letter code
is used for the other residues. Arrows represent the
directionality of cyclization. Ancovenin posseses neither the
head-to-tail lysinoalanine bridge nor the hydroxyaspartic
acid residue present in all other members of this group.13
the changes in the amino acid sequence compared to

that of cinnamycin (Figure 12) did not disturb the
The two components of lacticin 3147, LtnA1 and

LtnA2, were initially purified and characterized from
the producer strain L. lactis DPC3147 by Hill and
co-workers.183 Amino acid analysis of both LtnA1 and
A2 revealed the presence of Lan/MeLan residues as
well as D-Ala by chiral-phase GC. Furthermore, the
total number of Ala residues present in the two components was found to be greater than that predicted
from the genetic sequence of the prepeptides. Edman
sequencing of LtnA1 after derivatization with 1-pro-
gene sequences of both LtnA1/2 and lactocin S encode

Ser residues at the positions of the observed Dalanines, it is believed that they arise from dehydration followed by stereospecific hydrogenation.184,185
Although Lan/MeLan amino acids have been identified in the two-component lantibiotic systems plantaricin W,114 staphylococcin C55,113 and cytolysin L,115
their structural determination is not complete and
the position and nature of modified residues is only
speculative based on limited sequence homology to
known lantibiotics.
Figure 13. Modification of Dha, Dhb, Lan, and MeLan

by reduction with NaBD4 and nickel boride to generate a
product that is amenable for Edman degradation.112 This
method allows distinction between dehydro amino acids
and (methyl)lanthionines. Another method that achieves
this was reported by Smith et al.92
panethiol indicated the presence of Lan/MeLan and/

or Dha/Dhb residues. A genetically predicted Ser
residue at position 7 was found to correspond to a
D-Ala residue in the mature product suggesting
posttranslational conversion of Ser to D-Ala. The
Edman sequencing of LtnA2 was not possible even
after chemical modification and was attributed to the
presence of a 2-oxobutyryl residue at the N-terminus,
formed by the hydration-deamination reaction of a
Dhb residue at this position. Vederas and co-workers
recently completed the characterization of the structure of LtnA1 and A2 by a novel method involving
nickel boride (Ni2B), an in situ generated hydrogenation and desulfurization catalyst.112 Treatment of the
lantibiotics with Ni2B in the presence of NaBD4 in
CD3OD/D2O led to the desulfurization of Lan/MeLan
and incorporation of a deuterium atom at the -carbon of each of the constituent residues (Figure 13).
Simultaneous reduction of the Dha/Dhb residues
resulted in addition of two deuterium atoms across
the double bonds. The LtnA2 peptide was deblocked
prior to Ni2B treatment by the removal of two
N-terminal residues upon treatment with 1,2-diaminobenzene in an acetic acid/sodium acetate buffer.
Automated Edman sequencing of the deblocked
and reduced peptides along with NMR spectroscopy
were used to assign the structures of both peptides.
LtnA1 contains two Lan and two MeLan residues,
two dehydrobutyrines, and one D-Ala at position 6.
The overall structure resembles that of mersacidin
(Figure 6). LtnA2 was found to contain one Lan and
two MeLan residues besides two Dhb and two D-Ala
residues. The A2 peptide may have similarity with
lactocin S, the only other lantibiotic to contain
D-alanine but whose proposed structure has not yet
been unequivocally confirmed.184 Since the structural
Preliminary NMR solution structures of LtnA1 and

A2 show that they exist in different conformations.112
LtnA1 exhibits a globular shape resembling mersacidin, while LtnA2 is similar in structure to the type
AI lantibiotics, being elongated and screw-shaped.
The similarity in lanthionine bridging patterns of
LtnA1 and mersacidin, especially in the two Nterminal rings, have been invoked to suggest similar
conformational changes for LtnA1 that are observed
for mersacidin in the presence of the cell membrane
and lipid II (section 8.2).171
4. Biosynthesis of Lantibiotics
4.1. Lantibiotic Precursor Peptides
All lantibiotic precursor peptides (LanA) contain
a C-terminal structural region that undergoes posttranslational modification (propeptide) and a relatively long N-terminal leader sequence containing between 23 and 59 amino acid residues, which remains unaffected during biosynthesis. Whereas both
the leader sequence and propeptide region contain
serine and threonine residues, cysteines have only
been found in the propeptide segment. Comparisons
of the leader sequences of a large number of lantibiotics have revealed two different conserved motifs
(Figure 14),186 which has been proposed as the basis
for an alternative classification of lantibiotics81 than
that presented above (i.e., type AI, type AII, and type
B). In this organization by genetics rather than activity profiles or three-dimensional structure, the
class I lantibiotics all have a common FNLD motif
between positions -20 and -15 and usually a Pro at
position -2. The biosynthetic machinery that carries
out the posttranslational modifications in this class
consists of LanB and LanC. In contrast, class II peptides contain a characteristic GG or GA cleavage
site (historically termed the double Gly motif),186-188
contain multiple Asp and Glu residues, and are
processed by one modification enzyme (LanM). An
exception to this general rule are the salivaricins,
which contain the typical leader peptide signature
for the class II lantibiotics but which are actually
modified by SalB and SalC proteins.157 A few outlying
sequences are found in cinnamycin and mersacidin
that both have very long leader peptides, and in
lactocin S (Figure 14). The cleavage site of the
cinnamycin leader peptide has the AXA motif found
for recognition by type I signal peptidases of the
general secretory (sec) pathway.78
Chatterjee et al.
Figure 14. Sequence alignments of several prepeptides of lantibiotics showing conserved motifs discussed in the text in
red. The classification is based on consensus sequences present in the leader peptides as well as the modification process
that is catalyzed by two enzymes (LanB and LanC) for class I and by a single enzyme (LanM) in class II. It is clear that
within each class subclasses can be identified of compounds that are essentially natural variants. Within class I there is
high homology in the positions of Ser, Thr, and Cys residues that form the A- and B-rings with the exception of Pep5. In
class II, the ring structures of compounds whose leader sequences end in the so-called double glycine motif187,188 are all
very similar with the exception of salivaricin A and A1 and sublancin. Arrows indicate proteolytic processing sites. Accession
numbers: nisin A (P13068), subtilin (P10946), ericin S (AAL15569), epidermin (P08136), gallidermin (P21838), mutacin
1140 (O68586), mutacin I (AAG48564), Pep5 (CAA90023), lacticin 481 (P36499), SA-FF22 (AAB92600), variacin (A58711),
butyrivibriocin (AAC19355), mutacin II (AAC38144), salivaricin A (P36500), sublancin 168 (O34781), mersacidin (I40461),
cinnamycin (CAD60520), lactocin S (A55457).
The role of the leader sequences is at present unclear. With the exception of the leader peptide for cinnamycin,78 they lack the typical characteristics of the
sec-dependent transport signal sequences. Possible
functions that have been suggested include signaling
for export, protection of the producing strain by
keeping the peptides inactive, and providing scaffolds
for the posttranslational modification machinery.19,189
Precedent for all three functions can be found in the
literature on export proteins,190 hormones,191,192 class
II bacteriocins,193 and microcin biosynthesis,194 and
as discussed below all three functions appear important in lantibiotic biosynthesis.
In a recent study, a series of nonlantibiotic peptides
attached to the C-terminus of the NisA leader sequence were transported by NisT, suggesting secretion is directed by the leader peptide.195 Similarly,
alkaline phosphatase fused to the subtilin leader
peptide was exported in B. subtilis,196 a process that
was enhanced when the SpaT transporter was also
present. Analogous experiments in Escherichia coli
resulted in translocation of the fusion protein into
the periplasmic space.197 These studies support a role
of the leader peptide in recognition by the secretion
machinery and are also consistent with the extracellular membrane location of the NisP protease that
removes the leader peptide and the observation that
the leader of posttranslationally modified presubtilin
is removed by extracellular proteases.77 However, it
is unclear how general this statement is across the
lantibiotics family as several members contain cytoplasmic proteases that appear to remove the leader
peptide prior to transport (section 4.7).
A protective role of the leader sequence is consistent with studies on nisin,189,198,199 subtilin,77,200 lacticin 481,40 and mutacin II,201 showing that the
posttranslationally modified peptides with the leader
sequence still attached exhibit little to no biological
activity. NMR studies comparing posttranslationally
modified prenisin with its leader sequence still attached and mature nisin suggested that a different
interaction between the membrane and the N-terminal region of the modified propeptide in both
compounds accounts for the loss of antimicrobial
activity.202 In light of the subsequent discovery that
nisins activity is mediated by binding to lipid II,41 it
would be interesting to revisit this issue in membranes containing lipid II. Indeed, in the recent NMR
structure of nisin bound to lipid II, its N-terminal
Ile is located at the interface of the two molecules,
suggesting extension of the N-terminus may disrupt
complementarity.54
A number of intriguing in vivo studies have been
conducted with chimeras from the nisin and subtilin
leader and structural regions. While expression of the
nisin gene in a subtilin producing Bacillus strain did
not lead to nisin-related modified peptides, a chimera
consisting of the subtilin leader and nisin structural
gene sequences produced a fully processed product.203
This result suggested that the posttranslational
Figure 15. Sequence alignment of the leader sequences

of type AII lantibiotic and class II bacteriocin prepeptides.
LctA (lacticin 481), MutA (mutacin II), SalA (salivaricin
A), ScnA (streptococcin A-FF22), PedA (pediocin PA-1),
LcnGR (lactoccocin GR), LcnG (lactococcin G), LcnA
(lactococcin A). Completely conserved residues are highlighted in red while strongly conserved residues are in blue.
Alignment created with CLUSTAL W (v1.82). Entrez
accession numbers: LctA, P36499; MutA, JC6526; SalA,
P36500; ScnA, AAB92600; PedA, P29430; LcnA, A39443.
modification machinery of the host specifically recognized the leader sequence. However, when a similar chimera containing a subtilin leader and nisin
structural region was expressed in a nisin-producing
Lactococcus strain the structural region was processed.204 Furthermore, the leader sequence of several type AII lantibiotics have similarity to the leader
peptides of class II bacteriocins that are not posttranslationally modified (Figure 15).25,150 These observations appear to argue against a role of providing
a recognition motif for binding of the modifying
enzymes. At least one modification enzyme has been
demonstrated not to require the leader sequence.
Isolated EpiD, an oxidative decarboxylase involved
in the formation of AviCys (section 4.5), was able to
process peptides without the leader sequence.205
The importance of the conserved residues in the
leader peptides for proper posttranslational modifications and proteolytic processing has been probed by
site-directed mutagenesis for several lantibiotics.
These studies demonstrated a rather weak dependence of the maturation process on point mutations,
as the single mutants Pro(-2)Gly, Pro(-2)Val, Asp(-7)Ala, and Lys(-9)Leu (Figure 16A) as well as the
double mutants Ser(-10)Ala/Ser(-12)Ala and Val(-11)Asp/Val(-13)Glu still produced and secreted
nisin Z. Mutation of Arg(-1) to Gln in the nisA gene
for nisin Z resulted in production and excretion of
both the fully modified product and the posttranslationally modified product that still contained the
leader peptide.189 Apparently proteolytic processing
of the mutant peptide, which now actually has the
same residues in the -1 and -2 positions as subtilin
and Pep5, still occurs albeit with much reduced
efficiency. A similar result was obtained for Ala(4)Asp. Not all positions can tolerate substitutions,
however, because strains containing mutant genes
coding for Ser(-6)Leu, Asp(-15)Ala, Leu(-16)Lys,
and Phe(-18)Leu NisA did not produce any detectable products. Hence, the conserved F-(N/D)-L-(N/D/
E) motif in the class I leader peptides appeared
important for the biosynthetic machinery. However,
an analogous study on the leader sequence of Pep5
showed that the nonconservative mutations Phe(19)Ser and Glu(-16)Lys within this motif as well as
Asp(-6)Lys still resulted in respectable levels of Pep5
production (Figure 16B).206 These findings indicate
Figure 16. Sequence requirements of the leader peptides

of (A) nisin, (B) Pep 5, and (C) mutacin II as determined
by site-directed mutagenesis. In green are those mutants
that still result in full processing of the prepeptides, in blue
those that result in both mature lantibiotics and lantibiotics with their leader peptides still attached, and in red
the mutants that do not lead to lantibiotic production.
that the recognition of the leader peptide by the

processing enzymes is likely a complex process possibly involving recognition of tertiary structural elements rather than conserved residues. Early investigations using synthetic prepeptides and leader peptides did show that in structure-inducing hydrophobic
solvents, the peptides adopted amphiphilic helical
structures.207,208 In addition to the low substrate specificity in the leader peptide, several studies have indicated substrate promiscuity in the propeptide region. These efforts have resulted in engineering of
the structure of lantibiotics and are discussed in
section 7.
Two studies have investigated the leader sequence
requirement for the class II lantibiotics mutacin II
and lacticin 481. Similar to the studies of nisin and
Pep5, the investigation of mutacin II employed
expression of a mutated structural gene in a modified
producing strain.201 Replacement of the double Gly
motif in positions -1 and -2 with two Ala residues
resulted in complete abolishment of production of
bioactive peptides. Analysis of the intracellularly
accumulated peptides showed the presence of premutacin. Whether these peptides had been posttranslationally modified in the propeptide region is
not clear. Substitution of other conserved residues
with similar amino acids (Figure 16C) did not affect
mutacin production, and even replacement with
oppositely charged residues at positions Glu-8 and
Glu-13 only resulted in a decreased production of
mature mutacin II. Two mutants close to the propeptide (Ile(-4)Asp and Leu(-7)Lys), however, did fail
to produce mutacin and the prepeptides could not be
detected.
The recent in vitro reconstitution of lacticin 481

biosynthesis allowed the first examination of the
importance of the leader sequence and the structural
region for correct modification with isolated peptides.40 These studies demonstrate that the length
of the prepeptide is not critical for the modification
enzyme. For example, analogues of the prepeptide
LctA with attachments of extra amino acid residues
at either the N-terminus or the C-terminus did not
affect the activity of the bifunctional dehydratase/
cyclase LctM. In addition, LctA peptides truncated
at the C-terminus still provided the expected products upon LctM catalysis. Very interestingly, although removal of the entire leader sequence resulted
in loss of detectable modification by LctM, LctA
mutants that lacked the first three (LctA5-51), eight
amino acids (LctA10-51), or 11 amino acids (LctA1351) were fully processed by LctM (Figure 26). On the
other hand, a mutant that lacked the first 14 amino
acids (LctA16-51) was not a substrate (Chatterjee
and van der Donk, unpublished results). This observation suggests that the conserved residues in the
segment spanning residues 17-24 (Figure 14) are
essential for enzyme recognition. The highly conserved GG or GA sequence at the end of the leader
sequence was not critical for LctM activity. The single
point mutants His6-LctA(G23V), His6-LctA(A24D),
and His6-LctA(L20Q) were all dehydrated by LctM
(Chatterjee and van der Donk, unpublished results).
LctM also displayed high substrate promiscuity in
the propeptide region as a variety of LctA mutants
were processed (section 7.2).
4.2. The LanB Dehydratases

The selective dehydration of Ser and Thr residues
in the LanA structural region leading to Dha and
Dhb, respectively, is the key first reaction involved
in the biosynthesis of lantibiotics. For the members
of the type AI group, this modification is believed to
be carried out by the putative LanB dehydratases
based on in vivo genetic disruption studies. BLAST
searches do not reveal homology of the LanB enzymes
(120 kDa) with any other known proteins. Among
the LanB family, the overall sequence identity is only
around 30%. The low similarity might be due to the
significantly different prepeptide substrates and the
formation of products of vastly different three-dimensional structures. In cases in which the products are
structurally close, the dehydratase proteins also show
higher homology such as in subtilin and ericin S with
EriB sharing 83% identity with SpaB. The two
products have identical ring locations with only four
amino acid differences, and they contain identical
leader peptides in the respective precursors.88
The isolation of a dehydrated Pep5 peptide after
inactivation of pepC in a Pep5-producing strain
provided the first indirect evidence that the LanB
proteins are involved in the dehydration.75 More
recently, Dodd and co-workers generated a nisin A
variant (H27K, H31K) in which Ser33 is partially
dehydrated in about half of the processed product.
Overexpression of NisB in various L. lactis hosts
containing this mutated structural gene resulted in
elevated cellular levels of NisB as well as increased
Chatterjee et al.
efficiency of dehydration of Ser33, consistent with

dehydratase activity for NisB.209 Koponen et al.
reported the isolation of unmodified nisin precursor
NisA from strains in which NisB activity was impaired, whereas the dehydrated prepeptide was
recovered from strains lacking NisC activity, indicating the importance of NisB for dehydration in the
biosynthesis of nisin.210 The same result was obtained
by Kuipers et al. in studies expressing nisABT in a
nonproducing L. lactis strain that yielded dehydrated
prenisin without thioether rings.195 In accord with the
essential role of LanB proteins in carrying out the
first step in posttranslational modification, deletion
of spaB prevented subtilin production,68,211 stable
epiB mutants abolished epidermin formation,69 disruption of eriB in the ericin gene cluster eliminated
the production of both ericin A and ericin S,88 and
insertional inactivation of salB resulted in abrogation
of salivaricin A production.157
In vitro reconstitution of the enzymatic dehydration, either in cell-free extracts or with (partially)
purified proteins, has remained enigmatic to
date.75,209,210,212,213 Although a hydrophobicity plot
indicates that LanB proteins are rather hydrophilic
and contain no clear transmembrane domains, NisB
and SpaB were found to co-sediment with membrane vesicles, suggesting a membrane-associated
nature.214 Furthermore, for nisin and subtilin biosynthesis, yeast two-hybrid215,216 and immunoprecipitation215-217 experiments suggest that LanB forms a
membrane-associated multimeric complex with LanC
and LanT, the transport protein.
EpiB involved in epidermin biosynthesis in Staphylococcus epidermis was detected in both the cytoplasmic and the membrane fraction, indicating a
loose association with the cytoplasmic membrane. Initial attempts to express EpiB in E. coli resulted in
low protein production possibly due to differences in
codon usage. EpiB was then purified from the closely
related strain Staphylococcus carnosus using an improved staphylococcal expression system; however,
in vitro dehydration activity could not be detected.212
Recently, Xie et al. succeeded in expressing SpaB in
E. coli as a cytoplasmic protein with the aid of the
GroEL/ES molecular chaperones.217 Efforts to detect
dehydratase activity of purified SpaB in the presence
of the subtilin prepeptide (SpaS), SpaC, and a range
of potential cofactors or metal ions were, however,
unsuccessful. This lack of activity could be because
other components of the multimeric lantibiotic synthetase were absent. Consequently, detailed studies
on how the enzyme recognizes its substrate and carries out the multiple dehydrations are still lacking.
Recently, in vitro dehydration activity has been reported for a LanM enzyme as described in section 4.4.
4.3. The LanC Cyclases

Even though dehydro amino acids are less electrophilic than other Michael-type acceptors by virtue of
the built-in deactivating enamine moiety, nucleophilic additions to dehydro amino acids are wellknown and relatively fast. Hence, compared with the
dehydration reaction, the cyclization of a nucleophilic
thiolate onto an electrophilic dehydro amino acid is
relatively easy. The challenge for the cyclase enzymes

lies therefore not so much in chemical activation as
in the control of regio- and stereochemistry. When
biomimetic studies indicated that cyclization in short
peptide analogues of the ring systems encountered
in various lantibiotics occurred spontaneously to generate thioether rings of the correct stereochemistry,123-125 the question of the need for a dedicated
enzyme was raised. Early evidence of the requirement of the LanC proteins for in vivo cyclization was
reported in 1995 using the Pep5 biosynthetic system.75 In this work, a Pep5-producing strain of S.
epidermis was depleted of its ability to generate the
lantibiotic, but after supplementation with a plasmid
encoding the pepTIAPBC gene cluster the production
of fully modified Pep5 was restored. Subsequent
disruption of the pepC gene led to the formation of
the fully dehydrated prepeptide as well as fragments
thereof, but none contained the correctly cyclized
thioether bridges characteristic of the Pep5 structure.
It was proposed that PepC needs to bind the fully
dehydrated substrate in a specific conformation to
facilitate the correct addition of cysteines to their
dehydrated partners. These results were the first
experimental evidence implicating a LanC protein as
the cyclase involved in lantibiotic biosynthesis and
offered counterevidence to the idea of spontaneous
nonenzymatic cyclization of thioethers.
EpiC, involved in epidermin biosynthesis, has been
overexpressed and purified from S. epidermidis by
Kupke and Gotz.213 The enzyme showed no activity
with unmodified EpiA, as expected since it would not
be the substrate for the enzyme. Because dehydrated
EpiA was not available, in vitro cyclization activity
could not be tested. In these studies, it was noted that
plasmids encoding epiC mutants where a conserved
glycine residue was changed to a glutamate did not
restore epidermin production in strains lacking epiC.
The cause of inactivation of these mutants is still
unknown.
In a series of experiments similar to the work
probing the function of PepC, Koponen et al. engineered mutant strains of L. lactis lacking either the
nisB or nisC genes.210 A His-tagged nisin precursor
peptide was coexpressed simplifying the purification
of the products formed. Use of this tagged prepeptide
in a strain lacking the NisB gene showed that no
dehydrations or cyclizations took place. On the other
hand, the NisC deficient strain yielded a dehydrated
propeptide, but no cyclization products were observed. This was the first direct evidence in the nisin
system that NisC was responsible for the cyclization
of the thioether rings, either directly or by inducing
cyclization activity in NisB. NisC has been shown to
be localized at or in the cellular membrane by coimmunoprecipitation studies.215 Furthermore, yeast
two-hybrid studies indicated a specific interaction of
NisC with NisT, NisB, and itself.215 These findings
led to a putative model of nisin biosynthesis involving
a complex consisting of two NisT proteins, two NisC
polypeptides, and one NisB protein, all associated
with the cellular membrane (Figure 17). Similar
studies were conducted with the subtilin system,
yielding evidence for a multimeric biosynthetic com-
Figure 17. Proposed multi-enzyme complex involved in

nisin production on the basis of co-immunoprecipitation
and yeast two-hybrid studies.215
plex for the modification of the SpaS precursor

peptide (Figure 5).216 Small differences involved the
observation of an interaction between SpaB and SpaT
that was not observed for nisin as well as an
interaction of SpaB with itself suggesting an oligomeric structure.
Recently, the LanC enzymes involved in nisin and
subtilin production (NisC and SpaC) were cloned,
overexpressed in E. coli, and purified to homogeneity.
As isolated, the proteins are monomers and metal
analysis showed that each contains a stoichiometric
amount of zinc.218 The difference in oligomerization
state predicted by the two-hybrid studies and these
in vitro experiments with heterologously expressed
protein may be due to incorrectly folded protein in
either E. coli or yeast or both. It has been postulated
that the zinc may serve to activate the thiol substrates of the LanC enzymes,218 a role similar to that
in a number of other enzymes that catalyze thiol
alkylation.219 The zinc in these metalloenzymes is
believed to activate the thiol of their substrates by
lowering the pKa and enhancing the reactivity at
neutral pH. For example, the pH dependence of the
binding of a peptide substrate to farnesyltransferase
indicates that the pKa of the cysteine is lowered from
8.3 for the free peptide (GCVLS) to approximately
6.4 upon binding to the enzyme.220 Indeed, to sustain
a reasonable rate of lanthionine formation at neutral
pH, activation of the cysteine thiols of the substrate
by deprotonation is required. For instance, the rate
constant for the addition of free thiols to R,-unsaturated centers is 1010-fold decreased compared to the
corresponding thiolates.221 One noticeable feature
that is common among this family of zinc enzymes
that catalyze thiol alkylations is the presence of two
or more cysteine residues in the zinc coordination
sphere and an overall net negative charge. In accord
with this postulated role of the metal, extended X-ray
absorbance fine structure (EXAFS) analysis of SpaC
showed that the ligand environment surrounding the
zinc is comprised of two cysteines and possibly two
histidine residues, or one histidine and one water
molecule.218
The LanC proteins share low sequence similarity
(20-30%, Figure 18). Only very few amino acids
are strictly conserved, but these include two cysteine
residues (Cys284 and Cys330, NisC numbering) and
two histidines (His212, His331). These residues are
Chatterjee et al.
Figure 18. Partial sequence alignment of several LanC proteins and the C-terminal domains of a number of LanM proteins.
The putative metal ligands218 are in red font. Acccession numbers: SpaC-AAA22777, NisC-Q03202, EpiC-CAA44254, PepCCAA90026, LctM-AAC72258, MrsM-CAB60261, CinM-CAD60521.
also conserved in the C-terminal part of the LanM

proteins (section 4.4).66,155 Mutants of SpaC with
Cys303 and Cys349 changed to alanine residues were
analyzed by EXAFS or inductively coupled plasma
mass spectrometry (ICP-MS), confirming that they
indeed bind significantly less zinc than the wild type
protein and have only a single sulfur ligand.218 As
described above for the EpiC protein, activity determinations for the purified NisC and SpaC proteins
have been hampered by the inability to produce the
dehydrated prepeptides that are their putative substrates. Hence the effect of mutation of the Zn ligands
on activity has not been determined, and at present
the role for zinc is purely speculative and it cannot
be ruled out that the zinc in SpaC and NisC provides
structural integrity or acts as a Lewis acid for the
electrophilic activation of a carbonyl group.218
Interestingly, while the LanC enzymes are responsible for the formation of cyclic lanthionines and
methyllanthionines, the directionality of ring formation is not always uniform. The LanC enzymes
involved in the biosynthesis of type AI lantibiotics
such as nisin, subtilin, gallidermin, Pep5, and epidermin incorporate lanthionine rings solely in the
N-to-C terminal direction, i.e., the cysteine involved
in cyclization is always located on the C-terminal side
of its dehydroalanine or dehydrobutyrine reaction
partner.172 Conversely, in the case of some type B
lantibiotics such as cinnamycin (Figure 19),172 mersacidin,169 and the duramycins,108,110 some of the (methyl)lanthionine rings in the final lantibiotic structures result from cyclization of cysteines onto dehydrobutyrines that are located downstream (i.e., toward
the C-terminus).
The question of the origin of the stereoselectivity
of the cyclization step has been investigated in a
number of model studies after the initial work by
Toogood showed that biomimetic reactions featuring
the B-ring of epidermin provided the same stereochemical outcome as found in the natural product.123
Similar findings have been reported in two subse-
Figure 19. Schematic representation of the maturation

process of cinnamycin illustrating the different directions
of cyclization. The order of the cyclizations to form (methyl)lanthionines and lysinoalanine is not known nor are the
enzymes responsible for the formation of lysinoalanine and
-hydroxy aspartate.
quent studies on the biomimetic formation of lanthionine analogues of the subtilin and nisin B- and
E-rings, which like the epidermin B-ring contain four
amino acids.124,125 More recently, the actual methyl-
Scheme 1
Scheme 2
Figure 20. Attempted biomimetic synthesis of the A- and

B-rings of nisin. On the basis of the result in Scheme 2,
upon reduction of the disulfide bond in peptide 2, the Cys
at position 7 does form the A-ring, but the Cys at position
11 does not attack Dhb8 to generate the B-ring (dashed
arrow) but rather adds to Dha5.
lanthionine-containing B-ring of subtilin was prepared by nonenzymatic cyclization of a Cys onto a

Dhb residue,126 representing the first test of the
stereochemistry of biomimetic formation of methyllanthionines (Scheme 1).126 Through independent
synthesis of the natural stereoisomer, it was shown
that this Michael-type addition also occurred with
very high selectivity in favor of the naturally occurring diastereomer. Interestingly, when the cyclization
was carried out in the opposite direction (i.e., Cys
located N-terminal to the Dha or Dhb), the reaction
was not stereoselective.126,222 This finding suggests
that whereas the peptide substrates for type A
lantibiotics involving cyclization in the C-to-N terminus direction have a propensity to provide the
same stereochemistry observed in the natural products, the substrates of type B lantibiotics that undergo cyclization in the opposite direction do not have
such an intrinsic stereoselectivity. The major difference between these two different modes of cyclization
is whether an endocyclic (type A) or exocyclic enolate
(type B) is generated.126
Two studies have examined the regioselectivity of
biomimetic ring formation. Bradley and co-workers
investigated the biomimetic formation of the A-ring
of nisin (Scheme 2).124 After deprotection of the
cysteine of peptide 1, a Michael-type addition led to
two products in a 3:1 ratio. NMR analysis revealed
that these compounds both contained the connectivity
of the natural subtilin A ring, but with different
stereochemistry at the newly formed stereocenter.
Hence, whereas the correct regiochemistry was observed, the protonation of the enolate intermediate
is less selective in this example than that observed
for rings containing four amino acids (vide supra).
Zhu et al. recently addressed the chemo- and regioselectivity of the intramolecular Michael addition
of the precursor peptide to the A- and B-rings of nisin
(Figure 20).222 1H NMR analysis of the cyclization
products revealed that the product did not consist of
the A- and B-rings of nisin. Instead, the resonances
of the vinyl protons of the two dehydrobutyrines were
still present in the product, whereas the signals for
the vinyl protons of the two Dha residues were
absent. Hence, the much faster cyclization rate for
lanthionines compared to methyllanthionines126 prevents the biomimetic cyclization in which one Lan
(A-ring) and one MeLan (B-ring) would have been
formed (see nisin structure, Figure 2). The products
that were obtained were assigned the structures
depicted in Figure 20. The important conclusion from
these studies is that lantibiotic biosynthesis clearly
requires enzymatic control over the chemoselectivity
and/or processivity of the cyclization reactions, in
accordance with the observed genetic studies involving various LanC proteins. The much higher reactivity of the Dha residues compared to Dhb residues also
explains why nonenzymatic cyclization of dehydrated
prenisin and pre-Pep5 at elevated pH provided
products in which free cysteines were no longer
present but that did not have any biological activity.75,195
Intriguingly, a number of proteins with significant
sequence similarity to the LanC family of enzymes
has recently been discovered in mammalian erythrocytes. One of these, p40/GPR69A originally assigned
as a member of the G-protein-coupled receptor superfamily, has been postulated to be a peptidemodifying protein based on its sequence homology to
Chatterjee et al.
the LanC enzymes.223 Upon further characterization,

p40/GPR69A was renamed to LANCL1 for LanC-like
protein 1. Rat LANCL1 is highly expressed in the
testis and brain, and is possibly involved in the
immune surveillance of these particular organs,
although no exact function of this protein has been
determined.224 If the hypothesis that the zinc serves
an activating role to promote thiol alkylation proves
correct, the LANCL1 proteins likely carry out alkylation of a currently unknown thiol substrate because
the metal ligands are conserved. Since no proteins
with sequence similarity with the lantibiotic dehydratases have been reported in mammals, the electrophile would probably not be Dha or Dhb.
4.4. The LanM Bifunctional Enzymes

A novel gene designated lanM encoding a 9001000 amino acid protein is present in the gene
clusters of the class II lantibiotics (nomenclature as
per Figure 14).148 The C-termini of LanM proteins
have about 20-27% sequence identity with LanC
proteins including the conserved motifs with the
possible metal ligands (Figure 18). They show no
sequence homology to LanB proteins, and hence their
origin is unlikely to be from the fusion of lanB and
lanC genes.66 The unique sequences of LanM proteins
and the fact that no other candidates for catalyzing
the posttranslational modifications are present in the
gene clusters of class II lantibiotics led to the proposal
that they might be responsible for catalysis of both
dehydration and cyclization reactions.66,148 In support
of this hypothesis, disruption of lctM in the lacticin
481 biosynthetic gene cluster prevented the production of the mature lantibiotic.148,225 Similarly, mutacin
II production was not observed in the absence of
MutM, a frameshift in the gene for CylM eliminated
cytolysin formation,153,226 and LasM inactivation abolished the production of lactocin S.155 A molecular
interaction between the lacticin 481 prepeptide LctA
and LctM was observed using the yeast two-hybrid
system225 similar to the observations for the LanB
and LanC proteins with their LanA substrates discussed in sections 4.2 and 4.3. Interestingly, in the
gene cluster of the two-component lantibiotic lacticin
3147 two independent genes encoding LtnM1 and
LtnM2 are present (Figure 3). Disruption studies of
the ltnM genes showed that each prepeptide (LtnA1
and LtnA2) requires a dedicated LtnM for modification.71,227
Direct evidence for the role of LanM proteins was
recently provided by Xie et al. during the first
reconstitution of an active lantibiotic synthetase
(LctM) involved in the biosynthesis of lacticin 481.40
In this report, LctM was shown to convert the
prepeptide LctA into a 4-fold dehydrated species. The
product was characterized by MALDI-MS and highresolution FT-MS/MS demonstrating that the correct
cyclization reactions had also taken place (Figure 21)
thereby verifying that LctM is a bifunctional enzyme.
Removal of the leader sequence from the product with
the commercial protease Lys-C, which results in
lacticin 481 lacking its N-terminal Lys residue,
generated a bioactive peptide, whereas prior to proteolysis the product was devoid of any antimicrobial
Figure 21. The posttranslational maturation process of

lacticin 481. LctM catalyzes the dehydration of the underlined Ser and Thr residues in the propeptide region of LctA.
The sequence of the leader peptide is MKEQNSFNLLQEVTESELDLILGA, and in the substrate used for in vitro
reconstitution of the maturation process40 an N-terminal
His6-tag linker was attached with the sequence GSSHHHHHHSSGLVPRGSH. LctM also catalyzes the conjugate
addition of three Cys residues in a regiospecific manner to
three of the Dha and Dhb residues to generate three cyclic
thioethers, one methyllanthionine and two lanthionines.
The leader peptide is proteolytically removed by the
N-terminal protease domain of the LctT ABC-type transporter that excretes the final product (section 4.7).
activity at the concentrations tested. It should be

emphasized that although it is generally assumed
that the dehydration of all targeted Ser and Thr
residues in the prepeptide is completed before cyclization commences (e.g., Figures 4 and 21), at
present this has not been unambiguously established.
In an alternative model, the two active sites on LanM
(or the LanB and LanC proteins for type AI lantibiotics) could pass the substrate between them such
that dehydration of one particular Ser/Thr is immediately followed by ring formation before dehydration of the next Ser/Thr.
ATP and Mg2+ were required for LctM to carry out
the posttranslational modifications, although at
present the exact role of the cofactor is unknown. It
may activate the serines and threonines for elimination by phosphorylation of their hydroxyl groups, or
it may provide the energy for peptide translocation
during the series of dehydration and cyclization
reactions. ATP is converted into ADP in the process
and use of non-hydrolyzable ATP analogues did not
support catalysis. Although no dehydration is seen
in the absence of ATP, ATP hydrolysis is observed
in the absence of the peptide substrate (Xie and van
der Donk, unpublished results). Hence, either LctM
has ATPase activity or a contaminating protein is
present that catalyzes this background reaction.
Uncoupled ATP hydrolysis has also been reported for
the biosynthesis of microcin B17, a nonlantibiotic
posttranslationally modified antibiotic produced by
E. coli.228,229 The LctM-catalyzed process shown in
Figure 21 is a remarkable example of molecular
recognition as only four of the 14 serine and threonine residues are dehydrated without the presence
of a clear consensus sequence. Given this exquisite
control it is surprising, although not unexpected
given the results of in vivo mutagenesis studies on
other lantibiotics (section 7.1), that the purified LctM
demonstrated permissive substrate specificity processing a series of LctA mutants as well as Cterminally truncated LctA. The study of LctM not
only solved the long-standing question regarding the
exact function of the novel modification enzymes but
also potentially opened the door for future in vitro
lantibiotic engineering (section 7.2).40
4.5. The LanD Enzymes

The lantibiotics epidermin, gallidermin, cypemycin,
and mutacin 1140 possess the unusual amino acid
S-[(Z)-2-aminovinyl]-D-cysteine (AviCys) (Figures 1
and 6) at their C-terminus. The enzymes responsible
for the formation of AviCys are encoded by the lanD
genes. EpiD, involved in the formation of the AviCys
group in epidermin, was isolated by Kupke and coworkers in 1992230 from the native producer S.
epidermis Tu3298. The protein contains a stoichiometric noncovalently bound flavin mononucleotide
(FMN). In vivo experiments with His6-tagged EpiA231
as well as in vitro experiments with heterologously
expressed and purified EpiD and EpiA232 revealed the
loss of 46 Da from the substrate corresponding to the
loss of CO2 and two H atoms. This study represented
the first in vitro activity of an enzyme involved in
one of the posttranslational modifications of lantibiotics. The entire epidermin prepeptide EpiA, as well
as a fragment containing only the structural region
underwent decarboxylation, suggesting that the leader
sequence of EpiA was not necessary for EpiD action.
The substrate specificity of EpiD was probed by
constructing a library of heptapeptides containing
single amino acid mutations in the C-terminal sequence of EpiA (SFNSYCC).205,233 EpiD showed low
substrate specificity with only the C-terminal cysteine an absolute requirement. This Cys must be
present as a free thiol and possess a free carboxylate
for activity, suggesting that decarboxylation occurs
prior to ring formation. A general consensus sequence
for the final three amino acids at the C-terminal was
reported to be [V/I/L/F/W/Y/(M)]-[A/S/V/T/C/(I/L)]-C.
The mechanism of decarboxylation was investigated
by means of heteronuclear correlation NMR using a
model peptide KKSFNSYTC that was 13C-labeled at
the -carbon of the terminal cysteine.234 This experiment showed the formation of an unusual enethiol
in the product upon the action of EpiD. Addition of
this enethiol (pKa 6.0)235 to the Dha at position 19 of
epidermin, presumably catalyzed by EpiC, would
then yield AviCys (Figure 22).
Analysis of the effects of point mutations in an
MBP-EpiD fusion protein on the decarboxylation of
the substrate peptide SFNSYTC was used to identify
the active site residues of EpiD.236 On the basis of
these experiments and its complete conservation in
all other homologues, His67 was proposed as an
active site base in EpiD. An X-ray structure of EpiD
Figure 22. Proposed mechanism of EpiD-catalyzed formation of the enethiol intermediate and the putative EpiC
catalyzed formation of S-[(Z)-2-aminovinyl]-D-cysteine.237
and H67N-EpiD complexed to a pentapeptide substrate DSYTC showed that it exists as a dodecamer
consisting of tetrahedrally placed trimers.237 Each
monomer unit consists of a central parallel -sheet
domain of six strands flanked by nine helices in a
Rossmann-type fold (Figure 23a). The pentapeptide
substrate forms a parallel -sheet with a -strand
stretching from Phe149 to Ile151 in EpiD and forms
additional backbone hydrogen bonds with Asn117
and Asn14. On the basis of the proximity of the sulfur
of the terminal Cys of the substrate to N5 of FMN,
the authors proposed a mechanism featuring oxidation of the Cys to the thioaldehyde followed by
spontaneous decarboxylation to form a thioenolate,
as opposed to direct hydrogen removal from the CR
and C positions of the Cys (Figure 23b).
The enzyme MrsD involved in biosynthesis of the
S-[(Z)-2-aminovinyl]-(3S)-3-methyl-D-cysteine (AviMeCys) residue in mersacidin has also been purified and
characterized.238 Unlike EpiD, MrsD is a flavin
adenine dinucleotide (FAD)-containing enzyme with
a more stringent substrate requirement. Whereas
MrsA proved to be a substrate, neither the EpiAR30Q mutant nor a short peptide corresponding to
the C-terminal eight residues of MrsA were processed
by MrsD. However, similar to EpiD, mutation of the
conserved His75 to Asn was found to abolish MrsD
Figure 23. (A) X-ray structure of the EpiD His67Asn

mutant complexed with a pentapeptide substrate, DSYTC.237
A single monomer with the Rossmann-like fold is shown
as well as the distance between N5 of the cofactor FMN
and Cys-S. Figure generated using the program RASMOL.469 (B) On the basis of the proximity of S and N5 a
direct oxidation of the cysteine thiol has been proposed.237
activity, suggesting the role of an active site base in

decarboxylation or dehydrogenation. Like EpiD, MrsD
has a dodecameric structure with a Rossmann-type
fold and cofactor binding at the subunit interface.239
On the basis of their similarity with bacterial240 and
plant241 decarboxylases, EpiD and MrsD are members of the homooligomeric flavin-containing Cys
decarboxylase (HFCD) superfamily.
4.6. Other Posttranslational Modifications

Many lantibiotics undergo further posttranslational modifications in addition to the characteristic
Lan and MeLan ring formation and the aforementioned formation of AviCys and AviMeCys.24,163,242,243
Presumably, after removal of the leader peptides
nonenzymatic hydrolysis of Dhb at position 1 of
Pep5,7 and Dha at position 1 of lactocin S102,154 and
epilancin K796,145 leads to the formation of 2-oxobu-
Chatterjee et al.
tyryl and 2-oxopropionyl groups, respectively. Reduction of the 2-oxopropionyl functionality to the 2-hydroxypropionyl group has been reported in the case
of epilancin K796 and has been proposed for epicidin
280.72 A putative oxidoreductase EciO was hypothesized to be involved for the latter compound. The
stereochemistry of the reduction step is currently
unknown. Structural elucidation of cinnamycin and
the duramycins109,174,176 has shown the presence of
an erythro-3-hydroxy-L-aspartic acid resulting from
the hydroxylation of a genetically encoded L-Asp at
position 15 (Figures 6 and 12).172,56 This unusual
modification is also found in mammalian proteins,
such as the vitamin K-dependent glycoprotein, Protein C,244 and the epidermal growth factor (EGF)-like
domain in human plasma factor IX.245 The mammalian enzyme responsible for -hydroxylation of Asp
has been purified from native sources246and expressed in E. coli.247-249 The Asp -hydroxylase also
hydroxylates Asn residues to produce erythro-3hydroxy-L-asparagine. The enzyme is O2/Fe(II)/Rketoglutarate-dependent, and a stoichiometric amount
of CO2 is released per Asp hydroxylated.246 The role
of -hydroxylation in cinnamycin is currently still
uncertain, although this residue is essential for
recognition of its target phosphatidyl ethanolamine
(section 8.4).
The cinnamycin group also exhibits a head-to-tail
lysinoalanine bridge probably formed by addition of
the -amine of Lys19 to Dha6. Lysinoalanine is
commonly found in processed and unprocessed food
products such as eggs, meats, tortillas, and Chinese
noodles, as well as in body organs, where it is possibly
formed during the aging process.250 The formation of
lysinoalanine in these cases is due to chemical
dehydration of Ser and conjugate addition of Lys to
the resulting Dha and produces both diastereomers.250 It is unclear at present whether the lysinoalanine bridge in cinnamycin is formed by CinM
or by one of the genes of unknown function in its gene
cluster (section 2).
A number of additional modifications for which the
responsible enzymes are currently unknown have
been reported. The presence of D-Ala in place of
genetically encoded L-Ser is observed in lactocin S102
and both components of the two-component lantibiotic lacticin 3147 (Figure 6).112 The mechanism of
D-Ala formation may involve stereospecific hydrogenation of the dehydrated serine (Dha) by a hitherto
unknown enzyme. Subtilin has been shown to undergo NR-succinylation at late stages of cell growth
that leads to a reduction in its biological activity.128
Cypemycin isolated from Streptomyces contains a
number of unique modifications including bis-methylation at Ala1 (Me2N-Ala), the presence of alloisoleucine at positions 13 and 18, and a AviCys
involving residues 19 and 22 (Figure 9).103 Finally,
the lantibiotic sublancin 168 contains two unprecedented disulfide linkages and contains an additional
modification of currently unknown structure (Figure
9).105 The presence of a single disulfide bridge has
also recently been reported for the lantibiotic bovicin
HJ50, which awaits complete structural characterization.119
4.7. Proteases and Transporters

All lantibiotics known to date are ribosomally
produced as two-segment precursor peptides consisting of a leader region and a propeptide that undergoes modification by the LanB/C or LanM enzymes.
While still attached to the modified propeptide, the
leader region has been shown to negate biological
activity.40,77,198,200,201 The lanP or lanT genes in the
lantibiotic operon generally encode the protease
responsible for removal of the leader region.66,79 LanP
proteases vary in size depending upon the presence
or absence of an N-terminal sec-signal sequence and
a C-terminal cell wall anchor sequence. They all
share homology with the serine protease subtilisin,
especially in the sequence near the residues involved
in the catalytic triad (Asp, His, and Ser) and the Asn
involved in oxyanion hole formation. In the case of
the nisin protease (NisP), these residues were predicted to be Asp259, His306, Ser512, and Asn407,
respectively.198 Although the gene translation product
of nisP corresponds to 682 amino acids and has a
predicted mass of about 75 kDa, purification of NisP
yielded a protein of only 54 kDa.198 This loss of mass
was attributed to peptidase cleavage of an N-terminal
prosequence of 220 residues that contains the secdependent secretion signal and directs extracellular
transport of NisP. The C-terminal 30 residues of NisP
contain the sequence LPXTG, which is a consensus
sequence involved in anchoring surface proteins in
Gram-positive bacteria251 (Figure 17). The observed
in vitro proteolysis of a modified nisin precursor by
intact L. lactis NZ9800 producer cells or E. coli cells
expressing NisP, but not by cellular supernatants or
membrane-free extracts, supports this model.198 On
the other hand, disruption of the lanT gene in L.
lactis N8 led to a mutant strain that accumulated
fully processed nisin in the cytoplasm,74 suggesting
at first glance either that NisP is located at the
cytoplasmic side of the membrane or that other
cytoplasmic proteases in L. lactis N8 can also process
the posttranslationally modified NisA peptide. Since
the full-length NisP protein contains a sec-dependent
secretion signal, the first possibility is unlikely and
to date no intracellular proteases capable of removing
the leader have been identified. The cytoplasmic
fraction of this mutant strain was obtained by sonication and subsequent centrifugation, and hence the
posttranslationally modified prepeptide was probably
exposed to NisP during sample preparation leading
to proteolytic processing.
The substrate requirement of NisP has been tested
with a chimeric substrate consisting of the subtilin
leader sequence and nisin Z structural region.204
Although there is significant sequence similarity
between the nisin and subtilin leader peptides (Figure 14), the subtilin leader ends in a Gln instead of
the Arg found in nisin. The chimeric substrate was
fully processed in L. lactis to a subtilin leader-nisin
Z product, which did not undergo proteolytic removal
of the leader peptide. The importance of an Arg at
the cleavage site was also demonstrated in point
mutants of the NisA substrate in which Arg(-1) (P1)
and Asp(-4) (P4) were replaced, resulting in incomplete cleavage of the leader peptide.189 In a very
recent study, Kuipers and co-workers have shown

that in vivo NisP catalyzed removal of the leader
peptide occurs only upon formation of the thioether
rings in prenisin.195 Neither the unprocessed prepeptide nor the uncyclized dehydrated peptide were
substrates for NisP.
EpiP, the protease responsible for maturation of
epidermin, bears an overall 44% sequence identity
with NisP. Its gene sequence predicts a 99-residue
pre-prosequence that is absent from the enzyme
purified from the culture supernatant of S. carnosus.252 The C-terminal anchor sequence found in NisP
is absent in EpiP, suggesting the protein is not
covalently bound to the cell surface. Incubation of
culture supernatants with the epidermin precursor
EpiA resulted in cleavage between the Arg(-1) (P1)
and Ile1 (P1) residues, suggesting the protease is
excreted as a soluble extracellular protein. Tests of
substrate specificity revealed that EpiP was intolerant of mutation at Arg(-1) as it failed to cleave the
leader from an EpiA-R(-1)Q mutant. Subsequent
homology modeling based on the known crystal
structures of subtilisin and other serine proteases
predicted that the binding of NisP and EpiP to their
respective substrates is dominated by electrostatic
interactions at the P1 position.253 The proteases PepP
and ElkP involved in the maturation of Pep 575 and
epilancin K7,145 respectively, bear 44% sequence
identity to each other and only about 20% sequence
identity to NisP. The lack of an N-terminal sec-signal
sequence as well as a cell wall anchor sequence
suggests they are intracellularly localized. LasP, the
homologous protein from the lactocin S biosynthetic
cluster,155 also lacks the pre-pro sequence and is
probably localized within the cytoplasm.
The subtilin biosynthetic gene cluster does not
contain a dedicated protease. Experiments conducted
with subtilin-nisin Z prepeptide chimeras in B.
subtilis resulted in removal of the leader region,203
indicating that the substrate specificity of the protease(s) involved in subtilin production is more
relaxed than that of NisP, which did not remove the
leader peptide from this chimera as mentioned
before. The observation that subtilin could also be
obtained by cleavage of the leader peptide upon
incubation with culture supernatants from a nonproducing strain200 suggested the action of nonspecific
proteases in subtilin maturation. Entian and coworkers have recently shown that at least three
extracellular serine proteases, subtilisin (AprE), WprA,
and Vpr may activate subtilin to its mature form.77
The two-component lantibiotic cytolysin is comprised of the peptides CylLL and CylLs. They undergo
stepwise proteolysis starting in the cytoplasm of the
cell.115 The leader sequence of each peptide was
trimmed by an unknown protease to leave an identical six-residue tail attached to the structural region
of each component. Those peptides were then secreted to the extracellular medium where the serine
protease CylA (the cytolysin nomenclature differs
from the general classification of lantibiotic genes)
removed the last vestiges of the leader sequence and
generated the active lantibiotic. Mixtures of the
partly processed CylL peptides were found to have
Chatterjee et al.
Figure 24. Sequence alignment of the N-terminal domains of ABC-transporters involved in lantibiotic or class II bacteriocin
proteolytic processing. Lantibiotic transporters: LctT (lacticin 481), MutT (mutacin II), SalT (salivaricin A), ScnT
(streptococcin SA-FF22). Class II bacteriocin transporters: LcnC (lactococcin A), PedD (pediocin PA-1), LagD (lactococcin
G). Completely conserved residues are highlighted in blue with the proposed catalytically active residues in red.79 Alignment
created with CLUSTAL W (v1.82) and numbering based on LctT sequence. Accession numbers: LctT, AAC72259; MutT,
AAD01806; SalT, AAG32538; ScnT, AAB92603; LcnC, AAK04177; PedD, AAA25561; LagD, P59852.
biological activity only in the presence of CylA

establishing its role in the final proteolytic step. CylA
is a serine protease similar to EpiP in that it is
exported outside the cell but does not anchor to the
cell surface. It undergoes proteolytic activation by the
loss of a prosequence comprising of 95 amino acids.
Homology modeling of the interaction of CylA with
the posttranslationally modified CylL substrates
indicated a dominant electrostatic interaction between the Glu-P1 side-chain and His180 of the S1binding pocket of CylA.115
Lantibiotics are generally transported to the extracellular medium by the LanT family of ATPdependent transporters. There is some uncertainty
as to the absolute necessity of a LanT protein for all
lantibiotic systems, as evidenced by the nonrequirement of PepT in extracellular transport of Pep5 in
S. epidermis75 and the absence of a lanT gene for
epicidin 280.72 Conversely, in the case of nisin, it has
been shown that NisT is an absolute requirement for
extracellular transport. Deletion of the nisT gene in
L. lactis N8 led to a build-up of processed nisin in
the cytoplasm and no extracellular nisin was detected.74 NisT has relaxed substrate specificity as
demonstrated by its ability to transport not only
fusions of the leader region with processed or unprocessed forms of pronisin but also with peptide
fragments from enkephalin and angiotensin.195 Extracellular export of a protein consisting of the
subtilin leader sequence fused to alkaline phosphatase in the subtilin nonproducer strain B. subtilis
168 demonstrated recognition and transport in the
absence of SpaT, albeit with lower efficiency.196
The type AI group of ATP binding cassette (ABC)
transporters represented by NisT and SpaT are about
600 amino acids in length and bear significant
homology to hemolysin B-like ATP-dependent transport proteins present in a range of organisms.73,254
These proteins contain a hydrophobic N-terminal
domain, a six-helix membrane spanning domain, and
a C-terminal ATP-binding domain. In comparison,

most of the type AII LanT enzymes such as LctT,148
MutT,255 and SunT105 are about 700 residues long and
contain an extra N-terminal peptidase domain besides the membrane spanning and C-terminal ATPbinding domains. The absence of a lanP equivalent
in the gene clusters that encode these proteins
suggests a role for the N-terminal peptidase domain
in processing of the leader peptide concomitant with
export. Indirect evidence for this model is found in
the gene cluster of the type AII lantibiotic lactocin
S, which encodes an abbreviated 535 amino acid LasT
protein but also encodes a subtilisin-like LasP protein.155 Although the protease function of a LanT
protein has not been demonstrated, experiments with
homologous transport proteins (Figure 24) responsible for class II bacteriocin production have shown
peptidase activity in vitro and in vivo.79,256,257 Havarstein and co-workers reported in vitro cleavage of
both precursors of the two-component bacteriocin
lactococcin G by incubation with the N-terminal
domain of 150 amino acids of the transporter LagD.79
Investigation of the membrane topology of the lactococcin A transport protein (LcnC from L. lactis)
revealed that both the N-terminal protease domain
and the C-terminal ATP binding region were located
at the cytoplasmic side of the cell membrane.256 In
accord with this finding, overexpression of the Nterminal 164 amino acids of LcnC in the wild-type
lactococcin A producer along with the LcnA prepeptide led to the formation of mature LcnA in the
cytoplasm. The leader sequences of the lantibiotics
lacticin 481, mutacin II, streptococcin A-FF22, salivaricin A, and variacin exhibit significant similarity
to these class II bacteriocin leader sequences including the double-glycine type cleavage site (Figure 15)
where processing takes place. However, despite a
47% sequence identity with LctT and the presence
of conserved Cys and His residues that suggests they
are both cysteine proteases, the lactococcin A trans-
porter LcnC was not able to secrete lacticin 481 in

an L. lactis strain bearing the LctA and LctM
genes.258 This suggests that the LanT proteins recognize regions of the substrate other than the double
Gly motif alone.
5. Regulation of Lantibiotic Production

Lantibiotic-producing bacteria must maintain an
inherent balance between bacteriocin production and
immunity to their product. The production of these
compounds is regulated such that it typically takes
place late in the exponential growth phase. For
instance, Engelke et al. monitored the growth-dependent expression of prenisin and NisB by using
antibodies directed against the prepeptide and dehydratase.259 Prenisin expression was highest when
cells were in their mid-logarithmic growth phase; at
earlier times during growth nisin is produced and
excreted but remains adsorbed to the membrane and
is released when the pH drops below 5.5.260 NisB
expression was not observed in the first 4 h of growth
and increased in the late logarithmic state.259 For
many lantibiotics (e.g., nisin,259 subtilin,261 mersacidin,262 streptococcin A-FF22,263 and salivaricin A)157
production is controlled by a typical two-component
regulation system comprised of a receptor-histidine
kinase (LanK) and a transcriptional response regulator (LanR, e.g., Figure 5).264 In bacteria, these
systems are often involved in quorum sensing, the
intraspecies communication process that allows cells
to sense other organisms in their surroundings in a
cell-density-dependent manner.265-267 The receptorhistidine kinases are present on the cellular surface
and are involved in detecting extracellular changes,
leading to a signal cascade initiated by autophosphorylation of a histidine residue. A typical case is
exemplified by the nisin regulatory system259,268-270
in which the phosphoryl group from NisK is transferred to an aspartate on NisR that initiates its
binding to the nisA and nisF operators (see Figure 3
for transcriptional units). This in turn activates
transcription of the nisABTCIP operon involved in
nisin biosynthesis as well as nisFEG involved in selfimmunity (section 6).260,268 The nisRK genes themselves are under control of a separate nisR promoter.268 Similar to nisin, most lantibiotic producers
contain operons that include both the structural gene
for the prepeptide as well as the biosynthetic enzymes (e.g., see Figure 3). Typically, the DNA region
downstream of the structural gene allows limited
readthrough to the biosynthetic genes (leaky termination of transcription) to ensure the desired stoichiometry between the substrate for posttranslational modification and the modification machinery (e.g.,
refs 155 and 271).
Nisin,269 subtilin,272 and salivaricin A157 have been
shown to serve as the sensing molecules that trigger
the transcription of their prepeptides in an extracellular autoregulatory mechanism. In the case of nisin,
this system is extraordinarily efficient in that as few
as five nisin molecules are sufficient to activate
transcription.269 Strains of L. lactis with a four-base
pair deletion in the middle of nisA (nisA) such that
they could not produce nisin were used by Kuipers
et al. to monitor the effect of nisin A as an initiator

of nisA transcription.269 The authors showed that in
the absence of any external inducers, nisA was not
transcribed. Upon addition of nisin A, transcription
of nisA was observed in a concentration-dependent
manner. On the other hand, unmodified synthetic
nisin A precursor peptide did not induce transcription, providing evidence that the posttranslational
modifications within the nisin peptide are required
for induction of nisA expression. Subtilin, lacticin
481, and Pep5 did not stimulate transcription. The
structural requirements for induction were probed by
utilizing synthetic nisin A fragments.269 Truncated
peptides lacking the first two N-terminal residues or
composed of just the B- and C-, or the D- and E-rings
did not retain induction capacity, whereas a synthetic
fragment containing the first 11 residues including
the A- and B-rings was sufficient for transcriptional
activity. Nisin Z and several nisin Z mutants were
also shown to be transcriptional activators of nisA,
but not all mutants retained efficient signaling
capability (e.g., S3T in which ring A has a MeLan,
Val32Glu/Dha33Ser, and Val32Trp/Dha33Ser).269,273
In some cases, these nisin mutants stimulated transcriptional activity only at very high concentrations.273 Disrupted or strongly reduced signal transduction potency is important for lantibiotic engineering
(section 7) as mutants with weaker inductive properties may not be produced to the levels required to
trigger the autoregulatory mechanism to support
continued lantibiotic production.
Interruption of NisK expression yielded a strain of
L. lactis that did not produce nisin upon addition of
varying amounts of nisin A or Z, indicating its
involvement in the signal transduction cascade.269 To
evaluate the minimal requirements for efficient
induction of nisA transcription, the genes for nisR
and nisK were incorporated into the chromosome of
a nonlantibiotic producing L. lactis strain transformed with a plasmid carrying the gusA reporter
gene under control of the nisA promoter. Addition of
nisin to this strain resulted in gusA expression,
illustrating that NisK and NisR are the only requirements for nisin activation of gene expression. The
two-component regulatory system not only controls
expression from the nisA promoter but also from the
nisF promoter that regulates the nisFEG immunity
genes (section 6),260,268 whereas transcription from the
nisR promoter that controls expression of NisR and
NisK is nisin independent.268 Two sets of TCT direct
repeats located 39 and 107 bp upstream of the nisA
transcription start site have been proposed as the
putative binding site for NisR,274 and as described
below such binding has been established for the
analogous system involved in subtilin production.
The high efficiency of the nisin regulatory system has
found utility as a heterologous-controlled protein
expression system in food-grade lactic acid bacteria
(NICE ) nisin-controlled expression).275-287
Nisin-induced stimulation of its own biosynthesis
is not the only mechanism of enhancing nisA transcription. It has been reported that transcription from
the nisA promoter can also be induced in the absence
of externally supplied nisin in a carbon-source-
dependent fashion, with galactose and lactose enhancing transcription significantly.288 The nisRK
signal transduction system was not involved, and the
nisin and galactose/lactose induction regulators were
shown to compete for the same recognition site.
Galactose and lactose do not induce transcription
from the nisF promoter. A possible rationale for this
differential expression may be found in the presence
of two TCT-N8-TCT repeats upstream of the nisA
start site and only a single such repeat upstream of
the nisF start site.274
To understand the regulation of subtilin biosynthesis, experiments analogous to those performed
with nisin have been carried out.216 Deletions within
the spaR and spaK genes resulted in failure to
express SpaB and SpaC and eliminated subtilin
production. Complementation with a plasmid encoding the spaR gene sequence restored the ability to
produce the lantibiotic in the spaR mutant. Two
molecules of the SpaR protein have been shown to
bind to the spaS, spaB, and spaI promoter regions
(Figure 5), which contain a pentanucleotide repeat
separated by six nucleotides as the recognition motif
(spa-box)289 that is similar to that found in the nisA
and nisF promoters. This spa-box is located upstream
of the transcription initiation sites for all three
promoters (spaS, spaB, and spaI).271 While the subtilin and nisin systems are very similar in that both
possess a LanRK signal transduction pathway that
is autoinduced by the respective lantibiotic, an additional regulatory system governs subtilin biosynthesis. Stein et al. recently demonstrated a positive
regulatory system for spaR expression utilizing sigma
factor H (sigH gene), an endogenous regulator within
subtilin producing B. subtilis strains that is also
under cell-density-dependent control.272 An additional
B. subtilis regulator, AbrB, appears to negatively
regulate lantibiotic production as strains lacking
abrB exhibit a significant increase in the production
of subtilin.272
The similarity between the nisin and subtilin
regulatory systems has been illustrated via cross-talk
experiments involving the incorporation of SpaK into
the nisin induction system.55 A plasmid encoding a
reporter gene under control of the nisA promoter was
introduced into a bacterial strain that contained the
nisR gene on its chromosome. Upon introduction of
a plasmid containing the nisK gene sequence, a
completely functional nisin induction system was
established, which used nisin as a transcriptional
activator. Gene expression was also accomplished by
introduction of a plasmid encoding spaK and using
subtilin as inducer, illustrating that both SpaK and
NisK can phosphorylate NisR upon activation by
their respective lantibiotic. Furthermore, chimeric
NisK-SpaK proteins, in which the N-terminus corresponded to that of SpaK while the C-terminal
domain originated from NisK, can modulate the
specificity of the inducer.290 When this protein was
expressed in place of NisK in a L. lactis strain
equipped with the nisin signal transduction machinery, it resulted in a functional hybrid sensor kinase
that activated transcription of the nisA promoter in
the presence of subtilin. Not only do these results
Chatterjee et al.
provide evidence that the N-terminal portion of the

NisK protein is responsible for molecular interactions
with the inducer, but they also suggest the numerous
possibilities for protein engineering that are available
within the lantibiotic family.
Epidermin production in S. epidermis Tu3298 is
regulated by an accessory gene regulator (agr) quorum sensing system267 that is also responsible for the
transcriptional activation of many surface proteins
in various strains of staphylococci.291 The agr locus
contains the response regulator AgrA, a sensor kinase AgrC, a pheromone precursor (AgrD), and AgrB,
which is responsible for pheromone maturation. A
transcription profiling study identified a lantibiotic
locus that was under control of agr,292 but the
organism under investigation has not been shown to
produce a lantibiotic and the mechanism of epidermin
regulation by agr was not clear. Kies and co-workers
recently demonstrated that unlike the nisin and
subtilin systems in which interference with LanK/R
directly affects lantibiotic production by blocking
transcription of the biosynthetic genes, disruption of
the agr regulatory system does not interfere with the
epidermin biosynthetic proteins that introduce the
dehydro amino acids and thioether bridges.293 Instead, agr in S. epidermis controls removal of the
leader peptide from posttranslationally modified preepidermin. It was suggested that the cleavage of the
prosequence from EpiP, the protease responsible for
removing the leader sequence (section 4.7), may be
regulated by agr resulting in control over the activity
of EpiP.293
The epidermin gene cluster also contains the epiQ
gene that is essential for epidermin production.294-296
EpiQ belongs to the family of response regulators,
but the epidermin gene cluster does not contain a
gene for a corresponding receptor histidine kinase.
EpiQ is responsible for activating transcription from
the epiA, epiF, epiH, and epiT promoters that leads
to the expression of the proteins necessary for epidermin synthesis (epiABCD) and immunity (epiFEG,
epiT, epiH, see section 6). Interference with the
transcription of epiQ leads to complete loss of lantibiotic production. Peschel et al. showed that EpiQ
binds directly to an inverted repeat that is the
putative operator site for expression from the epiA
promoter.294
Regulation of production of the type B lantibiotic
mersacidin involves two LanR type proteins (MrsR1
and MrsR2) and one LanK sensor protein (MrsK2).297
The adjacent mrsR2 and mrsK2 genes (Figure 3)
encode proteins with significant similarity to the twocomponent regulatory systems present in nisin and
subtilin biosynthetic operons, whereas the MrsR1
protein is an additional response regulator. Mersacidin-producing Bacillus strains lacking mrsR1 were
unable to produce mersacidin but retained selfimmunity properties. On the other hand, inactivation
of mrsR2/K2 led to increased susceptibility to the
lantibiotic with biosynthesis remaining intact. In the
mrsR2/K2 knockout mutant, the mrsFGE genes
encoding an ABC-type transporter were not transcribed, suggesting they are involved in self-protection (section 6). These results provide evidence that
the mersacidin producer utilizes the MrsR2/K2 tandem to activate transcription of immunity genes and
the MrsR1 protein is responsible for promoting
biosynthesis of the lantibiotic.297 At present, the
stimulus for MrsR1 activation is unknown, nor is it
clear whether MrsR1 requires phosphorylation for
activity or which kinase would be involved. The
results described above do rule out MrsK2 as the
kinase. Other single regulatory proteins engaged in
lantibiotic biosynthesis that lack a corresponding
histidine kinase are the aforementioned EpiQ as well
as MutR, LasX, and LtnR (see below).
Despite the high homology in the structures of the
type AII lantibiotics, the regulation systems for these
compounds are quite diverse. The gene cluster of lacticin 481 lacks regulator genes corresponding to the
LanKR proteins. It was recently shown that lacticin
481 is regulated at the transcriptional level by pH
control of P1 and P3 promoters located upstream of
lctA in the lantibiotic operon.298 During growth L.
lactis produces lactic acid, which in turn leads to a
decrease of the pH of the growth medium from 7.0
to 5.8. This natural acidification correlates to the
amount of lacticin 481 that is produced. Medium that
was acidified to pH 5.8 using acetic acid prior to lacticin production, although resulting in a slower
growth rate, actually led to a higher production level
of lacticin 481, indicating that control of lantibiotic
expression is pH controlled, not lactic acid induced.
Since the lacticin 481 operon does not contain a
dedicated regulation system, transcription from the
P1 and P3 promoters is probably governed by a
general regulator.
Mutacin II production requires the MutR regulatory protein, but its biosynthetic gene cluster lacks
a sensor histidine kinase analogous to NisK or SpaK.
Although mutacin II is structurally very similar to
lacticin 481 (Figure 9), their biosynthetic regulation
systems are quite different. Mutacin II transcriptional regulation is controlled by the mutA and mutR
promoters, and the mutR gene299 encoding a protein
homologous to the family of Rgg (regulator gene of
glucosyltransferase) transcription regulators.300 Activation of the mutA promoter responsible for transcription of the mutAMTFEG operon is dependent on
MutR as well as currently unknown components in
the growth medium.301 Inactivation of the gene
encoding MutR eliminates transcription of all genes
in the mutA operon, including mutA, mutM, and the
immunity genes mutEFG.301 Although direct binding
of MutR to the mutA promoter site was not determined and the protein sequence does not show
obvious DNA binding motifs, the homology to Rgg is
consistent with its direct interaction with DNA.
Unlike mutacin II and lacticin 481, regulation of
the type AII compound salivaricin A in S. salivarius
UB1309 is much more similar to the type AI lantibiotics nisin and subtilin. It also autoregulates its
own production through a salKR two-component
response system.157 As mentioned in section 3.4, S.
pyogenes strains produce very close relatives such as
[Lys2,Phe7]-salivaricin A (salivaricin A1). Addition
of salivaricin A1 to the growth medium of S. salivarius UB1309 induced transcription of the salA
gene, showing that signaling not only occurs intraspecies but also interspecies.157
Dedicated repressors of lantibiotic gene expression
have only been described for the two-component lantibiotics lacticin 3147 and cytolysin, and for lactocin
S. McAuliffe et al. reported characterization of LtnR,
the first example of a repressor encoded in a lantibiotic biosynthetic operon (see also section 6).227
Biosynthesis of lacticin 3147 in L. lactis ssp. lactis
DPC3147 is under control of the constitutive Pbac
promoter that governs the transcription of the ltnA1A2M1TM2D operon. A second divergently transcribed transcriptional unit ltnRIFE (Figure 3) responsible for immunity is negatively regulated by
LtnR, which was shown to bind to the Pimm promoter
for this operon. A second interesting case of repression of lantibiotic production has been reported for
cytolysin.302 Mutational inactivation of either or both
cylR1 and cylR2 led to the constitutive expression of
high levels of lacZ placed under the control of the
cytolysin promoter pL in an engineered strain, whereas in the presence of both genes no expression was
observed unless the fully modified CylLS peptide was
added to the growth medium. Adding CylLL or incompletely processed CylLS or CylLL did not induce
transcription. These results indicate that CylR1 and
CylR2 together are needed for repression of the expression of the biosynthetic machinery. This block is
alleviated when CylLS reaches a certain threshold
concentration, which corresponded to about 107 colonyforming units per mL of the producer Enteroccus
faecalis.302 Hence, this is another example of quorum
sensing that leads to autoinduction of lantibiotic production and differs from that described above for
nisin, subtilin, and salivaricin A. CylR1 lacks homology to other known proteins and the CylR2 protein
is a member of the helix-turn-helix family of DNAbinding proteins. Therefore, this system is distinct
from other two-component signal transduction systems. A preliminary report of expression and crystallization of the transcriptional repressor CylR2 has
recently appeared.303 A third example of a repression
system is found for lactocin S, the production of which
is modulated by LasX, a protein that like MutR has
significant homology to Rgg-type proteins. Interestingly, LasX serves as an activator of the promoter
for transcription of the lasAMNTUVPJW operon and
as repressor of the overlapping promoter for the divergently transcribed bicistronic lasXY operon (Figure 3).304 It was proposed that this dual action may
aid in maintaining a steady rate of lactocin S production.
6. Self-Immunity of the Producing Strains

Any bacterial strain producing antimicrobial compounds that are active against closely related strains
must protect itself against its product. Although the
picture of these mechanisms is still emerging for the
lantibiotics, a number of recent studies have provided
the first insights for some family members. Early
work on the characterization of the nisin biosynthetic
gene cluster revealed the presence of the nisI
gene.259,305 The NisI protein (245 amino acids) shows
no homology with other proteins and has a hydro-
phobic N-terminus containing a consensus lipoprotein

sequence.305 It is membrane anchored by posttranslational removal of the first 19 amino acids and lipid
modification of the Cys at the new N-terminus.
Palmitoylation of NisI was confirmed by Qiao and coworkers199 in both E. coli and L. lactis strains using
[3H]-palmitic acid. SDS-PAGE and Western blotting
using anti-NisI antibodies showed that the protein
is localized extracellularly at the cytoplasmic membrane. More recently, Koponen et al. showed that a
significant percentage of NisI is secreted in a lipidfree form into the cytoplasm.306 The presence of
extracellular NisI may have a biological function,
complexing external nisin before it can aggregate at
the cell surface, thereby acting as an additional selfprotection mechanism. Expression of nisI in nisinsensitive L. lactis strains results in modestly increased levels of resistance,199,214,259,305 indicating that
the production of NisI is correlated to sensitivity to
exogenous nisin but that NisI by itself does not
impart full immunity. The extracellular membrane
association may facilitate direct interaction with
nisin at the exterior of the cell thereby possibly
inhibiting pore formation. Circular dichroism spectroscopy and biomolecular interaction analysis have
been used to demonstrate that NisI and nisin do
interact with each other,307 and interactions of purified His6-NisI with nisin but not subtilin were
detected by SDS-PAGE.308 The lack of strong sequence similarity among the LanI proteins has been
attributed to the specific nature of the recognition
between the LanI protein and the lantibiotic against
which it protects.305
Near wild-type levels of immunity could be conferred on nisin-sensitive strains via complementation
with a plasmid encoding NisI as well as the prepeptide (NisA) and the posttranslational modification
machinery.305 At the time, the need of nisin production for immunity was not understood, but with the
discovery that nisin is a quorum sensing molecule
that regulates its own production as well as transcription of immunity genes (section 5),269 the role of
nisin production in immunity was revealed. In support of this model, strains containing disruptions or
in-frame deletions of the nisB, nisC, and nisT genes
were highly sensitive to externally added nisin, but
upon preincubation with nisin, they regained about
20% of wild-type level immunity.307 These gene
disruptions led to the abolishment of nisFEG transcription in the absence of exogenous nisin, whereas
prior addition of nisin resulted in increased production of NisFEG, thereby implicating these proteins
in immunity. Since nisFEG transcription is under
control of the NisR/K regulation system (section 5),
these findings also explained why they are necessary
for inducing immunity and why strains of nisinproducing L. lactis with NisK mutations are sensitive
to nisin.309 A similar role of the final lantibiotic
product in self-protection was established for the
salivaricins.157 Collectively, these results indicate
that full nisin immunity requires three components:
nisin production and expression of the nisI and
nisFEG genes. Interruption of either nisI or nisFEG
results in strongly decreased immunity levels, indi-
Chatterjee et al.
cating that these proteins actually operate in a

cooperative manner.
The nisFEG genes were first identified by Siegers
and Entian in 1995.310 The NisE and NisF proteins
are homologous to members of the type B ABC
transporters of the HisP family.311-313 NisF contains
two potential ATP-binding sites, whereas NisE with
NisG is thought to form the integral membrane
segment of the transporter. NisG is a hydrophobic
protein with sequence similarity to the immunity proteins found for several channel-forming colicins.314,315 The latter proteins are believed to interact
directly with the pore-forming domains of the corresponding peptides. Investigations into the HisP and
MalK secretion systems showed that a complete
complex contains two hydrophobic membrane spanning subunits and two ATPase subunits.316 The
similarity with the MalFGK2 and HisMQP2 transport
systems involved in maltose317 and histidine318 transport, respectively, in E. coli suggests that the complex
for nisin (and other lantibiotics, see below and Figure
5) consists of NisF2EG. Individual disruptions of the
nisE, nisF, and nisG genes did not affect the ability
to produce nisin, indicating that their gene products
are not directly involved in lantibiotic biosynthesis,
but did lead to increased sensitivity to the lantibiotic.
Stein et al. reported that the genes involved in
nisin immunity in L. lactis could produce the same
phenotype in B. subtilis strains via coordinated
expression.308 B. subtilis cells expressing NisI showed
significant and comparable immunity to nisin as cells
containing the NisFEG transport proteins while cells
transformed with both nisI and nisFEG displayed full
nisin tolerance. Interestingly, nisin was associated
in much decreased levels with cells expressing NisFEG than with control cells, suggesting the NisFEG
proteins act by transporting nisin from the membrane into the extracellular space.
The proteins involved in self-protection of B. subtilis against subtilin were first characterized by Klein
and Entian in 1994.319 Like most other LanI proteins,
SpaI is a mostly hydrophilic lipoprotein with an
N-terminal hydrophobic domain postulated to anchor
the protein in the cell membrane. The SpaE, SpaF,
and SpaG proteins have a great deal of sequence
similarity to ABC-transporters located in the biosynthetic gene clusters of other lantibiotics including
NisE, NisF, and NisG (note: originally spaE and
SpaF were reported to be one orf, spaF,319 but they
were later shown to consist of two genes).272 Klein
and co-workers demonstrated that these proteins are
required for immunity and that their disruption does
not lead to the abolition of subtilin production by the
cell.319 In a further similarity to the nisin immunity
system, disruption of spaS led to subtilin-sensitive
producer strains, which is well explained by the same
type of autoregulation of expression of the immunity
genes as reported for nisin.272
Epidermin-producing S. epidermidis strains also
contain the epiFEG genes for self-immunity.295 Their
inactivation led to complete loss of immunity against
epidermin, and expression of epiFEG in an epidermin-sensitive strain conferred a significant level of
immunity. As mentioned above, it has been proposed
that the LanFEG proteins may scavenge lantibiotics

that have penetrated the cytoplasmic membrane and
secrete them into the extracellular medium.295,320 Experimental support for this model has been reported
by comparing epidermin concentrations in the supernatant of cells that do or do not express the epiFEG genes.320 The efficacy of releasing epidermin
from the cell surface was much higher than that with
nisin, showing specificity of the transport system. A
variation on this hypothesis proposed by Bierbaum
and co-workers321 has the LanEFG proteins detach
the lantibiotic from its cellular target lipid II (see section 8.2). This proposal is supported by the observation that lanEFG genes are found in the clusters for
nisin, epidermin, and mersacidin, which all interact
with lipid II but are absent from the cluster of Pep5,
which does not bind to lipid II.322 A corollary of this
hypothesis is that lantibiotics such as lacticin 481,
mutacin II, streptococcin A FF-22, and lacticin 3147,
which currently have not been reported to bind lipid
II but do contain lanEFG, also have a cellular target.
More recently, the accessory factors epiH (formerly
epiT)69,76 and gdmH that are unique to the epidermin and gallidermin gene clusters, respectively, were
shown to mediate immunity by controlling lantibiotic
secretion.323 Since the EpiT transport system in the
producer of epidermin is defective because of a 20bp deletion, studies on the role of EpiH and GdmH
have focused on gallidermin producing strains; gallidermin differs in only a single amino acid from
epidermin (section 3.2). The EpiH and GmdH proteins were first shown to aid secretion of epidermin
and gallidermin by GmdT, the one-component transport system responsible for lantibiotic export after
posttranslational modifications.296 The need for ancillary proteins for the assembly of active ABC-type
transporters is not uncommon and has been described for other systems.324 More recently, it was
shown that a plasmid containing only the gdmH gene
conferred a 2-fold decrease in sensitivity to gallidermin in the heterologous expression host S. carnosus,
illustrating that gdmH also has a role in immunity.323
Expression of the EFGHT proteins involved in gallidermin and/or epidermin production resulted in a
much higher level of immunity than expression of
only EpiFEG, and expression of the EpiEFGH proteins led to an immunity level equal to the additive
effects of the EpiFEG and GdmH proteins. These
studies demonstrated that while the GdmT protein
is not necessary to confer protection, it is required
for full cooperativity between the EFG and H proteins. These proteins did not confer any immunity
to nisin, providing further evidence for the high
specificity of the LanEFG proteins.323
The immunity mechanisms for bacterial strains
responsible for producing epicidin 280,72 Pep5,325 and
lactocin S155 are quite similar on a genetic level. The
operons for the biogenesis of all three compounds
contain lanI genes but lack genes that encode
LanEFG-type transporter proteins. The EciI and
PepI proteins share 74.2% sequence identity, and
EciI has significant similarity (42.1%) to the gene
product of ORF57 in the lactocin S biosynthetic gene
cluster.72 All of these proteins have a similar size and
charge distribution.326 PepI (69 amino acids) is characterized by an apolar N-terminus and a hydrophilic
C-terminal domain with a net positive charge. Early
studies implicated a role of PepA production in
immunity in the producing strain of Pep5,325 suggesting possibly a similar regulation of self-protection
by the mature product as described for nisin. Pag and
co-workers327 have subsequently shown, however,
that pepA transcription plays a different essential
role. They showed by complementation studies with
plasmids encoding either pepI or pepA that in trans
complementation is not sufficient to confer immunity
against Pep5, but when both genes are encoded on
the same plasmid immunity similar to that of the
wild-type strain was observed. This unusual observation was explained by a stabilizing effect on the PepI
mRNA by an inverted repeat located downstream of
the pepA gene sequence, which also functions as a
weak terminator that allows partial read-through to
the lanPBC genes.327 This inverted repeat is proposed
to form a hairpin that protects the PepI mRNA from
ribonucleases, possibly via direct binding. Hence,
pepA transcription is necessary for efficient expression of PepI. The position of this stabilizing inverted
repeat is not important, however, because placing the
terminator sequence upstream of pepI resulted in a
strain that was hyperimmune to Pep5. A dual role
of the leaky transcriptional terminator downstream
of a lanA has also been proposed for lactocin S,
although in this case it protects the lasA transcript.155
PepI confers cross immunity against epicidin 280
with which its shares 75% identity, the only such case
in lantibiotics producing strains that has been reported.242
Recently, the localization of PepI has been studied
using protein fusions with green fluorescent protein
(GFP).326 These PepI-GFP constructs revealed that
PepI is found at the cell wall-membrane interface.
Truncated proteins and site-directed mutants were
generated to determine the functional role of the two
domains of PepI. Introduction of charged amino acids
into the apolar N-terminus blocked export of PepI,
while shortening the C-terminal portion did not affect
the localization of PepI but reduced immunity. These
experiments illustrated that the two domains probably have distinct functions: the N-terminus serves
a role in localization of the protein, while the Cterminal end is involved in conferring immunity
against Pep5.
While investigating regulation systems utilized in
mersacidin producers, Guder and co-workers were
able to indirectly ascertain modes of self-protection
as well.321 In a mersacidin-producing strain, the
mrsR2/K2 genes were knocked out to investigate the
role of these gene products in regulation (section 5).
The resulting bacterial strain maintained its ability
to produce mersacidin, but its sensitivity to its
product increased significantly, although not as
dramatically as with NisI inactivation in L. lactis.
Analysis by reverse transcription-PCR revealed that
in this mutant the mrsEFG genes had not been
transcribed, therefore showing that lack of MrsEFG
protein expression leads to loss of immunity.321 These
proteins have significant similarity to the LanEFG
Chatterjee et al.
Figure 25. Representation of variants of nisin that have been reported. (A) Mutants that have been generated by sitedirected mutagenesis.48,136,273,331,335,337-339,378,389,470 Shown in black are mutants in the nisin Z background and in red mutants
in the nisin A background. (B) Truncation,273,392,416,471 contraction,48 or extension383,385 mutants were obtained either through
molecular biology or by chemical471 or proteolytic techniques.392
transporter systems described above, and it is believed that they actively extrude the lantibiotic.
In the gene cluster encoding the biosynthetic
proteins involved in lacticin 481 production three
genes (lctEFG) were identified with significant similarity to other lanEFG genes. Rince and co-workers
demonstrated that strains containing all three genes
were immune to lacticin 481, while the absence of
any one protein resulted in loss of immunity.328
Homologues of the lanFE genes were also identified
in the biosynthetic cluster of lacticin 3147 (ltnFE),329
but they do not appear to play an important role in
immunity. Instead, expression of ltnI is sufficient to
confer levels of immunity to sensitive strains that are
comparable to that of producing strains. LtnI is
predicted to be a protein of 116 amino acids and bears
no homology to any of the other LanI proteins.329
It is interesting to compare the various immunity
mechanisms in the lantibiotics. Nisin, subtilin, epidermin, lacticin 3147, streptococcin A FF-22,263 mutacin II,255 and lacticin 481 all have the lanFEG genes
(Figure 3), whereas Pep5, cytolysin,330 epicidin,72
lactocin S, and epidermin only require lanI for
immunity. Interestingly, the compounds that both
form pores and utilize lipid II as a docking molecule
to increase the efficiency of pore formation and also

interfere with peptidoglycan formation (nisin and
epidermin, section 8) all contain both lanI/lanH and
lanEFG genes.321 Mersacidin, which also inhibits cell
wall biosynthesis by binding to lipid II but does not
form pores, only contains the lanFEG genes. Further
investigation into the modes of action and selfprotection will shed more light onto the specific
mechanism of action of these proteins. For resistance
mechanisms of nonlantibiotic producing bacterial
strains against nisin, see section 9.
7. Lantibiotic Engineering
7.1. In Vivo Protein Engineering
The cloning of the gene clusters involved in the
biosynthesis of many lantibiotics laid the foundation
for genetic protein engineering aimed at in vivo production of novel compounds with potentially interesting properties. Many studies have indicated the feasibility of changing the molecular structures of lantibiotics by mutagenesis of the pre-lantibiotic genes.331
So far, engineering of the nisin structure has been
most extensively investigated (Figure 25), but engineered (heterologous) expression systems have also
been established for subtilin,137 Pep5,332 epidermin

and gallidermin,333 mutacin II,161 and mersacidin.334
In these studies, not only the biosynthetic machinery
but also the immunity factors had to be considered
for generating successful expression systems.
Replacement of Ser at position 5 by Thr in the
structural region of the nisZ gene led to the production of Dhb instead of Dha in mature nisin Z.335 The
variant exhibited an increased resistance to chemical
degradation albeit accompanied by a 2-10-fold reduction in bioactivity toward various indicator strains.
In contrast, replacement of Dha at position 2 in nisin
Z with Dhb (S2T mutation in NisA) resulted in a mutant that was twice as active as native nisin Z toward
certain sensitive strains.331 Replacement of Dha by
Dhb and vice versa has also been reported for other
lantibiotics. The gallidermin variant Dhb14Dha did
not exhibit any noticeable decrease in activity,331 and
the Dhb10Dha mutant of mutacin II also showed similar activity to wild-type.161 Very interesting is the
production of a nisin mutant with a Dhb residue at
position 18 in place of Gly after introduction of a Thr
codon in nisA,335 and the similar introduction of a
novel Dha in place of a Lys at position 18 in Pep 5.336
These studies clearly indicate low substrate specificity for the dehydratases involved. This low substrate
requirement has been further explored in a nonproducing L. lactis strain by overexpressing the nisin
dehydratase and transport proteins along with a
chimera consisting of the nisin leader peptide fused
at its C-terminus to the angiotensin heptapeptide
NRSYICP. A single dehydration was observed in the
exported chimera, suggesting that dehydratase function is independent of the nature of the structural
region as long as the leader is present.195
Generally, removal of dehydro amino acids from
lantibiotics reduces their biological activity. For
instance, removal of the Dha at position 5 by replacement with Ala eliminated nisin and subtilins activity
against outgrowing spores (section 8.3).136-138 Strains
expressing nisin in which either Dha33 was replaced
by Ala or both Dha5 and Dha33 were substituted
with Ala resulted in greatly reduced activity (about
1% of wild-type nisin producing strains). Unfortunately, it was not established whether this reduction
of activity was due to a less active antimicrobial
peptide or to poor processing and secretion of the
mutants and/or reduced expression in the engineered
system.337 Substitution of Dhb at positions 16 and 20
of Pep5 to Ala were also found to reduce its activity
against four test strains,336 and changing Dha16 in
mersacidin to Ile greatly reduced its activity toward
indicator strains such as M. luteus and S. pyogenes.334
Not all amino acid substitutions are tolerated by the
biosynthetic machinery, however. For example, attempts to generate the mutacin II mutant Dhb10Ala
did not lead to detectable mutacin production,161 and
replacement of Ser3, Ser19, or Cys22 that are involved in lanthionines in gallidermin also resulted
in loss of production.333 For those lantibiotics that
autoregulate their biosynthesis (section 5), the mutant structures may not be able to activate the twocomponent response system resulting in lack of
production.273
Alterations in the Lan and MeLan structures have

also been accomplished. Substitution of Ser at position 3 in nisin Z with Thr gave rise to MeLan instead
of Lan with a dramatic reduction in activity.331
Replacement of Thr13 with Cys produced a disulfide
in place of MeLan in nisin Z resulting in reduced
activity.338 Intriguingly, a fourth thioether bridge
(MeLan) was introduced between positions 16 and
19 in Pep5 by the mutation Ala19Cys.336 This methyllanthionine increased proteolytic stability against
the proteases chymotrypsin and Lys-C but also
resulted in a significant decrease in antimicrobial
activity. This decrease in activity may be due to
rigidification of the flexible central region that otherwise aids pore formation by Pep5. Other Pep5
analogues in which ring structures had been deleted
displayed a pronounced susceptibility toward proteolysis. Caufield and co-workers have mutated all
three Cys residues involved in ring formation in
mutacin II to Ala. While the Cys15Ala and Cys26Ala
mutants exhibited no antimicrobial activity, Cys27Ala
had low level activity (less than 10% of wild-type
mutacin II).161 Collectively, these studies on changing
the thioether bridges reiterate the importance of the
Lan/MeLan rings for antibiotic activity.
In addition to substitutions of the residues that are
posttranslationally modified, a large number of mutants have been reported in which other amino acids
in the polypeptide have been replaced (e.g., Figure
25 for nisin). Two nisin variants with higher solubility than the parent compound have been produced
by substitution of Asn27 or His31 with Lys.339 The
introduction of an extra Cys in nisin Z by van Kraaij
et al. (Ser5Cys and Met17Cys) resulted in a compound that required the presence of a reducing agent
for bioactivity.338 Natural selection may therefore
explain the absence of any lantibiotic structures with
free thiol groups, whereas the occurrence of unreacted Dha/Dhb residues is fairly common (Table 2).
Several nisin mutants have been reported in which
the residues in the so-called hinge region (Asn20,
Met21, Lys22) were changed,48,470 which resulted in
very interesting changes in the bactericidal activities
(section 8.2). Changing Glu4 of subtilin to Ile increased the biological activity 3-4-fold compared to
wild type and significantly slowed chemical modification of Dha5, a process that leads to loss of certain
biological activities of the compound (sections 8.2 and
8.3).136-138 The Leu6Val gallidermin mutant was
found to be twice as active as the wild-type against
M. luteus, while the mutants Dhb14Pro and Ala12Leu
showed increased resistance to proteolytic degradation.333 Two mutants of MrsA were expressed in an
engineered host,341 and the corresponding mersacidin
analogue Glu17Ala-mersacidin had strongly reduced
activity whereas Phe3Leu-mersacidin displayed activity closer to the wild-type lantibiotic. Heterologous
expression in Streptomyces lividans of the cin cluster
containing mutated cinA genes has resulted in the
production of Arg2Lys- and Phe10Leu-cinnamycin,
which correspond to duramycin and duramycin B
(section 3.6).78 Attempts to produce the sextuple
mutant R2A/Q3N/F7Y/F10L/F12W/V13S-cinnamycin
(duramycin C) were not successful.
Chatterjee et al.
Figure 26. Sequences of His6-LctA and its mutants used to investigate the minimal sequence requirements of the leader
peptide and the substrate specificity for posttranslational modifications in the propeptide. Truncations in the leader sequence
(black) and structural region (red) are depicted by dashed lines. Mutated residues are in yellow.
The important advancements in in vivo protein

engineering of the lantibiotics have greatly contributed to a better understanding of lantibiotic biosynthesis and antimicrobial activity. However, very few
mutant lantibiotics have been generated with improved antimicrobial activities, and none display enhanced activity for all test strains. A number of potential explanations could account for this. First, it
might not come as a surprise that Nature has already
optimized the biological activity of these compounds
using the same tools, i.e., mutagenesis with 20 amino
acids. Furthermore, the structural and functional
space that can be sampled using genetic engineering
of ribosomally synthesized proteins will likely always
be limited although great strides have been made to
overcome this impediment.342-344 Another contributing factor to the absence of more potent compounds
produced by genetically engineered lantibiotic producers may lie in the breakdown of self-immunity in
cases in which more active compounds are actually
generated. This might lead to either degradation of
the intermediates or shutdown of antibiotic production. Indeed, degradation products or incompletely
modified peptides are often observed even in cases
in which fully processed novel materials are isolated.137,336 Finally, it has been shown for several
lantibiotics that the prepeptide and/or the final
product fulfills a regulatory role in its production
(section 5).189,269,325,332 Structural variants, however,
may lack the ability to induce in vivo synthesis
resulting in reduced or abolished production.273
7.2. In Vitro Protein Engineering

In vitro engineering of the lantibiotic biosynthetic
processes has several conceptual advantages over
genetic protein engineering of lantibiotics. The structures of the prepeptides are not limited by the physiological amino acids, only by the ability to design
and synthesize the amino acids and incorporate them
into peptides using well-developed solid-phase synthesis and peptide ligation techniques. In addition,
peptide synthesis is particularly amenable to combinatorial techniques, thereby dramatically increasing the number of rapidly accessible substrate candidates. Because of the in vitro nature of the approach,
degradation of products is not a problem, nor will
cytotoxic or regulatory properties of the products be
a concern. This will permit exploration of the struc-
tural and functional tolerance of the biosynthetic enzymes in much greater detail because nonproteinogenic amino acids can be utilized in addition to the
natural amino acids. Finally, although speculative,
it may prove possible to use nonpeptide structures
in part of the substrates to produce even more stable
molecules.
While in vitro engineering of the posttranslationally produced oxazole and thiazole rings in the
bacteriocin microcin B17 by a microcin synthetase
complex has been demonstrated,345 similar attempts
to reconstitute an active lantibiotic synthetase in
vitro proved challenging until recently. In 2004, an
in vitro system for generation of the type AII lantibiotic lacticin 481 was the first example of its kind
(section 4.4).40 The prelacticin modifying enzyme
LctM was cloned from L. lactis CNRZ 481 and
heterologously expressed in E. coli. The prelacticin
peptide LctA was also expressed and purified with
an N-terminal His6-tag. The resulting functional in
vitro system was then exploited to test the substrate
specificity of LctM with His6-LctA derived substrates
(Figure 26). Engineering of LctA to obtain novel
substrates was achieved at the genetic level by
mutation and/or truncation of the lctA gene and also
posttranslationally by expressed protein ligation
(EPL)346,347 with an LctA(1-37)-intein-chitin binding
domain (CBD) fusion protein.
As expected, replacement of Thr48 with Ala resulted in only three dehydrations instead of the usual
four found in His6-LctA. An unexpected fifth dehydration was seen in the LctA-Cys49Ser mutant due
to the introduction of an extra Ser. Obviously, the
replacement of the Cys at this position also precluded
formation of the B-ring, but it was not anticipated
that the formation of an extra Dha at position 49 also
interfered with formation of the C-ring. Interestingly,
the LctA-Cys49Ala mutant underwent only two
(Thr33 and Ser35) of the possible four dehydrations
(Thr33, Ser35, Ser42 and Thr48), when a disulfide
bond was present between Cys38 and Cys50. Reduction of this incompletely dehydrated product with
DTT and reincubation with LctM led to the formation
of products with up to four dehydrations, demonstrating the ability of LctM to further process partially dehydrated products. A C-terminally truncated His-LctA(1-37) peptide that contained two of
the residues that undergo dehydration in the full
length substrate was found to be a substrate for LctM

and dehydration was localized by tandem MS to
Thr33 and Ser35. Furthermore, the His-LctA(1-38)
and His-LctA(1-38)Cys38Sec mutant peptides that
include the residues necessary for formation of the
A-ring were accepted by LctM resulting in dehydration and cyclization to form thioether and selenoether
rings, respectively. This example demonstrates the
potential of combining semi-synthetic substrates
generated by EPL with the substrate promiscuity of
LctM.
It should be noted that some advantages of genetic
engineering are lost in the in vitro strategy. One of
the biggest assets of molecular biology is that it
produces readily and rapidly renewable sources of
manipulated genes and organisms, which is not true
for chemically synthesized molecules and purified
enzymes. A potential solution for the problem of
rapidly generating mutant substrates with unnatural
amino acids at desired positions would be the utilization of the in vivo amber codon suppression methodology developed by Schultz and co-workers.348 It may
prove, however, that such in vivo use of unnatural
peptide substrates for lantibiotic engineering suffers
to an even higher degree from the drawbacks described in section 7.1 with respect to discovering
lantibiotics with new or improved biological activities.
The great promise of in vitro use of lantibiotic
synthases is not limited to the production of lantibiotic analogues. These enzymes may also find application in installing dehydro amino acids or lanthionine
rings into other synthetic targets. Cyclic lanthionine
containing peptides have found use as mimics of
natural products that contain disulfide bridges or as
structures that limit the conformational flexibility of
bioactive compounds.349 The lanthionine moiety also
provides higher chemical and proteolytic stability for
such analogues. There are reasons to be optimistic
that these type of compounds may be prepared using
lantibiotic synthases, although to date they have only
been produced by chemical synthesis. For instance,
Goodman and co-workers have demonstrated the
advantage of the structural constraints imposed by
Lan in a somatostatin analogue.350 Peptide cyclization on oxime resin (PCOR)351,352 was employed to
generate a conformationally rigid mimic of sandostatin, a somatostatin analogue, by replacing a disulfide
linkage with a thioether bridge. The lanthioninesandostatin analogue possessed enhanced receptor
selectivity and an increased half-life toward enzymatic degradation.350 Cyclic lanthionine containing
enkephalin analogues have also been synthesized and
the ,-dimethyl substituted compounds were selective for the -opiod receptor.353
In addition to using lanthionines themselves, several synthetic studies have incorporated lanthionine
analogues into biologically relevant compounds.350,353,354
Tabor and co-workers prepared the norlanthionine
analogue of the C-ring of nisin (Figure 27).355,356 The
analogue was synthesized by solid-phase peptide
synthesis in a linear fashion employing Fmoc-based
chemistry and a triply protected norlanthionine monomer. Deprotection of the allyl and Fmoc protecting
groups was followed by intramolecular amide bond
Figure 27. Solid-phase synthesis of the norlanthionine

analogue of the nisin C-ring.356
formation on the resin. Grieco and co-workers357 recently reported a variation on this concept. On-resin
cyclization of homologated lanthionines with varying
ring sizes was achieved by intramolecular amide
bond formation. A method for the synthesis of thioether-bridged peptides that yields diastereomerically
pure products was also developed by Yu and coworkers.358 The tert-butyldimethylsilyl (TBDMS) ether
of homoserine and tert-butyl disulfide protected Cys
were included in a 14-mer peptide by linear solidphase synthesis (Figure 28). Reaction with triphenylphosphine dichloride led to the conversion of the
homoserine to the corresponding chloride. Cleavage
from the resin and removal of the protecting groups
was followed by base-induced cyclization to afford a
homolanthionine-containing peptide. In addition to
the synthetic use of Lan and its analogues, dehydro
amino acids are also valuable synthons for further
manipulation when incorporated into peptides as
they constitute an electrophilic site for site-selective
ligation with external nucleophiles.359-361
Aside from preparation of lanthionine containing
structures by chemical synthesis, chemical modification strategies have been applied to natural lantibiotics. The single glutamic acid in actagardine (Figure
9) was converted selectively into a series of monocarboxamides in addition to variants that contained
amide functionalities at both Glu11 and the Cterminal carboxylate.362 Some of these semisynthetic
analogues displayed improved solubility and antibacterial activity.
Figure 28. Solid-phase synthesis of cyclic homolanthionine containing peptides.
8. Mode of Action
It has become clear in recent years that the mechanisms by which lantibiotics exert their antimicrobial
activities are more complex than initially thought.
For several type AI compounds, antibiotic activity
stems from more than one mechanism and may
include disruption of cell wall biosynthesis, inhibition
of spore outgrowth, and pore formation that may or
may not be aided by prior docking on cellular targets.
Nisin, for instance, and presumably also its close
structural analogues, uses all of the above modes of
action with the individual contributions depending
on the target organism. The currently known details
for these processes are discussed below.
8.1. Pore Formation in Model Membranes

Much of the work prior to the late 1990s focused
on the permeabilization of bacterial cell membranes
as the primary mode of action of nisin and other type
AI lantibiotics. Efflux of cellular components from
Clostridium butyricum upon treatment with nisin
was observed as early as 1960.363 Pore formation is
a widespread property of antibacterial peptides,364
and generally they are not expected to interact with
a specific chiral receptor in the target organism. Indeed, the observation that peptides consisting of all
D-amino acids provide the same activity as their
L-amino acid counterparts for the nonlantibiotics
magainin, cecropin, and melittin provided strong evidence for this supposition.365,366 As a result, numerous
studies focused on using cytoplasmic and artificial
membrane vesicles to study the mode of action of
Chatterjee et al.
nisin82 and other cationic type AI lantibiotics. Sahl

and Ruhr reported that the cytoplasmic membrane
is the primary target of nisin and that membrane
disruption results in efflux of metabolites and dissipation of vital ion gradients.367,368 Subsequent studies showed that membrane depolarization occurs in
a voltage dependent manner upon treatment with
nisin,367-372 subtilin,373 Pep5,374,375 epidermin,376 gallidermin,376 and streptococcin A-FF22.377 The threshold potential for depolarization of black lipid membranes (planar lipid bilayers) was in the 50-100 mV
range and required a trans-negative orientation (inside negative) with respect to the addition of nisin.
The higher the content of anionic phospholipids the
lower the threshold potential for dissipation of .376
It has been proposed that the transmembrane potential aids in nisins pore formation by effectively
pulling the charged amino acids at the N-terminus
into the membrane,82 a hypothesis that is supported
by the pore-forming activity of the Lys12Leu mutant
that is independent of a membrane potential.378 However, several studies have shown that a membrane
potential, while increasing membrane permeabilization, is not absolutely necessary for nisin-induced
leakage of certain solutes from negatively charged
vesicles.371,379-381 Moreover, nisin has been shown to
dissipate a transmembrane pH gradient in the absence of a transmembrane electrical potential in
sensitive Lactococcus cells and proteoliposomes.382
The pores that are assembled in the presence of a
membrane potential have been studied in detail and
are transiently formed with lifetimes of a few to
several hundred milliseconds and with diameters
ranging from 0.2 to 1-2 nm.368 Nisin and Pep5 form
pores that work only in one direction (rectifying)369,374
with the nisin pores somewhat anion selective,381
whereas gallidermin and epidermin form nonrectifying channels that are also more stable.376 The lipid
composition has a strong influence on the efficiency
of pore formation in model membrane systems such
as planar lipid bilayers and liposomes,370 indicating
a preferential interaction of nisin with anionic lipids.372,375,378,379,381,383,384 These were important findings
since Gram-positive bacteria have a high content of
anionic lipids in the membrane. Indeed in a monolayer study, it was shown that antimicrobial activity
correlated well with the nisin-anionic lipid interaction.383 In addition to the charge state of the membrane, several studies have reported an optimal
charge state for the nisin molecule itself.380,385 Efflux
of metabolites and depletion of the proton motive
force have also been demonstrated in other studies
focusing on the interactions of lantibiotics with artificial or bacteria-derived membranes.44,367,368,373,382,386-388
Two models have been proposed for the mechanism
of pore formation, the barrel-stave and the wedge
model. In the former, the cationic lantibiotic monomer (the stave) binds to the membrane surface
through electrostatic attraction, and after assembly
into a preaggregate, pores (barrel) are formed at a
certain membrane potential in which the lantibiotic
assumes a position perpendicular to the membrane.368 In the wedge model, surface bound nisin
molecules bind parallel to the membrane and produce
local strain leading to bending of the membrane such

that the lipid molecules together with nisin make up
the pore.372 In both models, the hydrophobic regions
of the amphiphilic nisin molecules presumably face
the membrane, whereas the hydrophilic face forms
the lumen of the pore. Although both models expected
a more or less perpendicular orientation of the molecule with respect to the membrane surface, fluorescence studies on antimicrobially active variants in
which tryptophans had been introduced at various
positions by mutagenesis379,389 suggested that nisin
adopts an overall stable parallel orientation.389 At
least three regions of the peptide were shown to
insert into membranes made up of anionic lipids with
the N-terminus inserted more deeply than the Cterminus. These results are similar to an assembly
of NMR structures obtained in DPC micelles (Figure
7).131,390 Combining the results of the fluorescence and
NMR studies, a model was proposed for nisins orientation in negatively charged membranes.389 In this
picture, nisin is relatively elongated and lies parallel
to the membrane surface with the positively charged
side-chains of lysines 12, 22, and 34 pointing out of
the lipid bilayer. Whereas this model represents the
most stable orientation, formation of the transient
pores most likely requires a conformational change
that allows the molecule to traverse the membrane.
During the pore forming process, the highly positively charged C-terminus of nisin (Lys22, Lys34,
His31) interacts initially with the anionic membrane
surface as demonstrated by binding studies, which
showed strongly reduced interaction of anionic lipids
with a Val32Glu mutant of nisin Z.381 Presumably,
the negative charge introduced in this mutant results
in electrostatic repulsion with the anionic headgroups
of the lipids. The weak binding also resulted in a
greatly reduced ability to release solutes from vesicles.
The importance of the C-terminus for binding to the
membrane was also shown by the strongly reduced
affinity of a nisin[1-12] fragment for anionic phospholipids.382 The binding isotherms with wt-nisin Z
and negatively charged membranes show biphasic
character, consistent with aggregation in the membrane.381 Although the initial binding to the membrane surface seems to involve the C-terminus of
nisin, studies with a variant of nisin Z in which a
short peptide (AspHis6) was fused to its C-terminus
showed that the C-terminus translocates across the
membrane, where it could be cleaved off behind
Lys34 by trypsin encapsulated in the lumen of
unilammellar vesicles.385 This translocation of the
C-terminus was correlated with pore-forming activity
and both activities were anionic lipid dependent.
Thus, after electrostatic binding of the C-terminal
region to the membrane surface, the peptide adopts
a membrane spanning orientation in which at least
part of the molecules that form the pore have their
C-terminus in the lumen of the vesicle. Alternatively,
some of the nisin molecules may translocate completely across the membrane as has been shown for
other nonlantibiotic pore-forming peptides such as
magainin 2.391 Several studies have been performed
to analyze the pore-forming capabilities of lantibiotic
mutants prepared by protein engineering (section
7.1). In these studies the antibacterial activity of

nisin is mainly affected by changes in the first three
rings of nisin,136,331,335,339 which is now believed to
originate from disrupted interactions with lipid II
(section 8.2).
8.2. Highjacking of Lipid II for Pore Formation

The in vitro studies using model membrane systems have provided much information regarding
their interaction with nisin and other lantibiotics, but
many observations could not be explained. For instance, nisins efficacy against intact cells (nM MIC)
was 3 orders of magnitude higher than pore formation in model membranes (M). Furthermore, although nisin[1-12] did not bind to anionic membranes, this fragment does antagonize nisins action
against intact cells.392 Other unexplained issues
involved nisins spectrum of biological activity, for
instance, L. monocytogenes has a high content of
anionic lipids (50-88%), and yet it is relatively
insensitive to nisin.393 Moreover, it was unclear why
mammalian membranes are only affected at millimolar concentrations of nisin.367,394 These apparent
discrepancies were explained when it was reported
by Breukink and Sahl and co-workers that nisin
interacts in a highly specific manner with lipid II
(undecaprenyl-pyrophosphoryl-MurNAc-(pentapeptide)-GlcNAc),41,322 the essential membrane bound
precursor for cell wall formation (Figure 29) that is
present in very different amounts in various microorganisms (e.g., E. coli 2000 molecules per cell,395
Micrococcus lysodeikticus 105 molecules per cell).396
In retrospect, several indications of nisins interference with peptidoglycan biosynthesis had been reported prior to the unequivocal demonstration that
it binds to lipid II. Linnet and Strominger showed
that nisin interferes with peptidoglycan biosynthesis
in an in vitro system made up of isolated bacterial
membranes.397 A subsequent study provided evidence
that this inhibition is caused by binding to the lipidassociated peptidoglycan precursors lipid I (lacking
the GlcNAc unit) and lipid II.398 For many years,
these early findings were not followed up on until a
series of recent studies that have firmly established
the interaction of nisin with lipid II.41,48-52,322 Mersacidin,399-401 epidermin,322 actagardine,167,322 and
probably lacticin 3147112 also inhibit peptidoglycan
biosynthesis. For mersacidin, epidermin, mutacin
1140,54 and actagardine, complex formation with lipid
II has been demonstrated, although unlike nisin,
mersacidin does not form pores. An NMR study on
the interaction of mersacidin with lipid II in SDS
micelles showed that it adopts a different structure
than that observed in the absence of lipid II.171
Interestingly, epidermin and nisin, but not mersacidin, have both been shown to cause an accumulation
of lipid I during in vitro peptidoglycan biosynthesis
assays, suggesting they may also interfere with the
conversion of lipid I into lipid II.322
Binding of antimicrobial substances to lipid II
interferes with peptidoglycan biosynthesis by physically sequestering the compound preventing utilization by transpeptidase and transglycosylase enzymes
that install the cross-linked network of the bacterial
Chatterjee et al.
Figure 29. The structure of lipid II and its incorporation into the peptidoglycan by transpeptidase and transglycosylase
enzymes. Lipid II is made up of an N-acetylglucosamine--1,4-N-acetylmuramic acid disaccharide connected to a C55-lipid
carrier undecaprenylpyrophosphate made up of eight Z-prenyl and three E-prenyl units.413 The muramic acid bears a
pentapeptide at O3 that contains a Lys for later cross-linking (or a meso-diaminopimelic acid in Gram-negative bacteria).
Bonds made by the transglycosylase are shown in red.
cell wall (Figure 29).402 Binding to lipid II is the

primary mode of action of the well-studied glycopeptide antibiotics vancomycin and teicoplanin,403,404 as
well as ramoplanin405-408 (see also the review by
Walker et al in this issue).409 As would be expected
for a common target, treatment of M. luteus with
ramoplanin prevented pore formation by nisin and
epidermin,322 and strongly decreased mersacidin
binding to the cell surface.401 Similarly, vancomycin
protected M. flavus cells against membrane leakage
induced by nisin but not by magainin.41 The lantibiotics must bind to a different part of lipid II than
vancomycin, which interacts with the L-Lys-D-Ala-DAla segment of the pentapeptide,403,404 since nisin and
mersacidin are active against vancomycin resistant
enterococci.41,399 Comparing existing NMR structural
data,169,410 McCafferty and co-workers suggested that
mersacidin and ramoplanin exhibit very similar
backbone conformations, which may be important for
their mode of binding lipid II.411 Binding of an
antibiotic to a complex biosynthetic intermediate like
lipid II has certain advantages over binding to a
single enzyme involved in peptidoglycan assembly
because changing the structure of lipid II is much
more demanding on a microbe than changing the
structure of the active site of one enzyme, thereby
decreasing the odds of bacterial resistance. For
instance, eight successive enzymes are required for
the biosynthesis of lipid II from UDP-GlcNAc.412,413
That resistance can nevertheless develop has been
demonstrated in vancomycin-resistance when bacteria change the D-Ala-D-Ala unit of lipid II to D-Alalactate.414,415 For resistance mechanisms to nisin, see
section 9.
Although nisin, vancomycin, teicoplanin, and ramoplanin interact with the same target, nisin is
unique in that it subsequently forms pores that
include lipid II as an essential constituent.51 For
instance, when lipid II is present in membranes,
nisins pore-forming efficiency is increased 1000-fold,
an increase that is not seen with other pore forming
peptides such as magainin.41 Interestingly, the afore-
mentioned increased efficiency of pore formation in

membranes made up of negatively charged lipids
observed in the absence of lipid II was completely
undone in its presence and the anion selectivity seen
in model systems disappeared upon addition of lipid
II.48 These observations suggest that in the presence
of lipid II the architecture of the pores is changed.
Lipid II-mediated pore formation is very effective as
only seven molecules of lipid II per vesicle (or 2 lipid
II per 105 phospholipid molecules) result in a dramatic decrease in the nisin concentrations needed to
release dyes from vesicles.41,322 The stoichiometry of
complex formation between nisin and lipid II is 1:1
in solution and in SDS micelles.48,50,54
The ability to produce lantibiotic mutants by sitedirected mutagenesis or chemical and enzymatic
truncation as described in section 7.1 (Figure 25) has
proven an extremely powerful asset in the study of
their modes of action. Deletion of the two C-terminal
amino acids (Dha33, Lys34) of nisin does not affect
antimicrobial activity.416 Mutation of Val32 to Lys or
Glu, thereby introducing an additional positive or
negative charge and also preventing dehydration of
Ser33,273 also had relatively little effect for the
Val32Lys mutation whereas the Val32Glu mutant
had about 4-fold decreased activity against certain
test strains.48 Hence, the C-terminus is relatively
unimportant for biological activity. This is in keeping
with the observation that epilancin K, which shares
a very similar C-terminal double-ring system with
nisin, does not appear to interact with lipid II.322 On
the other hand, several findings strongly suggested
that the N-terminus of nisin is essential for binding.
An inactive nisin fragment (nisin 1-12) was shown
to antagonize nisins bacteriocidal activity, suggesting
it competes for the same binding site.392 Furthermore,
epidermin and nisin both bind to lipid II,322 and they
have identical topologies of their A- and B-rings
(Figure 6). Moreover, chemical disruption of Dha5,
thereby opening the A-ring, results in more than 500fold reduction of biological activity,416 whereas complete removal of the D- and E-rings by proteolysis
Figure 30. Depiction of the cage-like structure of the Aand B-rings of nisin around the pyrophosphate group of
lipid II. The lipid II fragment containing the muramic acid
and pyrophosphate is shown in ball-and-stick format. The
A-ring is shown to the left of the pyrophosphate and the
B-ring is below the pyrophosphate.
results in only a 100-fold decrease in potency.392

Recent NMR studies on nisin in the presence of lipid
II provide the most direct support for the importance
of the N-terminus for binding. In SDS micelles
containing lipid II, nisin residues located in the
N-terminal region showed the largest perturbation
in chemical shift.50 A very recent exciting development is the determination of the solution structure
of a complex between nisin and a lipid II analogue.
The lipid II variant contained a shortened prenyl tail
that consisted of just 3 rather than 11 isoprene units.
The length of this isoprene tail does not affect poreforming activity.417 The structure of the 1:1 complex
was solved in DMSO revealing a cage-like binding
motif in which the N-terminus of nisin folds back onto
the A- and B-rings.54 The pyrophosphate moiety of
lipid II is bound within this cavity with intermolecular hydrogen bonds between the amides of Dhb2,
Ala3, Ile4, Dha5, and Abu8 on nisin and the oxygens
of the pyrophosphate group of lipid II (Figure 30). In
addition to the pyrophosphate group, MurNAc and
the first isoprene unit make up the binding site for
nisin recognition (Figure 31). This structure provides
explanations of various previous observations including studies with nisin mutants. Extension of its
N-terminus by the tetrapeptide IleThrProGln (Figure
25) significantly reduces the antimicrobial activity.204
Furthermore, it is in full agreement with the observations that chemical opening of the A-ring essentially eliminates all activity,416 and that changing
the Lan in the A-ring to a MeLan decreases the
affinity of the mutant nisin for lipid II 50-fold.48 The
NMR structure also reveals why nisin (and epidermin) binds to both lipid I and lipid II,322 which share
the recognition motif, and why nisin (and epidermin)
is active toward vancomycin resistant strains41,399 as
its binding site (the L-Lys-D-Ala-D-Ala segment of the
pentapeptide) on lipid II does not make contacts with
nisin. The structure also provides an important
lesson that it is not the side-chains on nisin that
impart its biological activity but rather the backbone.
Figure 31. Two views of the NMR structure in DMSO of

the 1:1 complex of nisin and a lipid II analogue with a
truncated prenyl tail.54 Figure generated using the program
RASMOL469 with PDB file 1UZT. Nisin is shown in cyan,
the prenyl chain of lipid II is shown in orange, the two
phosphate atoms of the pyrophosphate group are depicted
in white, and the muramic acid is presented in red. The
pentapeptide chain is shown in yellow and the GlcNAc in
magenta.
Hence, it rationalizes why to date no significantly

improved analogues have been reported in the bioengineering efforts (section 7). Whether this will also turn
out to be true for other lantibiotics remains to be
established.
The most interesting nisin mutants have changes
in the flexible segment between the C- and D-rings
(hinge region). Deletion of Asn20 and Met 21,
thereby decreasing the length of this linker to a single
residue, resulted in an 80% decrease of dye release
from lipid II-supplemented vesicles even at high nisin
concentrations (M).48 Replacement of these two
amino acids with prolines decreased the lipid II
promoted pore formation even further. The antimicrobial activities of these mutants against Streptococcus thermophilus were also strongly affected with
MIC values that increased 40- and 25-fold, respectively. However, their activity against Micrococcus
flavus was much less affected (3-9-fold), and a third
mutant, Met21Gly, showed comparable activity as wt
nisin Z against both strains despite having no pore
Chatterjee et al.
Figure 32. Proposed model for lipid-II mediated pore formation. The C-terminus of nisin is shown as residing in solution
upon initial binding of the N-terminus to lipid II, but it may also interact with the negatively charged headgroups of the
membrane.54 The pore structure has been shown to be made up of four lipid II and eight nisin molecules, but their
arrangement is not known and the shown architecture is therefore speculative.
forming activity. Hence, the binding to lipid II alone

constitutes a very potent antimicrobial activity in
vivo. Since pore formation by itself is also bactericidal, nisin truly has two distinct modes of action.
Using tryptophan fluorescence, it was demonstrated that lipid II changes the orientation of nisin
from parallel to perpendicular with respect to the
membrane surface.49 Furthermore, the use of pyrenelabeled lipid II showed that it is recruited into a
stable pore structure.51,53 The distance between two
labeled lipid II molecules was estimated to be about
18 51 with a pore diameter of 2 nm.52 A model has
been proposed to explain all of the experimental data
as shown in Figure 32.48,49,53 The N-terminal rings
of nisin bind to the disaccharide-pyrophosphate
region of lipid II, whereas the positively charged
C-terminus initially interacts with the headgroups
of the lipids in the membrane bilayer (not shown).
Multiple molecules of the lipid II-nisin complex51
subsequently aggregate and form a pore of defined
uniform structure. Whereas the stoichiometry of lipid
II to nisin in solution is 1:1,48,54 the stoichiometry in
the pore is 1:2 as it is made up of four lipid II and
eight nisin molecules.53 How the different stoichiometry in the pore affects the structure of the lipid IInisin complex that was determined by NMR spectroscopy (Figures 30 and 31) remains to be established.
However, a promising observation with respect to
establishing the structure of the pore at atomic
resolution involves the solubilization of the pore
complex using detergents such as n-octyl polyoxyethylene, n-octyl--D-glucopyranoside, and Tween 20
without any changes in the CD spectrum.53 The
stability of the pore complex has also been observed

in electrophysiology experiments. The lifetime of the
pore in the presence of lipid II is greatly enhanced
from milliseconds in its absence to seconds in its
presence.52 Furthermore, the threshold potential for
permeabilization was decreased from 100 mV in
the absence of lipid II to as low as 5-10 mV in its
presence, and unlike the studies discussed in section
8.1, when lipid II was present pores were formed by
applying either trans-negative or trans-positive membrane potentials.52 The high stability of the pore
complex is quite unique as other cationic pore forming antimicrobial peptides typically form transient
pores of low stability and without uniform structures.391,418
Mersacidin and actagardine also bind to lipid II,
but have no structural similarity with nisin and
epidermin (Figures 6 and 9). In doing so, these
compounds block the transglycosylation step of peptidoglycan biosynthesis (Figure 29).400 As mentioned
previously, mersacidin does not form pores once
bound to lipid II, which may explain the moderate
MIC values. However, the compound is very effective
in vivo against systemic staphylococcal infections,16,60
including methicillin resistant S. aureus (MRSA),473
as well as against vancomycin resistant enterococci.400 Hence, mersacidin and its close relative
actagardine may have potential chemotherapeutic
applications.
Except for nisin, epidermin, mutacin 1140,54 and
mersacidin, no other lantibiotics have been demonstrated to bind to lipid II, and some such as Pep5
and epilancin K7 have been specifically shown not
to interact with lipid I or lipid II.322 However, these

compounds still have activities against certain bacteria that are far greater than those of other pore
formers. For instance, Pep5 is active in low nanomolar concentrations against Staphylococcus simulans
and S. carnosus, which may indicate that it uses a
different high-affinity receptor/docking molecule for
its biological activity.322,327
8.3. Inhibition of Spore Outgrowth

In addition to their bactericidal activity against
vegetative cells, nisin, subtilin and sublancin also
inhibit the germination of spores from Bacillus and
Clostridium species.22,105 For nisin and subtilin, this
activity has been proposed to result from covalent
modification of a target on the spore coat by nucleophilic attack on Dha5.138 Indeed, reactive thiol groups
on the exterior of spores from Bacillus cereus react
with reagents such as S-nitrosothiols or iodoactetate,419 and nisin interferes with the modification
of these sulfhydryl groups. This suggests that nisins
target during its inhibition of spore outgrowth may
be these reactive thiol groups,420 but to date a
covalent mechanism has not been established. In
both subtilin137,138 and nisin,136 Dha5 has been implicated as the putative site of attack as replacement
by Ala via site-directed mutagenesis abolishes the
inhibition of spore outgrowth. Interestingly, these
mutations did not affect the growth inhibition activity
of subtilin and nisin against vegetative cells of
Bacillus cereus, and L. lactis and M. luteus, respectively. These studies clearly indicate that the inhibition of spore outgrowth is yet a different, third
distinct biological activity of these compounds with
a different structure-activity relationship. Given the
recent use of spores of Bacillus anthracis in bioterrorism and the similarities of the structure of the
spore coat in Bacillus species, these activities of
subtilin and nisin may have interesting future applications.
8.4. Other Biological Activities

In addition to having bactericidal and hemolytic activity,175 cinnamycin and the duramycins are potent
inhibitors of phospholipase A2 by sequestering its
phosphatidylethanolamine (PE) substrate.57,109,181,421,422
When phosphatidylcholine or other lipids are the
substrates of phospholipase, no inhibition and substrate binding is seen,57,175,421,423 indicating specific
recognition of PE. The binding stoichiometry deduced
from NMR investigations is 1:1181 and the binding
constant as determined by isothermal titration calorimetry is dependent on the lipid matrix, being 107108 M-1 in 1-palmitoyl-2-oleyl-sn-glycero-3-phosphocholine (POPC) and 106 M-1 in octyl glucoside
micelles.422,424 Since phospholipids such as PE reside
in the inner leaflet of plasma membranes, it was
unclear how cinnamycin accesses its target. A recent
study addressed this issue and demonstrated that the
compound induces transbilayer lipid movement, apparently in a PE-dependent fashion.425
Investigations of the conformation of cinnamycin
in SDS bilayers indicate the insertion of the lipo-
philic portion of the molecule into the bilayer due to

its interaction with the methylene (-CH2-) groups of
the lipid.179,180 Conformational changes in the presence of 1-dodecanoyl-sn-glycerophosphoethanolamine
(C12-LPE)180,340 in DMSO were induced primarily by
ionic interactions between the -hydroxy-Asp15 of
cinnamycin and the ammonium ion of glycerophosphoethanolamine, as well as via hydrophobic interactions
between the lipophilic portion of cinnamycin (Gly8,
Pro9, Val13) and the methine/methylene backbone
of the headgroup. A three-dimensional 1H NMR
structure of cinnamycin bound to 1-acetyl-sn-glycerophosphoethanolamine (C2-LPE) shows the formation of a rigid cylindrical complex, 11 in diameter, and 26 in length.340 The glycerophosphoethanolamine headgroup resided in the hydrophobic
pocket formed by residues Phe7 through Ala(S)14 of
cinnamycin. The high specificity of cinnamycin was
attributed to the limited space in the hydrophobic
pocket that prevents binding of larger molecules.340
In this study, C2-LPE and C12-LPE were thought
to bind cinnamycin in a fashion independent of their
acyl groups. However, more recent isothermal titration calorimetry (ITC) studies of cinnamycin binding
to diacyl-phosphoethanolamine of varying chain
lengths in both micellar and membrane environments indicate an optimal acyl chain length of eight
methylene units for binding.422 At least one acyl
group is necessary for binding and diacyl-PE binds
more strongly than lyso-PE.422 This appears to argue
in favor of an interaction between the lipophilic
portion of cinnamycin and the hydrocarbon chains
of diacyl-PE, which was not observed by 1H NMR
spectroscopy.
In addition to their other activities, nisin and Pep5
induce autolysis of certain staphylococcal strains
leading to cell wall break-down at the septa between
dividing cells. It is thought that the interaction of the
cationic lantibiotics with the negatively charged
teichoic and lipoteichoic acids displace and activate
N-acetyl-L-alanine amidase and N-acetylglucosaminidase enzymes that are usually associated with the
teichoic acids.426,427
9. Resistance Against Nisin

Microbial resistance to antibiotics is brought about
by the evolutionary pressure exerted by exposure of
bacteria to these compounds and is eventually unavoidable.26,428 The advantage conferred by the dual
mechanism of action of nisin (section 8) has attenuated the emergence of bacterial resistance to nisin,
but resistance has been induced in laboratory settings. In this section, the various mechanisms of resistance to nisin will be discussed. For a discussion
of self-immunity of the producing strains see section
6.
9.1. Gram-Negative Bacteria

In 1973, Linnet and Strominger observed that nisin
and subtilin do not exert a significant bactericidal
effect on intact Gram-negative organisms, such as E.
coli, even though they inhibit peptidoglycan biosynthesis in cell-free assays.397 Nisin also displays no
effects on the growth of the Gram-negative organisms

Serratia marcescens, Salmonella typhimurium, and
Pseudomonas aeruginosa.429 The nonsusceptibility of
Gram-negative bacteria to nisin (and subtilin) has
been attributed to the outer membrane that prevents
access of hydrophobic substances to the peptidoglycan
layer. Therefore, agents that disrupt the outer lipopolysaccharide (LPS) rich membrane and allow
nisin to access the inner membrane where lipid II is
present increase the susceptibility to nisin. For
instance, sensitization to nisin has been achieved by
treatment with chemical agents such as EDTA430,431
and trisodium phosphate432 as well as by temperature
shock.433 In each of these cases, susceptibility was
transient and was lost upon restoration of normal
growth conditions. Incubation of nisin with various
mutant strains of E. coli and Salmonella enterica
containing truncated LPS demonstrated that resistance was generally reduced in the absence of the
O-chain,434 which has been reported to inhibit antibiotic action.435 Interestingly, Pectinatus frisingensis,
an anaerobic microorganism responsible for spoilage
of beer, is nisin-sensitive.436 Resistance to nisin (Nis5000 strain) was induced in the laboratory by stepwise exposure to increasing nisin concentrations.
Sensitivity to nisin could not be induced by treatment
of Nis-5000 with EDTA in contrast to nisin-insensitive Gram-negative organisms, suggesting that the
LPS layer is not responsible for resistance of Nis5000. Chihib and co-workers have suggested an
alternate mechanism of nisin-resistance in this strain
that involves rigidification of the cellular membrane
by changes in the fatty acid composition of the
cytoplasmic membrane. This proposal is based on the
observation of a 2-fold decrease in unsaturated fatty
acids and an increase in saturated fatty acids in the
cell membrane of Nis-5000.436
9.2. Gram-Positive Bacteria

Nisin resistance in Gram-positive bacteria is often
developed in sensitive strains by repeated exposure
to increasing amounts of nisin. Such induced resistance (in a laboratory setting) is typically a complex phenotype arising from changes in the bacterial
cell wall and/or cell membrane, and sometimes
requiring divalent cations.437 Plasmid-mediated resistance438,439 as well as resistance due to changes in
expression levels of proteins such as a putative
penicillin binding protein,440,441 and a two-component
signal transduction system442 have also been observed as discussed below.
Resistance to nisin in several Bacillus strains was
reported by Jarvis in 1967 and ascribed to the
presence of nonproteolytic nisin inactivating enzymes.443 This followed in the wake of an earlier
proposal of an enzyme, nisinase, present in the cell
extracts of Streptococcus thermophilus, that was
capable of inactivating nisin but not subtilin.444
Jarvis noted that the treatment of nisin or subtilin
with cell-free extracts of Bacillus cereus and Bacillus
polymyxa led to loss of their antibiotic activity,
whereas antibiotics such as gramicidin S, polymyxin
B, and bacitracin were not affected, suggesting the
presence of lantibiotic-specific agents.443 The inacti-
Chatterjee et al.
vating activity of the B. cereus cell-free extracts was

dependent on Ca2+, Co2+, and Mg2+ ions and was
inhibited by EDTA. Partial purification of the nisininactivating fraction was achieved, but an actual
protein was never isolated.445 Acid hydrolysis of
inactivated and wild-type nisin before and after the
addition of methylmercaptoacetate and 14C-labeled
cysteine demonstrated a reduction in the number of
dehydrated residues in inactivated nisin. Furthermore, partial hydrolysis of inactivated nisin showed
the presence of Ala-Lys as well as pyruvate-Lys
(presumably formed from Dha33-Lys34). On the basis
of these results a putative dehydroalanine reductase
role was suggested for the enzyme involved.445
Several instances of resistance to nisin have been
attributed due to changes in the composition of the
cell membrane.393,446-449 The dissipative action of
nisin on both components of the proton motive force
(pH and ) was significantly reduced in a resistance-induced strain (NISr) of L. monocytogenes Scott
A (L. monocytogenes ATCC 700302).393 An evaluation
of the content of phosphatidylglycerol (PG) and
diphosphatidylglycerol (DPG) in the NISr strain
revealed an increase in the PG/DPG ratio to 7, from
a ratio of 5 in the wild type (wt) strain. On the basis
of the persistence of this ratio even in the absence of
exposure to nisin, the increased ratio was suggested
to be caused by a decreased activity of the enzyme
DPG synthetase, although this has not been experimentally verified. The higher negative charge density
of DPG possibly contributes to stronger binding of
the cationic nisin, and a greater ratio of PG/DPG may
therefore increase resistance by reducing binding
affinity. Other reported changes in the membrane
composition of NISr L. monocytogenes Scott A are a
lower ratio of C15/C17 long-chain fatty acids, a
significantly greater amount of PE, and lesser quantities of the anionic phospholipid cardiolipin compared to the wt strain.437 These changes probably also
result in reduced binding of the cationic nisin to the
less negative cell membrane, and may also inhibit
pore formation by increasing membrane rigidity.
Several studies have also reported constitutive
compositional changes in the cell wall of NISr cells
that persist in the absence of nisin from the growth
medium. For instance, NISr L. monocytogenes cells
exhibit increased resistance to lysozyme action and
the antibiotics gramicidin S and gentamycin while
displaying increased sensitivity to the cell wall
targeting antibiotics benzylpenicillin and ampicillin.437 The degree of resistance to nisin is dependent
on the presence of divalent cations (Mg2+, Ca2+, Mn2+,
Ba2+) and their sequestering by EDTA leads to
increased nisin sensitivity. The molecular identity of
the changes to the cell wall remain to be determined,
but no thickening of the cell wall was observed by
transmission electron microscopy (TEM) of NISr L.
monocytogenes 700302 cells.437 A proposed model for
nisin resistance in L. monocytogenes ATCC 700302
is shown in Figure 33.
Nisin resistance in L. monocytogenes446,447 and C.
botulinum 169 B448 has also been related to increased
membrane rigidity arising from the presence of more
straight-chained, monounsaturated, and saturated
Figure 33. Proposed model of nisin resistance437 in L.

monocytogenes that includes changes in the cell wall,
cytoplasmic membrane, and the requirement for divalent
cations. (A) In wt strains nisin passes through the cell wall,
binds to the cytoplasmic membrane via electrostatic interactions with anionic phospholipids, and initiates pore
formation. (B) In NISr cells, the permeability of the cell
wall toward nisin is reduced; possibly due to increased
amounts of teichoic acid as well as the presence of increased
quantities of Ca2+ ions. Furthermore, a reduced negative
charge density of the cytoplasmic membrane in NISr cells
leads to weaker binding of nisin molecules. Finally, membrane bound Ca2+ ions may repel the positively charged
nisin molecules.
fatty acids and less branched, polyunsaturated fatty

acids. Similarly, nisin-resistant variants of the pediocin producer, Pediococcus acidilacti UL5 exhibited
a significant increase in the amounts of monounsaturated C16:1 and C18:1 fatty acids.450 Interestingly,
nisin-sensitive L. monocytogenes Scott A cells grown
in the presence of 0.1% of the surfactant Tween 20
underwent changes in the cell membrane that did
not affect membrane fluidity yet doubled the nisin
binding capacity and increased their sensitivity to
nisin.449 While this result holds promise in reducing
the growth of nisin-sensitive L. monocytogenes, a
parallel study of the effect of Tween 20 on NISr L.
monocytogenes Scott A cells has not been reported to
date.
Transmission electron microscopy of a NISr strain
of Listeria innocua did show a thickened cell wall that
was less hydrophobic than that of wt, and showed
an increased resistance to hydrolysis by lysozyme and
mutanolysin.451 The effect of antibiotics such as carbenicillin, vancomycin, and D-cycloserine on the
growth of log-phase cultures of NISr L. innocua was
also attenuated in comparison to the wt.451 It was

suggested that cell wall changes in the NISr variants
may be due to the displacement/inhibition of autolysin by nisin and the activation of murein synthesis.
A similar thickening of the cell wall was also observed in TEM experiments on NISr strains of Streptococcus thermophilus INIA 463 after incubation at
37 C with nisin for only 2 h.452 However, contrary
to most other cases in which resistance persists
even after nisin is removed from the growth medium,447,448,450,453 this NISr strain lost resistance after
a single transfer to nisin-free medium.452 The involvement of the cell wall in acquisition of resistance
to nisin has also been reported in L. monocytogenes
F6861.454 NISr and wt cells became equally sensitive
to nisin upon removal of the cell wall by treatment
with lysozyme. Similar to the observations with L.
innocua451 these NISr L. monocytogenes cells were
less hydrophobic than wt when measured by their
affinity for n-hexadecane and retention during hydrophobic interaction chromatography.454 This reduced hydrophobicity of the cell surface may lead to
reduced binding of the hydrophobic nisin and thus
increased resistance.
Laboratory induced NISr cultures of Streptococcus
bovis, an opportunistic bacterium that resides in the
rumen and causes ruminal acidosis in cattle455 and
is associated with colon cancer in humans,456,457
contained more ampicillin-resistant cells, were unaffected by lysozyme, were less hydrophobic, and bound
less (cationic) cytochrome c, than wt S. bovis.453 After
incubation with nisin, the cell-free supernatant from
the NISr culture depleted potassium from a second
batch of nisin-sensitive cells, suggesting the supernatant did not inactivate the antibiotic nor was it
sequestered by the NISr cells. In contrast, cell-free
supernatants from nisin sensitive cells grown in the
presence of nisin, did not exhibit any K+-depletion
activity, suggesting that nisin is adsorbed to a much
larger degree on the sensitive strain.453
A reduction of the net negative charge of the cell
envelope has also been suggested to confer increased
resistance to nisin in B. subtilis.458 Expression of the
enzymes involved in biosynthesis of PE as well as
the D-alanylation of lipoteichoic (LTA) and wall
teichoic acids (WTA) is controlled in part by the
extracytoplasmic-function (ECF) X factor. Esterification of the glycerol moieties of teichoic acids with
D-alanine introduces free amine (-NH2) groups into
the cell wall and leads to a reduction in negative
charge. Similarly, the increased content of the zwitterionic PE molecule in place of anionic phospholipids
lowers the overall negative charge. Both the dlt
operon and pssAybfMpsd operon, which encode the
enzymes responsible for D-alanylation and PE biosynthesis, respectively, are preceded by promoters
recognized by the ECF X factor. Analysis of sigX,
dltA, pssA, and psd mutants showed an increased
sensitivity to nisin for all except the pssA mutant.458
The authors suggested that extracellular conditions
leading to activation of the X regulating factor could
enhance expression of the dlt operon. The direct
involvement of the dlt operon in resistance of Staphylococcus xylosus C2a and S. aureus Sa113 to the
lantibiotics nisin and gallidermin was demonstrated

by Peschel et al.459 Mutations of the dltA gene in S.
aureus by homologous recombination, and of dltA,
dltB, and dltD genes in S. xylosus by transposon
insertion resulted in no detectable incorporation of
D-alanine in either LTA or WTA of the mutant
strains. These dlt mutants displayed an 8-50-fold
increased sensitivity toward the cationic antimicrobials nisin and gallidermin, while no significant changes
in sensitivity toward the neutral peptide gramicidin
D was observed.459 The effect of preventing D-alanylation on the overall charge of the cell envelope has
been determined by the affinity toward the cationic
molecules cytochrome c and gallidermin and the
anionic green fluorescent protein (GFP). The dlt
mutants bound smaller quantities of GFP and greater
amounts of cytochrome c and gallidermin than the
wt strains, suggesting an increased overall negative
charge on the cell surface. D-Alanylation of LTA and
WTA was regained in the dlt mutants by expression
of the DltABCD proteins from the plasmid pRBdlt1.
The dlt mutant strains bearing the pRBdlt1 plasmid
regained wt-like resistance to nisin and gallidermin,
and wt S. aureus and S. xylosus transformed with
pRBdlt1 displayed a 1.4-1.6-fold increased resistance
to the two antibacterials. That this increased resistance was indeed due to increased D-alanylation of
LTA and WTA was confirmed by measuring the
molar ratio of D-alanine to phosphorus in the mutant
and wt strains, before and after the introduction of
pRBdlt1. Thus, reduction of the negative charge of
the cell envelope by esterification of LTA and WTA
with D-alanine in Gram-positive Staphylococcus and
Bacillus strains is one of the better understood
mechanisms of resistance to nisin.
Nisin binds to lipid II prior to formation of a nisinlipid II pore complex in the Gram-positive cell membrane (section 8). Kramer et al. recently investigated
the relation between resistance to nisin and the amount of membrane associated lipid II in M. flavus,
L. monocytogenes and their isogenic NISr variants.460
No significant differences were observed in the maximal amount of lipid II in the membranes of NISr
variants of M. flavus and L. monocytogenes and the
wt strains,460 indicating that resistance to nisin in
these strains is not related to lipid II levels. The authors observed that spheroplasts of NISr M. flavus
that lack the cell wall showed greatly enhanced dissipation of the membrane potential in the presence
of even 10 nM nisin, which had no detectable effect
on the intact NISr cells. This indicates that nisin resistance, at least in M. flavus, is determined by
changes in the cell wall and is independent of lipid
II levels.
Gravesen and co-workers have studied nisin resistance in L. monocytogenes 412 by analyzing changes
in gene expression in a spontaneous mutant using
restriction fragment differential display (RFDD).440
A 2-4-fold increase in MIC of nisin toward L.
monocytogenes 412N was associated with the increased expression of genes encoding a protein homologous to the glycosyltransferase domains of highmolecular-weight penicillin binding proteins (PBPs)
(pbp2229), a histidine protein kinase (hpk1021), and
Chatterjee et al.
an unknown protein (lmo2487).440 Gene disruption

studies demonstrated that PBP2229 and HPK1021
are directly involved in imparting nisin resistance
while LMO2487 levels did not affect nisin sensitivity.441 Moreover, the expression of PBP2229 was
dependent on the expression of HPK1021, and may
be controlled by a two-component signal transduction
system that includes HPK1021.441 Gravesen et al.
also observed a 1.8-fold increase in DltA expression
in L. monocytogenes 412N that might confer resistance by the same mechanism as reported for S.
aureus by Peschel and co-workers discussed previously.441,459 The authors proposed that nisin resistance in 412N is probably due to shielding of lipid II
from nisin through its binding to PBP2229.
The two-component signal transduction system
LisRK has been implicated in sensitivity of L. monocytogenes LO28 toward nisin and cephalosporins.442
Inactivation of the histidine-kinase LisK (not identical to HPK-1021 in L. monocytogenes 412N) led to
an increased resistance toward nisin. The direct role
of the regulatory protein LisR in sensitivity to nisin
was demonstrated by its overexpression in the
LO28lisK strain either constitutively or via nisin
controlled expression (NICE461). High level induction
of lisR could overcome the lisK mutation and impart
sensitivity to nisin.442 The precise nature of the gene
products controlled by the LisRK system is not
known, however, similar to the observations of Gravesen et al., they include proteins that bear homology
to a histidine kinase, a PBP, and a protein of
unknown function. Interestingly in the LO28 strain,
deletion of the lisK gene led to a greatly reduced
transcription of these genes as well as increased nisin
resistance,442 which is in contrast to the observations
of Gravesen et al for the 412N strain.440 The difference between the two studies may be due to different
growth phases of the cells that were studied.
Nisin resistance in the nisin nonproducer L. lactis
subspecies diacetylactis DRC3 is associated with a 60
kbp plasmid, pNP40462 containing a nisin resistance
(nsr) gene. Homologous sequences to the nisin resistance protein (predicted mass 35 kDa) were not detected in the nisin producer L. lactis 11454, nor were
homologous sequences found in the NBRF/PIR and
SWISS-PROT data banks at the time of publication
of this article (1991). Interestingly, a current BLAST
alignment (November 2004) reveals the presence of
a C-terminal conserved tail-specific protease domain
(residues 107-303). Tail-specific proteases are endopeptidases that bind their polypeptide substrates
at a nonpolar C-terminus prior to proteolysis.463,464
Whether this resistance protein actually causes proteolysis of nisin itself remains to be established since
its C-terminal six amino acids include Lys, His, and
Ser.
10. Summary and Outlook

The posttranslational modifications involved in
lantibiotics are unique in Nature and are essential
for their biological activities. After the pioneering
structural studies, the past decade has seen the
accumulation of a wealth of information about their
biogenesis from genetic studies. The biochemical
characterization of the proteins involved is only just

starting to be explored and with the successful
reconstitution of the dehydration, cyclization, and
oxidative decarboxylation reactions the near future
is likely to unveil many new exciting aspects of
enzymatic catalysis. In addition, the relaxed substrate specificity of the biosynthetic machinery will
see continuing exploration in vivo and in vitro to
engineer lantibiotic variants. These will provide
powerful tools to investigate the molecular mode(s)
of action of lantibiotics that may result in more
effective antimicrobials. The ubiquitous use of the
cyclic lanthionine structural motif by Gram-positive
organisms to build highly active compounds of very
diverse three-dimensional structure argues that it is
a natural privileged structure for constraining bioactive peptides. A related motif was recently determined in subtilosin A, a cyclic thioether containing
nonlantibiotic posttranslationally modified antimicrobial peptide in which a Cys sulfur is cross-linked
to the R-carbon of Phe and Thr.465,466 This example
further illustrates Natures use of the thioether crosslink to achieve a stable constrained conformer of a
bioactive peptide. Certainly, the recent revelations
of the recognition of molecular targets such as lipid
II and phosphatidyl ethanolamine with very high
affinities and specificities suggest that these structures are excellent starting points for design of
compounds with biological activities. Whether other
lantibiotics may have other specific molecular targets
is an open question that will likely be actively
investigated in years to come.
The low substrate specificity of the lantibiotic
biosynthetic machinery also bodes well for its use in
engineering of molecular architectures that are unrelated to lantibiotics. The recent in vivo studies by
Kuipers et al.195 as well as unpublished in vitro
results from our laboratory shows that nonlantibiotic
prepeptides fused to the leader peptide are substrates
for dehydration. Such installation of dehydroamino
acids provides orthogonal electrophilic handles for the
chemoselective and site-specific introduction of a wide
array of functionalities including biophysical probes,
thiosaccharides, and prenylthiols.359-361 Furthermore,
it may prove possible to prepare lanthionine analogues of natural cyclic peptides with impoved biological activity and/or stability using the biosynthetic
enzymes rather than traditional synthetic chemistry.
Clearly, the future of lantibiotic research and application of their biosynthetic machinery holds great
promise.
11. Abbreviations
ABC
Abu
Agr
Allo-Ile
Aloc
ATCC
ATP
AviCys
AviMeCys
ATP-binding cassette
l-R-aminobutyric acid
accessory gene regulator
allo-isoleucine
allyloxycarbonyl
American Type Culture Collection
adenosine triphosphate
S-[(Z)-2-aminovinyl]-D-cysteine
S-[(Z)-2-aminovinyl]-(3S)-3-methyl- D -cysteine

BLAST
Boc
C12-LPE
CBD
CD
Dhb
Dha
DMSO
DPC
DPG
DTT
ECF
EGF
EPL
ESI-MS
EXAFS
FAB-MS
FAD
FMN
Fmoc
FTMS/MS
GC
GFP
HFCD
Basic Local Alignment Search Tool

t-butyl carbamate
dodecanoyl-sn-glycerophosphoethanolamine
chitin binding domain
circular dichroism
dehydrobutyrine
dehydroalanine
dimethyl sulfoxide
dodecylphosphocholine
diphosphatidylglycerol
dithiothreitol
extracytoplasmic-function
epidermal growth factor
expressed protein ligation
electrospray ionization mass spectrometry
extended X-ray absorbance fine structure
fast atom bombardment mass spectrometry
flavin adenine dinucleotide
flavine mononucleotide
9-fluorenylmethoxycarbonyl
Fourier transform tandem mass spectrometry
gas chromatography
green fluorescent protein
homooligomeric flavin-containing Cys decarboxylase
HIF
hypoxia-inducible factor
HPLC
high performance liquid chromatography
ICP-MS
inductively coupled plasma-mass spectrometry
Lan
lanthionine
LanA
generic designation for precursor peptides for
lantibiotic biosynthesis
LanB
generic designation for dehydratases
LanC
generic designation for cyclases
LANCL
LanC-like protein
LanE
generic designation for component of ABC
transport protein involved in self-immunity
LanF
LanG
LanI
generic designation for lantibiotic immunity
proteins
LanK
generic designation for lantibiotic receptor
histidines kinase
LanM
generic designation of bifunctional enzymes
catalyzing both dehydration and cyclization
reactions
LanP
generic designation of proteases that remove
the leader peptides
LanR
generic designation for lantibiotic response
regulator protein
LanT
generic designation of ABC transporters that
excrete lantibiotics after biosynthesis
LPS
lipopolysaccharide
LTA
lipoteichoic acid
MALDI-MS matrix-assisted laser desorption/ionization
MBP
maltose binding protein
MeLan
methyllanthionine
MIC
minimal inhibitory concentration
MRSA
methicillin resistant Staphylococcus aureus
MS/MS
tandem mass spectrometry
NICE
nisin controlled expression
NMR
nuclear magnetic resonance
NRPS
nonribosomal peptide synthetase
ORF
open reading frame
PAGE
polyacrylamide gel electrophoresis
PCOR
peptide cyclization on oxime resin
PCR
polymerase chain reaction
PE
phosphatidylethanolamine

PG
POPC
Rgg
SDS
Sec
TBDMS
TEM
TFE
wt
WTA
phosphatidylglycerol
1-palmitoyl-2-oleyl-sn-glycero-3-phosphocholine
regulator gene of glucosyltransferase
sodium dodecyl sulfate
secratory pathway
tert-butyldimethylsilyl
transmission electron microscopy
trifluoroethanol
wild type
wall teichoic acid
12. Acknowledgments
This work was supported by the National Institutes
of Health (GM58822). We thank other laboratory
members who have contributed to our work on
lantibiotics, Nicole Okeley, Hao Zhou, Rashna Balsara, Matt Gieselman, Yantao Zhu, and Olga Averin,
and acknowledge fruitful collaborations with Dr. Neil
L. Kelleher and Leah Miller (UIUC) and Dr. Ninian
Blackburn (Oregon Graduate Institute).
13. Note Added In Proof

Several very recent studies have appeared in the
area of lantibiotics research. The gene cluster of
nukacin ISK-1 has been sequenced,472 and the activity of mersacidin against MRSA has been further
demonstrated.473 The morphogenic peptide SapB
from Streptomyces coelicolor is likely a lantibiotic,474
which further illustrates the breadth of natures
choice of the lanthionine as a stable residue to
contrain a peptide conformation. Interestingly, the
putative SapB biosynthetic machinery contains a
protein that has a kinase domain and a domain with
sequence homology to the LanC-proteins. The kinase
domain may be linked to phosphorylation of Ser/Thr
residues to be dehydrated. Of note, the LanC-like
domain lacks the putative metal binding residues. In
other recent work, the LctT protein has been shown
to be required for biosynthesis of lacticin 481,475 and
the substrate specificity of the NisB dehydratase and
the NisT transporter has been further defined.476
14. References
(1) Schnell, N.; Entian, K.-D.; Schneider, U.; Gotz, F.; Zahner, H.;
Kellner, R.; Jung, G. Nature 1988, 333, 276.
(2) Horn, M. J.; Jones, D. B.; Ringel, S. J. J. Biol. Chem. 1941, 138,
141.
(3) de Vos, W. M.; Jung, G.; Sahl, H.-G. In Nisin and Novel
Lantibiotics; Jung, G., Sahl, H.-G., Eds.; ESCOM: Leiden, 1991.
(4) Gross, E.; Morell, J. L. J. Am. Chem. Soc. 1971, 93, 4634.
(5) Gross, E.; Kiltz, H. H.; Nebelin, E. Hoppe-Seylers Z. Physiol.
Chem. 1973, 354, 810.
(6) Allgaier, H.; Jung, G.; Werner, R. G.; Schneider, U. Angew.
Chem., Int. Ed. Engl. 1985, 24, 1051.
(7) Kellner, R.; Jung, G.; Josten, M.; Kaletta, C.; Entian, K. D.; Sahl,
H. G. Angew. Chem. 1989, 101, 618.
(8) Kellner, R.; Jung, G.; Sahl, H. G. In Nisin and Novel Lantibiotics;
Jung, G., Sahl, H. G., Eds.; ESCOM: Leiden, The Netherlands,
1991.
(9) Kessler, H.; Steuernagel, S.; Gillessen, D.; Kamiyama, T. Helv.
Chim. Acta 1987, 70, 726.
(10) Kessler, H.; Steuernagel, S.; Will, M.; Jung, G.; Kellner, R.;
Gillessen, D.; Kamiyama, T. Helv. Chim. Acta 1988, 71, 1924.
(11) Wakamiya, T.; Fukase, K.; Naruse, N.; Konishi, M.; Shiba, T.
(12) Wakamiya, T.; Ueki, Y.; Shiba, T.; Kido, Y.; Motoki, Y. Tetrahedron Lett. 1985, 26, 665.
(13) Wakamiya, T.; Ueki, Y.; Shiba, T.; Kido, Y.; Motoki, Y. Bull.
Chem. Soc. Jpn. 1990, 63, 1032.
Chatterjee et al.
(14) Zimmermann, N.; Metzger, J. W.; Jung, G. Eur. J. Biochem.
1995, 228, 786.
(15) Zimmermann, N.; Jung, G. Eur. J. Biochem. 1997, 246, 809.
(16) Chatterjee, S.; Chatterjee, S.; Lad, S. J.; Phansalkar, M. S.;
Rupp, R. H.; Ganguli, B. N.; Fehlhaber, H. W.; Kogler, H. J.
(17) Morell, J. L.; Gross, E. J. Am. Chem. Soc. 1973, 95, 6480.
(18) Ross, K. F.; Ronson, C. W.; Tagg, J. R. Appl. Environ. Microbiol.
1993, 59, 2014.
(19) Jung, G. Angew. Chem., Intl. Ed. Engl. 1991, 30, 1051.
(20) Delves-Broughton, J.; Blackburn, P.; Evans, R. J.; Hugenholtz,
J. Antonie van Leeuwenhoek 1996, 69, 193.
(21) Rayman, M. K.; Aris, B.; Hurst, A. Appl. Environ. Microbiol.
1981, 41, 375.
(22) Hurst, A. Adv. Appl. Microbiol. 1981, 27, 85.
(23) Sahl, H.-G.; Jack, R. W.; Bierbaum, G. Eur. J. Biochem. 1995,
230, 827.
(24) Kupke, T.; Gotz, F. Antonie van Leeuwenhoek 1996, 69, 139.
(25) Klaenhammer, T. R. FEMS Microbiol. Rev. 1993, 12, 39.
Press: Washington, DC, 2003.
(27) Mahariel, M. A.; Stachelhaus, T.; Mootz, H. D. Chem. Rev. 1997,
26511.
(28) von Dohren, H.; Keller, U.; Vater, J.; Zocher, R. Chem. Rev. 1997,
97, 2675.
(29) Keating, T. A.; Walsh, C. T.; Cane, D. E.; Khosla, C. Curr. Opin.
Chem. Biol. 1999, 3, 598.
(30) Cane, D. E.; Walsh, C. T.; Khosla, C. Science 1998, 282, 63.
(31) Rogers, L. A. J. Bacteriol. 1928, 16, 321.
(32) Rogers, L. A.; Whittier, E. O. J. Bacteriol. 1928, 16, 211.
(33) Fleming, A. Br. J. Exp. Pathol. 1929, 10, 226.
(34) Nes, I. F.; Holo, H. Biopolymers 2000, 55, 50.
(35) Ingram, L. Biochim. Biophys. Acta 1970, 224, 263.
(36) Ingram, L. C. Biochim. Biophys. Acta 1969, 184, 216.
(37) Banerjee, S.; Hansen, J. N. J. Biol. Chem. 1988, 262, 9508.
(38) Buchman, G. W.; Banerjee, S.; Hansen, J. N. J. Biol. Chem. 1988,
263, 16260.
(39) Kaletta, C.; Entian, K. D. J. Bacteriol. 1989, 171, 1597.
(40) Xie, L.; Miller, L. M.; Chatterjee, C.; Averin, O.; Kelleher, N. L.;
van der Donk, W. A. Science 2004, 303, 679.
(41) Breukink, E.; Wiedemann, I.; van Kraaij, C.; Kuipers, O. P.; Sahl,
H.; de Kruijff, B. Science 1999, 286, 2361.
(42) Harris, L. J.; Fleming, H. P.; Klaenhammer, T. R. J. Food Prot.
1991, 54, 836.
(43) Montville, T. J.; Rogers, A. M.; Okereke, A. J. Food Prot. 1992,
55, 444.
(44) Bruno, M. E.; Kaiser, A.; Montville, T. J. Appl. Environ.
Microbiol. 1992, 58, 2255.
(45) Benkerroum, N.; Sandine, W. E. J. Dairy Sci. 1988, 71, 3237.
(46) Mead, P. S.; Slutsker, L.; Dietz, V.; McCaig, L. F.; Bresee, J. S.;
Shapiro, C.; Griffin, P. M.; Tauxe, R. V. Emerging Infect. Dis.
1999, 5, 607.
(47) Buzby, J. C.; Roberts, T. World Health Stat. Q. 1997, 50, 57.
(48) Wiedemann, I.; Breukink, E.; van Kraaij, C.; Kuipers, O. P.;
Bierbaum, G.; de Kruijff, B.; Sahl, H. G. J. Biol. Chem. 2001,
276, 1772.
(49) van Heusden, H. E.; de Kruijff, B.; Breukink, E. Biochemistry
2002, 41, 12171.
(50) Hsu, S. T.; Breukink, E.; de Kruijff, B.; Kaptein, R.; Bonvin, A.
M.; van Nuland, N. A. Biochemistry 2002, 41, 7670.
(51) Breukink, E.; van Heusden, H. E.; Vollmerhaus, P. J.; Swiezewska, E.; Brunner, L.; Walker, S.; Heck, A. J.; de Kruijff, B. J.
Biol. Chem. 2003, 278, 19898.
(52) Wiedemann, I.; Benz, R.; Sahl, H. G. J. Bacteriol. 2004, 186,
3259.
(53) Hasper, H. E.; de Kruijff, B.; Breukink, E. Biochemistry 2004,
43, 11567.
(54) Hsu, S. T.; Breukink, E.; Tischenko, E.; Lutters, M. A.; De
Kruijff, B.; Kaptein, R.; Bonvin, A. M.; Van Nuland, N. A. Nat.
Struct. Mol. Biol. 2004, 11, 963.
(55) van Kraaij, C.; de Vos, W. M.; Siezen, R. J.; Kuipers, O. P. Nat.
Prod. Rep. 1999, 16, 575.
(56) Fredenhagen, A.; Fendrich, G.; Marki, F.; Marki, W.; Gruner,
J.; Raschdorf, F.; Peter, H. H. J. Antibiot. (Tokyo) 1990, 43, 1403.
(57) Marki, F.; Hanni, E.; Fredenhagen, A.; van Oostrum, J. Biochem.
Pharmacol. 1991, 42, 2027.
(58) Kido, Y.; Hamakado, T.; Yoshida, T.; Anno, M.; Motoki, Y.;
Wakamiya, T.; Shiba, T. J. Antibiot. (Tokyo) 1983, 36, 1295.
(59) Chatterjee, S.; Chatterjee, D. K.; Jani, R. H.; Blumbach, J.;
Ganguli, B. N.; Klesel, N.; Limbert, M.; Seibert, G. J. Antibiot.
(Tokyo) 1992, 45, 839.
(60) Limbert, M. D.; Isert, D.; Klesel, N.; Markus, A.; Seibert, G.;
Chatterjee, S.; Chatterjee, D. K.; Jani, R. H.; Ganguli, B. N. In
Nisin and Novel Lantibiotics; Jung, G., Sahl, H.-G., Eds.;
ESCOM: Leiden, The Netherlands, 1991.
(61) Bierbaum, G.; Brotz, H.; Koller, K. P.; Sahl, H. G. FEMS
(62) Mota-Meira, M.; LaPointe, G.; Lacroix, C.; Lavoie, M. C. Antimicrob. Agents Chemother. 2000, 44, 24.

(63) Cleveland, J.; Montville, T. J.; Nes, I. F.; Chikindas, M. L. Int.
J. Food Microbiol. 2001, 71, 1.
(64) OSullivan, L.; Ross, R. P.; Hill, C. Biochimie 2002, 84, 593.
(65) Jack, R.; Bierbaum, G.; Heidrich, C.; Sahl, H.-G. BioEssays 1995,
17, 793.
(66) Siezen, R. J.; Kuipers, O. P.; de Vos, W. M. Antonie van
Leeuwenhoek 1996, 69, 171.
(67) Xie, L.; van der Donk, W. A. Curr. Opin. Chem. Biol. 2004, 8,
498.
(68) Klein, C.; Kaletta, C.; Schnell, N.; Entian, K.-D. Appl. Environ.
Microbiol. 1992, 58, 132.
(69) Augustin, J.; Rosenstein, R.; Wieland, B.; Schneider, U.; Schnell,
N.; Engelke, G.; Entian, K.-D.; Gotz, F. Eur. J. Biochem. 1992,
204, 1149.
(70) Hansen, J. N. Annu. Rev. Microbiol. 1993, 47, 535.
(71) McAuliffe, O.; Hill, C.; Ross, R. P. Microbiology 2000, 146, 2147.
(72) Heidrich, C.; Pag, U.; Josten, M.; Metzger, J.; Jack, R. W.;
Bierbaum, G.; Jung, G.; Sahl, H. G. Appl. Environ. Microbiol.
1998, 64, 3140.
(73) Chung, Y. J.; Steen, M. T.; Hansen, J. N. J. Bacteriol. 1992, 174,
1417.
(74) Qiao, M.; Saris, P. E. J. FEMS Microbiol. Lett. 1996, 144, 89.
(75) Meyer, C.; Bierbaum, G.; Heidrich, C.; Reis, M.; Suling, J.;
Iglesias-Wind, M. I.; Kempter, C.; Molitor, E.; Sahl, H.-G. Eur.
J. Biochem. 1995, 232, 478.
(76) Schnell, N.; Engelke, G.; Augustin, J.; Rosenstein, R.; Ungermann, V.; Gotz, F.; Entian, K.-D. Eur. J. Biochem. 1992, 204,
57.
(77) Corvey, C.; Stein, T.; Dusterhus, S.; Karas, M.; Entian, K. D.
Biochem. Biophys. Res. Commun. 2003, 304, 48.
(78) Widdick, D. A.; Dodd, H. M.; Barraille, P.; White, J.; Stein, T.
H.; Chater, K. F.; Gasson, M. J.; Bibb, M. J. Proc. Natl. Acad.
Sci. U.S.A. 2003, 100, 4316.
(79) Hvarstein, L. S.; Diep, D. B.; Nes, I. F. Mol. Microbiol. 1995,
16, 229.
(80) Jung, G. In Nisin and Novel Lantibiotics; Jung, G., Sahl, H. G.,
Eds.; ESCOM: Leiden, The Netherlands, 1991.
(81) Pag, U.; Sahl, H. G. Curr. Pharm. Des. 2002, 8, 815.
(82) Breukink, E.; de Kruijff, B. Biochim. Biophys. Acta 1999, 1462,
223.
(83) Mulders, J. W.; Boerrigter, I. J.; Rollema, H. S.; Siezen, R. J.;
de Vos, W. M. Eur. J. Biochem. 1991, 201, 581.
(84) de Vos, W. M.; Mulders, J. W.; Siezen, R. J.; Hugenholtz, J.;
Kuipers, O. P. Appl. Environ. Microbiol. 1993, 59, 213.
(85) Zendo, T.; Fukao, M.; Ueda, K.; Higuchi, T.; Nakayama, J.;
Sonomoto, K. Biosci. Biotechnol. Biochem. 2003, 67, 1616.
(86) Mattick, A. T. R.; Hirsch, A. Nature 1944, 154, 551.
(87) Fukase, K.; Kitazawa, M.; Akihiko, S.; Shimbo, K.; Fujita, H.;
Horimoto, S.; Wakamiya, T.; Shiba, T. Tetrahedron Lett. 1988,
29, 795.
(88) Stein, T.; Borchert, S.; Conrad, B.; Feesche, J.; Hofemeister, B.;
Hofemeister, J.; Entian, K.-D. J. Bacteriol. 2002, 184, 1703.
(89) Wescombe, P. A.; Tagg, J. R. Appl. Environ. Microbiol. 2003,
69, 2737.
(90) Israil, A. M.; Jack, R. W.; Jung, G.; Sahl, H. G. Zentralbl.
Bakteriol. 1996, 284, 285.
(91) Kellner, R.; Jung, G.; Hoerner, T.; Zaehner, H.; Schnell, N.;
Entian, K.-D.; Goetz, F. Eur. J. Biochem. 1988, 177, 53.
(92) Smith, L.; Novak, J.; Rocca, J.; McClung, S.; Hillman, J. D.;
Edison, A. S. Eur. J. Biochem. 2000, 267, 6810.
(93) Mota-Meira, M.; Lacroix, C.; LaPointe, G.; Lavoie, M. C. FEBS
Lett. 1997, 410, 275.
(94) Qi, F.; Chen, P.; Caufield, P. W. Appl. Environ. Microbiol. 1999,
65, 3880.
66, 3221.
(96) van de Kamp, M.; Horstink, L. M.; Van den Hooven, H. W.;
Konings, R. N. H.; Hilbers, C. W.; Frey, A.; Sahl, H.-G.; Metzger,
J. W.; van de Ven, F. J. M. Eur. J. Biochem. 1995, 227, 757.
(97) van den Hooven, H. W.; Lagerwerf, F. M.; Heerma, W.; Haverkamp, J.; Piard, J. C.; Hilbers, C. W.; Siezen, R. J.; Kuipers, O. P.;
Rollema, H. S. FEBS Lett. 1996, 391, 317.
(98) Pridmore, D.; Rekhif, N.; Pittet, A. C.; Suri, B.; Mollet, B. Appl.
Environ. Microbiol. 1996, 62, 1799.
(99) Krull, R. E.; Chen, P.; Novak, J.; Kirk, M.; Barnes, S.; Baker,
J.; Krishna, N. R.; Caufield, P. W. J. Biol. Chem. 2000, 275,
15845.
(100) Jack, R. W.; Carne, A.; Metzger, J.; Stefanovic, S.; Sahl, H. G.;
Jung, G.; Tagg, J. Eur. J. Biochem. 1994, 220, 455.
(101) Simpson, W. J.; Ragland, N. L.; Ronson, C. W.; Tagg, J. R. Dev.
Biol. Stand. 1995, 85, 639.
(102) Skaugen, M.; Nissenmeyer, J.; Jung, G.; Stevanovic, S.; Sletten,
K.; Abildgaard, C. I. M.; Nes, I. F. J. Biol. Chem. 1994, 269,
27183.
(103) Minami, Y.; Yoshida, K.; Azuma, R.; Urakawa, A.; Kawauchi,
T.; Otani, T.; Komiyama, K.; Omura, S. Tetrahedron Lett. 1994,
35, 8001.

(104) Turner, D. L.; Brennan, L.; Meyer, H. E.; Lohaus, C.; Siethoff,
C.; Costa, H. S.; Gonzalez, B.; Santos, H.; Suarez, J. E. Eur. J.
Biochem. 1999, 264, 833.
(105) Paik, S. H.; Chakicherla, A.; Hansen, J. N. J. Biol. Chem. 1998,
273, 23134.
(106) Kalmokoff, M. L.; Lu, D.; Whitford, M. F.; Teather, R. M. Appl.
(107) Benedict, R. G.; Dvonch, W.; Shotwell, O. L.; Pridham, T.;
Lindenfelser, L. A. Antibiot. Chemother. 1952, 2, 591.
(108) Hayashi, F.; Nagashima, K.; Terui, Y.; Kawamura, Y.; Matsumoto, K.; Itazaki, H. J. Antibiot. 1990, 43, 1421.
(109) Fredenhagen, A.; Maerki, F.; Fendrich, G.; Maerki, W.; Gruner,
J.; Van Oostrum, J.; Raschdorf, F.; Peter, H. H. In Nisin and
Novel Lantibiotics; Jung, G., Sahl, H.-G., Eds.; ESCOM: Leiden,
The Netherlands, 1991.
(110) Zimmermann, N.; Freund, S.; Fredenhagen, A.; Jung, G. Eur.
J. Biochem. 1993, 216, 419.
(111) Vertesy, L.; Aretz, W.; Bonnefoy, A.; Ehlers, E.; Kurz, M.;
Markus, A.; Schiell, M.; Vogel, M.; Wink, J.; Kogler, H. J.
Antibiot. 1999, 52, 730.
(112) Martin, N. I.; Sprules, T.; Carpenter, M. R.; Cotter, P. D.; Hill,
C.; Ross, R. P.; Vederas, J. C. Biochemistry 2004, 43, 3049.
(113) Navaratna, M. A.; Sahl, H. G.; Tagg, J. R. Appl. Environ.
Microbiol. 1998, 64, 4803.
(114) Holo, H.; Jeknic, Z.; Daeschel, M.; Stevanovic, S.; Nes, I. F.
Microbiology 2001, 147, 643.
(115) Booth, M. C.; Bogie, C. P.; Sahl, H.-G.; Siezen, R. J.; Hatter, K.
L.; Gilmore, M. S. Mol. Microbiol. 1996, 21, 1175.
(116) Dabard, J.; Bridonneau, C.; Phillipe, C.; Anglade, P.; Molle, D.;
Nardi, M.; Ladire, M.; Girardin, H.; Marcille, F.; Gomez, A.;
Fons, M. Appl. Environ. Microbiol. 2001, 67, 4111.
(117) Stoffels, G.; Nissen-Meyer, J.; Gudmundsdottir, A.; Sletten, K.;
Holo, H.; Nes, I. F. Appl. Environ. Microbiol. 1992, 58, 1417.
(118) Georgalaki, M. D.; Van den Berghe, E.; Kritikos, D.; Devreese,
B.; Van Beeumen, J.; Kalantzopoulos, G.; De Vuyst, L.; Tsakalidou, E. Appl. Environ. Microbiol. 2002, 68, 5891.
(119) Xiao, H.; Chen, X.; Chen, M.; Tang, S.; Zhao, X.; Huan, L.
Microbiology 2004, 150, 103.
(120) Sashihara, T.; Kimura, H.; Higuchi, T.; Adachi, A.; Matsusaki,
H.; Sonomoto, K.; Ishizaki, A. Biosci. Biotechnol. Biochem. 2000,
64, 2420.
(123) Toogood, P. L. Tetrahedron Lett. 1993, 34, 7833.
(124) Burrage, S.; Raynham, T.; Williams, G.; Essex, J. W.; Allen, C.;
Cardno, M.; Swali, V.; Bradley, M. Chem.-Eur. J. 2000, 6, 1455.
(125) Okeley, N. M.; Zhu, Y.; van der Donk, W. A. Org. Lett. 2000, 2,
3603.
(126) Zhou, H.; van der Donk, W. A. Org. Lett. 2002, 4, 1335.
(127) Chan, W. C.; Bycroft, B. W.; Leyland, M. L.; Lian, L. Y.; Yang,
J. C.; Roberts, G. C. FEBS Lett. 1992, 300, 56.
(128) Chan, W. C.; Bycroft, B. W.; Leyland, M. L.; Lian, L. Y.; Roberts,
G. C. Biochem. J. 1993, 291 (Pt 1), 23.
(129) Karaya, K.; Shimizu, T.; Taketo, A. J. Biochem. 2001, 129, 769.
(130) Van de Ven, F. J.; Van den Hooven, H. W.; Konings, R. N.;
Hilbers, C. W. Eur. J. Biochem. 1991, 202, 1181.
(131) Van den Hooven, H. W.; Doeland, C. C.; Van de Kamp, M.;
Konings, R. N.; Hilbers, C. W.; Van de Ven, F. J. M. Eur. J.
Biochem. 1996, 235, 382.
(132) Lian, L. Y.; Chan, W. C.; Morley, S. D.; Roberts, G. C.; Bycroft,
B. W.; Jackson, D. Biochem. J. 1992, 283, 413.
(133) Palmer, D. E.; Mierke, D. F.; Pattaroni, C.; Goodman, M.;
Wakamiya, T.; Fukase, K.; Kitazawa, M.; Fujita, H.; Shiba, T.
Biopolymers 1989, 28, 397.
(134) Furmanek, B.; Kaczorowski, T.; Bugalski, R.; Bielawski, K.;
Bogdanowicz, J.; Podhajska, A. J. J. Appl. Microbiol. 1999, 87,
856.
(135) Hillman, J. D.; Novak, J.; Sagura, E.; Gutierrez, J. A.; Brooks,
T. A.; Crowley, P. J.; Hess, M.; Azizi, A.; Leung, K.; Cvitkovitch,
D.; Bleiweis, A. S. Infect. Immun. 1998, 66, 2743.
(136) Chan, W. C.; Dodd, H. M.; Horn, N.; Maclean, K.; Lian, L. Y.;
Bycroft, B. W.; Gasson, M. J.; Roberts, G. C. Appl. Environ.
Microbiol. 1996, 62, 2966.
(137) Liu, W.; Hansen, J. N. J. Biol. Chem. 1992, 267, 25078.
(138) Liu, W.; Hansen, J. N. Appl. Environ. Microbiol. 1993, 59, 648.
(139) Freund, S.; Jung, G.; Gutbrod, O.; Folkers, G.; Gibbons, W. A.
In Nisin and Novel Lantibiotics; Jung, G., Sahl, H. G., Eds.;
(140) Freund, S.; Jung, G.; Gutbrod, O.; Folkers, G.; Gibbons, W. A.;
Allgaier, H.; Werner, R. Biopolymers 1991, 31, 803.
(141) Smith, L.; Zachariah, C.; Thirumoorthy, R.; Rocca, J.; Novak,
J.; Hillman, J. D.; Edison, A. S. Biochemistry 2003, 42, 10372.
(142) Sahl, H. G.; Brandis, H. J. Gen. Microbiol. 1981, 127 (Pt 2), 377.
(143) Kaletta, C.; Entian, K. D.; Kellner, R.; Jung, G.; Reis, M.; Sahl,
H. G. Arch. Microbiol. 1989, 152, 16.
(144) Meyer, H. E.; Heber, M.; Eisermann, B.; Korte, H.; Metzger, J.
W.; Jung, G. Anal. Biochem. 1994, 223, 185.

(145) van de Kamp, M.; van den Hooven, H. W.; Konings, R. N. H.;
Bierbaum, G.; Sahl, H.-G.; Kuipers, O. P.; Siezen, R. J.; de Vos,
W.; Hilbers, C. W.; van de Ven, F. J. Eur. J. Biochem. 1995,
230, 587.
(146) Freund, S.; Jung, G.; Gibbons, W. A.; Sahl, H. G. In Nisin and
Novel Lantibiotics; Jung, G., Sahl, H. G., Eds.; ESCOM: Leiden,
The Netherlands, 1991.
(147) Dufour, A.; Rince, A.; Hindre, T.; Haras, D.; Le Pennec, J. P.
Recent Res. Devel. Becteriol. 2003, 1, 219.
(148) Rince, A.; Dufour, A.; Le Pogam, S.; Thuault, D.; Bourgeois, C.
M.; Le Pennec, J. P. Appl. Environ. Microbiol. 1994, 60, 1652.
(149) Piard, J. C.; Muriana, P. M.; Desmazeaud, M. J.; Klaenhammer,
T. R. Appl. Environ. Microbiol. 1992, 58, 279.
(150) Piard, J. C.; Kuipers, O. P.; Rollema, H. S.; Desmazeaud, M. J.;
de Vos, W. M. J. Biol. Chem. 1993, 268, 16361.
(151) Novak, J.; Caufield, P. W.; Miller, E. J. J. Bacteriol. 1994, 176,
4316.
(152) Novak, J.; Kirk, M.; Caufield, P. W.; Barnes, S.; Morrison, K.;
Baker, J. Anal. Biochem. 1996, 236, 358.
(153) Woodruff, W. A.; Novak, J.; Caufield, P. W. Gene 1998, 206, 37.
(154) Mortvedt, C. I.; Nissen-Meyer, J.; Sletten, K.; Nes, I. F. Appl.
(155) Skaugen, M.; Abildgaard, C. I.; Nes, I. F. Mol. Gen. Genet. 1997,
253, 674.
(156) Hynes, W. L.; Ferretti, J. J.; Tagg, J. R. Appl. Environ. Microbiol.
1993, 59, 1969.
(157) Upton, M.; Tagg, J. R.; Wescombe, P.; Jenkinson, H. F. J.
Bacteriol. 2001, 183, 3931.
(158) Gonzalez, B.; Arca, P.; Mayo, B.; Suarez, J. E. Appl. Environ.
Microbiol. 1994, 60, 2158.
(159) Komiyama, K.; Otoguro, K.; Segawa, T.; Shiomi, K.; Yang, H.;
Takahashi, Y.; Hayashi, M.; Otani, T.; Omura, S. J. Antibiot.
(Tokyo) 1993, 46, 1666.
(160) Novak, J.; Chen, P.; Kirk, M.; Barnes, S.; Jablonsky, M. J.;
Holaday, S. K.; Krishna, N. R.; Baker, J.; Caufield, P. W. Protein
Eng. 1997, 10, 87.
(161) Chen, P.; Novak, J.; Kirk, M.; Barnes, S.; Qi, F.; Caufield, P. W.
Appl. Environ. Microbiol. 1998, 64, 2335.
(162) Kogler, H.; Bauch, M.; Fehlhaber, H. W.; Griesinger, C.; Schubert, W.; Teetz, V. In Nisin and Novel Lantibiotics; Jung, G.,
Sahl, H.-G., Eds.; ESCOM: Leiden, The Netherlands, 1991.
(163) Jack, R. W.; Sahl, H. G. Trends Biotechnol. 1995, 13, 269.
(164) Parenti, F.; Pagani, H.; Beretta, G. J. Antibiot. (Tokyo) 1976,
29, 501.
(165) Coronelli, C.; Tamoni, G.; Lancini, G. C. J. Antibiot. (Tokyo)
1976, 29, 507.
(166) Arioli, V.; Berti, M.; Silvestri, L. G. J. Antibiot. (Tokyo) 1976,
29, 511.
(167) Somma, S.; Merati, W.; Parenti, F. Antimicrob. Agents Chemother. 1977, 11, 396.
(168) Kettenring, J. K.; Malabarba, A.; Vekey, K.; Cavalleri, B. J.
Antibiot. (Tokyo) 1990, 43, 1082.
(169) Prasch, T.; Naumann, T.; Markert, R. L.; Sattler, M.; Schubert,
W.; Schaal, S.; Bauch, M.; Kogler, H.; Griesinger, C. Eur. J.
Biochem. 1997, 244, 501.
(170) Schneider, T. R.; Karcher, J.; Pohl, E.; Lubini, P.; Sheldrick, G.
M. Acta Crystallogr. Section D Biol. Cryst. 2000, 56 (Pt 6), 705.
(171) Hsu, S. T.; Breukink, E.; Bierbaum, G.; Sahl, H. G.; de Kruijff,
B.; Kaptein, R.; van Nuland, N. A.; Bonvin, A. M. J. Biol. Chem.
2003, 278, 13110.
(172) Kaletta, C.; Entian, K. D.; Jung, G. Eur. J. Biochem. 1991, 199,
411.
(173) Kessler, H.; Steuernagel, S.; Gillessen, D.; Kamiyama, T. Helv.
Chim. Acta 1987, 70, 726.
(174) Kessler, H.; Steuernagel, S.; Will, M.; Jung, G.; Kellner, R.;
Gillessen, D.; Kamiyama, T. Helv. Chim. Acta 1988, 71, 1924.
(175) Choung, S.; Kobayashi, T.; Inoue, J.; Takemoto, K.; Ishitsuka,
H.; Inoue, K. Biochim. Biophys. Acta 1988, 940, 171.
(176) Wakamiya, T.; Fukase, K.; Naruse, N.; Konishi, M.; Shiba, T.
(177) Gross, E. Adv. Exp. Med. Biol. 1977, 86B, 131.
(178) Kondo, S.; Sezaki, M.; Shimura, M.; Sato, K.; Hara, T. J.
Antibiotics (Tokyo) 1964, 17, 262.
(179) Kessler, H.; Seip, S.; Wein, T.; Steuernagel, S.; Will, M. In Nisin
and Novel Lantibiotics; Jung, G., Sahl, H.-G., Eds.; ESCOM:
Leiden, The Netherlands, 1991.
(180) Kessler, H.; Mierke, D. F.; Saulitis, J.; Seip, S.; Steuernagel, S.;
Wein, T.; Will, M. Biopolymers 1992, 32, 427.
(181) Wakamatsu, K.; Choung, S. Y.; Kobayashi, T.; Inoue, K.;
Higashijima, T.; Miyazawa, T. Biochemistry 1990, 29, 113.
(182) Garneau, S.; Martin, N. I.; Vederas, J. C. Biochimie 2002, 84,
577.
(183) Ryan, M. P.; Jack, R. W.; Josten, M.; Sahl, H. G.; Jung, G.; Ross,
R. P.; Hill, C. J. Biol. Chem. 1999, 274, 37544.
(184) Skaugen, M.; Nissen-Meyer, J.; Jung, G.; Stevanovic, S.; Sletten,
K.; Inger, C.; Abildgaard, M.; Nes, I. F. J. Biol. Chem. 1994,
269, 27183.
(185) Ryan, M. P.; Jack, R. W.; Josten, M.; Sahl, H.-G.; Jung, G.; Ross,
R. P.; Hill, C. J. Biol. Chem. 1999, 274, 37544.
Chatterjee et al.
(186) Nes, I. F.; Tagg, J. R. Antonie van Leeuwenhoek 1996, 69, 89.
(187) Dirix, G.; Monsieurs, P.; Dombrecht, B.; Daniels, R.; Marchal,
K.; Vanderleyden, J.; Michiels, J. Peptides 2004, 25, 1425.
(188) Dirix, G.; Monsieurs, P.; Marchal, K.; Vanderleyden, J.; Michiels,
J. Microbiology 2004, 150, 1121.
(189) van der Meer, J. R.; Rollema, H. S.; Siezen, R. J.; Beerthuyzen,
M. M.; Kuipers, O. P.; de Vos, W. M. J. Biol. Chem. 1994, 269,
3555.
(190) Tjalsma, H.; Antelmann, H.; Jongbloed, J. D.; Braun, P. G.;
Darmon, E.; Dorenbos, R.; Dubois, J. Y.; Westers, H.; Zanen,
G.; Quax, W. J.; Kuipers, O. P.; Bron, S.; Hecker, M.; Van Dijl,
J. M. Microbiol. Mol. Biol. Rev. 2004, 68, 207.
(191) Rockwell, N. C.; Krysan, D. J.; Komiyama, T.; Fuller, R. S. Chem.
Rev. 2002, 102, 4525.
(192) Rockwell, N. C.; Thorner, J. W. Trends Biochem. Sci. 2004, 29,
80.
(193) Sprules, T.; Kawulka, K. E.; Gibbs, A. C.; Wishart, D. S.; Vederas,
J. C. Eur. J. Biochem. 2004, 271, 1748.
(194) Madison, L. L.; Vivas, E. I.; Li, Y. M.; Walsh, C. T.; Kolter, R.
Mol. Microbiol. 1997, 23, 161.
(195) Kuipers, A.; De Boef, E.; Rink, R.; Fekken, S.; Kluskens, L. D.;
Driessen, A. J.; Leenhouts, K.; Kuipers, O. P.; Moll, G. N. J.
Biol. Chem. 2004, 279, 22176.
(196) Izaguirre, G.; Hansen, J. N. Appl. Environ. Microbiol. 1997, 63,
3965.
(197) Paul, L. K.; Izaguirre, G.; Hansen, J. N. FEMS Microbiol. Lett.
1999, 176, 45.
(198) van der Meer, J. R.; Polman, J.; Beerthuyzen, M. M.; Siezen, R.
J.; Kuipers, O. P.; de Vos, W. M. J. Bacteriol. 1993, 175, 2578.
(199) Qiao, M.; Immonen, T.; Koponen, O.; Saris, P. E. J. FEMS
(200) Stein, T.; Entian, K. D. Rapid Commun. Mass Spectrom. 2002,
16, 103.
(201) Chen, P.; Qi, F. X.; Novak, J.; Krull, R. E.; Caufield, P. W. FEMS
(202) van den Hooven, H. W.; Rollema, H. S.; Siezen, R. J.; Hilbers,
C. W.; Kuipers, O. P. Biochemistry 1997, 36, 14137.
(203) Chakicherla, A.; Hansen, J. N. J. Biol. Chem. 1995, 270, 23533.
(204) Kuipers, O. P.; Rollema, H. S.; de Vos, W. M.; Siezen, R. J. FEBS
Lett. 1993, 330, 23.
(205) Kupke, T.; Kempter, C.; Jung, G.; Gotz, F. J. Biol. Chem. 1995,
270, 11282.
(206) Neis, S.; Bierbaum, G.; Josten, M.; Pag, U.; Kempter, C.; Jung,
G.; Sahl, H. G. FEMS Microbiol. Lett. 1997, 149, 249.
(207) Beck-Sickinger, A. G.; Jung, G. In Nisin and Novel Lantibiotics;
Jung, G., Sahl, H.-G., Eds.; ESCOM: Leiden, 1991.
(208) Bycroft, B. W.; Chan, W. C.; Roberts, G. C. K. In Nisin and Novel
Lantibiotics; Jung, G., Sahl, H.-G., Eds.; ESCOM: Leiden, 1991.
(209) Sen, A. K.; Narbad, A.; Horn, N.; Dodd, H. M.; Parr, A. J.;
Colquhoun, I.; Gasson, M. J. Eur. J. Biochem. 1999, 261, 524.
(210) Koponen, O.; Tolonen, M.; Qiao, M.; Wahlstrom, G.; Helin, J.;
Saris Per, E. J. Microbiology 2002, 148, 3561.
(211) Chung, Y. J.; Hansen, J. N. J. Bacteriol. 1992, 174, 6699.
(212) Peschel, A.; Ottenwalder, B.; Gotz, F. FEMS Microbiol. Lett.
1996, 137, 279.
(213) Kupke, T.; Gotz, F. J. Bacteriol. 1996, 178, 1335.
(214) Engelke, G.; Gutowski-Eckel, Z.; Hammelmann, M.; Entian, K.D. Appl. Environ. Microbiol. 1992, 58, 3730.
(215) Siegers, K.; Heinzmann, S.; Entian, K.-D. J. Biol. Chem. 1996,
271, 12294.
(216) Kiesau, P.; Eikmanns, U.; Gutowski-Eckel, Z.; Weber, S.; Hammelmann, M.; Entian, K.-D. J. Bacteriol. 1997, 179, 1475.
(217) Xie, L.; Chatterjee, C.; Balsara, R.; Okeley, N. M.; van der Donk,
W. A. Biochem. Biophys. Res. Commun. 2002, 295, 952.
(218) Okeley, N. M.; Paul, M.; Stasser, J. P.; Blackburn, N.; van der
Donk, W. A. Biochemistry 2003, 42, 13613.
(219) Matthews, R. G.; Goulding, C. W. Curr. Opin. Chem. Biol. 1997,
1, 332.
(220) Hightower, K. E.; Huang, C. C.; Casey, P. J.; Fierke, C. A.
(221) Bednar, R. A. Biochemistry 1990, 29, 3684.
(222) Zhu, Y.; Gieselman, M.; Zhou, H.; Averin, O.; van der Donk, W.
A. Org. Biomol. Chem. 2003, 1, 3304.
(223) Bauer, H.; Mayer, H.; Marchler-Bauer, A.; Salzer, U.; Prohaska,
R. Biochem. Biophys. Res. Commun. 2000, 275, 69.
(224) Mayer, H.; Bauer, H.; Breuss, J.; Ziegler, S.; Prohaska, R. Gene
2001, 269, 73.
(225) Uguen, P.; Le Pennec, J. P.; Dufour, A. J. Bacteriol. 2000, 182,
5262.
(226) Gilmore, M. S.; Segarra, R. A.; Booth, M. C.; Bogie, C. P.; Hall,
L. R.; Clewell, D. B. J. Bacteriol. 1994, 176, 7335.
(227) McAuliffe, O.; OKeeffe, T.; Hill, C.; Ross, R. P. Mol. Microbiol.
2001, 39, 982.
(228) Milne, J. C.; Eliot, A. C.; Kelleher, N. L.; Walsh, C. T. Biochemistry 1998, 37, 13250.
(229) Milne, J. C.; Roy, R. S.; Eliot, A. C.; Kelleher, N. L.; Wokhlu, A.;
Nickels, B.; Walsh, C. T. Biochemistry 1999, 38, 4768.
(230) Kupke, T.; Stevanovic, S.; Sahl, H. G.; Gotz, F. J. Bacteriol. 1992,
174, 5354.

(231) Kupke, T.; Gotz, F. FEMS Microbiol. Lett. 1997, 153, 25.
(232) Kupke, T.; Kempter, C.; Gnau, V.; Jung, G.; Gotz, F. J. Biol.
Chem. 1994, 269, 5653.
(233) Schmid, D. G.; Majer, F.; Kupke, T.; Jung, G. Rapid Commun.
Mass Spectrom. 2002, 16, 1779.
(234) Kempter, C.; Kupke, T.; Kaiser, D.; Metzger, J. W.; Jung, G.
Angew. Chem., Intl. Ed. Engl. 1996, 35, 2104.
(235) Kupke, T.; Gotz, F. J. Biol. Chem. 1997, 272, 4759.
(236) Kupke, T.; Uebele, M.; Schmid, D.; Jung, G.; Blaesse, M.;
Steinbacher, S. J. Biol. Chem. 2000, 275, 31838.
(237) Blaesse, M.; Kupke, T.; Huber, R.; Steinbacher, S. EMBO J.
2000, 19, 6299.
(238) Majer, F.; Schmid, D. G.; Altena, K.; Bierbaum, G.; Kupke, T.
J. Bacteriol. 2002, 184, 1234.
(239) Blaesse, M.; Kupke, T.; Huber, R.; Steinbacher, S. Acta Crystallogr. Section D Biol. Cryst. 2003, D59, 1414.
(240) Kupke, T. J. Biol. Chem. 2001, 276, 27597.
(241) Hernandez-Acosta, P.; Schmid, D. G.; Jung, G.; Culianez-Macia,
F. A.; Kupke, T. J. Biol. Chem. 2002, 277, 20490.
(242) McAuliffe, O.; Ross, R. P.; Hill, C. FEMS Microbiol. Rev. 2001,
25, 285.
(243) Sahl, H. G.; Bierbaum, G. Annu. Rev. Microbiol. 1998, 52, 41.
(244) Drakenberg, T.; Fernlund, P.; Roepstorff, P.; Stenflo, J. Proc.
Natl. Acad. Sci. U.S.A. 1983, 80, 1802.
(245) Gronke, R. S.; VanDusen, W. J.; Garsky, V. M.; Jacobs, J. W.;
Sardana, M. K.; Stern, A. M.; Friedman, P. A. Proc. Natl. Acad.
Sci. U.S.A. 1989, 86, 3609.
(246) Wang, Q. P.; VanDusen, W. J.; Petroski, C. J.; Garsky, V. M.;
Stern, A. M.; Friedman, P. A. J. Biol. Chem. 1991, 266, 14004.
(247) Jia, S.; VanDusen, W. J.; Diehl, R. E.; Kohl, N. E.; Dixon, R. A.
F.; Elliston, K. O.; Stern, A. M.; Friedman, P. A. J. Biol. Chem.
1992, 267, 14322.
(248) Jia, S.; McGinnis, K.; VanDusen, W. J.; Burke, C. J.; Kuo, A.;
Griffin, P. R.; Sardana, M. K.; Elliston, K. O.; Stern, A. M.;
Friedman, P. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 7227.
(249) Korioth, F.; Gieffers, C.; Frey, J. Gene 1994, 150, 395.
(250) Friedman, M. J. Agric. Food Chem. 1999, 47, 1295.
(251) Schneewind, O.; Fowler, A.; Faull, K. F. Science 1995, 268, 103.
(252) Geissler, S.; Gotz, F.; Kupke, T. J. Bacteriol. 1996, 178, 284.
(253) Siezen, R. J.; Rollema, H. S.; Kuipers, O. P.; de Vos, W. M.
Protein Eng. 1995, 8, 117.
(254) Gentschev, I.; Goebel, W. Mol. Gen. Genet. 1992, 232, 40.
(255) Chen, P.; Qi, F.; Novak, J.; Caufield, P. W. Appl. Environ.
Microbiol. 1999, 65, 1356.
(256) Franke, C. M.; Tiemersma, J.; Venema, G.; Kok, J. J. Biol. Chem.
1999, 274, 8484.
(257) Venema, K.; Kok, J.; Marugg, J. D.; Toonen, M. Y.; Ledeboer,
A. M.; Venema, G.; Chikindas, M. L. Mol. Microbiol. 1995, 17,
515.
(258) Uguen, M.; Uguen, P. FEMS Microbiol. Lett. 2002, 208, 99.
(259) Engelke, G.; Gutowski-Eckel, Z.; Kiesau, P.; Siegers, K.; Hammelmann, M.; Entian, K.-D. Appl. Environ. Microbiol. 1994, 60,
814.
(260) Ra, S. R.; Qiao, M.; Immonen, T.; Pujana, I.; Saris, E. J.
Microbiology 1996, 142 (Pt 5), 1281.
(261) Klein, C.; Kaletta, C.; Entian, K. D. Appl. Environ. Microbiol.
1993, 59, 296.
(262) Altena, K.; Guder, A.; Cramer, C.; Bierbaum, G. Appl. Environ.
Microbiol. 2000, 66, 2565.
(263) McLaughlin, R. E.; Ferretti, J. J.; Hynes, W. L. FEMS Microbiol.
Lett. 1999, 175, 171.
(264) Stock, A. M.; Robinson, V. L.; Goudreau, P. N. Annu. Rev.
Biochem. 2000, 69, 183.
(265) Fuqua, C.; Winans, S. C.; Greenberg, E. P. Annu. Rev. Microbiol.
1996, 50, 727.
(266) Lyon, G. J.; Muir, T. W. Chem. Biol. 2003, 10, 1007.
(267) Podbielski, A.; Kreikemeyer, B. Int. J. Infect. Dis. 2004, 8, 81.
(268) de Ruyter, P. G.; Kuipers, O. P.; Beerthuyzen, M. M.; van AlenBoerrigter, I.; de Vos, W. M. J. Bacteriol. 1996, 178, 3434.
(269) Kuipers, O. P.; Beerthuyzen, M. M.; de Ruyter, P. G.; Luesink,
E. J.; de Vos, W. M. J. Biol. Chem. 1995, 270, 27299.
(270) Kleerebezem, M. Peptides 2004, 25, 1405.
(271) Kleerebezem, M.; Bongers, R.; Rutten, G.; de Vos, W. M.;
Kuipers, O. P. Peptides 2004, 25, 1415.
(272) Stein, T.; Borchert, S.; Kiesau, P.; Heinzmann, S.; Kloss, S.;
Klein, C.; Helfrich, M.; Entian, K. D. Mol. Microbiol. 2002, 44,
403.
(273) Van Kraaij, C.; Breukink, E.; Rollema, H. S.; Siezen, R. J.;
Demel, R. A.; De Kruijff, B.; Kuipers, O. P. Eur. J. Biochem.
1997, 247, 114.
(274) Chandrapati, S.; OSullivan, D. J. Mol. Microbiol. 2002, 46, 467.
(275) de Ruyter, P. G.; Kuipers, O. P.; de Vos, W. M. Appl. Environ.
Microbiol. 1996, 62, 3662.
(276) Kleerebezem, M.; Beerthuyzen, M. M.; Vaughan, E. E.; de Vos,
W. M.; Kuipers, O. P. Appl. Environ. Microbiol. 1997, 63, 4581.
(277) Kuipers, O. P.; de Ruyter, P. G.; Kleerebezem, M.; de Vos, W.
M. Trends Biotechnol. 1997, 15, 135.

(278) Eichenbaum, Z.; Federle, M. J.; Marra, D.; de Vos, W. M.;
Kuipers, O. P.; Kleerebezem, M.; Scott, J. R. Appl. Environ.
Microbiol. 1998, 64, 2763.
(279) Wegmann, U.; Klein, J. R.; Drumm, I.; Kuipers, O. P.; Henrich,
B. Appl. Environ. Microbiol. 1999, 65, 4729.
(280) Fernandez, M.; van Doesburg, W.; Rutten, G. A.; Marugg, J. D.;
Alting, A. C.; van Kranenburg, R.; Kuipers, O. P. Appl. Environ.
Microbiol. 2000, 66, 42.
(281) Sakamoto, K.; Margolles, A.; van Veen, H. W.; Konings, W. N.
J. Bacteriol. 2001, 183, 5371.
(282) Henrich, B.; Klein, J. R.; Weber, B.; Delorme, C.; Renault, P.;
Wegmann, U. Appl. Environ. Microbiol. 2002, 68, 5429.
(283) Neu, T.; Henrich, B. Appl. Environ. Microbiol. 2003, 69, 1377.
(284) Bermudez-Humaran, L. G.; Langella, P.; Commissaire, J.; Gilbert, S.; Le Loir, Y.; LHaridon, R.; Corthier, G. FEMS Microbiol.
Lett. 2003, 224, 307.
(285) Blatny, J. M.; Ertesvag, H.; Nes, I. F.; Valla, S. FEMS Microbiol.
Lett. 2003, 227, 229.
(286) Hickey, R. M.; Ross, R. P.; Hill, C. Appl. Environ. Microbiol.
2004, 70, 1744.
(287) Simoes-Barbosa, A.; Abreu, H.; Silva Neto, A.; Gruss, A.;
Langella, P. Appl. Microbiol. Biotechnol. 2004, 65, 61.
(288) Chandrapati, S.; OSullivan, D. J. FEMS Microbiol. Lett. 1999,
170, 191.
(289) Stein, T.; Heinzmann, S.; Kiesau, P.; Himmel, B.; Entian, K. D.
Mol. Microbiol. 2003, 47, 1627.
(290) Kleerebezem, M.; Quadri, L. E. Peptides 2001, 22, 1579.
(291) Dunman, P. M.; Murphy, E.; Haney, S.; Palacios, D.; TuckerKellogg, G.; Wu, S.; Brown, E. L.; Zagursky, R. J.; Shlaes, D.;
Projan, S. J. J. Bacteriol. 2001, 183, 7341.
(292) Dunman, P. M.; Murphy, E.; Haney, S.; Palacios, D.; TuckerKellogg, G.; Wu, S.; Brown, E. L.; Zagursky, R. J.; Shlaes, D.;
Projan, S. J. J. Bacteriol. 2001, 183, 7341.
(293) Kies, S.; Vuong, C.; Hille, M.; Peschel, A.; Meyer, C.; Gotz, F.;
Otto, M. Peptides 2003, 24, 329.
(294) Peschel, A.; Augustin, J.; Kupke, T.; Stevanovic, S.; Gotz, F. Mol.
Microbiol. 1993, 9, 31.
(295) Peschel, A.; Gotz, F. J. Bacteriol. 1996, 178, 531.
(296) Peschel, A.; Schnell, N.; Hille, M.; Entian, K.-D.; Gotz, F. Mol.
Gen. Genet. 1997, 254, 312.
(297) Guder, A.; Schmitter, T.; Wiedemann, I.; Sahl, H. G.; Bierbaum,
G. Appl. Environ. Microbiol. 2002, 68, 106.
(298) Hindre, T.; Le Pennec, J. P.; Haras, D.; Dufour, A. FEMS
(299) Sulavik, M. C.; Clewell, D. B. J. Bacteriol. 1996, 178, 5826.
(300) Sulavik, M. C.; Tardif, G.; Clewell, D. B. J. Bacteriol. 1992, 174,
3577.
65, 652.
(302) Haas, W.; Shepard, B. D.; Gilmore, M. S. Nature 2002, 415, 84.
(303) Razeto, A.; Giller, K.; Haas, W.; Gilmore, M. S.; Zweckstetter,
M.; Becker, S. Acta Crystallogr. Section D Biol. Cryst. 2004, 60,
746.
(304) Rawlinson, E. L.; Nes, I. F.; Skaugen, M. Biochimie 2002, 84,
559.
(305) Kuipers, O. P.; Beerthuyzen, M. M.; Siezen, R. J.; de Vos, W. M.
Eur. J. Biochem. 1993, 216, 281.
(306) Koponen, O.; Takala, T. M.; Saarela, U.; Qiao, M.; Saris, P. E.
J. FEMS Microbiol. Lett. 2004, 231, 85.
(307) Ra, R.; Beerthuyzen, M. M.; de Vos, W. M.; Saris, P. E. J.;
Kuipers, O. P. Microbiology 1999, 145, 1227.
(308) Stein, T.; Heinzmann, S.; Solovieva, I.; Entian, K.-D. J. Biol.
Chem. 2003, 278, 89.
(309) Saris, P. E.; Immonen, T.; Reis, M.; Sahl, H. G. Antonie van
(310) Siegers, K.; Entian, K. D. Appl. Environ. Microbiol. 1995, 61,
1082.
(311) Kraft, R.; Leinwand, L. A. Nucleic Acids Res. 1987, 15, 8568.
(312) Blight, M. A.; Holland, I. B. Mol. Microbiol. 1990, 4, 873.
(313) Garrido, M. C.; Herrero, M.; Kolter, R.; Moreno, F. EMBO J.
1988, 7, 1853.
(314) Song, H. Y.; Cramer, W. A. J. Bacteriol. 1991, 173, 2935.
(315) Pugsley, A. P. Mol. Gen. Genet. 1988, 211, 335.
(316) Higgins, C. F. Annu. Rev. Cell Biol. 1992, 8, 67.
(317) Panagiotidis, C. H.; Reyes, M.; Sievertsen, A.; Boos, W.; Shuman,
H. A. J. Biol. Chem. 1993, 268, 23685.
(318) Kerppola, R. E.; Shyamala, V. K.; Klebba, P.; Ames, G. F. J.
Biol. Chem. 1991, 266, 9857.
(319) Klein, C.; Entian, K. D. Appl. Environ. Microbiol. 1994, 60, 2793.
(320) Otto, M.; Peschel, A.; Gotz, F. FEMS Microbiol. Lett. 1998, 166,
203.
(321) Guder, A.; Schmitter, T.; Wiedemann, I.; Sahl, H. G.; Bierbaum,
(322) Brotz, H.; Josten, M.; Wiedemann, I.; Schneider, U.; Gotz, F.;
Bierbaum, G.; Sahl, H.-G. Mol. Microbiol. 1998, 30, 317.
(323) Hille, M.; Kies, S.; Gotz, F.; Peschel, A. Appl. Environ. Microbiol.
2001, 67, 1380.
(324) Fath, M. J.; Kolter, R. Microbiol. Rev. 1993, 57, 995.

(325) Reis, M.; Eschbach-Bludau, M.; Iglesias-Wind, M. I.; Kupke, T.;
Sahl, H. G. Appl. Environ. Microbiol. 1994, 60, 2876.
(326) Hoffmann, A.; Schneider, T.; Pag, U.; Sahl, H. G. Appl. Environ.
Microbiol. 2004, 70, 3263.
(327) Pag, U.; Heidrich, C.; Bierbaum, G.; Sahl, H. G. Appl. Environ.
Microbiol. 1999, 65, 591.
(328) Rince, A.; Dufour, A.; Uguen, P.; Le Pennec, J. P.; Haras, D.
Appl. Environ. Microbiol. 1997, 63, 4252.
(329) McAuliffe, O.; Hill, C.; Ross, R. P. Microbiology 2000, 146 (Pt
1), 129.
(330) Coburn, P. S.; Hancock, L. E.; Booth, M. C.; Gilmore, M. S. Infect.
Immun. 1999, 67, 3339.
(331) Kuipers, O. P.; Bierbaum, G.; Ottenwalder, B.; Dodd, H. M.;
Horn, N.; Metzger, J.; Kupke, T.; Gnau, V.; Bongers, R.; van den
Bogaard, P.; Kosters, H.; Rollema, H. S.; de Vos, W. M.; Siezen,
R. J.; Jung, G.; Gotz, F.; Sahl, H. G.; Gasson, M. J. Antonie van
(332) Bierbaum, G.; Reis, M.; Szekat, C.; Sahl, H. G. Appl. Environ.
Microbiol. 1994, 60, 4332.
(333) Ottenwalder, B.; Kupke, T.; Brecht, S.; Gnau, V.; Metzger, J.;
Jung, G.; Gotz, F. Appl. Environ. Microbiol. 1995, 61, 3894.
(334) Szekat, C.; Jack, R. W.; Skutlarek, D.; Farber, H.; Bierbaum,
(335) Kuipers, O. P.; Rollema, H. S.; Yap, W. M.; Boot, H. J.; Siezen,
R. J.; de Vos, W. M. J. Biol. Chem. 1992, 267, 24340.
(336) Bierbaum, G.; Szekat, C.; Josten, M.; Heidrich, C.; Kempter, C.;
Jung, G.; Sahl, H. G. Appl. Environ. Microbiol. 1996, 62, 385.
(337) Dodd, H. M.; Horn, N.; Gasson, M. J. Gene 1995, 162, 163.
(338) van Kraaij, C.; Breukink, E.; Rollema, H. S.; Bongers, R. S.;
Kosters, H. A.; de Kruijff, B.; Kuipers, O. P. Eur. J. Biochem.
2000, 267, 901.
(339) Rollema, H. S.; Kuipers, O. P.; Both, P.; de Vos, W. M.; Siezen,
R. J. Appl. Environ. Microbiol. 1995, 61, 2873.
(340) Hosoda, K.; Ohya, M.; Kohno, T.; Maeda, T.; Endo, S.; Wakamatsu, K. J. Biochem. (Tokyo) 1996, 119, 226.
(341) Szekat, C.; Jack, R. W.; Skutlarek, D.; Faerber, H.; Bierbaum,
(342) Forster, A. C.; Tan, Z. P.; Nalam, M. N. L.; Lin, H. N.; Qu, H.;
Cornish, V. W.; Blacklow, S. C. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 6353.
(343) Dedkova, L. M.; Fahmi, N. E.; Golovine, S. Y.; Hecht, S. M. J.
Am. Chem. Soc. 2003, 125, 6616.
(344) Starck, S. R.; Qi, X.; Olsen, B. N.; Roberts, R. W. J. Am. Chem.
Soc. 2003, 125, 8090.
(345) Li, Y. M.; Milne, J. C.; Madison, L. L.; Kolter, R.; Walsh, C. T.
Science 1996, 274, 1188.
(346) Muir, T. W.; Sondhi, D.; Cole, P. A. Proc. Natl. Acad. Sci. U.S.A.
1998, 95, 6705.
(347) Evans, T. C., Jr.; Xu, M. Q. Biopolymers 1999, 51, 333.
(348) Wang, L.; Schultz, P. G. Chem. Commun. 2002, 1.
(349) Paul, M.; van der Donk, W. A. Minirev. Org. Chem. 2005, 2, 23.
(350) O
sapay, G.; Prokai, L.; Kim, H. S.; Medzihradszky, K. F.; Coy,
D. H.; Liapakis, G.; Reisine, T.; Melacini, G.; Zhu, Q.; Wang, S.
H.; Mattern, R. H.; Goodman, M. J. Med. Chem. 1997, 40, 2241.
(351) O
sapay, G.; Goodman, M. J. Chem. Soc., Chem. Commun. 1993,
1955.
(352) O
sapay, G.; Zhu, Q.; Shao, H.; Chadha, R. K.; Goodman, M. Int.
J. Pept. Protein Res. 1995, 46, 290.
(353) Rew, Y.; Malkmus, S.; Svensson, C.; Yaksh, T. L.; Chung, N.
N.; Schiller, P. W.; Cassel, J. A.; DeHaven, R. N.; Goodman, M.
J. Med. Chem. 2002, 45, 3746.
(354) Kelland, J. G.; Arnold, L. D.; Palcic, M. M.; Pickard, M. A.;
Vederas, J. C. J. Biol. Chem. 1986, 261, 13216.
(355) Mohd Mustapa, M. F.; Harris, R.; Mould, J.; Chubb, N. A. L.;
Schultz, D.; Driscoll, P. C.; Tabor, A. B. Tetrahedron Lett. 2002,
43, 8363.
(356) Mustapa, M. F. M.; Harris, R.; Esposito, D.; Chubb, N. A. L.;
Mould, J.; Schultz, D.; Driscoll, P. C.; Tabor, A. B. J. Org. Chem.
2003, 68, 8193.
(357) Campiglia, P.; Gomez-Monterrey, I.; Longobardo, L.; Lama, T.;
Novellino, E.; Grieco, P. Tetrahedron Lett. 2004, 45, 1453.
(358) Yu, L.; Lai, Y. H.; Wade, J. V.; Coutts, S. M. Tetrahedron Lett.
1998, 39, 6633.
(359) Zhu, Y.; van der Donk, W. A. Org. Lett. 2001, 3, 1189.
(360) Gieselman, M. D.; Zhu, Y.; Zhou, H.; Galonic, D.; van der Donk,
W. A. Chembiochem 2002, 3, 709.
(361) Galonic, D.; van der Donk, W. A.; Gin, D. Y. Chem.-Eur. J. 2003,
24, 5997.
(362) Malabarba, A.; Pallanza, R.; Berti, M.; Cavalleri, B. J. Antibiot.
(Tokyo) 1990, 43, 1089.
(363) Ramseier, H. R. Arch. Mikrobiol. 1960, 37, 57.
(364) Epand, R. M.; Vogel, H. J. Biochim. Biophys. Acta 1999, 1462,
11.
(365) Wade, D.; Boman, A.; Wahlin, B.; Drain, C. M.; Andreu, D.;
Boman, H. G.; Merrifield, R. B. Proc. Natl. Acad. Sci. U.S.A.
1990, 87, 4761.
(366) Bessalle, R.; Kapitkovsky, A.; Gorea, A.; Shalit, I.; Fridkin, M.
FEBS Lett. 1990, 274, 151.
Chatterjee et al.
(367) Ruhr, E.; Sahl, H. G. Antimicrob. Agents Chemother. 1985, 27,
841.
(368) Sahl, H.-G. In Nisin and Novel Lantibiotics; Jung, G., Sahl, H.G., Eds.; ESCOM: Leiden, The Netherlands, 1991.
(369) Sahl, H. G.; Kordel, M.; Benz, R. Arch. Microbiol. 1987, 149,
120.
(370) Gao, F. H.; Abee, T.; Konings, W. N. Appl. Environ. Microbiol.
1991, 57, 2164.
(371) Garca Garcera, M. J.; Elferink, M. G.; Driessen, A. J.; Konings,
W. N. Eur. J. Biochem. 1993, 212, 417.
(372) Driessen, A. J.; van den Hooven, H. W.; Kuiper, W.; van de
Kamp, M.; Sahl, H. G.; Konings, R. N.; Konings, W. N. Biochemistry 1995, 34, 1606.
(373) Schueller, F.; Benz, R.; Sahl, H. G. Eur. J. Biochem. 1989, 182,
181.
(374) Kordel, M.; Benz, R.; Sahl, H. G. J. Bacteriol. 1988, 170, 84.
(375) Kordel, M.; Schuller, F.; Sahl, H. G. FEBS Lett. 1989, 244, 99.
(376) Benz, R.; Jung, G.; Sahl, H.-G. In Nisin and Novel Lantibiotics;
Jung, G., Sahl, H.-G., Eds.; ESCOM: Leiden, The Netherlands,
1991.
(377) Jack, R.; Benz, R.; Tagg, J.; Sahl, H. G. Eur. J. Biochem. 1994,
219, 699.
(378) Giffard, C. J.; Dodd, H. M.; Horn, N.; Ladha, S.; Mackie, A. R.;
Parr, A.; Gasson, M. J.; Sanders, D. Biochemistry 1997, 36, 3802.
(379) Martin, I.; Ruysschaert, J. M.; Sanders, D.; Giffard, C. J. Eur.
J. Biochem. 1996, 239, 156.
(380) Winkowski, K.; Ludescher, R. D.; Montville, T. J. Appl. Environ.
Microbiol. 1996, 62, 323.
(381) Breukink, E.; van Kraaij, C.; Demel, R. A.; Siezen, R. J.; Kuipers,
O. P.; de Kruijff, B. Biochemistry 1997, 36, 6968.
(382) Moll, G. N.; Clark, J.; Chan, W. C.; Bycroft, B. W.; Roberts, G.
C.; Konings, W. N.; Driessen, A. J. J. Bacteriol. 1997, 179, 135.
(383) Demel, R. A.; Peelen, T.; Siezen, R. J.; De Kruijff, B.; Kuipers,
O. P. Eur. J. Biochem. 1996, 235, 267.
(384) el-Jastimi, R.; Lafleur, M. Biochim. Biophys. Acta 1997, 1324,
151.
(385) van Kraaij, C.; Breukink, E.; Noordermeer, M. A.; Demel, R. A.;
Siezen, R. J.; Kuipers, O. P.; de Kruijff, B. Biochemistry 1998,
37, 16033.
(386) Okereke, A.; Montville, T. J. Appl. Environ. Microbiol. 1992, 58,
2463.
(387) Abee, T.; Rombouts, F. M.; Hugenholtz, J.; Guihard, G.; Letellier,
L. Appl. Environ. Microbiol. 1994, 60.
(388) Chikindas, M. L.; Novak, J.; Driessen, A. J.; Konings, W. N.;
Schilling, K. M.; Caufield, P. W. Antimicrob. Agents Chemother.
1995, 39, 2656.
(389) Breukink, E.; van Kraaij, C.; van Dalen, A.; Demel, R. A.; Siezen,
R. J.; de Kruijff, B.; Kuipers, O. P. Biochemistry 1998, 37, 8153.
(390) Van den Hooven, H. W.; Spronk, C. A.; Van de Kamp, M.;
Konings, R. N.; Hilbers, C. W.; Van de Ven, F. J. M. Eur. J.
Biochem. 1996, 235, 394.
(391) Matsuzaki, K. Biochim. Biophys. Acta 1999, 1462, 1.
(392) Chan, W. C.; Leyland, M.; Clark, J.; Dodd, H. M.; Lian, L. Y.;
Gasson, M. J.; Bycroft, B. W.; Roberts, G. C. FEBS Lett. 1996,
390, 129.
(393) Verheul, A.; Russell, N. J.; Van, T. H. R.; Rombouts, F. M.; Abee,
T. Appl. Environ. Microbiol. 1997, 63, 3451.
(394) Kordel, M.; Sahl, H. G. FEMS Microbiol. Lett. 1986, 34, 139.
(395) van Heijenoort, Y.; Gomez, M.; Derrien, M.; Ayala, J.; van
Heijenoort, J. J. Bacteriol. 1992, 174, 3549.
(396) Storm, D. R.; Strominger, J. L. J. Biol. Chem. 1974, 249, 1823.
(397) Linnett, P. E.; Strominger, J. L. Antimicrob. Agents Chemother.
1973, 4, 231.
(398) Reisinger, P.; Seidel, H.; Tschesche, H.; Hammes, W. P. Arch.
Microbiol. 1980, 127, 187.
(399) Brotz, H.; Bierbaum, G.; Markus, A.; Molitor, E.; Sahl, H.-G.
(400) Brotz, H.; Bierbaum, G.; Reynolds, P. E.; Sahl, H. G. Eur. J.
Biochem. 1997, 246, 193.
(401) Brotz, H.; Bierbaum, G.; Leopold, K.; Reynolds, P. E.; Sahl, H.
G. Antimicrob. Agents Chemother. 1998, 42, 154.
(402) Lazar, K.; Walker, S. Curr. Opin. Chem. Biol. 2002, 6, 786.
(403) Barna, J. C.; Williams, D. H. Annu. Rev. Microbiol. 1984, 38,
339.
(404) Williams, D. H. Nat. Prod. Rep. 1996, 13, 469.
(405) Somner, E. A.; Reynolds, P. E. Antimicrob. Agents Chemother.
1990, 34, 413.
(406) Lo, M. C.; Men, H.; Branstrom, A.; Helm, J.; Yao, N.; Goldman,
R.; Walker, S. J. Am. Chem. Soc. 2000, 122, 3540.
(407) Helm, J. S.; Chen, L.; Walker, S. J. Am. Chem. Soc. 2002, 124,
13970.
(408) McCafferty, D. G.; Cudic, P.; Frankel, B. A.; Barkallah, S.;
Kruger, R. G.; Li, W. Biopolymers 2002, 66, 261.
(409) Walker, S.; Helm, J. S.; Hu, Y.; Chen, L.; Rew, Y.; Shin, D.;
Boger, D. L. Chem Rev. 2005, 105, 449.
(410) Kurz, M.; Guba, W. Biochemistry 1996, 35, 12570.
(411) McCafferty, D. G.; Cudic, P.; Yu, M. K.; Behenna, D. C.; Kruger,
R. Curr. Opin. Chem. Biol. 1999, 3, 672.

(412) van Heijenoort, J. Glycobiology 2001, 11, 25R.
(413) van Heijenoort, J. Nat. Prod. Rep. 2001, 18, 503.
(414) Walsh, C. T.; Fisher, S. L.; Park, I. S.; Prahalad, M.; Wu, Z.
Chem. Biol. 1996, 3, 21.
(415) Healy, V. L.; Lessard, I. A.; Roper, D. I.; Knox, J. R.; Walsh, C.
T. Chem. Biol. 2000, 7, R109.
(416) Chan, W. C.; Bycroft, B. W.; Lian, L. Y.; Roberts, G. C. K. FEBS
Lett. 1989, 252, 29.
(417) Breukink, E.; van Heusden, H. E.; Vollmerhaus, P. J.; Swiezewska, E.; Brunner, L.; Walker, S.; Heck, A. J.; de Kruijff, B. J.
Biol. Chem. 2003, 278, 19898.
(418) Matsuzaki, K.; Murase, O.; Fujii, N.; Miyajima, K. Biochemistry
1995, 34, 6521.
(419) Morris, S. L.; Hansen, J. N. J. Bacteriol. 1981, 148, 465.
(420) Morris, S. L.; Walsh, R. C.; Hansen, J. N. J. Biol. Chem. 1984,
259, 13590.
(421) Choung, S. Y.; Kobayashi, T.; Takemoto, K.; Ishitsuka, H.; Inoue,
K. Biochim. Biophys. Acta 1988, 940, 180.
(422) Machaidze, G.; Seelig, J. Biochemistry 2003, 42, 12570.
(423) Navarro, J.; Chabot, J.; Sherrill, K.; Aneja, R.; Zahler, S. A.;
Racker, E. Biochemistry 1985, 24, 4645.
(424) Machaidze, G.; Ziegler, A.; Seelig, J. Biochemistry 2002, 41, 1965.
(425) Makino, A.; Baba, T.; Fujimoto, K.; Iwamoto, K.; Yano, Y.;
Terada, N.; Ohno, S.; Sato, S. B.; Ohta, A.; Umeda, M.; Matsuzaki, K.; Kobayashi, T. J. Biol. Chem. 2003, 278, 3204.
(426) Bierbaum, G.; Sahl, H. G. Arch. Microbiol. 1985, 141, 249.
(427) Bierbaum, G.; Sahl, H. G. J. Bacteriol. 1987, 169, 5452.
(428) Wright, G. D. Curr. Opin. Chem. Biol. 2003, 7, 563.
(429) Chung, K. T.; Dickson, J. S.; Crouse, J. D. Appl. Environ.
Microbiol. 1989, 55, 1329.
(430) Stevens, K. A.; Sheldon, B. W.; Klapes, N. A.; Klaenhammer, T.
R. Appl. Environ. Microbiol. 1991, 57, 3613.
(431) Boziaris, I. S.; Adams, M. R. Int. J. Food Microbiol. 1999, 53,
105.
(432) Carneiro de Melo, A. M.; Cassar, C. A.; Miles, R. J. J. Food Prot.
1998, 61, 839.
(433) Boziaris, I. S.; Adams, M. R. J. Appl. Microbiol. 2001, 91, 715.
(434) Ganzle, M. G.; Hertel, C.; Hammes, W. P. Int. J. Food Microbiol.
1999, 48, 37.
(435) Caroff, M.; Karibian, D. Carbohydr. Res. 2003, 338, 2431.
(436) Chihib, N.-E.; Crepin, T.; Delattre, G.; Tholozan, J.-L. FEMS
(437) Crandall, A. D.; Montville, T. J. Appl. Environ. Microbiol. 1998,
64, 231.
(438) Liu, C. Q.; Harvey, M. L.; Dunn, N. W. J. Gen. Appl. Microbiol.
1997, 43, 67.
(439) Froseth, B. R.; McKay, L. L. Appl. Environ. Microbiol. 1991, 57,
804.
(440) Gravesen, A.; Sorensen, K.; Aarestrup, F. M.; Knochel, S. Microb.
Drug Resist. 2001, 7, 127.
(441) Gravesen, A.; Kallipolitis, B.; Holmstrom, K.; Hoiby, P. E.;
Ramnath, M.; Knochel, S. Appl. Environ. Microbiol. 2004, 70,
1669.
(442) Cotter, P. D.; Guinane, C. M.; Hill, C. Antimicrob. Agents
Chemother. 2002, 46, 2784.
(443) Jarvis, B. J. Gen. Microbiol. 1967, 47, 33.
(444) Alifax, R.; Chevalier, R. J. Dairy Res. 1962, 29, 233.
(445) Jarvis, B.; Farr, J. Biochim. Biophys. Acta 1971, 227, 232.
(446) Ming, X.; Daeschel, M. A. J. Food Prot. 1993, 56, 944.
(447) Mazzotta, A. S.; Montville, T. J. J. Appl. Microbiol. 1997, 82,
32.
(448) Mazzotta, A. S.; Montville, T. J. Appl. Environ. Microbiol. 1999,
65, 659.
(449) Li, J.; Chikindas, M. L.; Ludescher, R. D.; Montville, T. J. Appl.

(450) Goulhen, F.; Meghrous, J.; Lacroix, C. J. Appl. Microbiol. 1998,
85, 387.
(451) Maisnier-Patin, S.; Richard, J. FEMS Microbiol. Lett. 1996, 140,
29.
(452) Garde, S.; Avila, M.; Medina, M.; Nunez, M. Int. J. Food
Microbiol. 2004, 96, 165.
(453) Mantovani, H. C.; Russell, J. B. Appl. Environ. Microbiol. 2001,
67, 808.
(454) Davies, E. A.; Falahee, M. B.; Adams, M. R. J. Appl. Bacteriol.
1996, 81, 139.
(455) Gill, H. S.; Shu, Q.; Leng, R. A. Vaccine 2000, 18, 2541.
(456) Biarc, J.; Nguyen, I. S.; Pini, A.; Gosse, F.; Richert, S.; Thierse,
D.; Van Dorsselaer, A.; Leize-Wagner, E.; Raul, F.; Klein, J.-P.;
Schoeller-Guinard, M. Carcinogenesis 2004, 25, 1477.
(457) Ellmerich, S.; Scholler, M.; Duranton, B.; Gosse, F.; Galluser,
M.; Klein, J. P.; Raul, F. Carcinogenesis 2000, 21, 753.
(458) Cao, M.; Helmann, J. D. J. Bacteriol. 2004, 186, 1136.
(459) Peschel, A.; Otto, M.; Jack, R. W.; Kalbacher, H.; Jung, G.; Gotz,
F. J. Biol. Chem. 1999, 274, 8405.
(460) Kramer, N. E.; Smid, E. J.; Kok, J.; De Kruijff, B.; Kuipers, O.
P.; Breukink, E. FEMS Microbiol. Lett. 2004, 239, 157.
(461) de Ruyter, P. G.; Kuipers, O. P.; de Vos, W. M. Appl. Environ.
Microbiol. 1996, 62, 3662.
(462) Froseth, B. R.; Herman, R. E.; McKay, L. L. Appl. Environ.
Microbiol. 1988, 54, 2136.
(463) Silber, K. R.; Keiler, K. C.; Sauer, R. T. Proc. Natl. Acad. Sci.
U.S.A. 1992, 89, 295.
(464) Beebe, K. D.; Shin, J.; Peng, J.; Chaudhury, C.; Khera, J.; Pei,
D. Biochemistry 2000, 39, 3149.
(465) Kawulka, K.; Sprules, T.; McKay, R. T.; Mercier, P.; Diaper, C.
M.; Zuber, P.; Vederas, J. C. J. Am. Chem. Soc. 2003, 125,
4726.
(466) Kawulka, K. E.; Sprules, T.; Diaper, C. M.; Whittal, R. M.;
McKay, R. T.; Mercier, P.; Zuber, P.; Vederas, J. C. Biochemistry
2004, 43, 3385.
(467) Skaugen, M.; Andersen, E. L.; Christie, V. H.; Nes, I. F. Appl.
(468) Gilmore, M. S.; Skaugen, M.; Nes, I. Antonie van Leeuwenhoek
1996, 69, 129.
(469) Sayle, R. A.; Milner-White, E. J. Trends Biochem. Sci. 1995, 20,
374.
(470) Yuan, J.; Zhang, Z. Z.; Chen, X. Z.; Yang, W.; Huan, L. D. Appl.
(471) Wakamiya, T.; Fukase, K.; Kitazawa, M.; Fujita, H.; Kubo, A.;
Maeshiro, Y.; Shiba, T. In Peptides; Rivier, J., Marshall, G., Eds.;
(472) Aso, Y.; Sashihara, T.; Nagao, J.; Kanemasa, Y.; Koga, H.;
Hashimoto, T.; Higuchi, T.; Adachi, A.; Nomiyama, H.; Ishizaki,
A.; Nakayama, J.; Sonomoto, K. Biosci. Biotechnol. Biochem.
2004, 68, 1663.
(473) Kruszewska, D.; Sahl, H. G.; Bierbaum, G.; Pag, U.; Hynes, S.
O.; Ljungh, A. J. Antimicrob. Chemother. 2003, 54, 648.
(474) Kodani, S.; Hudson, M. E.; Durrant, M. C.; Buttner, M. J.;
Nodwell, J. R.; Willey, J. M. Proc. Natl. Acad. Sci. U.S.A. 2004,
101, 11448.
(475) Uguen, P.; Hindre, T.; Didelot, S.; Marty, C.; Haras, D.; Le
Pennec, J. P.; Vallee-Rehel, K.; Dufour, A. Appl. Environ.
Microbiol. 2005, 71, 562.
(476) Rink, R.; Kuipers, A.; de Boef, E.; Leenhouts, K. J.; Driessen,
A. J. M.; Moll, G. N.; Kuipers, O. P. Submitted for publication.
CR030105V
Chem. Rev. 2005, 105, 685714
685
Thiopeptide Antibiotics
Mark C. Bagley,* James W. Dale, Eleanor A. Merritt, and Xin Xiong
School of Chemistry, Main Building, Cardiff University, Park Place, Cardiff, CF10 3AT, Wales, United Kingdom
Contents
1. Introduction
2. Isolation and Structure Elucidation
2.1. Piperidines and Dehydropiperidines
2.2. Dihydroimidazopiperidines
2.3. Trisubstituted Pyridines
2.4. Hydroxypyridine Thiopeptides
2.5. Unidentified Thiopeptides
3. Biosynthesis
4. Biological Properties
4.1. Ribosomal Inhibitors
4.2. Inhibition of Elongation Factor Tu
4.3. TipA Promotion
5. Total Synthesis
6. Future Perspectives
7. Acknowledgment
8. References
685
686
687
689
690
697
699
700
703
704
706
707
708
711
712
712
1. Introduction
The crisis currently facing antibacterial chemotherapy threatens to return our treatment of bacterial infections to the so-called dark age of a preantibiotic era with the alarming emergence of bacterial
strains resistant to conventional treatments.1 In the
face of this medical crisis, many resources have been
committed to improving the potency of existing
antibiotic classes, discovering new antibacterial agents
with novel modes of action, and understanding the
mechanisms of resistance that are adopted by different bacterial pathogens to overcome antibacterial
action. A discussion of the resistance mechanisms
used by antibiotic-producing organisms has been the
subject of a number of excellent reviews.2,3 Antibiotic
producers adopt different self-defense mechanisms in
order to avoid their own suicide, protecting themselves against extracellular drugs by inactivating
their antibiotic products, modifying the antibiotic
target sites, such as enzymes or ribosomes, or blocking the entrance of active compounds into the cell.
Characterizing the strategies used by either the
producers or related bacterial strains to avoid intoxication requires a detailed understanding of how each
antibiotic class functions as well as knowledge of the
biosynthetic machinery operating in the organism to
predict mechanisms of multi-drug resistance (MDR)
* To whom correspondence should be addressed. Phone: +44 29
2087 4029. Fax: +44 29 2087 4030. E-mail Bagleymc@cardiff.ac.uk.
prior to their clinical emergence in the design of new

or strategically modified treatments. This review
concentrates upon a single class of antibacterial
agent, the thiopeptide antibiotics, the first of which,
micrococcin, was isolated in 1948. Our understanding
of the biological properties of this agent, the parent
of the thiopeptides thiostrepton, discovered later in
1954, and other less well-studied members of this
antibiotic class has developed considerably in recent
years. Here we draw together structural information
on the thiopeptides, defining all of the known members of this family, to discuss the structural basis of
their biological properties and its relevance to how
producing and nonproducing bacterial strains resist
their action. Current synthetic technology for the
assembly of these complex targets in the chemical
laboratory is also presented along with advances that
these studies have made in our understanding of this
rapidly expanding family.
The thiopeptide antibiotics are naturally occurring,
sulfur-containing, highly modified, macrocyclic peptides, nearly all of which inhibit protein synthesis in
bacteria. These complex natural products, grouped
as thiazolyl peptides for Berdys classification of
antibiotics according to chemical structure,4 share a
number of common structural features: a tri- or
tetrasubstituted nitrogen heterocycle clustered in a
central polyazole domain that is part of a macrocyclic
framework consisting of modified heterocyclic residues, including thiazoles, oxazoles, and indoles, and
dehydroamino acids. These biologically active substances are secondary metabolites produced by actinomycetes, Gram-positive mycelial sporulating bacteria, largely of the genus Streptomyces that can be
subdivided into 29 different antibiotic families containing well over 76 structurally distinct entities.
Despite their chemical and taxonomical diversity,
many of them broadly seem to share a similar
biological profile, displaying almost no activity against
Gram-negative bacteria, whereas against Grampositive bacteria they are highly active inhibitors of
protein synthesis and are, in many cases, effective
against methicillin-resistant Staphylococcus aureus
(MRSA), a bacterial strain that is resistant to most
conventional treatments. This biological property, as
well as an increased understanding of their mode of
action and the failure of traditional therapies to
counter the emergence of bacterial resistance, has led
to renewed interest in this antibiotic class.
The development of screening programs to search
for new cyclic thiazolylpeptides with a similar bio-

Mark C. Bagley, born in 1968 in the United Kingdom, was educated at

the University of Oxford (D.Phil. in 1994 at the Dyson Perrins Laboratory,
Oxford, under the guidance of Professor L. M. Harwood). He worked with
Professor W. Oppolzer at the Universite de Gene`ve in Switzerland and
Professor C. J. Moody at Loughborough University and the University of
Exeter in the United Kingdom before taking an academic position at Cardiff
University in 1999, where he is now a Senior Lecturer in Organic
Chemistry. His main research interests concern heterocyclic chemistry,
the total synthesis of biologically active natural products, and microwaveassisted organic transformations.
James W. Dale, born in 1979 in England, obtained his B.S. degree at

Cardiff University, U.K., in 2000. He remained at Cardiff to study his Ph.D.
with Dr. Mark C. Bagley on thiopeptide antibiotics, graduating in 2004.
He now holds a postdoctoral research fellowship with Professor Richard
J. K. Taylor at York University, U.K., where he has been for 10 months
working toward the total synthesis of spirocyclic natural products. His
research interests include synthetic organic chemistry and natural product
chemistry.
logical profile has led to the rapid expansion of the

number of structurally distinct thiopeptide antibiotics
in recent years. Reinvestigating the chemical identity
of many known metabolites using modern methods
of structure elucidation has stimulated renewed
interest in their biological properties, including their
mode of action for the inhibition of bacterial protein
synthesis, biosynthetic mechanisms, and the origin
of resistance determinants, and has encouraged the
development of new and efficient methods for their
chemical synthesis and modification. This review
draws together all of these findings. Although our
technological capability has not yet managed to
provide us with facile synthetic access to many of
these complex antibiotics, or even verify the structure
of many of these agents, the total synthesis of
amythiamicin D, promothiocin A, and most recently
thiostrepton along with synthetic efforts toward
Bagley et al.
Eleanor A. Merritt, born in Cardiff, U.K., in 1982, obtained her M.Chem.

degree in Chemistry from Cardiff University in 2004. In October 2004
she began postgraduate studies in Dr. Mark C. Bagleys research group
in the School of Chemistry, Cardiff University, where she is working toward
the total synthesis of a range of thiopeptide antibiotics.
Xin Xiong was born in Liaoning, China, in 1977. He studied at Shenyang

Pharmaceutical University, China, where he obtained his B.S. degree in
Pharmaceutics in 1999. After receiving his Masters degree in Medicinal
Chemistry from Aston University, U.K., he joined Dr. Mark C. Bagleys
research group at the School of Chemistry, Cardiff University, for doctoral
studies in 2001. His current research interests are focused on synthetic
methodologies of heterocyclic compounds, especially pyridine derivatives,
and applications to thiopeptide natural product total synthesis.
micrococcin P1 may facilitate new studies of structurally diverse analogues in the future.
2. Isolation and Structure Elucidation

The individual chemical identity of structurally
distinct compounds isolated from natural sources and
classified as thiopeptide or thiazolylpeptide antibiotics is relatively diverse. The identification of a
plethora of unique structural motifs and unusual
functional groups assembled in a macrocyclic array
has provided us with a series of tantalizing heterocyclic chemical targets that a number of international
synthetic groups have found impossible to resist. Of
a more intriguing nature, despite many similarities
in biosynthetic origin, these secondary metabolites
have been isolated from a number of different strains
of actinomycetes, predominantly soil bacteria but also
from marine sources, and elicit a wide range of
different biological responses. The complexity of the
many mechanisms employed by pathogens to avoid
intoxication and used by these metabolites in order
Table 1. Thiopeptide Antibiotic Families Classified

According to Their Central Heterocyclic Domain
series a and b
bryamycin
(A-8506)a
Sch 18640
(68-1147)
siomycin
thiactina
thiopeptin
thiostrepton
series c
series d
Sch 40832 A10255
series e
glycothiohexide R
amythiamicin
MJ347-81F4
berninamycin
cyclothiazomycin
GE2270
GE37468
geninthiocin
methylsulfomycin
micrococcin
promoinducin
promothiocin
QN3323
radamycin
sulfomycin
thioactin
thiocillin
thiotipin
thioxamycin
YM-266183-4
multhiomycinb
nocathiacin
nosiheptide
S-54832
Figure 1. Structure of thiostrepton A and B.

b
Shown to be identical to thiostrepton. Shown to be

identical to nosiheptide.
to inhibit bacterial protein synthesis is only now

coming to light, facilitated by advances in crystallography, NMR spectroscopy, and our understanding
of Streptomyces transcriptional mechanisms and the
dynamic function of the bacterial ribosome. Although
further research may reveal that many of these
processes are interrelated, at present most can be
attributed to certain regions or motifs in metabolite
structure particular to certain families or groups of
families related by functional-group commonality.
For this reason, classifying thiopeptide antibiotics
according to structure, in particular, in the nature
of the central heterocyclic domain, also categorizes
their biological properties and is useful for highlighting structural relationships between the 29 different
antibiotic families identified to date. Examining the
structure of individual thiopeptides reveals that there
are essentially five distinct classes of these natural
products, assigned according to the oxidation state
of the central heterocyclic domain. Each class can be
further subdivided into families that group cyclic
peptides with a high degree of structural homology
or in some cases that were isolated from the same
antibiotic-producing organism (Table 1). It was Hensens who first suggested grouping thiopeptide antibiotics according to the structure and oxidation state
of the central heterocyclic domain to distinguish
between different constituents of the thiopeptins.5
Hensens classification system, in this review, has
been extended to describe the five distinct heterocyclic domains. Thus, the parent of the thiopeptide
antibiotics, thiostrepton as well as the siomycins are
classified as b series thiopeptides, whereas some of
the thiopeptin factors and Sch 18640 possess the fully
reduced a series central domain, tetrasubstituted
dehydropiperidine or saturated piperidine heterocycles, respectively. Clearly related to the series a
and b thiopeptides, the c series, which to date only
consists of a single antibiotic, Sch 40832, has an
unusual imidazopiperidine core of unique structure.
With increasing unsaturation, in line with Hensens
classification, by far the most prolific thiopeptide

class is the series d antibiotics, possessing a 2,3,6trisubstituted pyridine domain, which is shared by
19 different families including the first thiopeptide
to be isolated and identified, micrococcin. Finally
series e thiopeptides, such as nosiheptide, exhibit a
structurally related central motif, oxidized in comparison with their series d counterparts, containing
a tetrasubstituted hydroxypyridine.
2.1. Piperidines and Dehydropiperidines

All of the series a and b thiopeptide antibiotics
display antibacterial activity and, with a high degree
of structural homology, can be identified by their
piperidine or dehydropiperidine central heterocyclic
domain and bis-macrocyclic peptide backbone, containing quinaldic acid, thiazoline, dehydroalanine,
and dehydrodemethylvaline residues as well as a
number of thiazole heterocycles. There are four
families and 15 structurally distinct entities within
these two series, although the structural differences
are only very minor. Due to their complex nature,
extensive chemical degradation has been used to
determine structure as well as multiple NMR spectroscopic techniques and in some cases X-ray crystallographic data.
Thiostrepton (C72H85N19O18S5), sometimes called
thiostrepton A or A1, is often referred to as the parent
compound of the thiopeptide antibiotics (Figure 1).
First isolated from Streptomyces azureus in 1954,6-8
this secondary metabolite was found to be effective
against Gram-positive bacteria with activity comparable to that of the penicillins.9 However, despite a
very promising biological profile, thiostrepton has not
been developed for clinical use as resistance by the
bacterium develops before a therapeutic dose can be
reached, primarily as a consequence of its low aqueous solubility, a problem inherent with most of the
thiopeptide antibiotics.
Early developments in the identification of new
thiopeptide antibiotics paint a very confusing picture,
exemplified by thiostreptons story. Following isolation of this natural product in 1954, a new thiopep-
tide metabolite called bryamycin was isolated in 1955

from Streptomyces hawaiiensis, a strain of soil bacteria discovered in Hawaii.10 Bryamycin, often referred to by its trade name of A-8506, was shown in
1963 to be identical to both thiostrepton and thiactin
by extensive comparison of their hydrolysates, solving
just one of the ambiguities in chemical identity
inherent in early work on the isolation and identification of new thiopeptide antibiotics.11 Chemical
degradation has also been used to obtain thiazole and
quinoline fragments, determining the ratio of a
number of amino acid components of thiostrepton.12
The isolation of residues derived from (-)-alanine,
(-)-isoleucine, (-)-threonine, and (+)-cysteine, the
latter formed by hydrolysis of a thiazoline, inferred
the architecture of a number of structural motifs of
the natural product. However, Dorothy Crowfoot
Hodgkin made the first real breakthrough in thiopeptide structure determination using X-ray crystallographic methods on monoclinic crystals to confirm
previous structural hypotheses and elucidate the
absolute stereochemistry and constitution of thiostrepton,13 with the exception of the identity of the
dehydroalanine-containing side chain which was first
solved by Tori et al. on the basis of NMR experiments.14 Thiostrepton, later also isolated from Streptomyces laurentii,15 has been subjected to detailed 1H
and 13C NMR spectroscopic analyses by Hensens and
Albers-Schonberg,16 with reinvestigation in 1989 by
Floss et al., who used 2D NMR spectroscopic techniques on both unlabeled and biosynthetically multiple 13C-labeled samples to confirm the 1970 structural assignment.17 Furthermore, a recent investigation
by Hunter et al. solved the structure of a tetragonal
crystal form of thiostrepton using the anomalous
dispersive signal from sulfur collected at the Cu KR
wavelength and placed the coordinates in the public
domain.18 The closely related thiopeptide, known as
thiostrepton B (or A2) (C66H79N17O16S5), isolated from
S. azureus as a minor component with thiostrepton,
has received much less attention. Analysis by 13C
NMR spectroscopic methods elucidated the structure,
which indicates that thiostrepton B might be an
artifact from thiostrepton (Figure 1).19
The siomycins are a family of four thiopeptides
with structures closely related to that of thiostrepton.
The siomycin complex was first isolated from cultures
of Streptomyces sioyaensis in 1959 and found to be
active against Gram-positive bacteria and mycobacteria.20 Further purification of the crude preparation
10 years later21 found that this sulfur-containing
peptide antibiotic actually consisted of one major
constituent, siomycin A, and a number of minor
components, siomycin B, C, and D1, the latter of
which was only discovered 21 years after isolation of
the original antibiotic complex.22 Siomycin B (SIMB, C65H75N17O16S5) is derived from siomycin A (SIMA, C71H81N19O18S5) during storage, whereas siomycin
C(SIM-C,C72H82N18O19S5)andD1 (SIM-D1,C70H79N19O18S5)
are both true natural products of S. sioyaensis.
Degradative methods23 as well as a wide range of
spectroscopic experiments, in particular 13C and 1H
NMR studies,14,19,24 were carried out in order to
determine the relationship of the siomycin structure
Bagley et al.
Figure 2. Structure of the siomycins.
with the recently uncovered X-ray crystal structure

of thiostrepton13 and to explain the similar physicochemical and biological properties of all of these
antibiotics. The isolation and identification of chemical degradation products, by reduction, oxidation,
acidic hydrolysis, and ammonolysis, was supported
by extensive 1H, 13C, and nOe NMR spectroscopic
data as well as 15N NMR spectroscopy finally to
confirm that the siomycins differ from thiostrepton
only in a dehydroalanine-valine unit attached to the
quinaldic acid in place of an alanine-isoleucine residue (Figure 2).
The thiopeptins, produced by Streptomyces tateyamensis, were first characterized by Miyairi et al. and
inhibit Gram-positive bacteria with no significant
differences in their inhibition of protein synthesis in
cell-free Escherichia coli.25 Silica gel chromatography
separated the antibiotic complex into its main constituent, thiopeptin B, shown to be useful as a
growth-promoting feed additive for pigs and chickens26 and as a lactic-acidosis preventive in ruminants,27 and four minor components thiopeptin A1,
A2, A3, and A4. Unable to obtain a crystalline sample
suitable for X-ray crystallography, Hensens determined, on the basis of 1H and 13C NMR evidence, that
the thiopeptins actually consisted of two distinct
series of antibiotics, designated by the subscripts a
and b, but failed to isolate thiopeptin A2, whose
structure still remains undetermined. On the basis
of these studies, six distinct compounds were identified,thiopeptinA1a (C72H86N18O18S6),A1b (C72H84N18O18S6),
A3a(C65H79N17O15S6),A4a(C68H82N18O16S6),Ba(C71H84N18O18S6),
and Bb (C71H82N18O18S6), and two further components
that were analyzed as mixtures, thiopeptin A3b
(C65H77N17O15S6) and A4b (C68H80N18O16S6) (Figure 3).5
Figure 5. Structure of Sch 18640.
Figure 3. Structure of the thiopeptins.
Figure 4. Thiopeptin Ba hydrolysates 1.
The difference between the two series of thiopeptins

has its origin in the nature of, and in particular the
oxidation state of, the central heterocyclic domain,
the b series retaining the 1-piperidine moiety of
thiostrepton whereas the a series thiopeptin antibiotics (Ba, A1a, A3a, and A4a) possess a fully saturated
piperidine core, as confirmed by 13C NMR spectroscopic analysis, with relative and absolute stereochemistry defined by careful analysis of 1H NMR data
and consideration of biogenetic information.19,28
Structural assignments were supported by chemical degradation studies, the hydrolysis of the various
thiopeptin components yielding 2 mol of alanine and
1 mol of both threonine and valine.29 Furthermore,
although acidic hydrolysis of thiopeptin Ba resulted
in the isolation of two diastereoisomeric piperidines
1a,b (Figure 4) that differed in the configuration at
C-6, this was shown to be due to epimerization of the
(6S)-diastereoisomer (1a) during hydrolysis and thus
provided further confirmation of structure.30
Sch 18640 (C72H87N19O17S6), also referred to as 681147 complex, belongs to the series a thiopeptides
and was isolated as the major constituent of the
antibiotic complex produced by Micromonospora ar-
borensis.31 Differentiated from other related natural

products known at the time by TLC analysis, the
structure of Sch 18640 was established on the basis
of degradative and spectroscopic studies, employing
600 MHz 1H and 13C NMR spectroscopy in combination with plasma desorption mass spectrometry to
determine the molecular weight of the natural product. Comparing the 13C NMR spectrum of Sch 18640
with thiostrepton identified a piperidine rather than
dehydropiperidine central heterocyclic domain and
established that an amide linkage in thiostrepton had
been replaced by a thioamide function by observing
a characteristic downfield shift of 28.3 ppm in the
quaternary resonance of the thiocarbonyl group.
Accounting for these distinctions and supported by
amino acid analysis and hydrolysis experiments, the
close relationship between the structure of thiostrepton A and Sch 18640 is now firmly established
(Figure 5). Although no stereochemical investigation
of Sch 18640 has been carried out, the configuration
of the 18 stereogenic centers in this metabolite can
be inferred tentatively by comparison with other
series a and b thiopeptides.
2.2. Dihydroimidazopiperidines
Sch 40832 (C84H104N18O26S5) is the only example
of a series c thiopeptide and was isolated as the minor
component from the antibiotic complex, referred to
as 13-384 complex, produced by Micromonospora
carbonecea var. africana (ATCC 39149).32 Purified on
reverse-phase silica gel, its potent in vitro activity
was determined using a disc-diffusion agar plate
assay against Gram-positive bacteria. Analysis by IR
spectroscopy confirmed the presence of NH, OH, and
amide functional groups, amino acid analysis provided 1 mol of cysteine, 4 mol of threonine, and 1 mol
of lysine, FAB mass spectrometry determined the
molecular weight and molecular composition, and
NMR spectroscopic experiments, using a combination
of COSY, HMBC, and 13C techniques, elucidated the
connectivity of Sch 40832 (Figure 6) and established
its distinctiveness from both thiostrepton and Sch
18640. This thiopeptide possesses a unique and
unusual structure with a central domain consisting
of a fully saturated piperidine heterocycle fused to
Figure 6. Structure of Sch 40832.
an imidazoline ring derived from a modified thiazoline. In addition to the dihydroimidazo[1,5-a]piperidine, Sch 40832 contains a disaccharide moiety
attached to a threonine side chain, delineated in a
Hartman-Hahn (HOHAHA) experiment and tentatively assigned as -D-chromose A and B as well as a
deoxythiostreptine residue in the peptide backbone.
2.3. Trisubstituted Pyridines

Series d thiopeptides differ strikingly from the
piperidine or dehydropiperidine series a and b natural products. Predominantly they contain only one
macrocyclic peptide loop centered around a 2,3,6trisubstituted pyridine heteroaromatic domain clustered with thiazole and/or oxazole heterocycles with
a peptide side chain consisting of heterocyclic or
dehydrated amino acid residues attached at the
pyridine 6-position. Numerically, series d thiopeptides are the dominant class of thiopeptide antibiotics, with 19 families and over 49 distinct entities,
which show diversity in structure as well as biological
activity.
First discovered in 1948 from a strain of Micrococcus found in sewage from the city of Oxford and
reported in The British Journal of Experimental
Pathology,33 the isolate34 that was to be named
micrococcin35 is recorded as the first example of a
thiopeptide antibiotic, although no work to elucidate
the chemical structure of that material was ever
reported. An antibiotic with therapeutic activity later
obtained from the B. pumilus group of spore-bearing
bacillus, isolated from soil collected in East Africa,36
demonstrated a considerable degree of identity or
near-identity with the antibiotic isolated from Micrococcus, perhaps suggesting two members of a
closely related family, on the basis of their similar
properties and behavior, and so the new complex was
named micrococcin P, despite the taxonomic implications. This antibiotic actually consists of two distinct
components, present in the complex in ca. a 7:1 ratio
and designated micrococcin P1 and micrococcin P2,
respectively (alternatively written as micrococcin P1
and P2). Gratifyingly, with difficulties experienced
in obtaining the complex from these original sources,
Bagley et al.
micrococcin P1 has recently been obtained from food

borne Staphylococcus equorum WS2733 isolated from
French Raclette cheese,37 demonstrating how thiopeptide antibiotics can often be discovered from
seemingly unrelated sources. Work that spanned the
50-year period following initial isolation to elucidate
the structure of micrococcin P1 tells a remarkable
story, amazingly still unresolved, that, despite considerable spectroscopic advances and synthetic achievements, bears witness to a number of oversights and
the propagation of unsubstantiated hypotheses, accepted erroneously as the truth. With a lack of
structural data, the major isolated component, micrococcin P1 (also referred to as MP1), was characterized initially purely by examination of its acid
hydrolysates, which enabled a provisional molecular
weight estimate to be made.38 Analysis of the acidsoluble fraction gave a laevorotatory hydrochloride
that was identified conclusively as L-threonineHCl
by IR spectroscopy, thus confirming the identity and
stereochemistry of one of the residues of the natural
product. Further analysis of the acid-insoluble fraction established that micrococcin P1 must contain an
extended chromophoric system39 by isolating two
derivatives of the central heterocyclic domain identified as micrococcinic acid (2) and methyl micrococcinate (3) and separated an amino alcohol from the
acid-soluble hydrolysate of the peptide side chain.
Although this residue was assigned originally as
alaninol (2-aminopropan-1-ol),40 it was later identified as D-(R)-isoalaninol (2-hydroxypropylamine) from
13
C NMR spectroscopic data of the natural product.41
This study culminated in the Walker-Lukacs structure 4 of micrococcin P1 which, although a considerable advance, had assembled the order of the individual residues without evidence but rather based
upon an assumed structural homology with two other
thiopeptides, thiostrepton and nosiheptide. Prompted
by publication of the Walker-Lukacs structure,
Bycroft and Gowland separated micrococcin P1
(C48H49N13O9S6) and micrococcin P2 (C48H47N13O9S6)
and carried out their own NMR spectroscopic studies
and analyses of the acid hydrolysates of both micrococcin P1 and its sodium borohydride derivative.42 In
contrast to Walkers findings, only 1 mol of threonine
was produced in each of these hydrolysis experiments, leading to the proposal of the alternative
Bycroft-Gowland structure for both micrococcin P1
(5) and P2 (6) that accounted for their hydrolytic
behavior (Figure 7). The proposed structure was
accepted for over 20 years but was finally shown to
be erroneous when, in 1999, Ciufolini completed his
landmark total synthesis of MP1 and demonstrated
that this architecture did not correspond to that of
the natural product.43 This confusing situation has
been compounded further by Shins synthesis of two
epimeric substances of the Walker-Lukacs and Bycroft-Gowland structures, described as micrococcin
P44 and micrococcin P1 (7),45 respectively, containing
in each case an (S)-isoalaninol residue in the peptide
side chain in place of (R)-isoalaninol, established
unequivocally by combination of degradation studies40 and NMR spectroscopy.41 Ciufolini later drew
together all of these findings and validated the 1978
Figure 8. Structure of the thiocillins and thiopeptides

isolated from the marine sponge H. japonica.
Figure 7. Proposed structures for the micrococcins and

their chemical derivatives.
Bycroft-Gowland hypothesis with extensive NMR

studies, confirming that the order of residues in
micrococcin P1 was in accord with their original
proposal.46 It would appear, on the basis of these
results, that the difference between synthetic MP1
and the natural material, as examined by Bycroft and
Gowland in 1978, is purely stereochemical and
therefore must have its origin in either the configuration of the L-threonine-derived thiazole that forms
part of the central heterocyclic domain or in the (R)valine-derived thiazole in the peptide macrocycle,
both of which were proposed in the absence of reliable
experimental evidence. The stereochemical assignment of the latter was derived originally from chemical degradation studies, but the hydrochloride salt
of the key hydrolysate, isolated from the natural
product, was identified as both (+)- and (-)-2-(1amino-2-methylpropyl)thiazole-4-carboxylic acid in
separate studies,38 with a specific rotation that varied
for the latter between 0 and -20.8, leading the
authors to conclude that the stereochemistry of the
valine-derived residue is probably of the D-configuration.47 Considering these instrumental limitations

and Ciufolinis findings, it would appear that the
most probable structure for MP1 (8) and MP2 (9) is
at variance in the stereochemical assignment of this
unit, but it is hoped this longstanding mystery will
soon be solved by chemical synthesis.
A number of thiopeptides structurally related to
the micrococcins were isolated subsequently from the
Bacillus genus, including the thiocillins I and II,
isolated from the cultured broth of Bacillus cereus
G-15, and thiocillins II and III, isolated from Bacillus
badius AR-91.48 TLC analysis was used to differentiate between these cyclic peptides and both micrococcin P1 and P2 and provided a preliminary indication
that the thiocillins are in fact produced by other
Bacillus species, including Bacillus megatherium I-13
and strain AR-140, preliminarily identified with B.
pumilus, although these studies have not been substantiated with any other corroborating data. Structure analysis by 1H and 13C NMR spectroscopy
supported by evidence from chemical degradation
studies, including reduction, hydrogenolysis, and
hydrolysis, identified the individual structural components of all of these antibiotics and established
their connectivity and structural relationship with
the micrococcin thiopeptide antibiotics (Figure 8).49
Very recently the related cyclic thiazolylpeptides
YM-266183(C48H47N13O10S6)andYM-266184(C49H49N13O10S6) were found in the cultured broth of B. cereus
QN3323, isolated from the marine sponge Halichondria japonica, representing the first family of thiopeptide natural products to be derived from a marine
source.50 Extensive analytical studies were undertaken to elucidate their structure using high-resolution MALDI-TOF mass spectrometric studies to determine the molecular formulas, a wide range of NMR spectroscopic methods, in particular HMBC experiments, to establish the connectivity of individual
residues within these natural products, and ROESY
NMR correlations to indicate the Z-configuration of
propenyl groups in two dehydroamino acid units, one
in the macrocycle and one in the peptide side chain
(Figure 8).51 Both factors exhibit potent antibacterial
activity against staphylococci and enterococci includ-
Bagley et al.
Figure 9. Structure of the QN3323 factors.
ing multiple drug-resistant strains, the MIC for YM266183 and YM-266184 against S. aureus CAY 27701
(MRSA) being 0.78 and 0.39 g/mL, respectively, but
both are inactive against Gram-negative bacteria.
The thiocillins have also been isolated from other
cultures of the Bacillus genus that in addition yielded
the QN3323 compounds, a family of three thiopeptides denoted factors A, B, and Y1 with antibacterial
properties.52 Although the structure of these natural
products seems ambiguous, in particular relating to
the configuration of the propenylthiazole in the
peptide backbone, and requires the support of additional spectroscopic evidence, clearly these metabolites are closely related to the micrococcins and
thiocillins, also produced by species of Bacillus
(Figure 9).
Following isolation of the micrococcins, two series
d thiopeptide families, the berninamycins and sulfomycins, were identified as specific inhibitors of
bacterial protein synthesis in 1969. The berninamycins are a family of four metabolites, isolated from
the fermentation extract of Streptomyces bernensis,
shown to interfere with amino acid incorporation into
peptides.53 Initial structure determination studies on
berninamycin A (C51H51N15O15S), which it is proposed
corresponds to the purified berninamycin featured in
earlier reports, and berninamycin B (C51H51N15O14S),
carried out in 1975 by Liesch and Rinehart,54 used
NMR spectroscopic analyses of water-soluble sodium
and ammonium salt derivatives along with chemical
degradation studies by trifluoroacetolysis, sodium
borohydride reduction, and catalytic hydrogenolysis.54,55 The first structural hypothesis likened the
heterocyclic core of the berninamycins to that of its
acidic hydrolysate, berninamycinic acid (10), the
structure of which had been elucidated by X-ray
crystallographic studies56 and chemical synthesis.57
Following further investigation, a revised structure
for berninamycin A was postulated by Abe58 and
subsequently confirmed by Rinehart et al. in 1994.59
BerninamycinB,C(C48H48N14O14S),andD(C45H45N13O13S)
are minor components of the berninamycin complex
and were characterized at the same time by 13C NMR
spectroscopy and FAB mass spectrometry. The structure of berninamycin B has a valine residue in place
of a -hydroxyvaline unit found in the peptide macrocycle of berninamycin A, whereas berninamycin D
has two fewer dehydroalanine units in the pyridine-
Figure 10. Structure of the berninamycins.
6-carboxamide side chain (Figure 10). On the basis

of FAB mass spectrometry, the least abundant berninamycin component C, isolated in a quantity of only
about 1 mg from 1 g of the antibiotic complex, is
postulated to have a short peptide side chain containing only a single dehydroalanine residue. Stereochemical information can be garnered from biosynthetic studies using 13C-enriched L-valine and
threonine.60 As well as showing antibacterial activity,
the berninamycins also show tipA promoter activity
at nanomolar concentrations as regulators of gene
expression.61 A thiopeptide named neoberninamycin
was isolated from Micrococcus luteus, and results
from 1H NMR and mass spectrometric studies have
led to the conclusion that this compound is similar
in structure to berninamycin A but not identical,62
although the complete structure of neoberninamycin
has yet to be determined.
The sulfomycins, first isolated in the same year,
comprise a family of three cyclic peptides consisting
of sulfomycin I (C54H52N16O16S2), II (C54H52N16O15S2),
and III (C53H50N16O16S2). Sulfomycin I was obtained
from Streptomyces viridochromogenes subsp. sulfomycini ATCC 29776 and exhibits strong inhibitory
activity against Gram-positive bacteria,63 whereas all
three have been isolated from S. viridochromogenes
MCRL-0368. The structure of sulfomycin I was
determined using a combination of 1H and 13C NMR
spectroscopic techniques, FAB mass spectrometric
analysis,53 and chemical evidence. Acidic methanolysis provided a number of different fragments including dimethyl sulfomycinamate (11),64 the structure
of which has been confirmed by X-ray crystallographic studies and chemical synthesis (Figure 11).65,66
Sulfomycin II and III, isolated from subspecies
MCRL-0368, have closely related structures that vary
only in the nature of the side chain (R2) located on a
2-(2-aminoalkenyl)oxazole residue in the peptide
Figure 12. Structure of the A10255 factors.
Figure 11. Structure of the sulfomycins and dimethyl

sulfomycinamate (11).
backbone.67 These cyclic peptides share a common

oxazole-thiazole-pyridine-type d central heterocyclic domain and, as well as a prevalence of dehydroamino acids, contain an unusual alkoxythiazolylmethyl amide unit characteristic of this thiopeptide
family. All of the sulfomycins strongly inhibit the
growth of Gram-positive bacteria, including methicillin-resistant S. aureus, but are not active against
Gram-negative bacteria.
A number of related thiopeptide families, discovered subsequently, share the same oxazole-thiazolepyridine-type d domain, including the A10255 factors,
geninthiocin, methylsulfomycin, promoinducin, the
promothiocins, radamycin, thioactin, thiotipin, and
thioxamycin. The thiopeptide antibiotic complex
A10255, isolated from Streptomyces gardneri NRRL
15537,68 or cultures of NRRL 18260, a higher producing strain derived from NRRL 15922 by nitrosoguanidine mutagenesis,69 has been shown to exhibit strong
antimicrobial activity against Gram-positive bacteria
as well as promote growth and alleviate acidosis in
ruminants. This multicomponent complex (designated A10255B, -C, -E, -F, -G, -H, and -J), extracted
from the mycelia formed in submerged cultures of the
organism, was separated by chromatography to provide four major constituents, isolable in sufficient
quantity to elucidate structural data, factors B
(C53H48N16O15S3),E(C54H50N16O15S3),G(C52H46N16O15S3),
and J (C43H38N14O11S3). The ratio of factors produced
by the culture in the A10255 complex was found to
be dependent upon the medium, cobalt, or cobalamin
supplementation leading to a significant increase in
the BE/G ratio, which was consistent with an increase in methionine biosynthesis and provision of a
methyl donor for the biosynthetic methylation of
A10255G.70 A combination of collisionally induced
dissociation (CID) and FAB mass spectrometry, NMR
spectroscopic studies, and selective chemical degradation by reduction, methanolysis, acidic hydrolysis
and trifluoroacetolysis provided the complete structure of the major factors of A10255 (Figure 12).71
Three of these components are identical except for
the extent of methylation on a dehydroamino-acidderived oxazole residue in the peptide macrocycle,
factors G, B, and E being derived from dehydrobutyrine, dehydronorvaline, and dehydronorleucine residues, respectively. A10255J has a similar structure
to A10255G, with a masked dehydrobutyrine residue,
but differs by a single amidated dehydroalanine unit
in the side chain on the central pyridine domain. No
stereochemical information has been described for the
A10255 thiazolyl peptides regarding the Z/E-geometry of the 2-(1-prop-1-enyl)oxazole-4-carboxylate
residue or the (R/S)-configuration of the three stereogenic centers, although the successful incorporation
of the isotopic label of L-[1-13C]threonine in A10255
factors G, B, and E does indicate the stereochemistry
of this residue in all of these metabolites.72
The discovery of sensitive and specific thiopeptide
screening technology in the 1990s enabled Seto to
isolate a number of different but related antibiotic
families from microorganisms using a novel tipA
promoter inducing activity assay. Inserting the promoter (ptipA) of the tipA gene into a promoter probe
vector73 enabled thiostrepton-like compounds to be
identified by their ability to initiate transcription of
ptipA.74 In this manner, geninthiocin (C50H49N15O15S)
was isolated from Streptomyces sp. DD84 and its
structure determined by a series of spectroscopic
experiments using UV, IR, COSY, HMQC, HMBC,
and 1H and 13C NMR techniques. High-resolution
FAB mass spectrometry identified oxazole and thiazole heterocycles along with several unusual amino
acids (Figure 13) and distinguished this antibiotic
from neoberninamycin, a cyclic thiazolyl peptide of
unknown structure.62 Further analysis established
the L-configuration of the threonine residue in the
peptidic macrocycle by chiral-TLC and determined
the Z-configuration of the 2-(1-aminoprop-1-enyl)oxazole by nOe 1H NMR spectroscopic experiments.
Further screening experiments by Seto on cultures
of Streptomyces sp. SF2741 harvested two thiopeptidesfromthemycelialcake,promothiocinA(C36H37N11O8S2)
and promothiocin B (C42H43N13O10S2).75 A combina-
Bagley et al.
Figure 13. Structure of geninthiocin.

Figure 15. Structure of Promoinducin and Thiotipin.
Figure 14. Structure of the promothiocins.
tion of IR spectroscopy, high-resolution FAB mass

spectrometry, 2D NMR experiments, and amino acid
analyses, which provided 1 mol each of alanine,
glycine, and valine, elucidated the constitution of
both of the promothiocin factors but did not provide
any insight into the stereochemical assignment of
these antibiotics. However, the chemical synthesis of
promothiocin A by Moody and Bagley76,77 established
unequivocally the (S)-configuration of the three stereogenic centers in the natural product, and this was
later supported by degradation studies and molecular
modeling (Figure 14).78 Chiral-HPLC analysis of the
acid hydrolysates confirmed the L-configuration of
both alanine and valine, whereas the (S)-2-(1-amino1-ethyl)thiazole-4-carboxylate residue provided the
best calculated fit in a DADAS90 conformational
study. Minimum tipA promoter induction concentrations of promothiocins A and B were reported as 0.2
and 0.1 g/mL, respectively.75
Further investigation of the metabolites produced
by Streptomyces sp. SF2741, the producing strain of
the promothiocin antibiotics, isolated the thiazolyl
peptide promoinducin (C57H54N16O18S2) from the mycelial extract. Its structure was elucidated by 1H, 13C,
COSY, NOESY, HSQC, and 13C-decoupled HMBC
spectroscopic experiments as well as high-resolution
FAB mass spectrometry and contains a dehydroalanine tetrapeptide side chain and oxazole-thiazolepyridine central domain (Figure 15).79 The configuration of the L-threonine residue in the macrocycle
was established by chiral-TLC analysis of the acid
hydrolysate (6 N HCl, 110 C, 20 h), whereas the
geometrical configuration of the -methine and propenyl groups was established as Z in both cases from
NOESY data. This thiopeptide, related to the sulfomycins, promothiocins, and geninthiocin, shows activity against Gram-positive bacteria including M.
luteus, Streptococcus pneumoniae, Streptococcus pyogenes, and methicillin-resistant S. aureus at minimum inhibitory concentrations of 0.39, 0.1, 0.1, and
1.56 g/mL, respectively, and acts as a tipA promoter
at 40 ng/mL.
Thiotipin (C55H50N16O17S2), a series d thiopeptide
that is structurally related to promoinducin, was also
isolated by Seto from the mycelium of Streptomyces
sp. DT31 as a tipA promoter inducing substance.80
Structure elucidation using a combination of highresolution FAB mass spectrometry and 1H, 13C,
COSY, HMQC, and nOe NMR spectroscopic techniques, the latter of which assigned the Z-configuration of the propenyloxazole units, showed considerable structural homology with promoinducin with
only three points of variance, including the length of
the polydehydroalanine peptide side chain (Figure
15). The configuration of the L-threonine residue was
established by chiral TLC analysis of the acid hydrolysate, but these studies could not assign the
stereochemistry of the remaining unusual hydroxyaminoamide. Thiotipin was reported to show antibacterial activity against S. pneumoniae, S. pyogenes, and
M. luteus at the level of 3-6 g/mL and a minimum
induction concentration of 80 ng/mL for tipA promoter inducing activity.
Two closely related thiopeptides, thioxamycin
(C52H48N16O15S4) and its simpler derivative thioactin
(C43H40N14O11S4), were both isolated from the mycelium cake of Streptomyces sp. DP94 screening for tipA
promoter inducing activity,81 although the former has
also been found in the cultured broth of another
strain PA-46025.82 The distinct nature of thioxamycin
was apparent from its acidity, the free carboxylic acid
in the dehydroalanine side chain distinguishing the
structure of this metabolite from the related sulfomycins. Furthermore, treatment with aqueous acid
and analysis of the hydrolysates determined that
thioxamycin contained 1 mol of L-threonine, an (R)2-[1-amino-2-(methylthio)ethyl]oxazole unit, and two
other thiazole residues. Although no stereochemical
investigation has been carried out on thioactin,
tentatively it can be assumed to possess the same
Figure 18. Structure of radamycin.
Figure 16. Structure of thioactin and thioxamycin.
Figure 17. Structure of methylsulfomycin.
stereochemistry as 1H and 13C NMR spectroscopic

analyses of these metabolites connected their structures and confirmed the constitution of both natural
products (Figure 16). Minimum induction concentrations of thioxamycin and thioactin for tipA promoter
activity were 80 and 40 ng/mL, respectively, whereas
activity against aerobic Gram-positive bacteria including S. pyogenes C-203 and S. pneumoniae Type
I as well as against anaerobic bacteria such as
Streptococcus constellatus ATCC 27823 were 0.39,
0.78, and 6.25 g/mL, respectively.
Methylsulfomycin (C55H54N16O16S2), also sometimes
referred to as methylsulfomycin I,83 was first isolated
from Streptomyces sp. HIL Y-9420704. Mass spectrometry, chemical analysis, and COSY, 1H, and 13C
NMR spectroscopic experiments determined that this
thiopeptide differed structurally from sulfomycin by
only a single 5-methyl group in the oxazole of the
central domain (Figure 17).84 Curiously, this change
would appear to make methylsulfomycin much more
sensitive to oxygen, although its stability was improved sufficiently in structural analyses by the use
of degassed solvents. Further nOe NMR spectroscopic
experiments assigned the Z-configuration of both
double bonds in the dehydroamino acid residues of
the macrocyclic peptide.
Methylsulfomycin has also recently been isolated
from the fermentation broth of Streptomyces sp.
RSP9, where it was identified by comparison of its
NMR and mass spectra with previously reported
data, along with an unusual cyclic peptide, radamy-
cin (C48H47N15O11S3), a very strong inducer of the tipA

gene that possesses no antibacterial activity.85 The
structure of radamycin was elucidated by IR spectroscopy, high-resolution FAB mass spectrometry,
and 1D and 2D NMR spectroscopic experiments and
is closely related to that of methylsulfomycin, with
an oxazole-thiazole-pyridine central domain (Figure
18).86 The ability of radamycin to induce the tipA
promoter without itself having any antibacterial
activity, while being a curious feature, may provide
a number of future research opportunities for the
construction of inducible tipAP vectors lacking a
thiostrepton resistance gene (tsr) or in the industrialscale production of proteins, where the addition of
antibiotics may cause the selection of thiopeptideresistant strains.
A number of novel pyridine-containing cyclic peptide antibiotics were discovered in the 1990s from
screening programs designed to detect inhibitors of
protein synthesis. The GE2270 (MDL 62,879) class
of series d thiopeptides is the largest family of these
natural products with over 12 structurally related
components (Figure 19).87 The main factor of this
complex, GE2270A (C56H55N15O10S6), isolated in a
screening program designed to detect inhibitors of
protein synthesis from the fermentation broth of
Planobispora rosea ATCC 53773, was extracted from
the mycelium and purified by column chromatography on silica gel.88 The original structure, proposed
for this factor on the basis of chemical degradation,
UV, IR, and NMR spectroscopic studies89 and mass
spectrometric techniques,90 was revised upon further
experimentation, correcting the sequence of thiazole
amino acids in the cyclic peptide through analysis of
the hydrolysates91 with some assignment of absolute
stereochemistry.92 The isolation and characterization
of nine of the minor components, separated from the
GE2270 antibiotic complex by HPLC, was reported
in 1995 and facilitated by altering the fermentation
conditions, a study that showed that P. rosea modifies
the GE2270 backbone by introducing a variable
number of methylene units.93 The structures of the
individual factors, E, D1, D2, C1, C2a and C2b (not
to be confused with subscripts a and b used to denote
differences in the central domain in the thiopeptin
factors), B1, B2, and T were determined by 2D NMR
spectroscopy and differ from GE2270A in the nature
of the 5-substituent on two thiazole residues in the
peptide backbone, the asparagine N-substituent, and
Bagley et al.
Figure 20. Structure of GE37468A.
Figure 19. Structure of the GE2270 factors.
the oxidation state of the azole in the 6-pyridine side

chain. All of the GE2270 factors were found to inhibit
protein synthesis in Gram-positive microorganisms
and anaerobes,93 with particular activity noted against
Propionibacterium acnes (MIC for GE2270A against
L1014 ATCC 6919 < 0.004 g/mL) and Mycobacterium tuberculosis (MIC for GE2270A of 1 g/mL),88
as well as being quite active against some Gramnegative bacteria. These antibiotics inhibit bacterial
protein synthesis by acting specifically on the GTPbound form of Ef-Tu,94 the elongation factor required
for the binding of aminoacyl-tRNA to the ribosomal
A site, and so function in a fashion similar to
kirromycin-like antibiotics and pulvomycin, although
their spectrum of antibacterial activity is quite different.
The inhibition of bacterial protein synthesis through
binding to elongation factor Tu was subsequently
incorporated in a screening program to isolate antibiotics with a similar structure and mode of action
to GE2270A. In this fashion the thiazolylpeptide
GE37468A (C59H52N14O12S5) was obtained from the
fermentation broth of Streptomyces sp. ATCC 55365
isolated from a soil sample collected in Italy, although
the productivity and reproducibility of the isolate was
poor (typically < 10 mg/L).95 The selection of morphochromatic spontaneous phenotypes led to the
isolation of a stable high-yielding variant ATCC
55365/O/5, increasing antibiotic production dramatically with respect to the parent strain.96 Structure
elucidation by 1H and 13C NMR spectroscopy and
FAB mass spectrometric analysis on both the parent
compound and its hydrolysates showed that this
thiopeptide was related to both the amythiamicins
and the GE2270 factors, although the 3-thiazolyl

substituent on the pyridine in GE2270A is replaced
in GE37468A with a 3-(4-methyloxazolyl) group and
two extra dehydroalanine units are found in the
6-pyridine side chain along with an unusual 5-hydroxyproline residue in the peptide macrocycle (Figure 20).97 Two other thiopeptide factors have been
attributed to this family, GE37468B and -C; however,
no structural data have been published to date.98
GE37468A is highly active in vitro against Grampositive bacteria (MIC against P. acnes ATCC 6919
and S. aureus Smith is 2 and 16 g/L, respectively),
selective for prokaryotic protein synthesis by acting
on Ef-Tu in comparative studies in cell-free E. coli
and rat liver systems, and protects mice against S.
aureus infection.95
The amythiamicins were detected by a paper disk
diffusion screen based upon their in vitro antibacterial activity against S. aureus Smith and were obtained from the fermentation broth of Amycolatopsis
sp. MI481-42F4, isolated from soil samples collected
in Tokyo, Japan.99 These antibiotics, although inactive against most Gram-negative bacteria and fungi,
inhibit the growth of Gram-positive bacteria, including multi-drug-resistant strains such as S. aureus
MS9610 and methicillin-resistant S. aureus (MRSA),100 and show no signs of toxicity when administered intraperitoneally to mice at a dose of 100 mg/
kg.99 The structure of all of the components was
elucidated by a combination of NMR spectroscopic
techniques, chemical degradation, and FAB mass
spectrometry.101,102 Amino acid analyses of amythiamicin A (C50H51N15O8S6), B (C50H53N15O9S6), and C
(C50H50N14O9S6) found them to contain 1 mol of
glycine and 1 mol extra of both L-proline and L-serine,
determined by chiral HPLC, with respect to amythiamicin D (C43H42N12O7S6) (Figure 21). As the composition of the macrocyclic loop was determined to be the
same for all of these thiopeptides, differences in
amino acid analyses were attributed to variations in
the peptide side chain. Although the absolute configuration of the two valine-derived and one aspartate-derived thiazole residues in the amythiamicin
macrocycle could not be determined from the isolated
natural product, due to racemization under acidcatalyzed hydrolysis conditions, the chemical synthesis of amythiamicin D by Moody et al. has since
verified the structure and absolute stereochemistry
of this thiopeptide family, substantiated by X-ray
crystallographic data, confirming the L-stereochemistry of all constituent amino acids.103 Interest-
Figure 22. Cyclothiazomycin and its hydrolysates.

Figure 21. Structure of amythiamicin A-D.
ingly the amythiamicins are one of the few thiopeptides in this class that do not contain any dehydroalanine residues (along with GE2270 and the
thiocillins) and show an unusual mode of action for
the inhibition of bacterial protein synthesis, in common with the GE2270 family, binding to protein
elongation factor Tu (Ef-Tu).
Cyclothiazomycin (C59H64N18O14S7) is an unusual
series d thiazolylpeptide that possesses a number of
unique structural features. Isolated from the fermentation broth of Streptomyces sp. NR0516 from a soil
sample collected at Kanagawa, Japan, and purified
first by column chromatography and then by preparative HPLC,104 initial structure determination,
using high-resolution FAB mass spectrometry, elemental analysis, and 13C and 1H NMR spectroscopic
data, was supported by chemical degradation studies,
acidic hydrolysis generating an unusual pyridinecontaining -amino acid as lactam 12, the identity
of which has been verified by synthesis,105 and
saramycetic acid I (13).106 NOESY experimental data
elucidated both the structure and stereochemistry of
cyclothiazomycin, the latter supported by amino acid
analyses, and showed this unique series d thiazolylpeptide lacked the characteristic 2- and 3-azole
substituents on the central domain, containing instead an alanine-derived heterocyclic residue of (R)configuration, quaternary sulfide, and two macrocyclicpeptideloops(Figure22).107 Althoughnoantibacterial
data has been associated with cyclothiazomycin,
which also lacks the characteristic polydehydroalanine side chain implicated in tipA promoter activity,
this thiopeptide is still very worthy of note as a novel
and selective inhibitor of human plasma renin with
an IC50 of 1.7 M.104
2.4. Hydroxypyridine Thiopeptides

The series e thiopeptides all possess very closely
related structures, characterized by a 2,3,5,6-tetrasubstituted pyridine central heterocyclic domain
containing a 5-alkoxy or 5-hydroxy substituent. The
peptide backbone is divided into at least two macrocyclic loops and contains an indole or 1-hydroxyindole, connected in some cases by an S-thioester
linkage, as well as a glycosidic unit attached through
the anomeric position to a -amino acid residue, or
glutamate derivative, in the macrocycle. This series
consists of at least five thiopeptide families, although
considerable structural homology or near identity
exists between them, and contains over 12 structurally distinct compounds.
Nosiheptide (also known as RP9671, C51H43N13O12S6)
is one of the oldest known thiopeptide antibiotics
isolated from Streptomyces actuosus 40037 (NRRL
2954).108 Its general formula and its structural relationship to thiostrepton109 was first suggested by
combustion analyses and NMR spectroscopic experiments110,111 and improved subsequently by modification of the HSQC and HMBC pulse sequences112 and
chemical hydrolysis, which isolated and analyzed a
number of key fragments,113 although it was X-ray
crystallography that finally elucidated the structure114 and stereochemistry.115 Multhiomycin, isolated from Streptomyces antibioticus 8446-CC1,116 has
been shown by 13C NMR and IR spectroscopy as well
as thin-layer chromatography to be structurally
identical with nosiheptide,117 with a characteristic
thioester linkage at the macrocyclic bridgehead,
3-methylindole unit, hydroxyglutamate residue, and
dehydroalanine side chain (Figure 23). Nosiheptide
has been used as a feed additive to promote growth
in pigs and poultry118 and can be monitored in meat
and egg samples by liquid chromatography with
fluorescence detection.119 This antibiotic is very active
in vitro against Gram-positive bacteria (MIC 0.9 ng/
Bagley et al.
Figure 23. Structure of nosiheptide.

Figure 25. Structure of glycothiohexide R.
Figure 24. Structure of S 54832 A-I.
mL against S. aureus ATCC 6538 P) but inactive in

vivo in experimentally infected mice108 and selectively
inhibits protein synthesis in whole cells of Bacillus
subtilis and in E. coli lamelloplast by binding directly
to the ribosomes.120
Shortly thereafter the antibiotics S 54832 A-I
(C59H55N13O19S5), A-II, A-III, and A-IV (C59H57N13O19S5)
were isolated from a strain of Micromonosporaceae
in a Spanish soil sample, Micromonospora globosa,
and shown to be structurally distinct from nosiheptide on the basis of chromatographic evidence, amino
acid analyses, and UV and IR spectroscopic data.121
Although the data for S 54832 A-II and A-III proved
inconclusive, a structure for S 54832 A-I has been
proposed (Figure 24) which closely resembles that of
another recently discovered thiopeptide family, the
nocathiacins, varying only in the composition of the
glycosidic residue and one dehydrothreonine amino
acid in place of a methoxydehydrothreonine in the
peptide macrocycle. All of the members of the S 54832
family exhibit a growth-inhibiting effect toward
Gram-positive bacteria, including Staphylococci, Streptococci, Corynebacteria, and Mycobacteria in vitro
and against S. pyogenes and pneumoniae and S.
aureus in vivo.
The taxonomy, fermentation, and biological evaluation of the series e thiopeptide glycothiohexide R,
consisting of two structurally distinct components
LL-14E605 and O-methyl-LL-14E605 (Figure 25)
isolated from the fermentation broth of Sebekia
benihana (NRRL 21083),122 have been described.123
ThechemicalstructureofglycothiohexideR(C58H57N13O15S6)
was determined by extensive 2D NMR studies as well
as high-resolution FAB mass spectrometry and IR
spectroscopy. Closely related to the structure of
nosiheptide, glycothiohexide R in contrast possesses
a methoxydehydrothreonine unit, glycosidic aminodideoxypyranose moiety, carbinol methylene-substi-
Figure 26. Structure of MJ347-81F4 thiopeptides.
tuted indole, and modified glutamate residue with

additional 3-hydroxylation but lacks the dehydroalanine side chain often characteristic of the thiopeptide
antibiotics.124
Amycolatopsis sp. MJ347-81F4, isolated from soil
collected in Japan, produces two cyclic thiazolyl
peptide antibiotics, components A and B, the former
of which, as the major component, shows in vitro
activity against Gram-positive bacteria including
MRSA and Enterococcus faecalis with MICs typically
in the range from 0.006 to 0.1 g/mL.125 Furthermore,
component A showed poor antibacterial activity
against a thiostrepton-resistant mutant (L11) of B.
subtilis B-558 but was active against an amythiamicin-resistant mutant (V228A) of the same strain, with
a MIC of >100 and <0.19 g/mL, respectively,
indicating that the molecular target of this antibiotic
may well be the bacterial 50S ribosomal subunit. The
structures of both components A (C61H60N14O18S5)
and B (C60H58N14O18S5) were elucidated by chemical
degradation and spectroscopic analyses and are
reported to differ from glycothiohexide R by the
presence of a dehydroalanine side chain and an ester
linkage instead of an S-thioester as a result of
replacing the modified cysteine residue at the macrocyclic bridgehead with a serine (Figure 26).
The nocathiacins are cyclic thiazolyl peptide antibiotics isolated by fermentation of ATCC-202099 of
the genus Nocardia or the fungus Amicolaptosis sp.
Nocathiacin I (C61H60N14O18S5) is structurally identical to MJ347-81F4 component A (Figure 27) and
mediate improved the water solubility profile of the

nocathiacins while retaining good antibacterial activity.129
2.5. Unidentified Thiopeptides
Figure 27. Structure of nocathiacins I-IV.
displays potent activity in vitro and in vivo against

Gram-positive bacteria, including a number of antibiotic-resistant strains, by interacting directly with
the L11 protein and 23S RNA complex of the bacterial ribosome.126 In addition, nocathiacin I has been
reported to be more soluble at low pH than other
thiopeptide antibiotics. Although the structure and
stereochemistry of the nocathiacins could not be
determined by X-ray crystallography, extensive 2D
NMR studies have been carried out, including NOESY, HMBC and HMQC experiments on both 13C- and
15
N-labeled and unlabeled samples in combination
with metal chelate chiral capillary electrophoresis.
These studies established that nocathiacin I (and
hence MJ347-81F4 A) contains L-threonine and determined the conformation of the antibiotic in solution in order to computationally model the interaction
between thiazolylpeptides and the bacterial ribosome.
The structures of three other factors, nocathiacin II
(C61H60N14O17S5), III (C52H43N13O16S5), and IV (C58H57N13O17S5), have been reported (Figure 27), the latter
of which can be derived chemically from nocathiacin
I by dehydroalanine cleavage under mild conditions
using iodomethane and hydroiodic acid in THF at 45
C.127 Nocathiacin I and its structural surrogate
nocathiacin IV have been used as leads for the
development of a parenterally administered broadspectrum antibiotic through chemical modification of
the natural material.128 The direct incorporation of
2-hydroxy- or 2-(dialkylamino)ethyl groups in the
amide side chain (R1) of synthetic analogues by
condensing nocathiacin IV with glycolaldehyde followed by reduction of the Amadori-rearranged inter-
Despite the considerable structural information

that is known about many thiopeptide antibiotics, the
notorious difficulty in obtaining X-ray crystallographic data on individual factors has meant that,
above and beyond the minor structural and stereochemical ambiguities and connectivity discrepancies,
there are a number of metabolites that have been
isolated that would appear to belong to the thiazolyl
peptide class but have not been fully characterized.
Many of these unidentified thiopeptide antibiotics
have only been subjected to preliminary analyses
using techniques such as IR and UV spectroscopy,
mass spectrometry, and combustion and have not
been examined in NMR spectroscopic experiments or
by X-ray crystallography. The biological properties
of these compounds suggest that they belong to the
thiopeptide class, but the lack of sufficient experimental data means that even a tentative proposal of
structure is not possible. These compounds were
understood to be structurally distinct from other
known thiopeptide antibiotics, although with the
rapid expansion of isolated and identified metabolites
in recent years it cannot be known for certain that
named compounds, isolated in early studies, represent structurally distinct factors or new thiopeptide
families without the benefit of further experimentation that in view of the difficulty in obtaining the
original organism seems unlikely to be carried out.
The kimorexins (90-GT-302), isolated from Kitasatosporia kimorexae,130,131 pepthiomycin A and B
isolated from the culture broth of Streptomyces
roseospinus,132 the antifungal antibiotic saramycetin
(also referred to as X-5079C or Sch 43057)133,134 active
against systemic mycoses,135 and sporangiomycin,
isolated from a soil sample collected in Argentina
(strain B987) containing Planomonospora parontospora var. antibiotica,136 all fall into the category of
structurally ambiguous or unidentified thiopeptide
antibiotics isolated from actinomycetes. Although the
classification of these compounds as thiazolyl peptides was made on the basis of good experimental
evidence, in particular their high sulfur content as
indicated by combustion and mass spectrometric
analyses, no structures have been proposed for any
of these metabolites. The 1H NMR spectrum of
saramycetin indicated that this metabolite contains
dehydroalanine residues and a number of thiazoline
heterocycles, and chemical degradation studies suggest a considerable degree of structural identity with
cyclothiazomycin. Biological data would indicate that
some of these unidentified antibiotics belong to the
series a or b thiopeptide class, with similar spectra
of activity for the inhibition of bacterial protein
synthesis being notably inactive against thiostreptonresistant strains. However, in the absence of 2D NMR
studies or X-ray crystallographic data, the structural
classification of many of these compounds will remain
largely unsupported.
The structurally ambiguous metabolite neoberninamycin, produced by M. luteus,62 is active against
Gram-positive and Gram-negative anaerobes, with a

spectrum of activity that is similar to the berninamycins. Analytical data, from mass spectrometry and
1H NMR studies, suggested that this antibiotic shares
a considerable degree of structural homology with the
berninamycins but also demonstrated that this thiopeptide is distinct from the known members of the
berninamycin family (neoberninamycin Mr 1131;
berninamycin Mr 1146).
Multhiomycin was first isolated in 1970 from S.
antibioticus 8444-CC1 and found to have a Mr of 1043
by isothermal distillation. Its empirical formula,
C44H45N11O11S5, was confirmed by elemental analysis
and clearly indicated its thiopeptide nature.116 The
mode of action of multhiomycin is similar to other
thiopeptide antibiotics, blocking bacterial protein
synthesis by inhibiting the transfer of amino acyltRNA to the A site of the bacterial 50S ribosomal
subunit. Later in 1977 the same group claimed that
multhiomycin and nosiheptide were identical. Degradation of multhiomycin provided 1 mol each of
dehydroalanine, cysteine, and threonine, thiazole
heterocycles, and a weakly acidic fragment. The
empirical formula of multhiomycin was later reported
as C51H43N13O12S6, without any comment on its
discrepancy with earlier reports.117 On the basis of
the available evidence, multhiomycin and nosiheptide
would appear to be identical, although no known
thiopeptide fits the original description of multhiomycin, indicating that the material first isolated in
1970 may yet prove to be a unique metabolite of the
thiopeptide class.
3. Biosynthesis
Antibiotic-producing organisms can adopt a number or combination of different strategies to defend
themselves against extracellular drugs and thus
avoid self-intoxication, including modification of the
drug binding site, drug inactivation/sequestration, or
establishing membrane permeability barriers, with
an efficient efflux and exclusion mechanism.2,3 In
actinomycetes resistance determinants are commonly
linked to antibiotic production genes, with coregulation involving divergent promoters, overlapping transcripts, and/or polycistronic transcripts. The regulation of antibiotic production can be linked to the
regulation of resistance, the downregulation of an
enzyme-modifying gene product achieved by the
appropriate use of a weak promoter. However, for
many of the thiopeptide producers it is not clear how
genes encoding resistance and antibiotic biosynthesis
enzymes came to congregate in the same cell. A
detailed understanding of the biosynthesis of these
antibiotics and its genetic origin could help uncover
actinomycetes resistance determinants and allow the
prediction of novel resistance mechanisms prior to
their emergence.
The biosynthesis of a number of thiopeptides has
been investigated by following the incorporation of
isotopically labeled amino acids in order to determine
the origin of many of the unusual heterocyclic structural motifs inherent in these antibiotics. Replacing
some of the atoms of amino acid precursors with 13C,
14C, deuterium, or tritium and examining the incor-
Bagley et al.
poration of these labels in the metabolite has been

used not only to indicate the biosynthetic pathways
operating in the producing organism but also to
validate stereochemical hypotheses, confirm structural identity, and suggest biomimetic routes for their
chemical synthesis, the latter of which has been put
to good use in separate studies by Nicolaou137 and
Moody138 for the laboratory synthesis of the central
heterocyclic domain of series a/b and d thiopeptide
targets, respectively. The various components all
originate from amino acids heavily modified by the
organism to elaborate the complex heterocyclic structural arrays, although notable biosynthetic differences have been found in different bacterial strains.
Distinct structural similarities exist between the
thiopeptide antibiotics and a number of other oxazoline and thiazoline peptide natural products for which
the nature and function of the molecular machinery
responsible for the biosynthesis of heterocyclic components from amino acid precursors is well described,139 and so one might expect that the biogenesis of cyclic thiazolylpeptides is well understood also.
Indeed, as a result of a number of key biosynthetic
studies, much is known about the origin of many of
the components, but the mechanisms of these multistep processes and the mode of assembly of the
various modified components, poorly characterized in
the past, remain salient points for discussion that are
only now being unraveled.
The biosynthesis of thiostrepton was investigated
by administering isotopically labeled precursors,140
including cysteine, serine, isoleucine, threonine, methionine, and tryptophan, to cultures of S. azureus
ATCC 14921 or S. laurentii ATCC 31255, the latter
of which gave better antibiotic yields. The amino acid
origin of all components was demonstrated, the
threonine and butyrine residues were both formed
from threonine, whereas isoleucine was the precursor
to thiostreptine, a dihydroxylated derivative further
elaborated at the C-terminus to a thiazole heterocycle. Experiments with (S)-[1,2-13C2]- and (S)-[2,313
C2]serine demonstrated the incorporation of this
amino acid into thiazole, thiazoline, alanine, and
dehydroalanine residues, the origin of the Z-hydrogens in the latter being the pro-R -hydrogens of
serine according to feeding experiments with (2S,3S)[3-13C,2H1]serine (Figure 28). The 2,3,4,5-carbons of
the tetrahydropyridine core also originate from serine
in a tail-to-tail condensation that probably proceeds
via the corresponding dehydroalanine moieties, suprafacial addition to the terminal dehydroalanine
being followed by anti addition of two hydrogens, one
to the Si face at C3 and the second to the Re face at
C6, which is provided by an adjacent cysteine (Scheme
1).
The quinaldic acid residue was shown to originate
from tryptophan and methionine.140 The first step in
this sequence was shown to be the formation of
2-methyltryptophan from tryptophan and (S)-adenosylmethionine (AdoMet)141 by a methyl transfer that
proceeds surprisingly with retention.142 Ring expansion by cleavage of the N1/C2 bond and cyclization
onto the tryptophan R-position was confirmed by
labeling studies with (S)-[1,2-13C,indole-15N]tryp-

Scheme 2. Biosynthesis of the Quinaldic Acid
Residue of Thiostrepton
Figure 28. Serine labeling studies on thiostrepton.

Scheme 1. Biosynthesis of the Dehydropiperidine
Domain of Thiostrepton
tophan that generated thiostrepton with 13C enrichment in the quinaldic acid carboxyl group (Scheme
2).140a The intermediacy of 4-(1-hydroxyethyl)quinaldic acid (HEQ) was demonstrated by the albeit low
incorporation of tritiated HEQ, adding further support for a ring-expansion mechanism in the biosynthesis of the quinaldic acid residue.140
Consideration of this evidence indicates that thiostrepton is generated by the modification of a linear
peptide, containing at least one residue, (S)-cysteine,
and possibly more of unnatural configuration, presumably generated in a postsynthetic modification.
In these biosynthetic operations the peptide chain
must fold back upon itself to facilitate the cycloaddition that generates the series b domain and establishes the large macrocycle. The amide nitrogen of
the side chain probably arises from an additional
carboxy-terminal serine removed in an oxidative
process, although this has not been shown experimentally. Subsequent attachment of the quinaldic
acid140c and epoxide ring opening with the N-terminus would establish the second macrocycle and
complete the skeletal assembly of the antibiotic.
One of the first thiopeptides to be studied biosynthetically was nosiheptide.143 In a similar fashion, by
feeding radioactive and stable-isotope-labeled amino
acid precursors to cultures of the producing organism,
the origin of many of its unusual components was
verified. Dehydroamino acid residues are formed by
the anti elimination of water from either serine or
threonine, thiazole heterocycles are produced from
cysteine with loss of the pro-3R hydrogen in the
oxidation step, and the terminal amide nitrogen in
the side chain is derived from an additional serine
residue, removed except for its nitrogen during
processing. The central hydroxypyridine domain is
produced by the tail-to-tail condensation of two serine
residues, situated nine amino acids apart in the
peptide chain, and incorporates the carboxyl group
of an adjacent cysteine in an overall process that
formally can be represented as a cycloaddition, a
mechanism which was proposed originally by Bycroft
and Gowland.42 Related to the corresponding biosynthesis of the dehydropiperidine domain of thiostrepton (Scheme 1), loss of water from the vinylogous
carbinolamine, aromatization by elimination of ammonia or some additional amino-terminal residue,
and subsequent hydroxylation would complete the
biosynthesis of the central domain (Scheme 3).
The indolic acid moiety is derived from tryptophan,
although the mechanism of its production remains
unclear. This residue is apparently attached to the
peptide backbone at a late stage in the biosynthesis
Bagley et al.
Scheme 3. Biosynthesis of the Hydroxypyridine

Domain of Nosiheptide
Scheme 4. Biosynthesis of Indolic Acid Moiety of

Nosiheptide
Figure 29. 13C incorporation into sulfomycin I using
isotopically labeled precursors.
and subsequently hydroxylated on the 4-methyl substituent to complete the macrocyclic lactone, although
whether the attachment of the indolecarboxylic acid
occurs before or after 4-methylation is not apparent.
Evidence for the order of events comes from feeding
experiments, which show the efficient incorporation
of 3-methylindole-2-carboxylic acid and 3,4-dimethylindole-3-carboxylic acid but not 4-(hydroxymethyl)3-methylindole-3-carboxylic acid (Scheme 4).140c,143
The biosynthetic origin of A10255G, -B, and -E has
been investigated in S. gardneri and confirms the
amino acid origin of all of the components.72 The
incorporation of (R,S)-[1-13C]serine and [2-13C]glycine
was found in 15 of the 17 amino acid residues,
suggesting the conversion of glycine to [2,3-13C]serine.
Biosynthetic studies on berninamycin A, using first
14
C-labeled144 and then 13C-enriched amino acids,60
confirmed many of these findings, such as indicating
that dehydroalanine residues are formed by the
dehydration of serine, oxazoles are generated by the
cyclodehydration of serine- or threonine-containing
dipeptide units, and the thiazole is derived by the
cyclodehydration of a cysteine residue onto the carboxyl group of a neighboring serine. The origin of the
central heterocyclic domain was confirmed in both
of these studies as a tail-to-tail condensation of two
serine residues, feeding experiments with (R,S)-[313
C]- and (R,S)-[1-13C]serine in the biosynthesis of
berninamycin providing additional support for the
original Bycroft-Gowland proposal of micrococcin
domain biogenesis.42 The isolation of minor metabo-
lites containing dehydroalanine side chains of varying

length provide some indication as to the origin of the
terminal carboxamide nitrogen in berninamycin,
suggesting that the cleavage of a serine or dehydroalanine unit is responsible for the biosynthesis of
this group in many of the thiopeptides. Incorporation
of (S)-[1-13C]valine and the observation that berninamycin with no significant radioactivity is produced
by the growth medium of S. bernensis administered
with (R,S)-[3-14C]--hydroxyvaline verifies that the
hydroxyvaline residue in the macrocyclic backbone
originates from valine and is hydroxylated after the
peptide is assembled.
Isotopic labeling studies on the biogenesis of sulfomycin I (U-102408) by Streptomyces arginensis
demonstrated a number of unusual features not
observed in the biosynthesis of other thiopeptides.145
In accordance with related studies, fortifying the
fermentation medium of S. arginensis with 13C- or
2
H-labeled threonine, serine, glycine, or methionine
indicates that thiazoles and oxazoles are derived by
the cyclocondensation of the corresponding amino
acid with an adjacent carbonyl group and that
dehydroalanine residues originate from the dehydration of serine (Figure 29). Additionally, [2-13C]glycine
was found to be incorporated into both the 2- and
3-positions of dehydroalanine residues, indicating
that glycine is used as a precursor for the biosynthesis of serine, mediated by serine hydroxymethyltransferase with a tetrahydrofolate cofactor. On this
basis [2-13C]glycine incorporation was observed at
four of the positions in the pyridine motif, supporting
a tail-to-tail condensation route to the central domain
of these antibiotics, although glycine-serine interconversion might also have been responsible for the
lack of incorporation of [3-2H]- or [3-3H]serine into
sulfomycin I. Unusually, in a number of experiments,
threonine was incorporated into sites labeled by
serine, although the converse incorporation was not
observed. The 2-amino-4-hydroxy-2-pentenoic acid
Figure 30. 13C incorporation into GE2270A using isotopically labeled glycine or serine.
and methoxyglycine residues were found to originate

from threonine and glycine, respectively, by the
appendage of an additional one-carbon unit with
incorporation from [2-13C]glycine or methioninemethyl-13C in both cases.
Further evidence that the methylene group of
glycine is an effective source of methylating equivalents in other actinomycete strains has been obtained
in a biosynthetic study of GE2270A by P. rosea.146
Incorporation of [2-13C]glycine, with enrichment at
C-2 and C-3 of serine-derived residues, at C-4 and
C-5 of the oxazoline, as well as C-2 through C-5 of
the pyridine and to C-, N-, and O-methyl groups
(Figure 30), supports the efficient conversion of
glycine into serine and the use of a cellular methyl
donor by the organism. Furthermore, it was noted
that the two 5-substituted thiazole residues were
derived from cysteine with subsequent methylation
and not by incorporation of an S-analogue of threonine. Addition of [2-13C]acetate did show enrichment
at all of the positions of the asparagine residue,
consistent with its conversion into oxaloacetate and
transamination to the amino acid, but also demonstrated the poor utilization of this carbon source
under these experimental conditions.
The manner that these antibiotics are assembled
in the organism has been a matter of some conjecture,
although since broadly speaking they all share a
similar architecture it could be anticipated that the
biosynthetic machinery responsible for biogenesis in
different strains has many similarities, and thus the
blueprint which specifies the amino acid sequence
and requisite postsynthetic modifications should be
highly conserved. The construction of the peptide
from its amino acid precursors may occur either by
a ribosomal process or by a template-directed nonribosomal enzymatic process, the latter of which
would seem the most likely.140a Either the peptide
synthetase generates the requisite linear peptide
which is subsequently modified by individual separate enzymes or the whole operation proceeds upon
one or several multienzyme complexes that modify
the individual amino acid components as the peptide
chain or fragments thereof are assembled, both of
which have been shown to operate in the biogenesis
of other oxazole or thiazole peptide natural products.139 The number and organization of iterated
modules would dictate the size and structural composition of the final antibiotic, each module activating
a certain amino acid in closely coupled domains in
nonribosomal peptide synthetase (NRPS) multimodular templates. A putative NRPS gene fragment
that probably encodes a module of the micrococcin
P1 synthetase complex has been identified in the
producing strain S. equorum WS2733, representing
an adenylation (A) domain for generation of the
corresponding amino acyl adenylate organized into
a condensation-adenylation-thiolation-condensation module that was selective for threonine.37b This
finding supports the hypothesis that the biosynthesis
of the thiopeptide antibiotics occurs nonribosomally
and may provide the basis for the characterization
of thiopeptide gene clusters and the future manipulation of NRPS templates for the targeted engineering
of new antibiotics.
4. Biological Properties
To interpret the biological relevance of an antibiotic
purely in terms of its ability to inhibit the growth of
competing organisms would constitute a considerable
oversight. The biological challenge is only initiated
by the discovery of a new antibacterial agent, although the importance of many targeted screening
programs developed to identify novel metabolites
with specific binding properties should not be understated. However, in order to develop agents of clinical
importance, an in depth understanding on the mode
of action of microbial products must be gained,
starting with identification of the biological target,
followed by the site and nature of binding in order
to establish which essential cellular function is being
inhibited, a challenging problem in itself in the past
with ribosomal inhibitors, and most importantly a
determination of how the target organism can counter
the designed purpose of an antibiotic and so become
resistant to its action. Clues to processes employed
by emerging resistant bacterial strains and indeed
insights into mechanisms that may develop in the
future can be gained by studying the survival strategies adopted by the antibiotic-producing organisms
themselves to avoid self-intoxication for it may well
be the case that these organisms are the source of
some resistance determinants, a hypothesis with farreaching consequences for the characterization of
novel resistance mechanisms prior to their clinical
emergence and in the rational design of advantageous agents.
The mode of action of an antibiotic involves its
interaction with a specific receptor either within the
cell or associated with the cell surface. By modifying
these antibiotic target sites, the organism can weaken
or even prevent drug-receptor interactions and
achieve very high levels of resistance. Recent experimentation has led to a number of discoveries on the
origin of the biological properties of antibiotics of the
thiopeptide class and has increased our understanding of the organisms that produce them considerably.
With advancements in our structural knowledge of
the bacterial ribosome147 and new insights into its
function,148 along with the ready availability of many
bacterial genomes149 and evermore sophisticated
computational methods, strategies for the modification of known antibiotics, development of existing or
novel antibacterial targets, or discovery of new
classes of agent by structure-based drug design have
never been so well developed.150
The thiopeptide antibiotics largely inhibit the
growth of Gram-positive bacteria, although the activity of some of these metabolites as antifungal or
anticancer agents, against Gram-negative bacteria,
as renin inhibitors, or against Plasmodium falciparum, the malaria parasite, has also been reported.
Despite considerable structural homology the site and
mode of action for these antibiotics actually varies
in different thiopeptide families and can be categorized, broadly, into two classes: those that bind to a
region of the 23S ribosomal RNA (rRNA) known as
the L11 binding domain (L11BD) and those that bind
to a protein (Ef-Tu) complex involved in the elongation cycle.
The antibacterial activity of the thiopeptide antibiotics in vitro is comparable to that of the penicillins,
with little or no adverse toxicological effects in
mammalian cells, disrupting protein synthesis in the
bacterial cells protein factory, the ribosome. Prokaryotic and eukaryotic ribosomes interpret the information in messenger RNA (mRNA) and use it to
assemble the corresponding sequence of amino acids
in a protein.151 Although bacterial and mammalian
ribosomes do exhibit many structural similarities,
they differ considerably in size, the latter being about
30% larger and containing so-called expansion sequences in the rRNA as well as a number of additional ribosomal proteins. The job of the ribosome
is translation, that is to read each codon of the mRNA
in turn and match it with the anticodon of the
corresponding transfer RNA (tRNA) bound amino
acid, assembled by the respective synthetase, and
thus build up the protein, residue by residue, that it
encodes. All ribosomes are composed of two subunits
of unequal size: the bacterial ribosome, with a
relative sedimentary rate of 70S, consisting of a large
50S and a small 30S subunit. These two subunits are
associated through noncovalent interactions and
organize to give a ribonucleoprotein particle 2.6-2.8
MDa in size, with a diameter of 200-250 , that
functions as a platform for bacterial protein synthesis. In the eubacteria E. coli each subunit consists of
proteins and rRNA fragments: the small 30S subunit
containing 21 proteins (S1-S21) and a 16S rRNA
strand, whereas the 50S subunit comprises 34 proteins (L1-L34) and two strands of rRNA, the 23S and
5S (Figure 31).
The sites on the ribosome involved in the sequential construction of the nascent protein from the
individual amino acid components are denoted as the
A site, where aminoacyl-tRNA (aa-tRNA) containing
the next amino acid residue docks on instruction from
the codon of its corresponding mRNA, the P site,
where the growing peptide chain waits in readiness
to form the next peptide bond, and the E site, which
receives the tRNA for its exit at the end of the
sequence (Scheme 5). Once bacterial protein synthesis has been initiated by interaction of the 3 end of
the 16S rRNA in the 30S subunit with a complemen-
Bagley et al.
Figure 31. Composition of the 70S bacterial ribosome.147,152

Color key: (left) small 30S subunit, proteins S1-S21 (blue)
and 16S rRNA strand (pink); (right) large 50S subunit,
proteins L1-L34 (blue), 23S rRNA (pink), and 5S rRNA
(yellow). (Credit to David S. Goodsell of The Scripps
Research Institute.)
tary sequence on mRNA,148 the initiator tRNA binds

directly to the P site (Scheme 5a) and each aa-tRNA
in accordance with its corresponding codon is delivered to the A site as a ternary complex (Scheme 5b)
formed by combination with the elongation factor Tu
(Ef-Tu) and GTP. GTP hydrolysis causes a conformational change in the ternary complex that releases
Ef-TuGDP from the ribosome to be recycled back to
Ef-TuGTP, leaving the aa-tRNA bound in the A site
(Scheme 5c). The next peptide bond is then formed
on the large ribosomal subunit by the transfer of the
peptide to the A site, generating a peptidyl-tRNA
while leaving behind its own tRNA in the P site
(Scheme 5d). Translocation of the peptidyl-tRNA
from the A site back to the P site is then mediated
by a different elongation factor, Ef-G (Scheme 5e),
which vacates the A site and moves the deacylated
tRNA to the E site ready for exit (Scheme 5f). When
the next aa-tRNA-containing ternary complex binds
(Scheme 5g), the tRNA docked in the E site is
released and the protein elongation cycle repeats
itself (Scheme 5c) until protein termination factors
liberate the finished peptide and dissociate the ribosome.
4.1. Ribosomal Inhibitors

Seven different classes of antibiotics in clinical
practice target the bacterial ribosome, including the
aminoglycosides, tetracyclines, macrolides, and streptogramins. Ribosomal inhibitors that interact with
rRNA may exhibit a favorable resistance profile in
the clinic as most pathogens have multiple copies of
the rrn operons that encode rRNA; thus, resistanceinducing mutations are rarely dominant.150
Many thiopeptide antibiotics interfere with bacterial protein synthesis on the ribosome, although the
precise inhibitory mechanism operating for many of
these agents has not been established. The most
closely studied mode of action of all of the thiopeptides is that of thiostrepton, which has been applied
in rational structure-based drug design in an attempt
Scheme 5. Overview of Protein Translationa
a The small 30S subunit is depicted in yellow, and the 50S

subunit is depicted in blue. A, P, and E denote sites on the
ribosome that can be occupied by tRNA. The A site is where aatRNA binds, the P site is where peptidyl-tRNA binds before peptide
bond formation, and the E site is the exit site for deacylated tRNA.
Translation cycle consists of (a) initiator tRNA binds in the P site;
(b) aa-tRNAEf-TuGTP is delivered to the A site; (c) aa-tRNA is
bound in the A site; (d) peptide is transferred to the A site with
formation of next peptide bond; (e) translocation of peptidyl-tRNA
from the A site back to the P site is mediated by Ef-GGTP; (f)
deacylated tRNA waits in the E site; (g) aa-tRNAEf-TuGTP is
delivered to the A site, releasing the deacylated tRNA from the E
site. (Reprinted with permission from ref 151. Copyright 2003
Wiley-VCH Verlag GmbH & Co. KgaA.)
to address problems with this antibiotics low solubility and poor bioavailability.153 In vivo thiostrepton
inhibits the binding of the aminoacyl-tRNA-containing ternary complex to the ribosomal A site.154 The
energy for protein translation is provided by the
action of the elongation factors Ef-Tu and Ef-G, these
hydrolysis reactions taking place on the large ribosomal subunit at a GTPase center located on a
double-hairpin structure within domain II of 23S
rRNA,155 where ribosomal protein L11 and the pentameric complex L10(L12)4 assemble cooperatively,
and on a ribotoxin hairpin loop within domain VI of
23S rRNA.156 The action of thiostrepton inhibits
peptide elongation, probably by impeding a conformational change within protein L11, when bound in
a region of the 23S rRNA known as the L11 binding
domain (L11BD).157 The RNA-binding domain of
protein L11 recognizes an rRNA tertiary structure

that is stabilized by thiostrepton,158 the antibiotic
preventing one or more conformational transitions
critical for stimulating the GTPase action of the
elongation factors,159 necessary to drive the directional movement of transfer and messenger RNA on
the ribosome.160
Both thiostrepton and micrococcin inhibit the
growth of the malaria parasite P. falciparum, inhibiting organellar protein synthesis by targeting the
large subunit encoded by a 35-kb organelle, which is
one of two extrachromosomal DNAs possessed by the
parasite.161 Thiostrepton has been shown to bind to
malarial plastid rRNA162 and has a 50% inhibitory
concentration (IC50) of 3.2 nM, whereas growth
inhibition by micrococcin has an IC50 of 35 nM.163
Two types of resistance mechanisms have been
observed to thiostrepton in Gram-positive bacteria,
whereas Gram-negative organisms are completely
resistant to its action, as thiostrepton cannot penetrate the bacterial cell. In Gram-positive organisms
one resistance mechanism is defined by the absence
of a protein homologous with L11 in E. coli and
designated BM-L11164 in Bacillus megaterium and
BS-L11165 in B. subtilis, giving rise to proteindeficient ribosomes with a reduced affinity for thiostrepton in vitro but that retain substantial protein
synthetic activity. Mutants of B. subtilis were found
to be resistant to both thiostrepton and sporangiomycin, further substantiating the hypothesis that
these antibiotics inhibit protein synthesis by an
identical mechanism and that the mutation alters the
site on the 50S ribosomal subunit that is responsible
for antibiotic binding.166 In a similar fashion a
mutant strain of B. megaterium possessed an altered
form of protein BM-L11, causing the strain to be
resistant to the action of micrococcin.167 However, the
tRNA uncoupled hydrolysis reaction catalyzed jointly
by the ribosome and the protein factor Ef-G, which
is inhibited by thiostrepton, is markedly stimulated
by micrococcin upon organisms sensitive to this
drug,168 a disparity worthy of note in considering the
seemingly common mode of action of thiopeptide
ribosomal inhibitors. Both thiostrepton and micrococcin bind to the GTPase region in domain II of 23S
rRNA, and in so doing they alter the accessibility of
adenosine (A)-1067 and A1095 in the 23S rRNA
toward chemical reagents, indicating that thiopeptide
ribosomal inhibitors interact directly with these
nucleotides.169 These two drugs had different effects
on the chemical reactivity of A1067 in a terminal loop
of E. coli ribosomes in vitro: micrococcin enhanced
its reactivity, whereas thiostrepton protected the N-1
position reducing reactivity, a difference that correlates with the opposite effects of the two antibiotics
on GTPase activity.170 This can be rationalized in a
model for the tRNA uncoupled system where the
dissociation of Ef-GGDP from the ribosome is the
rate-limiting step of Ef-G-dependent GTP hydrolysis.171 Thiostrepton and micrococcin act by increasing
the dissociation rates of both Ef-GGTP and Ef-G
GDP, although micrococcin with lesser potency has
a weaker dissociating effect on Ef-GGTP than thiostrepton. In this model the action of micrococcin
would increase the turnover of Ef-G, increasing the

rate of GTP hydrolysis, whereas Ef-GGTP dissociation induced by thiostrepton would be too rapid to
allow for GTP hydrolysis, reducing the rate, thus
explaining the apparently different inhibitory effects
of the two drugs. The structural basis for the contrasting activities of ribosome-binding thiazole antibiotics has been studied using NMR and a thiostrepton-resistancemethyltransferaseassay,whichrationalized
the different binding profiles of thiostrepton, nosiheptide, siomycin, and micrococcin based upon the
interaction of the quinaldic acid residue in thiostrepton, or an equivalent aromatic group, with the A1067
residue located in the L11BD.172 However, it may well
be the case that the key step in protein synthesis
inhibition differs from organism to organism and
depends on cellular growth conditions, although
clearly the conformational constraint of protein L11
perturbs the function of the ribosomal factor-guanosine nucleotide complexes and thus inhibits cell
growth.171
Actinomycetes antibiotic producers have to adopt
a resistance mechanism to defend themselves against
their own self-intoxication.2 In general, Streptomyces
are quite sensitive to thiostrepton and yet the producing organism, S. azureus, is totally unaffected by
the drug.3 An RNA-pentose methylase enzyme173
produced constitutively from its own promoter is
responsible for the autoimmunity in thiostrepton
producers,174 and this has been demonstrated in vitro
upon ribosomal RNA from other bacteria. The methylase enzyme, thiostrepton-resistance methylase,
introduces a single methyl group into the A1067
residue of the 23S rRNA in E. coli to give a modified
2-O-methyladenosine-containing ribosome that is
completely resistant to the antibiotic,175 a phenomenon that has been observed in 23S rRNA mutants
of Halobacterium halobium.176 Base changes at position 1159 in halobacteria, which corresponds to
A1067 of the E. coli rRNA, from A to U or G as well
as base methylation causes high-level resistance to
thiostrepton.177 The thiostrepton-resistance (tsr) gene
hybridizes with a sulfomycin resistance determinant
from S. viridochromogenes ATCC 29776,178 encodes
the 23S rRNA A1067 methyltransferase in S. laurentii, and is located within a cluster of ribosomal
protein operons not clustered with genes that encode
the biosynthetic enzymes.179 The tsr gene product has
been overexpressed from S. azureus in E. coli to
characterize the enzymic reaction and establish that
recognition is dependent upon the secondary structure of the ribosomal hairpin loop that contains
nucleotide 1067.180 The study of thiostrepton-resistant mutants is thus not only of relevance to account
for the origin and determinants of antibiotic resistance, but also as a valuable tool in establishing the
complex function and specific dynamic processes in
operation during bacterial protein synthesis to provide new leads and focus for structure-based drug
design.
RNA-pentose methylation confers resistance in
other actinomycetes, including S. laurentii, another
thiostrepton producer, P. parontospora, from which
sporangiomycin was isolated, and S. sioyaensis, the
Bagley et al.
producer of the siomycins.181 Radiographical studies

on the inhibition of bacterial protein synthesis by
siomycin supports many of the mechanistic findings
for thiostrepton and lends credence to a common, or
very closely related, mode of action that operates for
all thiopeptide ribosomal inhibitors. The incorporation of radioactive amino acids and base pairs into
the nascent peptide and mRNA, respectively, was
studied in the presence of siomycin.182 Although the
action of the antibiotic prevented the incorporation
of radioactive amino acids into B. subtilis cells, 14C
base pairs were incorporated into the mRNA, indicating that the agent acted as a ribosomal rather
than transcription inhibitor, which was selective for
Gram-positive and mycobacteria. Antibiotic binding
at the ribosomal G site could cause a distortion of
the A site, which would explain why siomycin inhibits
translocation of the peptidyl-tRNA from the A to the
P site, and prevents the binding of both Ef-GGTP
and the ternary complex aa-tRNAEf-TuGTP to the
ribosome at the G and A sites, respectively.183
Actinomycetes that produce thiopeptide metabolites with very different chemical structures have
been shown to adopt the same resistance mechanism
and give rise to antibiotics that function by a common
mode of action. The growth of S. bernensis, the
producer of the berninamycins, is totally resistant to
the action of thiostrepton, even though structurally
these two antibiotics are grouped in a different
thiopeptide series.53 Similarly, nosiheptide has been
shown to partially inhibit both the enzymatic binding
of aminoacyl-tRNA to the ribosome and the simultaneous hydrolysis of GTP but only when present at
quite a high molar excess over the ribosomes.184
Despite differences in antibiotic structure, all of these
compounds bind to the complex of 23S RNA and
protein L11 and affect various functions of the
ribosomal A site, although thiostrepton is considerably more potent in its action than nosiheptide.185
The specific pentose-methylation of the 23S rRNA
renders the ribosomes of S. bernensis and S. actuosus
totally refractory to berninamycin and nosiheptide,
respectively, and prevents the binding of thiostrepton,186 indicating both a common mode of action of
these drugs and a common mechanism of self-defense
employed by the respective producing organisms.
4.2. Inhibition of Elongation Factor Tu

Elongation factor Tu (Ef-Tu) is the most abundant
protein of the bacterial cell and participates in
peptide elongation by mediating the recognition and
delivery of noninitiator aminoacyl-tRNA, as a ternary
complex with GTP, to the mRNA codon in the
acceptor site of the bacterial ribosome. When its
GTPase action is triggered, Ef-TuGDP, which does
not bind aa-tRNA, dissociates from the ribosomal
complex, leaving behind the aa-tRNA in the ribosomal A site in readiness for peptide translocation.
Many types of antibiotics act by binding to Ef-Tu,
which either prevents the dissociation of Ef-TuGDP
from the ribosomal complex, as is the case for the
polyketide kirromycin, or inhibits the binding of aatRNA to Ef-TuGTP, exemplified by pulvomycin.
The thiopeptide GE2270A (MDL 62,879) inhibits
bacterial protein synthesis at an IC50 of 0.4 M by
binding to Ef-Tu at a distinct site from kirromycin

and so differs in its mode of action from thiopeptide
ribosomal inhibitors.187 GE2270A interacts directly
with the GTP-bound form of Ef-Tu, and this inhibits
the formation of a stable ternary complex by preventing the binding of aa-tRNA to Ef-TuGTP.94 Although
it does not change the GTPase activity of Ef-Tu,
antibiotic binding does prevent the ribosomal catalysis of this process and slows down the dissociation of
the Ef-TuGTP complex to even a greater degree than
aa-tRNA. GE2270A forms a strongly bound 1:1 molar
complex with Ef-Tu, which can be observed by a
mobility shift in an electric field, only reversed by
protein denaturation. The strong affinity between E.
coli Ef-TuGDP and GE2270A has been explained by
X-ray diffraction studies of the complex at a resolution of 2.35 , which showed, in addition to 18
protein-antibiotic van der Waals interactions (<3.5
), a highly unusual salt bridge between Arg 223 and
Glu 259 that folds over the GE2270A side chain when
bound in a pocket formed by three segments of amino
acids, Glu 215-Arg 230, Thr 256-Leu 264, and Asn
273-Leu 277, located in the second domain of Ef-Tu
GTP.188 The exchange of guanine nucleotides induces
conformational changes in Ef-Tu complexes, as confirmed by proteolytic cleavage experiments.187 The
superposition of the Ef-TuGDPGE2270A and Ef-Tu
GTP structures suggests that antibiotic binding
causes steric constraints that prevent complete formation of the GTP conformation. Thus, it seems
likely that on binding to domain II GE2270A impedes
the GDP to GTP conformational change and prevents
the formation of an Ef-TuGTPaa-tRNA ternary
complex by blocking the aa-tRNA binding site in the
antibiotic-bound complex.188
This mode of action has been shown to operate for
the GE2270 complex, GE37468, and the amythiamicins and has been used to develop a screening
program for thiopeptide antibiotics with similar
binding properties by selecting activities antagonized
by exogenous Ef-Tu.100,189 Inhibitors of this GTPdependent translation factor have been shown to
possess antimalarial activity in blood cultures of P.
falciparum, the most active being amythiamicin A
with an IC50 of 0.01 M, and to bind to recombinant
P. falciparum plastid Ef-Tu, indicating that endogenous plastid protein synthesis is a potential target
for thiopeptides that inhibit prokaryotic protein
synthesis by this mechanism.190
Not much is known about how the producers of EfTu binding thiopeptides avoid self-intoxication, although resistance mechanisms are complicated by
the fact that some actinomycetes possess more than
one tuf gene to encode Ef-Tu translation factors. In
P. rosea, the GE2270A producer, both the tuf1 and
tuf3 genes encode Ef-Tu-like proteins, the former
encoding the regular elongation factor EF-Tu1.191
This bacterial protein is totally resistant to GE2270A,
10 times more resistant to kirromycin than Ef-Tu1
of Streptomyces coelicolor, and not at all resistant to
pulvomycin, an antibiotic that also inhibits the
formation of the aa-tRNAEF-TuGTP complex. Kirromycin resistance is accounted for by the replacement of Tyr 160 with a Gln at the kirromycin binding
site, the interface of domains 1 and 3 of Ef-Tu in its

GTP-bound conformation. The mutations that confer
GE2270A resistance would appear to map to a
number of amino acids in domain 2, located at the
GE2270A binding pocket,188 in close proximity to
residue 226 at the domain 1-2 interface, which is part
of the binding site of the 3-end of aa-tRNA. Ef-Tu
mutants of B. subtilis that were resistant to Ef-Tubinding thiopeptide antibiotics also showed significant changes of conserved amino acids, which may
account for their similar resistance behavior.100,192
4.3. TipA Promotion

Actinomycetes multi-drug-resistance mechanisms
often rely upon transport proteins or modifying
enzymes and transcriptional regulatory proteins to
recognize multiple drugs and respond to stressresponse signals. Streptomyces lividans 1326, a strain
that is not known as a thiazolylpeptide producer,
reacts to thiostrepton by inducing resistance to a
number of structurally heterogeneous antibiotics
with the accumulation of thiostrepton-induced proteins (Tip), two of which, TipAL (253 amino acids)
and TipAS (144 amino acids), the latter corresponding to the C-terminal region of TipAL, are in-frame
translation products of the same gene, tipA.73 TipAS,
the independently translated thiopeptide binding
domain present in vast molar excess (>20:1) over
TipAL, renders the organism resistant by sequestering the drug in the cytosol and modulating a drugdependent positive feedback loop that controls its
own expression.193 Their transcription is induced by
a number of thiopeptide antibiotics, including thiostrepton, promothiocin, and nosiheptide, by complex
formation with TipAL in the absence of added cofactors, which activates transcription of a monocistronic
mRNA from the tipA promoter (ptipA) and so elicits
autogenous expression of its own promoter.193 The
TipAL protein is a dimer in solution with an Nterminal domain (residues 1-109), containing both
a helix-turn-helix that binds DNA (residues 5-25)
and a long coiled-coil dimerization region (residues
74-109) and a C-terminal domain (residues 110253), represented by TipAS, for thiopeptide recognition (Figure 32A). The formation of a covalent bond
between cysteine 214 (not cysteine 207)61 and a
dehydroalanine residue in the thiopeptide irreversibly generates a ligandTipAL complex (Figure 32B)
and enhances the affinity of the bound protein for
its operator site, inducing the recruitment of RNA
polymerase (RNAP) to the promoter ptipA194 and
activating its transcription at least 200-fold.73 On
binding thiostrepton, conformational changes within
TipAL enhance its association with ptipA and lower
the rate of dissociation from the binding site, increasing the affinity of RNAP for ptipA in an alternative
mechanism of transcriptional activation.195 By analogy with the mercury-resistance regulator (MerR)
protein, the ligand-bound TipAL dimer activates
transcription and increases the affinity of the ligandbound TipAL to the tipA promoter by inducing folding
of the unstructured N-terminal part of apo TipAS, a
globin-like R-helical fold with a deep surface cleft, and
an unfolded N-terminal region that is the linker to
Bagley et al.
Table 2. Thiopeptide ptipA Induction Activity
Figure 32. Structure of the folded part of apo TipAS. (Left)

Ribbon representation showing R-helices, N- and C-termini,
and ligand binding residue C214 as a space-filling model.
(Right) Color map of ligand-induced chemical shift changes
on TipAS-thiostrepton complex formation. Red indicates
strongly affected residues, orange indicates moderately
affected residues, and blue indicates weakly affected
residues. The thiopeptide antibiotic shown alongside the
TipAS-ligand binding site is thiostrepton. (Reprinted with
permission from ref 196. Copyright 2003 European Molecular Biology Organization.)
Scheme 6. Suggested Mechanism of
TipA-Induction by TipALThiopeptide Complex
Formationa
a Cartoon color coding: thiopeptide antibiotic, red; DNA, yellow;

TipAL, green, blue, and brown. (Reprinted with permission from
ref 196. Copyright 2003 European Molecular Biology Organization.)
the DNA-binding domain of TipAL.196 TipAL and

other MerR regulators bind as dimers to an inverted
repeat sequence located within the spacer region of
their promoters. In contrast to the MerR regulatory
proteins, TipAL can bind to its target site and
activate transcription in the absence of the ligand,
elevated external osmolarity causing an increase in
intracellular negative DNA supercoiling that enhances ptipA expression.197 On binding thiopeptide
antibiotics the N-terminal coiled-coil linker parts of
the TipAL complex become rigid. This conformational
change clamps the DNA-binding domains of the
dimer and twists the DNA helix of the promoter,
bringing the two consensus recognition sequences
into alignment to allow RNAP to initiate transcription (Scheme 6).
A number of novel thiopeptide antibiotics have
been identified from actinomycete metabolite libraries by screening for ptipA-inducing activities in a
specific microbiological disk assay, including geninthiocin,74 promoinducin,79 the promothiocins,75 thioxamycin,81 thioactin, and thiotipin.80 Minimum induction concentrations (Cmins) for ptipA induction activity
were found to vary depending upon the structure of
thiopeptide
Cmin/nM
promothiocin B
geninthiocin
berninamycin A
thiostrepton A
promothiocin A
promoinducin
thiotipin
thioactin
thioxamycin
A10255G
thiostrepton B
cyclothiazomycina
amythiamicin Aa
GE2270Aa
promothiocin MOa
0.63
1.0
1.0
1.4
24
30
32
38
63
66
67
>700
>800
>1000
3300
Contains no dehydroalanine residues.
the thiopeptide ligand (Table 2) with the series d

thiopeptides promothiocin B, geninthiocin, and berninamycin A being the most active.61 Promothiocin
derivatives without dehydroalanine residues (promothiocin MO: methyl ester) as well as thiostrepton
B, which does not possess the characteristic polydehydroamino-acid-containing side chain, did not form
TipASantibiotic complexes but nevertheless were
found to induce both the promoter and the synthesis
of TipAS protein, confirming that covalent attachment was not required for stable interaction in vitro
or in vivo and that the macrocyclic thiopeptide
contained a TipA recognition motif. This autogenously controlled antibiotic resistance system responds to these antibiotics by synthesizing a single
mRNA that includes TipAS to sequester the antibiotic and thus limit activation of the TipAL-dependent
promoter. It may well be the case that these proteins,
in addition to providing a low-level antibioticresistance system, regulate resistance to other stressresponse signals, such as those induced by heavy
metals or changes in redox potentials, indicating a
biological relevance far beyond growth inhibition in
competing organisms.
5. Total Synthesis
In recent years significant advances toward the
synthesis of many of the thiopeptide antibiotics and
their unusual heterocyclic or heavily modified constituent components have been made. The structural
complexity of many of these antibiotics means that
efforts toward their total synthesis rarely result in
success. Despite, or perhaps because of, the significant challenge, considerable effort has been directed
toward a number of thiopeptide families in recent
years, culminating in the total synthesis of promothiocin A and more recently amythiamicin D and
thiostrepton. Considerable progress has been made
toward the acidic hydrolysates of many thiopeptide
antibiotics, including dimethyl sulfomycinamate,65,66
berninamycinic acid,57 and micrococcinic acid,198 as
well as useful building blocks for the synthesis of
heterocyclic components of, among others, thiostrepton,199 nosiheptide,200-202 glycothiohexide R,203 the
promothiocins,204 sulfomycins,66,205 amythiamicins,206
berninamycins,207 cyclothiazomycin,105,208,209 A10255,210
GE2270A,211 and the micrococcins,212 the latter being

perhaps the most unusual thiopeptide targets that
have witnessed three separate total syntheses, none
of which have yielded a natural product.43-45 However, with significant advances made in this decade
toward the expedient preparation of some of the more
complex members of this antibiotic family, it can be
presumed with some degree of certainty that major
synthetic achievements will be made in this area
soon.
The first chemical synthesis of a thiopeptide natural product was the preparation of promothiocin A
by Moody.76,77 This landmark synthesis featured the
little-used Bohlmann-Rahtz heteroannulation reaction to establish the central oxazole-thiazole-pyridine domain 17 by saponification, amide formation,
thionation, and Hantzsch thiazole synthesis. The
Bohlmann-Rahtz synthesis of pyridine 16 from ethyl
3-amino-3-(4-oxazolyl)propenoate 14, prepared from
(S)-N-tert-butoxycarbonylalaninamide by dirhodium(II)-catalyzed carbenoid insertion into the amide
N-H followed by cyclodehydration with triphenylphosphine-iodine under basic conditions, saponification,
homologation using the magnesium enolate of ethyl
potassium malonate and enamine formation, and
1-benzyloxybut-3-yn-2-one (15), obtained by Grignard
addition and propargylic oxidation, proceeds in two
steps by initial Michael addition at 50 C and
subsequent double-bond isomerization-cyclization at
140 C in the absence of solvent (Scheme 7). The
elongation of the linear peptide by N- and C-terminus
functionalization followed by macrolactamization
under basic conditions via the pentafluorophenylester
gave macrocycle 18. Benzyl ether cleavage with boron
trichloride dimethyl sulfide complex gave alcohol 19,
which was oxidized to carboxylic acid 20 using
o-iodoxybenzoic acid (IBX) in DMSO followed by
treatment with sodium chlorite and then coupled
with an O-protected serinamide derivative using 1-(3dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDCI). Protodesilylation and dehydration
by mesylate formation followed by treatment with
triethylamine installed the dehydroalanine side chain,
established the first total synthesis of one of the
thiopeptide antibiotics, and verified the (S)-stereochemistry of all three stereogenic centers in the
metabolite isolated from Streptomyces sp. SF2741.
Both Shin and Ciufolini made considerable progress
toward the synthesis of the micrococcins, with the
preparation of micrococcin isomers by the former44,45
and the synthesis of the Bycroft-Gowland structure
5 of micrococcin P1 by the latter.43 In Ciufolinis
approach the central domain 21 was established by
the heteroannulation of 1,5-diketone 22 (Figure 33),
generated by Michael addition and cyclized in two
steps by treatment with ammonium acetate followed
by oxidation with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ).212 Subsequent incorporation of the
(R)-isoalaninol side chain to give tris(thiazolyl)pyridine 24, coupling with the N-terminus of modified
pentapeptide 23, and finally macrolactamization by
hydrolysis and acid-catalyzed deprotection of the Cand N-terminal protecting groups, respectively, followed by treatment with diphenylphosphoryl azide

Scheme 7. Total Synthesis of Promothiocin A
(DPPA) gave the Bycroft-Gowland structure 5 in 23

steps and an overall yield of 5.7%. However, although
the spectra of the synthetic material and natural
micrococcin P were very similar, they were not
identical, their diastereoisomeric relationship being
shown subsequently in NMR and computational
studies.46
An alternative route to the series d domain by
Moody used a biomimetic strategy for the synthesis
of the 2,3,6-tris(thiazolyl)pyridine 27 of the amythiamicins, elaborated to amythiamicin D to establish the
(S)-stereochemistry of the three stereogenic centers
of the natural product (Scheme 8).103 The formal azaDiels-Alder cycloaddition of dehydroalanine dienophile 25 and 2-azadiene 26 proceeded in modest yield
by microwave irradiation138 in toluene at 120 C for
12 h. Elongation of the central pyridine domain 27
gave linear peptide 28, which was cyclized by libera-
Bagley et al.
Scheme 9. Biomimetic Synthesis of Series a or b
Piperidine Domain
Figure 33. Intermediates in the synthesis of the BycroftGowland structure 5 of micrococcin P1.
Scheme 8. Total Synthesis of Amythiamicin D
tion of both C- and N-termini using trifluoroacetic

acid (TFA) followed by treatment with diphenylphosphoryl azide (DPPA) and Hunigs base in DMF to give
the macrocyclic natural product amythiamicin D.
An elegant biomimetic strategy has been used by
Nicolaou to establish the dehydropiperidine domain
of thiostrepton, and this, with the stereoselective
synthesis of the quinaldic acid-containing macrocycle213 and construction of all requisite components,214,215 led to the highly convergent total synthesis of this complex antibiotic.216 The regioselective
and endo-selective hetero-Diels-Alder dimerization
of 2-azadiene 31, obtained from thiazolidine 30 by
treatment with silver carbonate, proceeded without
facial selectivity to give dehydropiperidine 33 as a
1:1 mixture of diastereomers in a cascade sequence
with in situ lysis of imine intermediate 32 and release
of aldehyde 29 to be recycled to dimerization precursor 30 (Scheme 9).137 Stereospecific reduction of
cycloadduct 33 using sodium cyanoborohydride generated piperidine 34 and demonstrated that a bio-
Scheme 10. Total Synthesis of Thiostrepton
mimetic heterodimerization approach can provide

expedient access to the central domain of either series
a (34) or b (33) thiopeptide antibiotics.
The application of this work in the total synthesis
of thiostrepton was realized by capturing the free
amino group of the dehydropiperidine intermediatewith the acyl chloride of an azide derivative of
alanine (35) to produce imine 36 exclusively as a
diastereomeric mixture (Scheme 10).215 After peptide
elaboration, closure of the thiazoline-containing macrocycle was successful for only one (37) of two
monoacids, formed by the action of Me3SnOH in 1,2dichloroethane, to give the desired macrocycle 38
following reduction with PMe3-H2O and treatment
with HATU-HOAt-iPr2NEt. The construction of the
two macrocyclic domains was effected by attachment
of a phenylseleno-disubstituted peptide 40,217 as a
precursor for the dehydroalanine subunits in the side
chain, to acid 39, followed by N-terminal deprotection, coupling with quinaldic acid linear peptide 41,
and macrolactonization under Yamaguchi conditions
with 2,4,6-trichlorobenzoyl chloride.216 The synthesis
was completed by tBuOOH-mediated oxidation of all
three phenylseleno groups in bis-macrocycle 42,
which brought about spontaneous selenoxide syn
elimination, followed by silyl ether deprotection with
hydrogen fluoride-pyridine, with concomitant elimination to form the thiazoline-conjugated Z double
bond. This landmark route gave synthetic thiostrepton with identical physical properties to an authentic
sample, constituting a highly convergent and stereoselective synthesis of a complex thiopeptide antibiotic
that may pave the way for related synthetic studies
in the future.
6. Future Perspectives
Recent years have seen many developments in our
understanding of the chemistry and biology of the
thiopeptide antibiotics. Targeted screening programs
have isolated an ever-increasing number of actinomycete thiazolylpeptide metabolites obtained from
various sources. Alongside this increase in diversity,
analytical methods, in particular X-ray crystallography and NMR techniques, well suited to these macrocyclic natural products have evolved to elucidate
thiopeptide structure and stereochemistry with much
greater certainty, removing many of the structural
ambiguities inherent in earlier work in the area.
Considerable advances have been made in our understanding of the dynamic function of the bacterial
ribosome, the mode of action and site of binding of
thiopeptide ribosomal inhibitors, the inhibition of
organellar protein synthesis by these agents in P.
falciparum, and the manner in which these metabolites are assembled in the organism, and it is suspected that many more revelations in these areas will
be forthcoming. New insights into multi-drug-resistance systems in bacteria have revealed the stress
responses of actinomycetes to thiopeptide antibiotics
and their role and its structural basis in regulating
gene expression. Furthermore, significant progress

in the chemical synthesis of complex molecular
architectures found in this family of antibiotics has
been made, with the total synthesis of promothiocin
A, amythiamicin D, and the stunningly complex
thiostrepton, making it appear likely that further
success will be enjoyed in the synthesis of similarly
challenging metabolites and derivatives thereof in the
future to optimize their biological function in the
computer-assisted design of analogue structures.
With bacterial evolution threatening to overthrow our
current antibiotic regime, we can be certain that
despite 50 years of discoveries the thiopeptide antibiotics will continue to enjoy increasing attention
from a wide variety of scientific consortia whose work
will continue to surprise us with its innovation,
tenacity, ambition, and strategic relevance.
7. Acknowledgment
We thank the Engineering and Physical Sciences
Research Council (EPSRC), Vernalis (Granta Park,
Abington, Cambridge, U.K.), the Great BritainChina Educational Trust, the Royal Society, and the
Henry Lester Trust Ltd. for generous financial support and Dr. Phil A. Lowden, Prof. Chris J. Moody,
and Dr. Justin Bower for helpful discussions.
8. References
(1) Niccolai, D.; Tarsi, L.; Thomas, R. J. Chem. Commun. 1997,
2333.
(2) Cundliffe, E. Annu. Rev. Microbiol. 1989, 43, 207.
(3) Cundliffe, E. Br. Med. Bull. 1984, 40, 61.
(4) Berdy, J. Adv. Appl. Microbiol. 1974, 18, 309.
(5) Hensens, O. D.; Albers-Schonberg, G. Tetrahedron Lett. 1978,
3649.
(6) Pagano, J. F.; Weinstein, M. J.; Stout, H. A.; Donovick, R.
Antibiot. Ann. 1955/56, 554.
(7) Vandeputte, J.; Dutcher, J. D. Antibiot. Ann. 1955/56, 560.
(8) Steinberg, B. A.; Jambor, W. P.; Suydam, L. O.; Soriano, A.
Antibiot. Ann. 1955/56, 562.
(9) (a) Jones, W. F., Jr.; Finland, M. Antibiot. Chemother. 1958, 8,
387. (b) Kutscher, A. H.; Seguin, L.; Rankow, R. M.; Piro, J. D.
Antibiot. Chemother. 1958, 8, 576.
(10) Cron, M. J.; Whitehead, D. F.; Hooper, I. R.; Heinemann, B.;
Lein, J. Antibiot. Chemother. 1956, 6, 63.
(11) Bodanszky, M.; Dutcher, J. D.; Williams, N. J. J. Antibiot., Ser.
A 1963, 16, 76.
(12) Bodanszky, M.; Fried, J.; Sheehan, J. T.; Williams, N. J.; Alicino,
J.; Cohen, A. I.; Keeler, B. T.; Birkhimer, C. A. J. Am. Chem.
Soc. 1964, 86, 2478.
(13) Anderson, B.; Hodgkin, D. C.; Viswamitra, M. A. Nature 1970,
225, 233.
(14) Tori, K.; Tokura, K.; Okabe, K.; Ebata, M.; Otsuka, H. Tetrahedron Lett. 1976, 185.
(15) Trejo, W. H.; Dean, L. D.; Pluscec, J.; Meyers, E.; Brown, W. E.
J. Antibiot. 1977, 30, 639.
(16) Hensens, O. D.; Albers-Schonberg, G. J. Antibiot. 1983, 36, 832.
(17) Mocek, U.; Beale, J. M.; Floss, H. G. J. Antibiot. 1989, 42, 1649.
(18) Bond, C. S.; Shaw, M. P.; Alphey, M. S.; Hunter, W. N. Acta
Crystallogr. 2001, D57, 755.
(19) Tori, K.; Tokura, K.; Yoshimura, Y.; Terui, Y.; Okabe, K.; Otsuka,
H.; Matsushita, K.; Inagaki, F.; Miyazawa, T. J. Antibiot. 1981,
34, 124.
(20) Nishimura, H.; Okamoto, S.; Mayama, M.; Otsuka, H.; Nakajima, K.; Tawara, K.; Shimohira, M.; Shimaoka, N. J. Antibiot.,
Ser. A. 1961, 14, 255.
(21) Ebata, M.; Miyazaki, K.; Otsuka, H. J. Antibiot. 1969, 22, 364.
(22) Tokura, K.; Tori, K.; Yoshimura, Y.; Okabe, K.; Otsuka, H.;
Matsushita, K.; Inagaki, F.; Miyazawa, T. J. Antibiot. 1980, 33,
1563.
(23) Ebata, M.; Miyazaki, K.; Otsuka, H. J. Antibiot. 1969, 22, 434.
(24) Tori, K.; Tokura, K.; Yoshimura, Y.; Okabe, K.; Otsuka, H.;
Inagaki, F.; Miyazawa, T. J. Antibiot. 1979, 32, 1072.
(25) Miyairi, N.; Miyoshi, T.; Aoki, H.; Kohsaka, M.; Ikushima, H.;
Kunugita, K.; Sakai, H.; Imanaka, H. Antimicrob. Agents
Chemother. 1972, 1, 192.
Bagley et al.
(26) Mine, K.; Miyairi, N.; Takano, N.; Mori, S.; Watanabe, N.
(27) Muir, L. A.; Barreto, A. U.S. Patent 4,061,732,1977; Chem. Abstr.
1978, 88, 58574y.
(28) Hensens, O. D.; Albers-Schonberg, G. J. Antibiot. 1983, 36, 814.
(29) Muramatsu, I.; Motoki, Y.; Aoyama, M.; Suzuki, H. J. Antibiot.
1977, 30, 383.
(30) Motoki, Y.; Muramatsu, I. Pept. Chem. 1979 (Pub. 1980), 13;
Chem. Abstr. 1981, 94, 103798m.
(31) Puar, M. S.; Ganguly, A. K.; Afonso, A.; Brambilla, R.; Mangiaracina, P.; Sarre, O.; MacFarlane, R. D. J. Am. Chem. Soc. 1981,
103, 5231.
(32) Puar, M. S.; Chan, T. M.; Hegde, V.; Patel, M.; Bartner, P.; Ng,
K. J.; Pramanik, B. N.; MacFarlane, R. D. J. Antibiot. 1998, 51,
221.
(33) Su, T. L. Br. J. Exp. Path. 1948, 29, 473.
(34) Kelly, B. K.; Miller, G. A.; Whitmarsh, J. M. Brit. Patent 711,593, 1954; Chem. Abstr. 1955, 49, 573.
(35) Heatley, N. G.; Doery, H. M. Biochem. J. 1951, 50, 247.
(36) Fuller, A. T. Nature 1955, 175, 722.
(37) (a) Carnio, M. C.; Holtzel, A.; Rudolf, M.; Henle, T.; Jung, G.;
Scherer, S. Appl. Environ. Microbiol. 2000, 66, 2378. (b) Carnio,
M. C.; Stachelhaus, T.; Francis, K. P.; Scherer, S. Eur. J.
Biochem. 2001, 268, 6390.
(38) Brookes, P.; Fuller, A. T.; Walker, J. J. Chem. Soc. 1957, 689.
(39) Brookes, P.; Clark, R. J.; Fuller, A. T.; Mijovic, M. P. V.; Walker,
J. J. Chem. Soc. C 1960, 916.
(40) Mijovic, M. P. V.; Walker, J. J. Chem. Soc. C 1960, 909.
(41) Walker, J.; Olesker, A.; Valente, L.; Rabanal, R.; Lukacs, G. J.
Chem. Soc., Chem. Commun. 1977, 706.
(42) Bycroft, B. W.; Gowland, M. S. J. Chem. Soc., Chem. Commun.
1978, 256.
(43) Ciufolini, M. A.; Shen, Y.-C. Org. Lett. 1999, 1, 1843.
(44) Shin, C.; Okumura, K.; Shigekuni, M.; Nakamura, Y. Chem. Lett.
1998, 139.
(45) Okumura, K.; Ito, A.; Yoshioka, D.; Shin, C. Heterocycles 1998,
48, 1319.
(46) Fenet, B.; Pierre, F.; Cundliffe, E.; Ciufolini, M. A. Tetrahedron
Lett. 2002, 43, 2367.
(47) Dean, B. M.; Mijovic, M. P. V.; Walker, J. J. Chem. Soc. 1961,
3394.
(48) Shoji, J.; Hinoo, H.; Wakisaka, Y.; Koizumi, K.; Mayama, M.;
Matsuura, S.; Matsumoto, K. J. Antibiot. 1976, 29, 366.
(49) Shoji, J.; Kato, T.; Yoshimura, Y.; Tori, K. J. Antibiot. 1981, 34,
1126.
(50) Nagai, K.; Kamigiri, K.; Arao, N.; Suzumura, K.-I.; Kawano, Y.;
Yamaoka, M.; Zhang, H.; Watanabe, M.; Suzuki, K. J. Antibiot.
2003, 56, 123.
(51) Suzumura, K.-I.; Yokoi, T.; Funatsu, M.; Nagai, K.; Tanaka, K.;
Zhang, H.; Suzuki, K. J. Antibiot. 2003, 56, 129.
(52) Kamigiri, K.; Watanabe, M.; Nagai, K.; Arao, N.; Suzumura, K.;
Suzuki, K.; Kurane, R.; Yamaoka, M.; Kawano, Y. PCT Int. Appl.
072617, 2002; Chem. Abstr. 2002, 137, 246602.
(53) Reusser, F. Biochemistry 1969, 8, 3303.
(54) Liesch, J. M.; Millington, D. S.; Pandey, R. C.; Rinehart, K. L.,
Jr. J. Am. Chem. Soc. 1976, 98, 8237.
(55) Liesch, J. M.; Rinehart, K. L., Jr. J. Am. Chem. Soc. 1977, 99,
1645.
(56) Liesch, J. M.; McMillan, J. A.; Pandey, R. C.; Paul, I. C.;
Rinehart, K. L., Jr.; Reusser, F. J. Am. Chem. Soc. 1979, 98,
299.
(57) Kelly, T. R.; Echavarren, A.; Chandrakumar, N. S.; Koksal, Y.
(58) Abe, H.; Kushida, K.; Shiobara, Y.; Kodama, M. Tetrahedron Lett.
1988, 29, 1401.
(59) Lau, R. C. M.; Rinehart, K. L. J. Antibiot. 1994, 47, 1466.
(60) Lau, R. C. M.; Rinehart, K. L. J. Am. Chem. Soc. 1995, 117,
7606.
(61) Chiu, M. L.; Folcher, M.; Katoh, T.; Puglia, A. M.; Vohradsky,
J.; Yun, B.-S.; Seto, H.; Thompson, C. J. J. Biol. Chem. 1999,
274, 20578.
(62) Biskupiak, J. E.; Meyers, E.; Gillum, A. M.; Dean, L.; Trejo, W.
H.; Kirsch, D. R. J. Antibiot. 1988, 41, 684.
(63) Egawa, Y.; Umino, K.; Tamura, Y.; Shimizu, M.; Kaneko, K.;
Sakurazawa, M.; Awataguchi, S.; Okuda, T. J. Antibiot. 1969,
22, 12.
(64) Abe, H.; Takaishi, T.; Okuda, T.; Aoe, K.; Date, T. Tetrahedron
Lett. 1978, 2791.
(65) Kelly, T. R.; Lang, F. J. Org. Chem. 1996, 61, 4623.
(66) Bagley, M. C.; Dale, J. W.; Xiong, X.; Bower, J. Org. Lett. 2003,
5, 4421.
(67) Kohno, J.; Kameda, N.; Nishio, M.; Kinumaki, A.; Komatsubara,
S. J. Antibiot. 1996, 49, 1063.
(68) Boeck, L. D.; Berry, D. M.; Mertz, F. P.; Wetzel, R. W. J. Antibiot.
1992, 45, 1222.
(69) Boeck, L. D.; Favret, M. E.; Wetzel, R. W. J. Antibiot. 1992, 45,
1278.
(70) Favret, M. E.; Boeck, L. D. J. Antibiot. 1992, 45, 1809.
(71) Debono, M.; Molloy, R. M.; Occolowitz, J. L.; Paschal, J. W.;
Hunt, A. H.; Michel, K. H.; Martin, J. W. J. Org. Chem. 1992,
57, 5200.
(72) Favret, M. E.; Paschal, J. W.; Elzey, T. K.; Boeck, L. D. J.
Antibiot. 1992, 45, 1499.
(73) Murakami, T.; Holt, T. G.; Thompson, C. J. J. Bacteriol. 1989,
171, 1459.
(74) Yun, B.-S.; Hidaka, T.; Furihata, K.; Seto, H. J. Antibiot. 1994,
47, 969.
47, 510.
(76) Moody, C. J.; Bagley, M. C. Chem. Commun. 1998, 2049.
(77) Bagley, M. C.; Bashford, K. E.; Hesketh, C. L.; Moody, C. J. J.
Am. Chem. Soc. 2000, 122, 3301.
(78) Yun, B.-S.; Fujita, K.-i.; Furihata, K.; Seto, H. Tetrahedron 2001,
57, 9683.
(79) Yun, B.-S.; Seto, H. Biosci. Biotechnol. Biochem. 1995, 59, 876.
(80) Yun, B.-S.; Hidaka, T.; Furihata, K.; Seto, H. Tetrahedron 1994,
50, 11659.
47, 1541.
(82) Matsumoto, M.; Kawamura, Y.; Yasuda, Y.; Tanimoto, T.;
Matsumoto, K.; Yoshida, T.; Shoji, J. J. Antibiot. 1989, 42, 1465.
(83) Nadkarni, S. R.; Mukhopadhyay, T.; Jayvanti, K.; Vijaya Kumar,
E. K. Eur. Patent 96111156,4, 1998; Chem. Abstr. 1998, 128,
153208.
(84) Vijaya Kumar, E. K. S.; Kenia, J.; Mukhopadhyay, T.; Nadkarni,
S. R. J. Nat. Prod. 1999, 62, 1562.
(85) Holgado, G. G.; Rodrguez, J. C.; Hernandez, L. M. C.; Daz, M.;
Fernandez-Abalos, J. M.; Trujillano, I.; Santamara, R. I. J.
Antibiot. 2002, 55, 383.
(86) Rodrguez, J. C.; Holgado, G. G.; Sanchez, R. I. S.; Canedo, L.
M. J. Antibiot. 2002, 55, 391.
(87) Selva, E.; Beretta, G.; Montanini, N.; Goldstein, B. P.; Denaro,
M. Eur. Patent 359,062, 1990; Chem. Abstr. 1990, 113, 38919w.
(88) Selva, E.; Beretta, G.; Montanini, N.; Saddler, G. S.; Gastaldo,
L.; Ferrari, P.; Lorenzetti, R.; Landini, P.; Ripamonti, F.;
Goldstein, B. P.; Berti, M.; Montanaro, L.; Denaro, M. J. Antibiot.
1991, 44, 693.
(89) Kettenring, J.; Colombo, L.; Ferrari, P.; Tavecchia, P.; Nebuloni,
M.; Vekey, K.; Gallo, G. G.; Selva, E. J. Antibiot. 1991, 44, 702.
(90) Colombo, L.; Tavecchia, P.; Selva, E.; Gallo, G. G.; Zerilli, L. F.
Org. Mass Spectrom. 1992, 27, 219.
(91) Tavecchia, P.; Gentili, P.; Kurz, M.; Sottani, C.; Bonfichi, R.;
Lociuro, S.; Selva, E. J. Antibiot. 1994, 47, 1564.
(92) Tavecchia, P.; Gentili, P.; Kurz, M.; Sottani, C.; Bonfichi, R.;
Selva, E.; Lociuro, S.; Restelli, E.; Ciabatti, R. Tetrahedron 1995,
51, 4867.
(93) Selva, E.; Ferrari, P.; Kurz, M.; Tavecchia, P.; Colombo, L.; Stella,
S.; Restelli, E.; Goldstein, B. P.; Ripamonti, F.; Denaro, M. J.
Antibiot. 1995, 48, 1039.
(94) Anborgh, P. H.; Parmeggiani, A. EMBO J. 1991, 10, 779.
(95) Stella, S.; Montanini, N.; Le Monnier, F.; Ferrari, P.; Colombo,
L.; Landini, P.; Ciciliato, I.; Goldstein, B. P.; Selva, E.; Denaro,
M. J. Antibiot. 1995, 48, 780.
(96) Marinelli, F.; Gastaldo, L.; Toppo, G.; Quarta, C. J. Antibiot.
1996, 49, 880.
(97) Ferrari, P.; Colombo, L.; Stella, S.; Selva, E.; Zerilli, L. F. J.
Antibiot. 1995, 48, 1304.
(98) Stella, S.; Montanini, N.; Le Monnier, F. J.; Colombo, L.; Selva,
E.; Denaro, M. PCT Int. Appl. WO 94 14,838, 1994; Chem. Abstr.
1995, 122, 128491r.
(99) Shimanaka, K.; Kinoshita, N.; Iinuma, H.; Hamada, M.; Takeuchi, T. J. Antibiot. 1994, 47, 668.
(100) Shimanaka, K.; Iinuma, H.; Hamada, M.; Ikeno, S.; S.-Tsuchiya,
K.; Arita, M.; Hori, M. J. Antibiot. 1995, 48, 182.
(101) Shimanaka, K.; Takahashi, Y.; Iinuma, H.; Naganawa, H.;
Takeuchi, T. J. Antibiot. 1994, 47, 1153.
(102) Shimanaka, K.; Takahashi, Y.; Iinuma, H.; Naganawa, H.;
Takeuchi, T. J. Antibiot. 1994, 47, 1145.
(103) Hughes, R. A.; Thompson, S. P.; Alcaraz, L.; Moody, C. J. Chem.
Commun. 2004, 946.
(104) Aoki, M.; Ohtsuka, T.; Yamada, M.; Ohba, Y.; Yoshizaki, H.;
Yasuno, H.; Sano, T.; Watanabe, J.; Yokose, K. J. Antibiot. 1991,
44, 582.
(105) Bagley, M. C.; Xiong, X. Org. Lett. 2004, 6, 3401.
(106) Aoki, M.; Ohtsuka, T.; Itezono, Y.; Yokose, K.; Furihata, K.; Seto,
H. Tetrahedron Lett. 1991, 32, 217.
(107) Aoki, M.; Ohtsuka, T.; Itezono, Y.; Yokose, K.; Furihata, K.; Seto,
H. Tetrahedron Lett. 1991, 32, 221.
(108) Benazet, F.; Cartier, M.; Florent, J.; Godard, C.; Jung, G.; Lunel,
J.; Mancy, D.; Pascal, C.; Renaut, J.; Tarridec, P.; Theilleux, J.;
Tissier, R.; Dubost, M.; Ninet, L. Experientia 1980, 36, 414.
(109) Depaire, H.; Thomas, J.-P.; Brun, A.; Olesker, A.; Lukacs, G.
Tetrahedron Lett. 1977, 1403.
(110) Depaire, H.; Thomas, J.-P.; Brun, A.; Olesker, A.; Lukacs, G.
Tetrahedron Lett. 1977, 1397.
(111) Depaire, H.; Thomas, J.-P.; Brun, A.; Hull, W. P.; Olesker, A.;
Lukacs, G. Tetrahedron Lett. 1977, 1401.

(112) Gasmi, G.; Massiot, G.; Nuzillard, J. M. Magn. Reson. Chem.
1996, 34, 185.
(113) Depaire, H.; Thomas, J.-P.; Brun, A.; Lukacs, G. Tetrahedron
Lett. 1977, 1395.
(114) Prange, T.; Ducruix, A.; Pascard, C.; Lunel, J. Nature 1977, 265,
189.
(115) Pascard, C.; Ducruix, A.; Lunel, J.; Prange, T. J. Am. Chem. Soc.
1977, 99, 6418.
(116) Tanaka, T.; Endoj, T.; Shimazu, A.; Yoshida, R.; Suzuki, Y.;
O
h take, N.; Yonehara, H. J. Antibiot. 1970, 23, 231.
(117) Endoj, T.; Yonehara, H. J. Antibiot. 1978, 31, 623.
(118) Casteels, M.; Bekaert, H.; Buysse, F. X. Rev. Agric. (Brussels)
1980, 33, 1069; Chem. Abstr. 1981, 94, 101869m.
(119) Horii, S.; Oku, N. J. AOAC Int. 2000, 83, 17.
(120) Tanaka, T.; Sakaguchi, K.; Yonehara, H. J. Antibiot. 1970, 23,
401.
(121) Keller-Juslen, C.; Kuhn, M.; DeLisle King, H. Ger. Offen. 2,921,148, 1979; Chem. Abstr. 1980, 93, 6124k.
(122) Northcote, P. T.; Williams, D.; Manning, J. K.; Borders, D. B.;
Maiese, W. M.; Lee, M. D. J. Antibiot. 1994, 47, 894.
(123) Steinberg, D. A.; Bernan, V. S.; Montenegro, D. A.; Abbanat, D.
R.; Pearce, C. J.; Korshalla, J. D.; Jacobus, N. V.; Petersen, P.
J.; Mroczenski-Wildey, M. J.; Maiese, W. M.; Greenstein, M. J.
Antibiot. 1994, 47, 887.
(124) Northcote, P. T.; Siegel, M.; Borders, D. B.; Lee, M. D. J. Antibiot.
1994, 47, 901.
(125) Sasaki, T.; Otani, T.; Matsumoto, H.; Unemi, N.; Hamada, M.;
Takeuchi, T.; Hori, M. J. Antibiot. 1998, 51, 715.
(126) (a) Constantine, K. L.; Mueller, L.; Huang, S.; Abid, S.; Lam, K.
S.; Li, W.; Leet, J. E. J. Am. Chem. Soc. 2002, 124, 7284. (b) Li,
W.; Leet, J. E.; Ax, H. A.; Gustavson, D. R.; Brown, D. M.;
Turner, L.; Brown, K.; Clark, J.; Yang, H.; Fung-Tomc, J.; Lam,
K. S. J. Antibiot. 2003, 56, 226. (c) Leet, J. E.; Li, W.; Ax, H. A.;
Matson, J. A.; Huang, S.; Huang, R.; Cantone, J. L.; Drexler,
D.; Dalterio, R. A.; Lam, K. S. J. Antibiot. 2003, 56, 232.
(127) Regueiro-Ren, A.; Ueda, Y. J. Org. Chem. 2002, 67, 8699.
(128) Naidu, B. N.; Li, W.; Sorenson, M. E.; Connolly, T. P.; Wichtowski, J. A.; Zhang, Y.; Kim, O. K.; Matiskella, J. D.; Lam, K.
S.; Bronson, J. J.; Ueda, Y. Tetrahedron Lett. 2004, 45, 1059.
(129) Hrnciar, P.; Ueda, Y.; Huang, S.; Leet, J. E.; Bronson, J. J. J.
Org. Chem. 2002, 67, 8789.
(130) Yeo, W.-H.; Kim, S.-K.; Kim, S.-S.; Yu, S.-H.; Park, E. K. J.
(131) Yeo, W.-H.; Kim, S.-K.; Kim, S.-S.; Yu, S.-H.; Park, E. K. J.
(132) Mizuno, K.; Hamada, M.; Maeda, K.; Umezawa, H. J. Antibiot.
1968, 21, 429.
(133) Cooper, R.; Truumees, I.; Barrett, T.; Patel, M.; Schwartz, J.;
Puar, M.; Das, P.; Pramanik, B. J. Antibiot. 1990, 43, 897.
(134) Baudet, P.; Cherbuliez, E. Helv. Chim. Acta 1964, 47, 661.
(135) Berger, J.; Sternbach, L. H.; Muller, M.; Lasala, E. R.; Grunberg,
E.; Goldberg, M. W. Antimicrob. Agents Chemother. 1962, 436.
(136) Thiemann, J. E.; Coronelli, C.; Pagani, H.; Beretta, G.; Tamoni,
G.; Arioli, V. J. Antibiot. 1968, 21, 525.
(137) Nicolaou, K. C.; Nevalainen, M.; Safina, B. S.; Zak, M.; Bulat,
S. Angew. Chem., Int. Ed. 2002, 41, 1941.
(138) Moody, C. J.; Hughes, R. A.; Thompson, S. P.; Alcaraz, L. Chem.
Commun. 2002, 1760.
C. T. Nat. Prod. Rep. 1999, 16, 249.
(140) (a) Mocek, U.; Zeng, Z.; OHagan, D.; Zhou, P.; Fan, L.-D. G.;
Beale, J. M.; Floss, H. G. J. Am. Chem. Soc. 1993, 115, 7992.
(b) Priestley, N. D.; Smith, T. M.; Shipley, P. R.; Floss, H. G.
Biorg. Med. Chem. 1996, 4, 1135. (c) Smith, T. M.; Priestley, N.
D.; Knaggs, A. R.; Nguyen, T.; Floss, H. G. J. Chem. Soc., Chem.
Commun. 1993, 1612.
(141) Frenzel, T.; Zhou, P.; Floss, H. G. Arch. Biochem. Biophys. 1990,
278, 35.
(142) Floss, H. G.; Lee, S. Acc. Chem. Res. 1993, 26, 116.
(143) Mocek, U.; Knaggs, A. R.; Tsuchiya, R.; Nguyen, T.; Beale, J.
M.; Floss, H. G. J. Am. Chem. Soc. 1993, 115, 7557.
(144) Pearce, C. J.; Rinehart, K. L., Jr. J. Am. Chem. Soc. 1979, 101,
5069.
(145) Fate, G. D.; Benner, C. P.; Grode, S. H.; Gilbertson, T. J. J. Am.
Chem. Soc. 1996, 118, 11363.
(146) De Pietro, M. T.; Marazzi, A.; Sosio, M.; Donadio, S.; Lancini,
G. J. Antibiot. 2001, 54, 1066.
(147) Ban, N.; Nissen, P.; Hansen, J.; Moore, P. B.; Steitz, T. A. Science
2000, 289, 905.
(148) Ramakrishnan, V. Cell 2002, 108, 557.
(149) Payne, D. J.; Holmes, D. J.; Rosenberg, M. Curr. Opin. Invest.
Drugs 2001, 2, 1028.
(150) Knowles, D. J. C.; Foloppe, N.; Matassova, N. B.; Murchie, A. I.
H. Curr. Opin. Pharmacol. 2002, 2, 501.
(151) Wilson, D. N.; Nierhaus, K. H. Angew. Chem., Int. Ed. 2003,
42, 3464.
(152) (a) Carter, A. P.; Clemons, W. M.; Broderson, D. E.; MorganWarren, R. J.; Wimberly, B. T.; Ramakrishnan, V. Nature 2000,
407, 340. (b) Schluenzen, F.; Tocilj, A.; Zarivach, R.; Harms, J.;
(153)
(154)
(155)
(156)
(157)
(158)
(159)
(160)
(161)
(162)
(163)
(164)
(165)
(166)
(167)
(168)
(169)
(170)
(171)
(172)
(173)
(174)
(175)
(176)
(177)
(178)
(179)
(180)
(181)
(182)
(183)
(184)
(185)
Gluehmann, M.; Janell, D.; Bashan, A.; Bartels, H.; Agmon, I.;
Franceschi, F.; Yonath, A. Cell 2000, 102, 615. (c) Wimberly, B.
T.; Broderson, D. E.; Clemons, W. M., Jr.; Morgan-Warren, R.
J.; Carter, A. P.; Vonrhein, C.; Hartsch, T.; Ramakrishnan, V.
Nature 2000, 407, 327. (d) Protein Data Bank (PDB) ID: 1FFK,
1FKA, and 1FLG. Credit also to David S. Goodsell, The Scripps
Research Institute for diagrams.
Bower, J.; Drysdale, M.; Hebdon, R.; Jordan, A.; Lentzen, G.;
Matassova, N.; Murchie, A.; Powles, J.; Roughley, S. Bioorg. Med.
Chem. Lett. 2003, 13, 2455.
Cundliffe, E. Biochem. Biophys. Res. Commun. 1971, 44, 912.
(a) Ryan, P. C.; Lu, M.; Draper, D. E. J. Mol. Biol. 1991, 221,
1257. (b) Conn, G. L.; Draper, D. E.; Lattman, E. E.; Gittis, A.
G. Science 1999, 284, 1171.
Egebjerg, J.; Douthwaite, S. R.; Liljas, A.; Garrett, R. A. J. Mol.
Biol. 1990, 213, 275.
Xing, Y.; Draper, D. E. Biochemistry 1996, 35, 1581.
(a) Blyn, L. B.; Risen, L. M.; Griffey, R. H.; Draper, D. E. Nucleic
Acids Res. 2000, 28, 1778. (b) Wimberly, B. T.; Guymon, R.;
McCutcheon, J. P.; White, S. W.; Ramakrishnan, V. Cell 1999,
97, 491.
Porse, B. T.; Leviev, I.; Mankin, A. S.; Garrett, R. A. J. Mol.
Biol. 1998, 276, 391.
Rodnina, M. V.; Savelsbergh, A.; Katunin, V. I.; Wintermeyer,
W. Nature 1997, 385, 37.
McConkey, G. A.; Rogers, M. J.; McCutchan, T. F. J. Biol. Chem.
1997, 272, 2046.
Clough, B.; Strath, M.; Preiser, P.; Denny, P.; Wilson, I. R. J.
M. FEBS Lett. 1997, 406, 123.
Rogers, M. J.; Cundliffe, E.; McCutchan, T. F. Antimicrob. Agents
Chemother. 1998, 42, 715.
Cundliffe, E.; Dixon, P.; Stark, M.; Stoffler, G.; Ehrlich, R.;
Stoffler-Meilicke, M.; Cannon, M. J. Mol. Biol. 1979, 132, 235.
Wienen, B.; Ehrlich, R.; Stoffler-Meilicke, M.; Stoffler, G.; Smith,
I.; Weiss, D.; Vince, R.; Pestka, S. J. Biol. Chem. 1979, 254, 8031.
Bazzicalupo, M.; Parisi, B.; Pirali, G.; Polsinelli, M.; Sala, F.
Spedding, G.; Cundliffe, E. Eur. J. Biochem. 1984, 140, 453.
Cundliffe, E.; Thompson, J. Eur. J. Biochem. 1981, 118, 47.
Rosendahl, G.; Douthwaite, S. Nucleic Acids Res. 1994, 22, 357.
Egebjerg, J.; Douthwaite, S.; Garrett, R. A. EMBO J. 1989, 8,
607.
Porse, B. T.; Cundliffe, E.; Garrett, R. A. J. Mol. Biol. 1999, 287,
33.
Lentzen, G.; Klinck, R.; Matassova, N.; Aboul-ela, F.; Murchie,
A. I. H. Chem. Biol. 2003, 10, 769.
Cundliffe, E.; Thompson, J. Nature 1979, 278, 859.
Cundliffe, E. Nature 1978, 272, 792.
Thompson, J.; Schmidt, F.; Cundliffe, E. J. Biol. Chem. 1982,
257, 7915.
Mankin, A. S.; Leviev, I.; Garrett, R. A. J. Mol. Biol. 1994, 244,
151.
Hummel, H.; Bock, A. Biochimie 1987, 69, 857.
Nakanishi, N.; Oshida, T.; Yano, S.; Takeda, K.; Yamaguchi, T.;
Ito, Y. Plasmid 1986, 15, 217.
Smith, T. M.; Jiang, Y.-F.; Shipley, P.; Floss, H. G. Gene 1995,
164, 137.
Bechthold, A.; Floss, H. G. Eur. J. Biochem. 1994, 224, 431.
Thompson, J.; Cundliffe, E. J. Bacteriol. 1980, 142, 455.
Tanaka, K.; Watanabe, S.; Tamaki, M. J. Antibiot. 1970, 23, 13.
Watanabe, S. J. Mol. Biol. 1972, 67, 443.
Tanaka, T.; Sakaguchi, K.; Yonehara, H. J. Biochem. 1971, 69,
1127.
Cundliffe, E.; Thompson, J. J. Gen. Microbiol. 1981, 126, 185.
Bagley et al.
(186) Thompson, J.; Cundliffe, E.; Stark, M. J. R. J. Gen. Microbiol.
1982, 128, 875.
(187) Landini, P.; Soffientini, A.; Monti, F.; Lociuro, S.; Marzorati, E.;
Islam, K. Biochemistry 1996, 35, 15288.
(188) Heffron, S. E.; Jurnak, F. Biochemistry 2000, 39, 37.
(189) Selva, E.; Montanini, N.; Stella, S.; Soffientini, A.; Gastaldo, L.;
Denaro, M. J. Antibiot. 1997, 50, 22.
(190) Clough, B.; Rangachari, K.; Strath, M.; Preiser, P. R.; Wilson,
R. J. M. Protist 1999, 150, 189.
(191) Mohrle, V. G.; Tieleman, L. N.; Kraal, B. Biochem. Biophys. Res.
Commun. 1997, 230, 320.
(192) Sosio, M.; Amati, G.; Capellano, C.; Sarubbi, E.; Monti, F.;
Donadio, S. Mol. Microbiol. 1996, 22, 43.
(193) Holmes, D. J.; Caso, J. L.; Thompson, C. J. EMBO J. 1993, 12,
3183.
(194) Chiu, M. L.; Folcher, M.; Griffin, P.; Holt, T.; Klatt, T.;
Thompson, C. J. Biochemistry 1996, 35, 2332.
(195) Chiu, M. L.; Viollier, P. H.; Katoh, T.; Ramsden, J. J.; Thompson,
C. J. Biochemistry 2001, 40, 12950.
(196) Kahmann, J. D.; Sass, H.-J.; Allan, M. G.; Seto, H.; Thompson,
C. J.; Grzesiek, S. EMBO J. 2003, 22, 1824.
(197) Ali, N.; Herron, P. R.; Evans, M. C.; Dyson, P. J. Microbiology
2002, 148, 381.
(198) Kelly, T. R.; Jagoe, C. T.; Gu, Z. Tetrahedron Lett. 1991, 32, 4263.
(199) (a) Higashibayashi, S.; Hashimoto, K.; Nakata, M. Tetrahedron
Lett. 2002, 43, 105. (b) Higashibayashi, S.; Kohno, M.; Goto, T.;
Suzuki, K.; Mori, T.; Hashimoto, K.; Nakata, M. Tetrahedron
Lett. 2004, 45, 3707.
(200) Koerber-Ple, K.; Massiot, G. Synlett 1994, 759.
(201) Umemura, K.; Tate, T.; Yamaura, M.; Yoshimura, J.; Yonezawa,
Y.; Shin, C. Synthesis 1995, 1423.
(202) Umemura, K.; Noda, H.; Yoshimura, J.; Konn, A.; Yonezawa,
Y.; Shin, C. Bull. Chem. Soc. Jpn. 1998, 71, 1391.
(203) Bentley, D. J.; Fairhurst, J.; Gallagher, P. T.; Manteuffel, A. K.;
Moody, C. J.; Pinder, J. L. Org. Biomol. Chem. 2004, 2, 701.
(204) Moody, C. J.; Bagley, M. C. Synlett 1998, 361.
(205) Kayano, T.; Yonezawa, Y.; Shin, C. Chem. Lett. 2004, 33, 72.
(206) Bagley, M. C.; Dale, J. W.; Jenkins, R. L.; Bower, J. Chem.
Commun. 2004, 102.
(207) Yamada, T.; Okumura, K.; Yonezawa, Y.; Shin, C. Chem. Lett.
2001, 102.
(208) Shin, C.; Okabe, A.; Ito, A.; Ito, A.; Yonezawa, Y. Bull. Chem.
Soc. Jpn. 2002, 75, 1583.
(209) Endoh, N.; Yonezawa, Y.; Shin, C. Bull. Chem. Soc. Jpn. 2003,
76, 643.
(210) Umemura, K.; Ikeda, S.; Yoshimura, J.; Okumura, K.; Saito, H.;
Shin, C. Chem. Lett. 1997, 1203.
(211) Okumura, K.; Saito, H.; Shin, C.; Umemura, K.; Yoshimura, J.
Bull. Chem. Soc. Jpn. 1998, 71, 1863.
(212) (a) Ciufolini, M. A.; Shen, Y. C. J. Org. Chem. 1997, 62, 3804.
(b) Suzuki, S.; Yonezawa, Y.; Shin, C. Chem. Lett. 2004, 33, 814.
(213) Nicolaou, K. C.; Safina, B. S.; Funke, C.; Zak, M.; Zecri, F. J.
Angew. Chem., Int. Ed. 2002, 41, 1937.
(214) Nicolaou, K. C.; Nevalainen, M.; Zak, M.; Bulat, S.; Bella, M.;
Safina, B. S. Angew. Chem., Int. Ed. 2003, 42, 3418.
(215) Nicolaou, K. C.; Safina, B. S.; Zak, M.; Estrada, A. A.; Lee, S.
H. Angew. Chem., Int. Ed. 2004, 43, 5087.
(216) Nicolaou, K. C.; Zak, M.; Safina, B. S.; Lee, S. H.; Estrada, A.
A. Angew. Chem., Int. Ed. 2004, 43, 5092.
(217) Nicolaou, K. C. Personal communication. The stereochemistry
of Boc-Sec(Ph) residues, derived from L-serine, is (R) as shown
in Scheme 10.
CR0300441
Chem. Rev. 2005, 105, 715738
715
Molecular Mechanisms Underlying Nonribosomal Peptide Synthesis:

Approaches to New Antibiotics
Stephan A. Sieber and Mohamed A. Marahiel*
Philipps-Universitat Marburg, Fachbereich Chemie/Biochemie, Hans-Meerwein-Strasse, 35032 Marburg, Germany
Received March 30, 2004
Contents
1. Introduction
1.1. NRPS Synthesis
1.2. Product Diversity Assisted by NRPS
2. NRPS Factory
2.1. Activation by the Adenylation Domain
2.2. Intermediates Transport by the Peptidyl
Carrier Protein
2.2.1. Misacylation and Regeneration
2.3. Peptide Elongation by the Condensation
Domain
2.4. Editing Domains
2.4.1. Epimerization
2.4.2. Methylation
2.4.3. Further Modifications
2.5. Peptide Release
2.5.1. Diversity by Cyclization
2.6. Quaternary Architecture
2.7. NRPSs in Higher Eucaryotes
3. Approaches to New Antibiotics
3.1. Genetic Engineering Approaches
3.2. Chemoenzymatic Approaches
3.2.1. Chemoenzymatic Potential of TE
Domains
3.2.2. Chemoenzymatic Route to New Drugs
and Peptide Antibiotics
3.2.3. Expanding the TE Tool Box
3.2.4. Synthetic Utility of TEs: Chemical vs
Enzymatic Cyclization
4. Conclusions
5. Acknowledgments
6. References
715
715
716
718
719
721
722
722
723
723
724
724
725
726
728
729
729
729
731
731
733
733
735
736
736
737
1. Introduction
1.1. NRPS Synthesis
Research into bioactive natural products began
when A. Fleming discovered the antibiotic activity
of the peptide derivative penicillin produced by the
fungal host organism Penicillium notatum.1 Since
then microorganisms have attracted considerable
attention as a new source for pharmaceutical agents,
* To whom correspondence should be addressed. Phone:
+49-6421-2825722. Fax: +49-6421-2822191. E-mail: marahiel@
chemie.uni-marburg.de.
and screening of microbial extracts has afforded a

very large number of new compounds with antimicrobial, antiviral, immunosuppressive, and antitumor
activities. These secondary metabolites were optimized for their dedicated function over eons of
evolution and now represent promising scaffolds for
the development of new drug leads.
Among these substances small peptide molecules
represent a large subclass of bioactive natural products, which contain unique structural features such
as heterocyclic elements, D-amino acids, and glycosylated as well as N-methylated residues. Moreover,
in contrast to proteins produced by ribosomal synthesis, small peptide products contain not only the
common 20 amino acids but also hundreds of different building blocks, suggesting a nonribosomal origin
of biosynthesis. In the 1970s Lipmann et al. reported
a nucleic-acid-independent synthesis of the peptide
antibiotics gramicidin S and tyrocidine A from Bacillus sp. by large enzyme complexes similar to fatty
acid synthases.2 In the following years more and
more peptidic natural products were shown to be
assembled by such large enzymes, referred to as
nonribosomal peptide synthetases (NRPS). Significant progress has been made in the past decades
toward understanding the molecular principles of
bioactive peptide synthesis in microorganisms, and
this has been the subject of extensive recent reviews.3-6 While research first focused on elucidating
the chemical structure of the assembled molecules
and characterizing the architecture of the associated
multienzymes, later work focused on the identification of their biosynthetic gene clusters. More recent
biochemical investigations also revealed high-resolution structures of some of the central catalytic core
enzymes. Moreover, genetic and chemoenzymatic
approaches were developed to reprogram natural
peptide sequences by the combined action of rational
enzyme design and chemical peptide synthesis followed by subsequent enzyme catalysis. This review
aims to give a global overview of our understanding
of natural nonribosomal peptide synthesis and of
progress in genetically engineered and chemoenzymatic synthesis of nonribosomal peptide products. In
the natural synthesis section major emphasis will
be given to recent progress made on structure determination and mechanistic predictions. The genetic
and chemoenzymatic section will complement this
section by providing approaches to novel peptide
antibiotics.

Stephan A. Sieber was born in Marburg, Germany, in 1976. He studied

Chemistry at the Universities of Marburg and Birmingham, England.
Fascinated by the great potential of combining enzymatic and chemical
synthesis, he joined the group of Professor Mohamed A. Marahiel, where
he obtained his diploma in 2001. During his graduate studies in the lab
of Professors Marahiel and Christopher T. Walsh at Harvard Medical
School, Boston, MA, he worked on a chemoenzymatic approach to cyclic
peptides. In 2004 he obtained his Ph.D. degree and received the FriedrichWeygand award for his thesis work. During his undergraduate and
graduate studies he was a fellow of the Studienstiftung des Deutschen
Volkes. Soon after graduation he joined the group of Professor Benjamin
F. Cravatt at The Scripps Research Institute in La Jolla, CA, for
postdoctoral work for which he received a DFG Emmy-Noether Stipend.
Mohamed A. Marahiel studied Chemistry at the Universities of Cairo

(Egypt) and Gottingen (Germany). In 1977 he obtained his Ph.D. degree
in Biochemistry and Microbiology from the University of Gottingen.
Subsequently, he received an assistant professors position at the Technical
University of Berlin, where in 1987 he obtained his Habilitation in
biochemistry. Three years later he moved to the Philipps-Universitat as a
professor of biochemistry in the Chemistry Department. He was a DFG
fellow in 1978 and 1986 at the John Innes Institute in Norwich (U.K.) and
at the Biolabs, Harvard University, respectively. Since 2004 he has been
a member of the LeopoldinasDeutsche Akademie der Naturforscher.
His present research focuses on the structurefunction relationship and
elucidation of reaction mechanisms of modular peptide synthetases
involved in the nonribosomal synthesis of peptide antibiotics. His group
is also interested in studying the cold shock response in soil bacteria as
well as other stress-induced proteins.
1.2. Product Diversity Assisted by NRPS

Natural peptide products synthesized by NRPSs
can be grouped according to their biological activities.
A major class comprises antibiotic and antifungal
activities including, for example, the peptides tyrocidine, bacitracin, surfactin, pristinamycin, vancomycin, and fengycin.3,7-15 Their biological functionality is strictly associated with their chemical
Sieber and Marahiel
structure, which constrains the peptide sequence in

its biologically active conformation and ensures specific interaction with a dedicated molecular target.
This structural rigidity is achieved by either cyclization or oxidative cross-linking of side chains, which
contribute to stability. Moreover, the great diversity
of chemical modifications, such as incorporation of
fatty acid chains, D-amino acids, glycosylated amino
acids, and heterocyclic rings, adds much to these
specific interactions (Figure 1).
For many natural peptide products the relation
between structural features and biological activities
has been investigated. Some of these, such as bacitracin from Bacillus licheniformis and gramicidin S
form Bacillus brevis, act also nonspecifically as
membrane-inserting cationic hydrophobic species.16-19
In addition to the macrocyclic ring, bacitracin contains a small thiazoline ring that supports the specific
cation-dependent complexation of the membrane
phosphate moiety of the C55 lipid phosphate. This
complexation leads to inhibition of the lipid cycle.3,20,21
Amphiphilic lipopeptides, such as the surface tension
reducing agent surfactin and the antifungal mycosubtilin, both produced by Bacillus subtilis, are also
thought to penetrate and disrupt the cell membrane, a process where the lipo chain seems to play
a key role.22-24 In addition to the exceptional surfactant power provided by its amphiphilic sequence, surfactin has been reported to exhibit hemolytic, antiviral, antibacterial, and antitumor properties.10 Also, in the case of the cytotoxic molecule
syringomycin from Pseudomonas syringae, which
exhibits toxicity against plant tissues, the amphiphatic nature of the polar peptide head and hydrophobic fatty acid tail allows insertion into the plant
plasma membrane and formation of transmembrane pores, permitting ions to flow freely across
the membrane.25,26 Moreover, as seen for the Streptomyces lipopeptidelactones CDA and daptomycin,
metal ions such as Ca2+ trigger antibiotic activity
also.27-30 These complexes may exert their antibacterial activity through membrane seeking, surfaceactive behavior. Another close relationship between
peptide structure and function is observed for the
glycopeptide antibiotics of the vancomycin family
produced by Streptomyces. Vancomycin is a linear
heptapeptide whose backbone is constrained by oxidative cross-linking. This unique structure sequesters
substrate peptidoglycan-D-Ala-D-Ala termini units
with five hydrogen bonds and shuts down the
transpeptidation reaction.12 A different cellular target
is attacked by the antibiotic pristinamycin from
Streptomyces pristinaespiralis, which blocks polypeptide translation by binding the 50S subunit of bacterial ribosomes at 23S rRNA sites. Investigations
revealed interaction with the ribosome via the 3-hydroxy picolinic acid residue of pristinamycin, emphasizing the importance of nonproteinogenic residues
for antibiotic activity.31
Cyclosporin produced by Tolypocladium niveum
exhibits immunosuppressive and toxic properties due
the formation of a specific complex with cyclophilin
which inhibits, in turn, the protein phosphatase
calcineurin, responsible for T-cell activation.32-34
Figure 1. Natural peptidic products. A selection of nonribosomally synthesized peptides. Characteristic structural features
are highlighted.
Cyclosporin is highly lipophilic, and 7 of its 11 amino

acids are N-methylated. This high degree of methylation protects the peptide from proteolytic digestion
but complicates chemical synthesis due to low coupling yields and side reactions.35 In an iron-deficient
environment some bacteria such as E. coli, B. subtilis,
and Vibrio cholerae synthesize and secrete ironchelating molecules known as siderophores that
scavenge Fe3+ with picomolar affinity, important for
host survival.36,37 Three catechol ligands derived from
2,3-dihydroxybenzoyl (DHB) building blocks in bacillibactin, enterobactin, and vibriobactin complex iron
by forming intramolecular octahedra.
Many nonribosomal peptide products presented
here show distinct chemical modifications, important
to specifically interact and inhibit certain cellular
functions, which are essential for survival. The high
toxicity of the peptide products could therefore also
become a problem for the producer organism unless
strategies for its own protection and immunity have
been coevolved with antibiotic biosynthesis. This
immunity is achieved by several strategies including
efflux pumps, temporary product inactivation, and

modifications of the target in the producer strain.3
The latter strategy is used by vancomycin-producing
Streptomycetes by changing the D-Ala-D-Ala terminus
of the peptidoglycan pentapeptide precursor to a
D-Ala-D-lactate terminus, which reduces binding affinity to vancomycin 1000-fold.12
Due to their exceptional pharmacological activities,
many compounds such as cyclosporin and vancomycin have been synthesized nonenzymatically.38,39 Regio- and stereoselective reactions require the use of
protecting groups as well as chiral catalysts. Moreover, macrocyclization and coupling of N-methylated
peptide bonds are difficult to achieve in satisfying
yields, indicating an advantage of natural vs synthetic strategies. Structural peculiarities of these
complex peptide products suggested early on a nucleicacid-independent biosynthesis facilitated by multiple
catalytic domains expressed as a single multidomain
protein. The diverse chemical reactions mediated by
distinct enzymatic units will be the focus of the
following sections.
Sieber and Marahiel
Figure 2. Surfactin assembly line. The multienzyme complex consists of seven modules (grey and red) which are specific
for the incorporation of seven amino acids. Twenty-four domains of five different types (C, A, PCP, E, and TE) are responsible
for the catalysis of 24 chemical reactions. Twenty-three reactions are required for peptide elongation, while the last domain
is unique and required for peptide release by cyclization.
2. NRPS Factory
Although structurally diverse, most biologically
produced peptides share a common mode of synthesis, the multienzyme thiotemplate mechanism.2,6,40
According to this model peptide bond formation takes
place on large multienzyme complexes, which simultaneously represent template and biosynthetic machinery. Sequencing of genes encoding NRPSs of
bacterial and fungal origin provided insights into
molecular architecture and revealed a modular organization.6 A module is a distinct section of the
multienzyme that is responsible for the incorporation
of one specific amino acid into the final product.3,6,41
It is further subdivided into a catalytically independent set of domains responsible for substrate recognition, activation, binding, modification, elongation,
and release. Domains can be identified at the protein
level by characteristic highly conserved sequence
motifs. Thus far, 10 different domains are known
within NRPS templates which catalyze independent
chemical reactions and will be introduced in more
detail in the following sections. As an example to
illustrate basic principles, Figure 2 shows a prototype
NRPS assembly line for the cyclic lipoheptapeptide

surfactin.42
The carboxy group of amino acid building blocks
is first activated by ATP hydrolysis to afford the
corresponding aminoacyl-adenylate. This reactive
intermediate is transferred onto the free thiol group
of an enzyme-bound 4-phosphopantetheinyl cofactor
(ppan), establishing a covalent linkage between
enzyme and substrate. At this stage the substrate
can undergo modifications such as epimerization or
N-methylation. Assembly of the final product then
occurs by a series of peptide bond formation steps
(elongation) between the downstream building block
with its free amine and the carboxy-thioester of the
upstream substrate. The ppan cofactor facilitates the
ordered transfer of thioester substrates between
catalytically active units with all intermediates covalently tethered to the multienzyme until the product is released by the action of the C-terminal
thioesterase (TE) domain (termination). This strategy
minimizes side reactions as well as diffusion times.
Type I polyketide synthases (PKS) and fatty acid
synthases (FAS) similarly display a multienzymatic
Figure 3. Domain-catalyzed reactions. Domains in action are indicated in red. (A) Recognition and activation of a dedicated
amino acid with ATP by the A domain. (B) Covalent attachment of the activated aminoacyl adenylate onto the free thiol
group of the PCP-bound ppan cofactor. (C) Peptide elongation by the C domain which catalyzes an attack of the nucleophilic
amine of the acceptor substrate onto the electrophilic thioester of the donor substrate.
organization and catalyze a repetitive reaction cycle

involving decarboxylative condensation of smaller
acyl groups.43,44
2.1. Activation by the Adenylation Domain

Each nonribosomal peptide synthesis is initiated
by specific recognition and activation of the relevant
dedicated amino acid from a pool of substrates by the
ca. 550 amino acid long adenylation domain (A
domain). For example, each amino acid found in the
heptapeptide surfactin is directly selected for incorporation into the growing peptide chain by one of the
seven A domains of the surfactin synthetase (Figure
2). Substrate activation is achieved in a two-step
chemical reaction. First, after binding of the cognate
amino acid, the enzyme catalyzes the formation of
an aminoacyl adenylate intermediate at the expense
of Mg2+-ATP and release of PPi (Figure 3A). Second,
the amino acid-O-AMP oxoester is converted into
a thioester by a nucleophilic attack of the free thiolppan cofactor of the adjacent PCP domain, which will
be discussed below (Figure 3B). This mechanism
resembles amino acid activation catalyzed by aminoacyl-tRNA synthetases, although these enzyme
families share neither sequence nor structural simi-
larity.45 Many A domains can be heterologously

expressed in E. coli, and their activity as well as
substrate specificity can be assayed in vitro by an
equilibrium ATP-PPi exchange assay with radiolabeled PPi.46
Sequence alignments of A domains revealed early
on an adenylate (AMP) binding motif that is conserved in a superfamily of so-called adenylate-forming enzymes, which include 4-coumerate-CoA
ligases, acetyl-CoA synthetases, and oxidoreductases.47 Several crystal structures of members of this
family have been solved. These include the oxidoreductase luciferase from Photinus pyralis,48 the
acetyl-CoA synthetase (Acs) of primary metabolism,49 the phenylalanine-activating A domain (PheA)
of the first module of gramicidin S synthetase of B.
brevis,50 and the 2,3-dihydroxybenzoate (DHB) activating A domain (DhbE) of B. subtilis.51 All crystal
structures possess low sequence identity but exhibit
an almost identical fold. They consist of a large N-terminal subdomain and a small C-terminal subdomain
(the term subdomain means a stable tertiary fold
within the structure) with the active site at the junction between them. All NRPS A domains share ca.
30-60% sequence identity,47 which allowed identifi-
cation of 10 core motifs that serve as functional

anchors. In combination with the A-domain crystal
structures it could be shown that many amino acid
residues of these motifs are responsible for, i.e., ATP
binding, hydrolysis, and adenylation of the substrate
carboxy moiety.52 Cocrystallization of PheA with
phenylalanine and AMP revealed specific interactions of the AMP phosphate moiety with a conserved
Lys517 (Figure 4A). The substrate binding site has
Figure 4. Crystal structures of catalytic core domains. (A)

Crystal structure of the phenylalanine-activating A domain
(PheA) of the first module of gramicidin S synthetase of B.
brevis. (B) Structure of the PCP domain derived from the
third module of the B. brevis tyrocidine synthetase. (C)
Crystal structure of VibH, a stand alone C domain of the
V. cholerae vibriobactin synthetase.
a channel-like entrance, and phenylalanine is bound

in a hydrophobic pocket with the carboxy group
interacting with Lys517 and the R-amino group with
Asp235. Mutation of these residues confirmed their
relevance for catalysis.53,54 Nature uses the important
catalytic function of the conserved lysine residue in
related acetyl-CoA synthetase (Acs) for regulation
of acetyl-CoA production in primary metabolism.55
Similar to A domains, Acs synthesizes acetyl-CoA
from acetate, ATP, and CoA through an acetyl-AMP
intermediate. This synthesis is regulated by posttranslational acylation of Lys609, which blocks the
formation of the adenylate intermediate. Reactivation
of the acetylated enzyme requires the NAD-dependent protein deacetylase activity of a CobB Sir2
protein type as shown recently in S. enterica.55 This
mode of regulation was speculated to modulate
Sieber and Marahiel
activity for all adenylate-forming enzymes including

NRPS A domains.
The large diversity of building blocks found in
nonribosomally produced peptides corresponds to the
variations in structure and sequence of A-domain
binding pockets as seen for the aryl-acid-activating
A-domain DhbE.51 The substrate pocket is shallower
than that in PheA, and 2,3-dihydroxybenzoic acid is
coordinated by its two hydroxy groups, which interact
with Ser240 and Asn235, and by the carboxy moiety,
which interacts with two residues Lys517 and His234.
This binding mode contrasts with the situation in
PheA where a second interaction is provided to the
R-amino group and reflects differences in proteinogenic vs nonproteinogenic amino acid activation. By
comparing the residues lining the binding pockets of
PheA and DhbE with the corresponding residues in
other A domains, general rules for substrate specificity were developed. These rules, also referred to as
the nonribosomal code, were tested by mutations
in specificity conferring residues in different A domains, which lead to the conclusion that substrate specificity can be predicted with fairly high
accuracy.52,56-58
The first crystal structures of adenylate-forming
enzymes primarily provided insight into the adenylation half reaction, while a recent crystal structure
of Acs bound to adenosine-5-propyl phosphate and
coenzyme A (CoA) has also shed light on the second
reaction step, thioester formation with coenzyme A.49
This is particularly interesting since aminoacylAMP transfer onto the ppan cofactor of NRPS has
largely been the subject of speculation until now.6,50,51
It was observed that the C-terminal domain of Acs
is rotated by ca. 140 relative to its conformation in
DhbE and PheA. This domain movement exposes a
new set of residues to the active site and relocates
Lys609, important for the adenylation half reaction,
as it coordinates the carbonyl group of acetate. This
conclusion was also supported by experiments in
which acetylation of Lys609 was shown to inhibit
adenylation but had no effect on catalysis of the
second half reaction, the acylation of CoA. This result
suggests the existence of two sets of catalytic residues
corresponding to two different reaction steps and to
two different structural orientations.49,55 Moreover,
the structural rearrangement appropriately positions
the CoA thiol for nucleophilic attack on the acetylAMP intermediate. Since crystal structures only
provide snapshots of individual states in a dynamic
multiple-step process, the findings suggest that members of the adenylate-forming family of enzymes
adopt two different orientations to catalyze their two
half reactions.
In contrast to ribosomal protein synthesis, which
shows very accurate proof reading, less stringent substrate selection and incorporation is observed for
some nonribosomal peptides.52 However, new studies
on the mode of substrate selection of NRPS A domains have revealed an intrinsic ATPase activity
which is enhanced in the presence of noncognate
amino acid substrates.59 In turn, less pronounced
variations in ATPase activity are observed in A domains with relaxed amino acid specificity. The cyclic
Figure 5. Phosphopantetheinylation: Apo to holo enzyme conversion. (A) The phosphopantetheine moiety of coenzyme
A (red) is covalently attached to an invariant serine residue of PCP by Sfp, a dedicated phosphopantheteine transferase.
(B) Crystal structure of Sfp with its substrate coenzyme A and Mg2+.
decapeptide tyrocidine consists of a mixture of four

compounds that vary in two positions,60 whereas
about 30 variants are known for the immunosuppressive drug cyclosporin.32 With this simple strategy the
diversity of natural products can be readily increased.
2.2. Intermediates Transport by the Peptidyl

Carrier Protein
The only NRPS domain without autonomous catalytic activity is the peptidyl carrier protein (PCP),
also referred to as thiolation domain (T). The protein
comprises ca. 100 amino acids and is located downstream of the A domain. Within the NRPS assembly
line PCP is responsible for transportation of substrates and elongation intermediates to the catalytic
centers. As discussed in the previous section, the A
domain catalyzes the transfer of the activated aminoacyl-adenylate substrate onto the terminal cysteamine thiol group of the ppan cofactor bound to
PCP in its second half reaction (Figure 3B). In
nonribosomal peptide synthesis the combination of
A domain and PCP is defined as an initiation module
since both domains are required to activate and
covalently tether the first building block for subsequent peptide synthesis. The activity of recombinant
A domains with adjacent holo-PCPs can be assayed
in vitro by an aminoacylation assay with ATP and
radioactive amino acids.46 In contrast to ribosomal
protein synthesis with tRNA-bound ester intermediates, nonribosomal peptide synthetases exploit more
reactive PCP-thioesters. This difference in reactivity
is due to the lower mesomeric stabilization of the
thioester, which forms less stable pdp double bonds
and emphasizes that the initial activation energy
provided by the A domain is preserved here for
subsequent catalytic reactions such as condensation,
hydrolysis, and cyclization (see below).61
A 20 long prosthetic ppan moiety of coenzyme A

is covalently tethered to the side chain of a strictly
conserved PCP serine residue and serves as a crane
for building-block delivery.62-65 Transfer of ppan onto
apo-PCP is catalyzed by NRPS-specialized 4-phosphopantetheinyl transferases such as Sfp and Gsp
from B. subtilis and B. brevis, respectively53,66,67 (Figure 5A). The conversion of inactive apo-PCP into its
active ppan-PCP holo form was monitored in vitro
with recombinant Sfp and PCPs from the surfactin
synthetase. These studies revealed the very low
selectivity of Sfp for the carrier proteins.68 Sfp was
shown to efficiently phosphopantetheinylate not only
apo-PCPs from various NRPS templates but also acyl
carrier proteins from fatty acid and polyketide synthases.66,69 Furthermore, recent results suggest a
broad acyl-CoA tolerance. Various synthetic peptidyl-CoAs were covalently attached to apo-PCPs
under catalysis of Sfp, albeit with reduced efficiency.
This relaxed specificity of Sfp has proved useful for
preparative applications, which will be presented below.70,71 Insights into how Sfp mediates binding and
protein recognition were provided by a crystal structure in complex with its substrate CoA72 (Figure 5B).
The structure of the 224 amino acid comprising Sfp
monomer shows a pseudo-2-fold symmetry which
divides the protein into two similar folds of almost
identical size. The CoA substrate is bound in a bent
conformation within a pocket formed by the two Sfp
halves. The 3-phospho-5-ADP moiety of CoA is well
defined in the electron density map and coordinated
by several Sfp residues and Mg2+, while the main
part of the ppan arm shows no interactions with Sfp
and points out into bulk solvent. This unique type of
CoA coordination is in complete agreement with the
observed binding tolerance for peptidyl-CoA substrates (and acyl-CoA) in which the peptide compo-
nent presumably does not interact with the enzyme

(section 3.2.3).68 PCP cofactor modification was suggested to occur via Sfp-Glu151-mediated deprotonation of the serine hydroxy group of PCP and its
subsequent nucleophilic attack on the -phosphate
of CoA, leading to holo-PCP and 3,5-ADP as a byproduct. Several mutational analyses of Sfp confirmed this type of CoA binding and revealed several
residues in a loop region between 4-R5 in the PCP
binding region.73
The first solution structure of a PCP was solved
using NMR spectroscopy with the PCP of the third
module of the B. brevis tyrocidine synthetase74 (Figure 4B). PCP is a distorted four-helix bundle with
an extended loop between the first two helices, which
is probably important for interaction with Sfp. The
invariant serine residue, the site of cofactor binding,
is located at the interface between this loop and the
second helix. The cofactor shows no interactions with
the protein and is accommodated in the solvent. No
peptide-binding pocket was observed, which is in
agreement with the lack of substrate selectivity. The
fold is well defined between residues 8 and 82, and
the structural core was defined to be a region spanning 37 amino acids in both directions from the
conserved serine. PCPs have a function similar to
acyl carrier proteins (ACP) from fatty acid and
polyketide synthases. Sequence homologies between
PCPs and ACPs only exist in the immediate neighborhood of the invariant serine residue, although all
proteins possess an almost identical structural
fold.75-77 The most obvious difference between ACPs
and PCPs is the overall charge of the proteins. While
ACPs have predominantly acidic side chains on their
surface, the PCPs surface is much less polar. This
corresponds to the charges of the corresponding
ppan-transferases AcpS and Sfp. While AcpS exclusively primes ACP and not PCP, it was observed that
Sfp is promiscuous enough to serve both ACP and
PCP. A crystal structure of a complex from B. subtilis
ACP and AcpS77 throws light on the importance of
the charged residues and explains why AcpS only
interacts with ACP. Acidic residues of the ACP helix
3 interact specifically by the formation of salt bridges
with the first helix of AcpS. This positions the serine
residue in the right orientation for the priming
reaction. In PCPs one of the interacting acidic
residues is exchanged by a basic residue, which might
explain the inability of AcpS to prime PCPs. On the
basis of these results a novel hybrid PCP was
constructed by replacing the B. brevis TycC3 PCP
helix 2 with the corresponding helix of B. subtilis
ACP which contains the interacting residues.78 This
hybrid enzyme was stoichiometrically phosphopantetheinylated in vitro by both AcpS and Sfp.
2.2.1. Misacylation and Regeneration

The promiscuity of Sfp also causes undesired
misacylation of PCPs within NRPS assembly lines
since not only CoA but also acyl-CoAs can serve as
cofactor donors. To regenerate these misprimed
NRPS templates, nature has developed specific enzymes which catalyze hydrolysis of the undesired acyl
group. These so-called thioesterase II domains (TEII)
Sieber and Marahiel
were shown to be specific for acyl-PCPs in in vitro

assays, while there was no hydrolysis observed for
acyl-ACPs, which are essential in primary fatty acid
metabolism.79 Comparison of the catalytic properties
of TEII-mediated aminoacyl- or peptidyl-PCP hydrolysis vs acetyl-PCP showed a strong preference
for the latter substrate, which indicates that this
proofreading enzyme is important for NRPS activity.
Deblocking of misacetylated PCPs was also confirmed
by TEII knockout studies.
2.3. Peptide Elongation by the Condensation

Domain
After activation and covalent binding of the first
amino acid substrate by the A-PCP initiation module, peptide synthesis proceeds by stepwise condensation with amino acid building blocks bound to PCPs
of the downstream elongation modules (C-A-PCP)n.
Peptide bond formation is mediated by a ca. 450
amino acid long condensation domain (C domain).
The C domain catalyzes the nucleophilic attack of the
downstream PCP-bound acceptor amino acid with its
free R-amino group on the activated thioester of the
upstream PCP-bound donor amino acid or peptide80
(Figure 3C).
Biochemical characterization of different C domains from the B. brevis tyrocidine and E. coli
enterobactin synthetases revealed insights into their
substrate specificities. Probing substrate specificity
in the natural synthetase is difficult because the
upstream donor and downstream acceptor substrates
are defined by restrictive A domains. To directly
evaluate C-domain specificity, various aminoacylCoA substrates were synthesized and attached via
an Sfp-catalyzed reaction to the apo-PCPs of a
minimal, bimodular NRPS enzyme composed of module 1 from B. brevis gramicidine synthetase and
module 2 from tyrocidine synthetase. Mischarging of
PCPs from both modules in condensation assays
revealed that the C domain of tyrocidine module 2
seems to possess an acceptor position for the downstream PCP-bound nucleophile that discriminates
against the noncognate D-enantiomer as well as
differences in the side chain. By contrast, low sidechain selectivity was observed for the donor position
of the upstream PCP-bound electrophile. Interestingly, a preference for cognate D-enantiomers was
observed.80 This was confirmed by further investigations with the C domain of tyrocidine elongation
module 5, which revealed that the donor position
exclusively selects a tetrapeptide with the cognate
D-configuration of the C-terminal residue for condensation reactions.71 This shows that besides A domains
and TEII domains C domains also represent a
selectivity filter in nonribosomal peptide synthesis.
Selection of the correct downstream nucleophile by
the acceptor position prevents the formation of
product mixtures and facilitates peptide synthesis in
a directed manner. Assays with amino acids attached
to N-acetylamine (SNAC) support the previously
mentioned results and show that these soluble mimics of the ppan arm can serve as substrates for C
domains, allowing the first kinetic studies at the
acceptor site on C domains. Analogous experiments
at the donor site have not been possible however due

to low turnover.81
Sequence alignments of several C domains revealed
a highly conserved HHXXXDG motif that is also
found in acyltransferases such as chloramphenicol
acetyltransferase (CAT), NRPS epimerization, and
heterocyclization domains.82,83 Mutations in residues
His147 in the conserved motif of the C domain of
tyrocidine module 2 and His138 in enterobactin
synthetase module F abolished activity in condensation assays,84,85 providing evidence of an active role
of the conserved histidine in catalysis. A recent
crystal structure of VibH revealed insights into the
architectural organization of this enzyme82 (Figure
4C). VibH of the Vibrio cholerae vibriobactin synthetase is a very unusual NRPS C domain since it
catalyzes peptide bond formation between a PCPbound DHB donor and a freely diffusible norspermidine acceptor substrate.86 Moreover, VibH represents
one of the few NRPS C domains which are not
covalently attached to a NRPS module. VibH is a
monomer with two pseudodimeric domains consisting
of an RR sandwich. The conserved HHXXXDG motif
is located in a loop at the interface between the two
domains, which provides access to His126 from two
different faces of the enzyme. His126 could function
as a base to activate the free R-amino-group of the
downstream acceptor substrate for nucleophilic attack on the upstream thioester. By analogy to structural data on CAT, it was postulated that the
C-terminal face of VibH would bind the donor DHB
substrate with its ppan arm extending into a solvent
channel. Norspermidine could then nucleophilically
attack the thioester from the opposite N-terminal
site. In other NRPS C domains with donor and
acceptor presented on PCPs, both ppan arms would
be extended into the solvent channel from opposing
open ends, which would represent binding sites for
the two PCP proteins. In contrast to the tyrocidine
C domain and EntF C domain, mutation of the same
conserved histidine in the C domain VibH resulted
in little reduction in catalytic activity, which may
reflect its specialized catalytic requirements.82
The heterocyclization domain (Cy), which can
replace C domains in NRPS templates and catalyzes
peptide elongation as well as heterocyclization by a
more complex mechanism, is structurally and mechanistically related to the C domain. Five-membered
heterocyclic rings such as oxazoline in vibriobactin
and thiazoline in bacitracin are common structural
features of nonribosomal peptides and important for
chelating metals or interaction with proteins, DNA,
or RNA87 (Figure1). The first reaction step promoted
by Cy domains is the nucleophilic attack of a PCPbound cysteine, threonine, or serine acceptor substrate onto the thioester of the donor substrate.
Recent mutational studies of the bacitracin Cy domain demonstrated that the free R-amino group is
the nucleophile in this step, as observed for C
domains.88 In the next step the side chain hydroxy
or thiol group carries out a nucleophilic attack onto
the R-carbonyl C atom of the donor amino acid,
forming a five-membered heterocyclic ring, which is
subsequently dehydrated to form the final oxazoline
or thiazoline product. Further insights into the

mechanism of catalysis were provided by the vibriobactin synthetase, which contains two adjacent Cy
domains. Analysis revealed that Cy domain 2 was
responsible for the condensation step while Cy domain 1 carried out the heterocyclization and dehydration steps. This result indicates separate mechanisms for catalysis of condensation and heterocyclization.86 These investigations as well as more general
structural similarities indicate that Cy domains are
evolutionarily specialized C domains.
2.4. Editing Domains

While the amino acid is covalently tethered onto
the PCP, several editing domains can carry out
further modifications to increase the diversity of the
final product. Incorporation of D-amino acids and
methylated amide bonds increases the stability of the
peptide against proteolytic digest and also favors
population of unique conformations important for
biological activity.
2.4.1. Epimerization
Almost every nonribosomally synthesized peptide
contains D-configurated amino acids to a various
extent. NRPSs utilize two different strategies for
their incorporation. The most common route involves
epimerization of L-amino acids by integrated 450amino-acid-long epimerization domains (E).89 The
latter promote epimerization of the CR-carbon of the
PCP-tethered aminoacyl substrate to afford a D/L
equilibrium.90 Racemization in vitro can either occur
from L to D or D to L. Rapid quench kinetics revealed
that this equilibrium is achieved within seconds.91
Specific incorporation of only the D-amino acid into
the growing peptide chain is ensured by the enantioselective donor site of the downstream condensation
domain.80 This principle is also used in the surfactin
synthetase in modules 3 and 6 to incorporate D-Leu
twice in the final product. The combination of D- and
L-amino acids contributes to the unique conformation
of surfactin that is important for its biological activity.10 A second strategy of D-amino acid incorporation
is often observed in fungal systems:32 The A domain
of cyclosporin synthetase, for example, exclusively
incorporates D-Ala, which is provided by an external
racemase.92
Biochemical characterization of E-domain substrate specificity also revealed that noncognate amino
acids were racemized but with lower efficiency.93
Further studies showed that artificial E-domain
constructs without a preceding C domain (as observed in the native initiation module) could epimerize aminoacyl-PCP. In contrast, identical constructs
with a preceding cognate C domain (as in an elongation module) did not show epimerization activity for
the bound aminoacyl-S-ppan substrate. This observation led to the conclusion that C domains tightly
bind aminoacyl-PCP in the acceptor site until condensation occurs. The resulting peptidyl-PCP has a
lower binding affinity for the acceptor site and is then
transferred to the subsequent E domain or next C
domain.71,94 These investigations contributed to an
Figure 6. Directionality of peptide elongation. Reaction

sequence (1-5) within an NRPS elongation module (shown
in gray) containing an epimerization domain. AA ) amino
acid; AAX ) upstream amino acid or peptidyl chain.
understanding of timing and directionality in nonribosomal peptide synthesis (Figure 6).

Although a crystal structure of an E domain is not
yet available, sequence alignments indicate their
similarity to C and Cy domains.82 C and E domains
share a conserved HHXXXDG motif, the second His
of which seems to be involved in catalysis.83,90 Similar
to C domains, it is assumed that this residue de- and
subsequently reprotonates the CR carbon atom (one
base mechanism). Sequence alignments and structural comparison with the VibH C domain revealed
that E domains have an insertion at the C-terminal
end of the solvent channel, indicating that this face
may be blocked and PCP binding may occur from the
N-terminal face only.82
Moreover, it was shown that E domains play a
crucial role in NRPS protein-protein recognition,
mainly in bacterial systems.95 For example, two
subunits in surfactin synthetase have terminal E
domains which communicate intermolecularly (in
trans) with the C domain of the adjacent subunit
(Figure 2). Moreover, product formation between
two modules of the tyrocidine synthetase in trans
was only observed if the free-standing module 1
(A-PCP-E) harbored the cognate E domain. A
recent sequence analysis of the operon-encoding
linear gramicidin A revealed the presence of 7 E
domains, one of which surprisingly attached to a
glycine-incorporating module. Since this E domain,
which carries mutations in core motifs, is at an in
trans junction of the synthetase, it may only be required for mediating protein-protein recognition.96
2.4.2. Methylation
Some nonribosomal peptides such as cyclosporin,97
enniatin,98 actinomycin,99 and pristinamycin11,100 have
N-methylated peptide bonds. This modification is
introduced by a ca. 420 amino acid comprising
N-methylation domain (N-Mt) which is inserted into
the accompanying A domain. The N-Mt domain
catalyzes the transfer of the S-methyl group of
S-adenosylmethionine (SAM) to the R-amino group
Sieber and Marahiel
of the thioesterified amino acid releasing S-adenosylhomocysteine as a reaction byproduct. In comparison to other NRPS domains discussed previously, less
is known about N-Mt domains. It was shown for
actinomycin synthetase a valine-activating A domain
in module 2 could be replaced by a methyl-valineactivating A-N-Mt domain. This construct promoted peptide bond formation with acyl threonine,
which is in agreement with the observed relaxed
donor site specificity of C domains. In the absence of
the appropriate acceptor only methyl-valine was
observed, indicating that N-methylation occurs before
peptide bond formation.99
SAM-dependent C-methylating domains (C-Mt) are
also known. In yersiniabactin synthetase a thiazoline
ring is C-methylated.101 Recently, a new type of SAMdependent methyl transferases was identified. Melithiazol synthetase from the myxobacterium Melittangium lichenicola lacks the conserved SAM binding
signature sequence of N-Mt domains.102 This enzyme
is involved in an unusual methylation of a carboxy
acid to form an ester, which represents the last step
in melithiazol biosynthesis.
2.4.3. Further Modifications

Besides epimerization and methylation further
modifications can be introduced into the peptide
sequence. The oxidation state of oxazoline and thiazoline rings can be altered under catalysis of additional oxidation (Ox) or reduction (R) domains. The
ca. 250 amino acids comprising oxidation domains are
observed, e.g., in the NRPS modules of bleomycin,
epothilone, or myxothiazol synthetases.103-105 Two
different organizations of the flavine-mononucleotide (FMN) containing domains have been reported.
In myxothiazol synthetase one Ox domain is Cterminally fused to the PCP while another Ox domain
is incorporated within the A domain.105 Recent biochemical characterization of a recombinant Ox domain from the epothilone synthetase revealed that
this enzyme retains autonomous activity, catalyzing
the oxidation of thiazoline to thiazole. Molecular
oxygen was required in these experiments to reoxidize reduced FMN.104 Similar experiments were
carried out with an Ox domain from bleomycin.106
Recently, it was shown that an in-frame deletion of
an Ox domain from myxothiazol synthetase did not
alter the final product from a thiazole into a thiazoline derivative. It was therefore speculated that the
other Ox domain of the synthetase would oxidize both
thiazolines.107 An interesting oxidation strategy is
also realized in the biosynthesis of melithiazol and
myxothiazol.102,107 Glycine is incorporated as the last
amino acid into the myxothiazol precursor and subsequent hydroxylation by a monoOx domain leads to
the release of the myxothiazol amide and PCP-bound
glyoxylic acid. While the terminal amide of myxothiazol is the final product, the terminal amide of
melithiazol is processed further by enzymatic hydrolysis and methylation (see N-methylation).
Nature has also developed an opposite strategy
which allows reduction of heterocycles by addition of
two electrons as seen in one of the rings in yersiniabactin and pyochelin. Reduction is catalyzed by
Figure 7. Peptide release by the TE domain. Depending on the identity of the NRPS template, product release can be
carried out either by the external nucleophile water to give the linear acid product (A), as observed in case of the vancomycin
TE, or by an internal nucleophile to yield a cyclic product as seen for tyrocidine TE (B).
NADPH-dependent R domains which catalyze the

reduction of thiazoline into thiazolidine.108 R domains
are also involved in peptide release as discussed later.
A key characteristic of many nonribosomally produced peptides such as surfactin, mycosubtilin, and
fengycin is the N-terminally fused lipo acid. Thus far,
it is not well understood how this lipoinitiation is
mediated. Since all initiation modules in these synthetases start with a C domain it is assumed that
an acyl transferase provides the acyl chain for the C
domain donor site.13,42
Another modification of the N-terminal peptide end
is N-formylation catalyzed by a N-formyltetrahydrofolate-dependent formyltransferase domain (F) as
observed in anabaenopeptilid 90-A and linear gramicidine A.96,109
2.5. Peptide Release

All catalytic domains discussed so far are repeating
units of the enzymatic template and contribute to the
synthesis of a linear peptide molecule tethered to the
multienzyme. To reactivate the multienzyme for a
next synthesis cycle the mature peptide has to be
cleaved once it reaches the end of the assembly line.
This reaction is usually accomplished by a ca. 280
amino acid long thioesterase domain (TE domain)
fused to the C-terminal module, also referred to as a
termination module. In the last step of peptide
assembly an active site serine of the TE domain
carries out a nucleophilic attack on the PCP-peptidyl
thioester to form a covalent acyl-enzyme intermediate. Depending on the NRPS template and hence on
the TE domain, this intermediate can either be released by hydrolysis as a linear acid or by an intramolecular reaction with an internal nucleophile to
give a cyclic peptide (Figure 7). Hydrolytic release is
observed for peptides such as vancomycin, whose
peptide backbone is constrained by further postsyn-
thetic oxidative cross-linking reactions, whereas the

tyrocidine and surfactin backbones become directly
constrained by TE-mediated macrocyclization (Figure
2 and Figure 7A).12,110
Alternatively, peptide release can also occur concomitant with reduction of the carboxy group catalyzed by the NADPH-dependent reduction domain (R)
to give linear aldehydes or alcohols such as in the
yeast biosynthetic pathway for the essential amino
acid lysine111 and in the biosynthesis of linear gramicidin A in B. brevis96 or by head-to-tail condensation
mediated by the C domain as seen in cyclosporin
synthetase. Many other cyclization strategies have
been uncovered which give rise to a large and diverse
set of cyclic or cyclic branched molecules with distinct
biological activities. Thioesterase domains that catalyze a cyclization reaction are also referred to as
peptide cyclases.
The broad variety of cyclization strategies is encoded by the three-dimensional architectures of TE
domains. Recent access to the crystal structure of the
surfactin TE (Srf TE) domain has afforded deeper
insight into the mechanism of cyclization.112 The
crystal structure represents a very prominent R,hydrolase fold which is characteristic of a large family
of proteases, lipases, and esterases (Figure 8A). The
R,-hydrolase fold also provides a stable scaffold
for the active sites of a variety of other TE domains which share only low sequence identities of
10-15%.4,113 This high flexibility in primary sequence
reflects the broad spectrum of activities mediated by
these enzymes which need a rigid fold to ensure
precise alignment for catalytic action. The crystal
structure of Srf TE revealed several catalytic residues
and regions which were assumed to play a crucial
role in catalysis. A putative PCP domain interaction
site ensures docking and presentation of the ppanbound substrate in the active site via a cleft in the
Figure 8. Structure of Srf TE. (A) Crystal structure of

Srf TE (R,-hydrolase fold) together with an upstream PCP
domain (shown in yellow) at a putative interaction site. A
modeled posphopantetheine arm points from the invariant
PCP serine residue (*) into the Srf TE active site cavity
(circled). (B) Magnification of the binding pocket reveals
the catalytic triad (Ser80, His207, and Asp107) shown in
yellow. The cavity is lined predominantly by hydrophobic
residues with the exception of two positively charged side
chains of Lys111 and Arg120 (yellow).
surface of the TE (Figure 8A). The active site cavity

is bowl shaped and lined with predominantly hydrophobic and aromatic residues with the exception of
two positively charged residues Lys111 and Arg120
(Figure 8B). This hydrophobic environment matches
the hydrophobic peptide sequence of surfactin, while
the two positive charges in the active site were
postulated to mediate recognition and alignment by
coordination of Glu1 and Asp5 in the surfactin
sequence. Mutations of these residues to Ala confirmed their role in catalysis.110 The three residues
Ser80, His207, and Asp107 exhibit the right geometry and distance required for a catalytic triad.
Mutation of all three residues confirmed that serine
forms a covalent acyl-enzyme intermediate after
deprotonation by His.110 This acyl-enzyme intermediate was first identified by mass spectrometry in the
enterobactin TE after mutation of the triad residue
Sieber and Marahiel
His to Ala. No such behavior was observed for the

same mutation in Srf TE, which indicates different
reactivity profiles in different TE domains. Better
understanding of the Srf TE substrate-binding mode
was then achieved by cocrystallization studies with
a boronic acid inhibitor.110 The cocrystal structure
confirmed previous mutational studies and revealed
binding of the boronic acid by the triad residue Ser80.
Moreover, the structure revealed distinct recognition
and binding of the C-terminal residues Leu7 and
D-Leu6 of the surfactin peptide in hydrophobic pockets while the rest of the peptide seemed to be less
ordered and constrained by the enzyme. Cyclization
of lipopeptides may be triggered by a pronounced
hydrophobic pocket which is present in the Srf TE
active site close to the predicted fatty acid position.
Binding of the fatty acid in this pocket might ensure
a precise positioning of the -hydroxy group required
for a nucleophilic attack on the serine-peptide oxoester. This assumption is supported by investigations
with the CDA TE domain where longer fatty acid
chains increased not only cyclization yields dramatically but also the regioselectivity of the cyclization
reaction.70
Although all members of the R,-hydrolase enzyme
family possess a similar structural fold, the mode by
which the acyl-enzyme intermediate breaks down
can be substantially different. While some TE domains such as vancomycin TE and lipases catalyze
hydrolysis of the enzyme-substrate oxoester to give
linear acids, other TE domains catalyze cyclization.
A critical stage of the two different paths is encoded
by the oxyanion hole. Two amide bonds stabilize the
negatively charged tetrahedral intermediate in the
release reaction. The structural integrity in the Srf
TE domain is ensured by a rigid proline residue
adjacent to one of the stabilizing amide bonds. This
proline is highly conserved in cyclizing TE domains.
In contrast, a flexible glycine is at this position in
hydrolyzing lipases, which may ease the entry of
water during a critical step of catalysis. Mutation of
Pro26 to Gly in the Srf TE confirmed this hypothesis,
dramatically increasing the amount of hydrolysis
product formed. This residue may therefore represent
a switch between cyclic or linear products in different
enzyme families.
2.5.1. Diversity by Cyclization

Macrocyclization is a key structural feature of
many nonribosomal peptide products which constrains the flexible peptide chain in a biologically
active conformation. Rigidity in the peptide backbone
facilitates specific interactions with dedicated cellular
targets (section 1). In turn, many linear analogues
of cyclic peptides display no or only diminished
activities.114 The huge variety of cyclic nonribosomal
peptides is achieved by a diverse set of TE domains.
Their high degree of specificity allows cyclization
reactions to occur in the presence of other nucleophiles without the use of protecting groups. Contrary
to the situation in solution where the peptide chain
has to find the right conformation for cyclization by
various rotations, the TE domain active site guides
the folding of the peptide chain. TE domains can
Figure 9. Cyclization strategies. The majority of cyclization reactions within NRPS templates are catalyzed by TE domains
(red). In some cases, also C domains and R domains (gray) are involved in cyclic peptide release.
specifically select one residue from a source of nucleophiles to catalyze regio- and stereoselective cyclization reactions. These nucleophiles always attack
to the C-terminal end of the peptide; no side-chainto-side-chain or N-terminal-amine-to-side-chain cyclizations have ever been observed. In basic head-totail cyclizations, as seen for tyrocidine (B. brevis), the
free N-terminal amine is connected to the C-terminus, yielding a lactam product (Figure 9). Surfactin
and mycosubtilin (B. subtilis) are examples of a
branched chain lactone and lactam, respectively
(Figure 9). The TE mediates ring closure by connecting a -hydroxy fatty acid for surfactin and a -amino
fatty acid for mycosubtilin to the peptide C-terminus.
Both lipoheptapeptides share similarities in size,
activity, and mode of synthesis as well as in the
precursor -keto fatty acid. In the case of surfactin
the ketone is reduced to a hydroxy group, while in
mycosubtilin several catalytic domains convert the
ketone into an amino group. This processing of the
same precursor in different ways leads to an increase
in structural diversity. A change in the nucleophile
from a hydroxy group to an amine seems to alter the
chemoselectivity of the corresponding TE. Experiments revealed that it is not possible to cyclize a
-amino analogue of either surfactin or mycosubtilin
directly with Srf TE, which demonstrates that alternative nucleophiles are not tolerated.110 Similar
results were observed for CDA TE.70 Srf TE is specific
only for the (R)-configured hydroxy fatty acid, emphasizing a high degree of chemo-, stereo-, and
regioselectivity also observed for the analogous fengycin, CDA, and syringomycin TE domains.70,115,116
Besides functionalized fatty acid residues, amino acid
side chains can also be involved in cyclization. The
TE domains of fengycin, syringomycin, and CDA
synthetases specifically select dedicated tyrosine,
serine, and threonine side-chain nucleophiles for
connection with the C-terminus (Figure 9).
In many NRPSs the modular enzymatic template
is collinear with the peptide product sequence. In
these linear type A NRPS assembly lines TE domains
only catalyze one reaction step, either cyclization or
hydrolysis of the linear precursor.5 However, in
iterative NRPS type B templates, the TE domains
have an additional function which allows the enzyme
to repeat the collinear synthesis once or twice. In this
case the TE has to count the monomers stalled at
the end of the assembly line and initiates release by
cyclic dimer or trimer formation only once the desired
length is achieved. This strategy is observed for
gramicidin S, enterobactin, and bacillibactin peptides
(Figure 9). Less is known about the mechanism and

structure of iterative TEs. Detailed mass spectrometric analysis was carried out for the last module
of the enterobactin assembly line (EntF) containing
a C-terminal TE domain which catalyzes cyclotrimerization of three 2,3-dihydroxybenzoyl serine (DHBSer) units to give the cyclic trilactone enterobactin.117
It was demonstrated that this TE is involved in two
reactions: acyl chain growth and cyclization.
Macrocyclization is not exclusively mediated by TE
domains. In cyclosporin A the final peptide bond is
formed by a putative condensation domain118 instead
of a TE domain, emphasizing that nature developed
additional enzyme species capable of catalyzing
product release by cyclization. A PCP-C didomain
can also catalyze oligomerization as observed for the
trilactone enniatin.119 Recently, a new type of headto-tail macrocyclization was reported for nostocyclopeptide from the terrestrial cyanobacterium Nostoc
sp. The C-terminal residue of the linear peptide is
reduced by the action of an R domain to give an
aldehyde, which is then intramolecularly captured
by the R-amino group of the N-terminal amino acid
residue to give a cyclic imine120 (Figure 9).
2.6. Quaternary Architecture

Many enzymes catalyzing sequential metabolic
reactions aggregate by noncovalent linkage of identical or nonidentical subunits to form multienzyme
complexes. The fatty acid synthases (FAS) of eukaryotes and the modular polyketide synthases (PKS) are
well-known examples of a class of enzymes that forms
complexes composed of two identical subunits (Figure
10).121-123 A double-helical structure model for modu-
Figure 10. Quaternary structures of NRPS, PKS, and

FAS multienzyme complexes. While FAS and PKS fold into
homodimeric enzymes with domains communicating across
the dimeric interface, no such interaction was found for
several NRPSs (* except VibF).
lar PKS has been proposed.123 According to this model

the enzyme subunits are orientated head-to-head and
folded in an interwound helical manner. Ketosynthase (KS), acyl carrier protein (ACP), and acyltransferase (AT) domains form a core which is necessary
for the observed interaction between KS and ACP
domains from different strands (Figure 10). Optional
domains such as the ketoreductase domain (KR) are
accommodated in outside loops. The model for FAS
enzymes also requires functional interactions be-
Sieber and Marahiel
tween both subunits to explain the results of crosslinking studies.124,125

The related organization and chain elongation logic
of PKS and NRPS as well as the existence of
naturally occurring hybrids that produce natural
products such as epothilone has led to the assumption
of a similar quaternary structure for both enzyme
families.44 In contrast to PKS and FAS enzymes, not
much was known about the quaternary organization
of NRPS for a long time. Although high-resolution
structures of NRPS core domains are now available,
structural insights into multidomain organization
have remained elusive. Structural differences in the
oligomeric states of NRPS and PKS enzymes were
first revealed by comparison of their respective
thioesterase domains.112,113 While the crystal structure of the erythromycin PKS TE domain clearly
shows a leucine-rich hydrophobic dimer interface,
only a monomeric structure is observed for the
surfactin NRPS TE domain. A monomeric organization has been similarly postulated82 for the NRPS C
domain VibH.
The first global analysis of the quaternary architecture of NRPS was carried out with various NRPS
multidomain modules derived from tyrocidine, gramicidin S, enniatin, and enterobactin synthetases using
strategies which were previously successfully applied
to establish a dimeric interaction in FAS and PKS
enzymes.126,127 Biophysical methods such as gel filtration, chemical cross-linking, and analytical ultracentrifugation revealed a monomeric organization of all
enzymes investigated. For the larger dimodular
enzyme TycB2-3 (A-PCP-C-A-PCP-E) a dimeric
species was observed during ultracentrifugation at
high nonphysiological concentrations, which was
speculated to be an evolutionary relic. Biochemical
experiments such as mutant complementation and
affinity tag labeling also supported the monomeric
state of the tested NRPS enzymes.126 On the basis of
these results, an overall monomeric structure of
NRPSs was suggested. The existence of NRPS-PKS
hybrids lead to the assumption that the PKS portion
of the protein dimerizes while the NRPS part is in a
monomeric form. In contrast to dimeric FAS and
PKS, the monomeric structure of many NRPS enzymes emphasizes that there is no mechanistic
requirement for them to function as dimers. In a
recent publication Smith and co-workers showed that
there is no such mechanistic requirement for FAS128
either, in contrast to previous belief. Full inactivation of one polypeptide chain in a dimer did not
abolish FAS activity, indicating that one functional
subunit in a heterodimer is sufficient for product
assembly. The dimeric state therefore seems to
contribute predominantly to enzyme stability and
integrity, since monomeric FASs are functionally
inactive.
Recent ultracentrifugation and mutant complementation studies of the NRPS module VibF, which
has the unusual domain organization Cy-Cy-A-CPCP-C, revealed a dimeric state.129 In contrast to
cold labile FAS, the VibF dimers dissociate at elevated temperatures, which suggests a different
pattern in dimer breakdown and reformation. Since
the accompanying free-standing C-domain VibH was

shown to be monomeric, two oligomeric states in the
vibriobactin biosynthetic template can be assumed,
similar to the postulated situation in NRPS-PKS
hybrid enzymes. This may also indicate that NRPS
enzymes are able to display two different modes of
structural organization. Knowledge of the quaternary
structure of NRPSs is not only important for understanding reactions occurring on or between enzymes
but may also contribute to engineering of more
efficient new hybrid enzymes. A structural model of
intramolecular domain interactions within a monomer must await high-resolution structures of a multidomain system.
2.7. NRPSs in Higher Eucaryotes

Until recently, NRPSs were only observed in
bacteria, fungi, and yeast, but two recent publications
provide evidence that a NRPS-like assemblage is also
formed in higher eukaryotes.130,131 In Drosophila the
three-domain multienzyme Ebony (A-PCP-AS) was
shown to be involved in histamine neurotransmitter
metabolism at the photoreceptor synapse of the eye.
Its postulated function is to ensure rapid removal of
histamine from the synaptic cleft, which is essential
to excite the postsynaptic cell by disinhibition. Histamine is trapped by Ebony via peptide bond formation with -alanine. Experimental data with a 879
amino acid long recombinant form of Ebony revealed
a novel two-step reaction mechanism involving amino
acid activation by an A domain, covalent attachment
to the ppan group of the PCP, followed by peptide
bond formation via a C-terminal domain with an as
yet unknown mechanism.130 In vitro assays showed
that the Ebony A domain exclusively activates -alanine and transfers it subsequently to PCP. The novel
C-terminal AS domain presumably then catalyzes the
nucleophilic attack of primary amines such as histamine onto the -alanine thioester. Given the existence of Ebony in Drosophila, NRPSs in higher
organisms cannot be excluded. This conclusion was
confirmed by a recent publication of Kato and coworkers, who described a 1100 amino acid long
multienzyme from mouse, U26, which contains an A
domain, PCP, and seven pyrroloquinoline quinone
(PQQ) binding motifs.131 In mammals U26 seems to
be involved in lysine degradation by oxidation of
2-aminoadipic 6-semialdehyde by PQQ to 2-aminoadipic acid. Interestingly, in yeast lysine is synthesized
by the reverse pathway from the NRPS enzyme Lys2
and Lys5. Lys2 has NADPH-dependent 2-aminoadipic acid reductase activity, and Lys5 catalyzes 4phosphopantetheinylation of Lys2.111 A homologue of
Lys5 has been identified in humans, raising the
question as to whether a structural analogue of Lys2
exists in animals. In addition, the human 4-phosphopantetheinyl transferase was capable of priming
prokaryotic PCP and ACP domains, indicating that
in humans a single enzyme with broad specificity is
responsible for all posttranslational priming reactions.132 These recent developments indicate that
NRPS activity has been preserved throughout evolution to higher eukaryotes.
3. Approaches to New Antibiotics

One of the current challenges in NRPS research is
to re-engineer natural products in order to increase
or alter their biological activities. Research of the past
few years has revealed that NRPS and NRPS-PKS
hybrids can produce biologically active compounds
and have outstanding potential for new drug discovery. A prominent example is the mixed NRPS-PKS
product epothilone which is a promising candidate
for combating cancer.133-135 One goal is to improve
natural-product-based drugs by rational protein engineering of these enzymes (e.g., by module and
domain swapping). Recently, a second strategy has
evolved which exploits a combination of chemistry
and enzymology to create libraries of novel compounds. Solid-phase peptide chemistry is a wellestablished method to produce linear peptides in good
yield. Cyclization of these molecules, however, poses
some challenges,38 which can often be solved by enzymatic means. The chemoenzymatic approach can also
contribute to a better understanding of NRPS natural
product assembly in general and of individual domains. Knowledge gained from such studies is essential for further biological or chemical engineering.
3.1. Genetic Engineering Approaches

Nature utilizes the modular assembly line methodology to produce a large set of small bioactive
peptides with a huge variety of building blocks and
modifications. The order and domain composition of
modules are the result of a careful selection during
evolution to synthesize a peptide molecule with the
best bioactivity. As a consequence, several hundred
NRPS building blocks are known which are incorporated by the same number of specific modules. Once
the logic and mechanisms of NRPS assembly had
been explored, interest developed in rationally redesigning the NRPS template to synthesize new peptide
products.
Redesign was initially attempted at the genetic
level, and several strategies have been examined to
alter product outcome, including exchange of A-PCP
units, artificial fusion of modules, module swaps, and
deletions. The first reported genetic re-engineering
experiment involved the terminal module of the
surfactin synthetase which incorporates leucine in
the natural system. This module exhibits the domain
composition C-A-PCP-TE. To alter the amino acid
specificity, the activating and covalent attachment
domains (A-PCP) were exchanged by A-PCP units
from bacterial and fungal origin with various amino
acid specificities. Novel surfactin variants with aliphatic (Val), charged (Orn), and aromatic (Phe)
residues at position 7 were created and confirmed by
mass spectrometry.136 All these new variants displayed the same hemolytic activity as the native
surfactin. However, low yields of the peptide products
(0.1-0.5% in comparison to the parent strain) were
observed probably because of the high selectivity of
C domains in the acceptor site for the cognate amino
acid substrate.80,81 Moreover, disruption of essential
amino acids at domain borders could cause additional
problems, and further attempts to obtain variants by
Sieber and Marahiel
Figure 11. NRPS engineering. (A) Schematic representation of a linker region (ca. 15 amino acids) between individual
domains. (B) Construction of a bimodular hybrid NRPS template derived from module 2 and module 10 (shown in red) of
the tyrocidine synthetases Tyc B and Tyc C. Modules were defined as C-A-PCP.
domain swapping of surfactin synthetase module 2

were unsuccessful.137 Domain borders generally seem
to be crucial determents of multienzyme activity.
Biochemical studies, sequence analysis, and structural information revealed linker regions between
NRPS domains50,74,82,138 which correspond to short
nonfunctional stretches of ca. 15 amino acids bearing
predominantly small and hydrophilic side chains
with an almost random distribution. The sequence
flexibility and their location between domains make
such linkers suitable targets for artificial fusions
without disrupting enzymatic integrity (Figure 11A).
This strategy was first tested for tyrocidine synthetase by fusion of Pro-activating module 2 with
Orn-activating module 9 or Leu-activating module
10138 (Figure 11B). These two hybrid enzymes were
then incubated with D-Phe-activating module 1 to
yield artificial D-Phe-Pro-Orn and D-Phe-Pro-Leu
tripeptides. Catalyzed tripeptide release was only
observed when a TE domain was present at the
C-terminal end of the last module. The catalytic
potential of TE domains to increase the rate of
product formation in engineered synthetases was
explored in more detail by fusion of six different TEs
to module 2 of the tyrocidine synthetase. All TEs
were active in hydrolyzing dipeptide products from
the enzymatic template, albeit with variable turnover
rate after incubation with module 1.139 Interestingly,
an artificial NRPS-PKS hybrid enzyme, which was
generated by fusion of the DEBS TE, also yielded
active protein, demonstrating that engineering within
the two enzyme families can be successful. The
promiscuity of TE domains at hydrolyzing diverse
products in artificial synthetases was further utilized
in a recent approach for production of the dipeptide
R-Asp-Phe, which is a precursor of the sweetener
Aspartame. Asp-Phe represents the first example in
which redesign of a NRPS was guided by a specific
application. Although many NRPS biosynthetic clusters have been sequenced, no biosynthetic template
is known where Asp- and Phe-activating modules are
organized one after the other. Consequently, six
hybrid NRPS enzymes were created in which A-PCP
or A-PCP-C units of Asp-activating module 5
derived from the surfactin synthetase were fused via
their linker regions to A-PCP, C-A, A-PCP, or A
units of Phe-activating modules 1 or 3 of the tyrocidine synthetase.140 The best hybrid enzyme composed, of A-PCP from surfactin module 5, C-A from
tyrocidine module 3, and PCP-TE from tyrocidine

module 10, displayed good chemoselectivity for the
desired product R-Asp-Phe. However, the turnover
rate of 0.7 min-1 limits the practicality of technical
applications and requires further optimization.
Many natural peptide products display small heterocyclic elements which are associated with their
bioactivity. Synthesis and incorporation of heterocyclic compounds such as thiazoline or oxazoline are
therefore desired for the synthesis of new pharmaceutical lead compounds. Recently, a genetic approach to these compounds was introduced by construction of new hybrid NRPSs using heterocyclization domains (Cy).88 In this study the Ile-activating A-PCP module 1 of bacitracin synthetase
was fused either to a Thr-activating Cy-A-PCP
module of mycobactin synthetase or to a Cys-activating Cy-A(Ox)-PCP module of myxothiazole synthetase.88 The latter module carries an oxidation
domain which further modifies the thiazoline product
to give a thiazole. To ensure product release, both
hybrids were again equipped with the tyrocidine TE
domain at the C-terminus. As predicted from the
selectivity of the corresponding A domains, the two
expected products are Ile-Thr-oxazoline and oxidized
Ile-Cys-thiazole. In case of the latter, Ile-Ser-oxazole
was the preferred product, demonstrating tolerance
for different amino acid substrates. The model studies
show that heterocyclization can be achieved in engineered synthetases, although low yields of the corresponding products are typically encountered.
In addition to module and domain swapping, insertion and deletion of modules within their defined
linker regions could contribute to alterations in the
product sequence. To this end, deletion of the Leuincorporating module 2 from surfactin synthetase
was engineered.141 Genetic manipulation and fermentation in the natural producer strain B. subtilis
afforded the predicted surfactin product deprived of
the second Leu residue in about 10% yield (in
comparison to the parent strain). This represents a
major improvement compared to initial engineering
studies on surfactin module 2 and 7 and points to
the importance of precise linker surgery.136,137 This
conclusion is further supported by results obtained
from an exchange of Leu-activating module 2 of
surfactin synthetase with Leu-activating module 10
of tyrocidine synthetase. Using the same linker
strategy as in the deletion studies, surfactin produc-
tion could be improved to a 19% yield (in comparison

to the parent strain).
The recombination of whole modules represents a
rather drastic intervention in NRPS biosynthesis
which usually results in reduced catalytic efficiency
and product yield. A more conservative strategy
involves manipulating the A domains specificity
through point mutations of substrate-coordinating
amino acid residues according to the nonribosomal
code (see A-domain section). For example, the substrate specificity of the Glu-activating module 1 of
surfactin synthetase was rationally altered in this
way. A single point mutation changed the specificity
from Glu to Gln without a decrease in catalytic
efficiency. A second specificity change in module 5
from Asp to Asn yielded the expected surfactin
derivative in vivo.57
The exploration and definition of linker regions and
C- and A-domain specificities has enabled substantial
progress in the field of genetic NRPS engineering.
However, the ultimate goal of mixing and matching
diverse modules to synthesize peptides of any desired
sequence has not yet been achieved. High-yield
production of bioactive peptides with unusual structural elements such as D-amino acids and heterocyclic
elements by fermentation would be an economic
alternative to expensive chemical synthesis.
3.2. Chemoenzymatic Approaches

One goal of modern drug design is to identify new
pharmacophores by rapid synthesis and bioactivity
screens. NRPS peptides are promising scaffolds for
such drug leads and have attracted much attention
for applications in medicine, as the example of
cyclosporin illustrates. Genetic manipulation is one
way to create potential new NRPS drug leads, but it
requires much labor and effort to generate the
desired peptide product. To ensure rapid synthesis
of large peptide libraries, chemical solid-phase synthesis seems to be superior. However, a limitation of
this technique is the weak tendency of linear peptides
to macrocyclize by chemical means due to the high
entropic cost of populating the conformation with the
right geometry cyclization. As a consequence, chemical synthesis often suffers from low cyclization
yields.35,142,143 A recent chemoenzymatic approach
combines the strength of synthetic peptide synthesis
with the strength of regio- and stereoselective TEdomain cyclization in high yields.144,145 Once a new
lead drug has been identified by this rapid synthetic
method, genetic engineering may provide the enzymatic synthesis template to allow high-yield fermentation of the desired drug.
3.2.1. Chemoenzymatic Potential of TE Domains

To evaluate the chemoenzymatic potential of TE
domains, characterization of substrate specificity,
tolerance, and enzymatic restrictions have to be
performed. These specific TE-domain investigations
cannot be performed with the whole multienzyme
complex because its large size causes preparative
problems. Instead, an easy in vitro assay system was
devised by cloning TE domains from the tyrocidine
and surfactin synthetases and producing them as
isolated enzymes. To test weather these excised TEs

are active outside their natural synthetase context,
their cognate decapeptide and heptapeptide substrates were synthesized by chemical means. In this
approach the complete linear enzymatic peptide
synthesis machinery, composed of a repeating set of
catalytic domains, was replaced by solid-phase peptide synthesis (Figure 12A). Both methods share a
similar strategy of precursor activation and tethering
on a solid support, which facilitates rapid and ordered
synthesis of desired products in high yields. One
advantage of chemical synthesis is the huge diversity
which can be incorporated into the linear peptide
chain in order to create a variety of substrate
analogues for biochemical studies of excised TE
domains. In the natural synthetase peptide substrates are activated as ppan thioesters which ensure
acylation of the TE domain active site. To provide
similar recognition and activation for the artificial
system, the thiol component of the natural ppan
cofactor N-acetylcysteamine (SNAC) was attached to
the C-terminus of a synthetic peptide (Figure 12B).
Incubation of artificial tyrocidine and surfactin SNAC
substrates with the cognate TE domains resulted in
cyclization and hydrolysis.110,115,146 The cyclic product
was formed in as high as 85% yield, and the turnover
rate for tyrocidine TE was 59 min-1, sufficient for
more specific investigations.
A set of experiments was designed to evaluate the
general utility of TE domains for catalyzing diverse
cyclization reactions. With regard to general utility
as a cyclization tool, broad substrate tolerance would
be desirable. Enzymatic recognition elements of Srf
TE and Tyc TE were first investigated by systematic
alteration of each amino acid in the hepta- and
decapeptide sequences. In the case of tyrocidine, each
amino acid in the substrate was replaced by alanine
and the resulting 10 SNAC variants were incubated
with the TE. This amino acid scan revealed a broad
substrate tolerance. Only amino acids at the C- and
N-termini seem to be recognized, leaving space for
alterations in the middle of the peptide (Figure 13).
Identity and stereochemistry of the N-terminal D-Phe
is essential for enzyme activity, although an exchange of the free amine by a hydroxy group was
tolerated and led to macrolactonization. Alterations
in peptide length gave cyclic hexa- and dodecapeptides, and cyclodimerization of two pentapeptides was
also tolerated.115 Tyc TE thus seems to be a very
promiscuous cyclization catalyst, tolerant to changes
which are desired in the generation of new peptide
products. Alterations of the tyrocidine peptide backbone yielded a minimal recognition model for explaining cyclization activity.147 Similar investigations
on the substrate tolerance of Srf TE also showed the
importance of the N- and C-terminal residues in the
peptide110 (Figure 13). Cocrystallization with a boronic acid inhibitor confirmed the biochemical data,
providing evidence for two hydrophobic binding pockets which can accommodate the two C-terminal Leu
residues of surfactin. In contrast to Tyc TE, however,
alterations in the nucleophile and substrate length
were not tolerated. The -sheet content of the two
peptides may provide an explanation for the observed
Sieber and Marahiel
Figure 12. Chemoenzymatic cyclization. (A) The NRPS multienzyme machinery required for peptide elongation is replaced
by solid-phase peptide synthesis, and the TE domain is used as an isolated enzyme. (B) Recognition of the artificial substrate
by the enzyme is ensured by the ppan cofactor mimic SNAC (highlighted).
Figure 13. Substrate tolerance of tyrocidine and surfactin TE domains. Residues marked in gray (tyrocidine 2-8 and
surfactin 2-5) can be substituted by alanine or diaminopropionic acid, respectively. Those shown in red are essential for
substrate recognition by the TEs.
differences. Tyrocidine has a high -sheet content,

and cyclization is promoted by preorganization of the
peptide backbone. Cyclization of tyrocidine can there-
fore also occur without enzymatic catalysis but with

lower efficiency.148 The reduced number of -sheets
in surfactin as well as the branched chain cyclization
mode, which requires larger fatty acids for improved

activity,70 may account for the observed differences
in substrate tolerance. The broader specificity of Tyc
TE makes it a promising candidate for biosynthetic
applications.
polyketides with therapeutic potency,151 while the

second method allows the peptide backbone to be
modified postsynthetically by chemical metathesis.152
Natural hybrids such as epothilone and bleomycin
already exhibit cytostatic activity which could be
further optimized via the chemoenzymatic route.
3.2.2. Chemoenzymatic Route to New Drugs and Peptide

Antibiotics
3.2.3. Expanding the TE Tool Box
Many cyclic peptide antibiotics act on bacterial cells

by insertion into the membrane, followed by disruption of osmotic and ionic regulation. Studies with
small synthetic peptides revealed that alternating Dand L-amino acids as well as the presence of positively charged side chains contribute much to the
overall activity against negatively charged prokaryotic cell membranes.149 Many nonribosomal peptides
such as tyrocidine exhibit low selectivity for prokaryotic vs eukaryotic cells, which limits their application
as antibiotic drugs. To improve the preference for
bacterial targets as well as the spectrum of activity
against common bacterial pathogens, Walsh and coworkers introduced a solid-phase combinatorial synthesis of novel tyrocidine analogues.150 A library of
peptides with natural and nonnatural amino acid
substituents introduced at two positions was constructed on a solid support (PEGA resin) by parallel
synthesis. Enzymatic on-resin cyclization and subsequent analysis of the antibiotic activity of the
reaction products against B. subtilis revealed potency
for those peptides carrying a positively charged
D-amino acid at the D-Phe4 position with 30-fold
selectivity for bacterial membranes. Two of the best
analogues also gained activity against Gram-negative
organisms (Figure 14A). The improved tyrocidine
variants which have been identified via this combinatorial chemoenzymatic approach can now be translated back into a modular engineered NRPS template
for large-scale production by fermentation.
The chemoenzymatic potential of Tyc TE was
subsequently used to generate small molecules with
different therapeutic potential: peptide inhibitors of
integrin receptors and hybrid peptide/polyketides
(Figure 14B and C). Many natural ligands for integrins contain an Arg-Gly-Asp (RGD) sequence motif
that is believed to be important for receptor interaction. Peptides containing the RGD motif are potent
inhibitors and therefore potential therapeutic leads.
Because interaction with the receptor is improved by
incorporation of -turns, a large number of changes
in Tyc TE was required to produce the desired
inhibitor substrate. In the end, only the Tyc TE
minimal recognition elements were retained. Although 7 of 10 cognate residues were replaced in
some cases with amino acids of opposite stereochemistry, Tyc TE was still capable of cyclizing the
substrate, albeit with reduced yield.114 Products were
shown to be inhibitors of ligand binding by integrin
receptors with cyclization as an important contributor to nanomolar potency. The broad substrate tolerance of Tyc TE was further utilized to mediate
cyclization of hybrid polyketide-tyrocidine substrates
as well as (E)-alkene-dipeptide isostere peptidomometics. The first approach contributes significantly to the synthesis of novel hybrid peptide/
In previous sections the utility of chemoenzymology

was illustrated predominantly through studies with
tyrocidine thioesterase. To expand the scope of macrocyclization catalysts, thioesterases from other NRPS
enzymes were cloned and overexpressed. Surprisingly, no activity was observed for fengycin, mycosubtilin, and syringomycin TEs upon incubation with
SNAC substrates, indicating limitations in the
chemoenzymatic potential of these cyclases.153 Two
reasons might account for the observed inactivity of
these TE domains. First, the enzymes could have
been misfolded and were unable to reconstitute their
activity after heterologous expression. Alternatively,
substrate presentation by the short SNAC leaving
group might have been insufficient. To rule out the
second possibility, a new strategy was employed
which allowed Sfp-catalyzed loading of peptidylCoAs onto apo PCPs mimicking the natural substrate
presentation as close as possible (Figure 14A). Remarkably, Sfp was promiscuous enough to tolerate
peptidyl-CoA substrates instead of CoA and acetylCoA. This observation is explained by the crystal
structure, which shows specific interactions with the
adenine base but the ppan arm pointing into solution
(Figure 5B). This binding mode ensures enough space
and freedom for the attached peptide chain. Loading
fengycin-CoA onto the fengycin PCP-TE didomain
indeed gave rise to both cyclization and hydrolysis
activity, which had not been previously observed with
SNACs. This result demonstrated that the PCPtethered ppan arm is necessary to direct the substrate into the enzyme active site and guarantees
appropriate alignment for nucleophilic attack of the
active site serine.153 This approach additionally illustrates the high regioselectivity of fengycin TE.
Recently, other C and R domains have been examined with respect to the possibility of bypassing
NRPS specificity by directly loading peptidyl-CoA
on any apo-PCP within an assembly line.71,96 However, the single turnover nature of the reaction has
proved to be a limitation (Figure 14A). After product
release, the cofactor ppan remains attached to the
PCP-TE didomain cyclase, which blocks further Sfpcatalyzed transfer of additional peptidyl-CoAs onto
ppan-PCP.
To force multiple rather than single turnover cycles
a new strategy was developed to expand the utility
of peptidyl-CoA loading. It is known that amino acid
thioesters undergo trans-thioesterification reactions
when exposed to thiol-containing compounds.154 In
principle, such a thioester exchange reaction between
the free ppan-PCP thiol and a soluble thioesterpeptide substrate could enable both chemical reloading of substrate onto the ppan-PCP-TE didomain154,155 and natural substrate presentation by the
ppan arm. Inspired by expressed protein ligation,155
Sieber and Marahiel
Figure 14. New strategies of substrate presentation. (A) PCP-TE ensures natural substrate interaction. Synthetic
peptidyl-CoA (e.g., fengycin) is loaded onto apo PCP-TE by Sfp to give peptidyl-S-ppan-PCP-TE. Subsequently, the
peptide substrate is transferred onto an invariant serine residue of the TE active site, which is then released by cyclization
(via tyrosine3 in case of fengycin). (B) Activity-based enzyme acyclation. The active site serine of the TE domain is directly
acylated by a reactive peptidyl-thiophenol substrate. The acyl-enzyme intermediate is then captured by an intramolecular
nucleophilic attack to yield the cyclic product (e.g., fengycin).
the peptide was activated as a thioester with thiophenol, which has favorable leaving-group properties. In
the course of these studies it became obvious that
soluble fengycin thiophenol directly acylated the TE
active site serine rather than the free ppan thiol
(Figure 14B).116 The rapid direct acylation of the
active site serine confirmed the autonomous activity
of the excised enzyme. Moreover, this result showed
that contrary to previous belief, natural cofactor
recognition elements as displayed in SNAC or CoA
substrates are not necessary for enzyme acylation but
can be replaced by a suitable reactive leaving group.
The results of these experiments further suggested
that nature might have developed peptide cyclases
with different catalytic activities. While tyrocidine TE
and surfactin TE show activity directly with SNAC
substrates, mycosubtilin, fengycin, and syringomycin
TEs appear to be completely inactive. A 15-fold
increase in catalytic activity was observed for Srf TE
when the SNAC leaving group was replaced by
thiophenol. Recently, a similar increase in activity
was reported for CDA-thiophenol compared to CDASNAC for CDA cyclase.70 The thiophenol leaving
group increases the velocity of the acylation step.
While the acylation step depends on substrate presentation, as seen in the peptidyl-CoA experiments,
deacylation is an intrinsic property of the acylenzyme intermediate, as shown by comparable cyclization-to-hydrolysis ratios irrespective of whether
SNAC or thiophenol substrates are used. Undesired
hydrolytic byproducts were observed in thiophenolbased in vitro studies, presumably due to spontaneous cleavage of the highly activated substrate in
solution. The ratio of cyclization to hydrolysis was
most favorable when cognate thiophenol substrates
that fit precisely into the enzyme active site were
used. Not much is known about the hydrolysis rate
of natural NRPS templates, but it is likely that the
multienzyme complexes also produce hydrolyzed
byproducts to a certain extent. In general, however,
selective enzyme acylation can be achieved for dedicated peptide cyclases with substrates activated with
a variety of leaving groups.
3.2.4. Synthetic Utility of TEs: Chemical vs Enzymatic

Cyclization
To evaluate the utility and potential of enzymecatalyzed cyclization reactions, a comparison with
established chemical methods is necessary. In principle, the chemical formation of macrocyclic rings is
difficult because of energetically disfavored ecliptic
and transannular interactions.61 In solution, only a
few conformers have the right geometry to allow
intramolecular attack of a nucleophilic group on the
C-terminal carboxy group. The entropic costs of
populating these few productive conformations by
several C-C bond rotations are high and therefore
disfavored. To ensure regioselective cyclization, undesired competing nucleophiles such as hydroxy or
amino groups have to remain protected while the
desired nucleophile needs to be deprotected, which
requires orthogonal protecting-group strategies. Side
reactions, including intermolecular peptide bond
formation and subsequent cyclo-oligomerization, may
predominate since peptide bonds are usually transconfigured and favor a higher population of linear
precursors. To minimize intermolecular reactions,
high dilution conditions are applied (10-4-10-5 M)
which make large-scale reactions difficult. Alternatively, a peptide can be cyclized while it is still
attached to the resin. Because the peptide chains are
physically separated, intramolecular reactions are
favored.156 Turn-inducing elements such as D-amino
acids, proline, glycine, or N-alkyl amino acids can also
favor cyclization in solution, illustrating the importance of preorganization of the linear precursor for
efficient ring closure, something that has also been
observed for some NRPS products.147 Another problem of chemical cyclization is the activation of the
C-terminal carboxy group without amino acid racemization. Coupling reagents such as BOP and TBTU
permit rapid cyclization but suffer from C-terminal
racemization.156 Less reactive coupling reagents minimize racemization but prolong reaction times. Better
results were achieved with HOAt and DPPA.142,157-159
Typically, with the above-mentioned chemical cyclization methods a 30-40% yield of cyclic peptide
products can be obtained.35,142,143,157 Reaction times
range from several hours to days.
By contrast, enzymatically catalyzed cyclization
reactions do not require protecting groups or high
dilution conditions due to enzymatic specificity. In
the literature, enzymatic methods for head-to-tail
peptide cyclization have been predominantly reported. Cyclization of linear peptide esters was first
described for the subtilisin mutant subtiligase.160
Subtiligase cyclizes peptide esters longer than 12
residues with yields of 30-88% in a regioselective
head-to-tail fashion. Hydrolysis and dimerization are
observed byproducts. Head-to-tail cyclization without byproducts was reported for an intramolecular
cyclization using split-inteins, allowing the generation of backbone-cyclized peptides in vitro and in
vivo.161
Several NRPS enzymes, including surfactin, mycosubtilin, fengycin, and syringomycin TE, were
shown to regiospecifically catalyze branched chain
cyclization between one dedicated nucleophile and
the activated C-terminal peptide residue in the
presence of other potential competing nucleophiles.
No oligomerization or C-terminal racemization was
observed. The advantage of enzymatic vs chemical
cyclization is illustrated by the example of tyrocidine
A synthesis. While chemical, on-resin cyclizations
typically occur in only 30% yield, enzymatic cyclization gives 85% product.146,162 Because ring formation
competes with the production of linear hydrolytic
byproducts due to competing nucleophilic attack of
water molecules, typical yields of TE-mediated cyclization reactions range from 40% to 91%; observed
reaction times are several minutes to hours.
Since these recombinant peptide cyclases are usually embedded in a hydrophobic multienzyme complex, their production as isolated TE or PCP-TE
domains may enhance the exposure of the active site
to water. Moreover, substrate analogues used in in
vitro studies often lack structural features, like longchain fatty acyl chains, which are important for a
perfect fit into the enzyme binding pocket as seen for

CDA but are difficult to incorporate synthetically.70
The latter point is illustrated by in vivo studies with
a genetically engineered surfactin synthetase lacking
module 2.141 While it was not possible to detect
hydrolyzed surfactin product in the supernatant of
the wild-type producer cells, hydrolysis was observed
for the shorter surfactin variant produced by the
engineered strain. This result indicates that hydrolysis can also occur in vivo, where the TE domain is
embedded in a functional multienzyme complex.
Therefore, presentation of the dedicated substrate
seems to be a key step to minimize hydrolysis by TE
domains. Moreover, artificial thioester leaving groups
such as SNAC and thiophenol are short mimics of
the natural cofactor ppan-PCP and may therefore
be less effective of blocking water from the active site
and hence more susceptible to hydrolysis. The nucleophilic side reaction with water could be minimized by enzyme catalysis in organic solvents.163 The
high enzymatic selectivity for cognate substrates can
also be a disadvantage when the cyclization of
substrate analogues is desired. Substitutions of residues, especially at the C- and N-terminal ends of the
peptide sequence, can decrease or completely abolish
cyclization yields. Nevertheless, with the current set
of active peptide cyclases, diverse cyclization reactions such as head-to-tail and branched chain lactamization and lactonization of various substrates
can already be performed.
In comparison to organic synthesis, the chemoenzymatic approach minimizes time and side reactions
and maximizes purity and yield of cyclic peptide
production. The current chemoenzymatic applications
of TEs are summarized in Figure 15. Biochemical
prerequisites such as enzyme characterization and
leaving-group technology have been solved which will
Figure 15. TE domains as versatile catalysts with a

potential for the synthesis of new bioactive compounds by
a variety of strategies. With recent progress in charging
TE domains with activated substrates, a large catalytic
toolbox of enzymes awaits chemoenzymatic application for
the generation of novel antibiotics. Substrates can be
presented to the TE either bound to a PCP or an artificial
solid support or by soluble thioester leaving groups.
Sieber and Marahiel
now allow for applications such as formation of

various cyclic peptide libraries.150 In the future,
further exploration of structurally diverse peptide
cyclases will help to generate a tool kit of cyclization
catalysts which in combination with protein evolution
are likely to broaden the applications and increase
the utility of enzymatic peptide cyclization. These
proteins may rival the utility of lipases as catalysts
for stereoselective transformations of diverse substrate molecules.
4. Conclusions
NRPSs are highly sophisticated natural nanomachines that were optimized for the biosynthesis of
compounds that cannot be produced by the ribosomal
machinery and were selected during evolution for
diverse structures and for broad biological activities.
Recently, a wealth of information about the threedimensional structure of several NRPS core domains
in combination with detailed biochemical, chemoenzymatic, and genetic studies has not only facilitated
the construction of hybrid NRPS but also accelerated
the speed by which such bioactive cyclic peptides can
be produced. However, some NRPS global structural
aspects remain still elusive. One current challenge
is the crystallization of modules comprising a multidomain structure that can provide information
about domain interaction and the overall architecture
within these building blocks of such megaenzymes.
Moreover, such structural information could contribute to a precise definition of interdomain linker
regions and possible protein-protein interaction sites
between the catalytic domains during the concerted
action of this assembly line mechanism. This knowledge will have a direct influence on the success of
rational engineering attempts, which at present
suffer from the lack of this information. In contrast
to the well-studied essential domains (A, C, PCP, TE),
very little is known about the mechanisms of chemical reactions catalyzed by tailoring domains such as
peptide heterocyclization, N-methylation, oxidation,
reduction, formylation, epimerization, etc. The enzymatic domains carrying out these reactions act
within the NRPS assembly line in high precision and
efficiency, a fact that makes them attractive for
synthetic applications. Chemoenzymatic cyclization
by TE domains has already proven this notion and
is now established for a set of excised TE domains.
Future research will show if this new single-domain
catalysis is suitable and potent enough to identify
novel drug leads by large cyclic library screens. The
utility certainly depends on enzymatic substrate
tolerance, turnover, and product yield. In many cases
this will need to be optimized by directed protein
evolution efforts. Enzyme engineering will further
show if other NRPS core and tailoring domains will
exhibit the same tolerance in vitro for their desired
chemical reactions
5. Acknowledgments
We thank all the members of our group for helpful
discussions and support. This work was supported
by the Studienstiftung des deutschen Volkes (SAS)

and Deutsche Forschungsgemeinschaft (MAM).
6. References
(1) Flemming, A. Br. J. Exp. Path. 1929, 10, 226.
(2) Lipmann, F.; Gevers, W.; Kleinkauf, H.; Roskoski, R., Jr. Adv.
Enzymol. Relat. Areas Mol. Biol. 1971, 35, 1.
Press: Washington, DC, 2003.
(4) Schwarzer, D.; Finking, R.; Marahiel, M. A. Nat. Prod. Rep. 2003,
20, 275.
(5) Mootz, H. D.; Schwarzer, D.; Marahiel, M. A. Chembiochem 2002,
3, 490.
(6) Marahiel, M. A.; Stachelhaus, T.; Mootz, H. D. Chem. Rev. 1997,
97, 2651.
(7) Zasloff, M. Nature 2002, 415, 389.
(8) Mootz, H. D.; Marahiel, M. A. J. Bacteriol. 1997, 179, 6843.
(9) Simlot, M. M.; Pfaender, P.; Specht, D. Hoppe-Seylers Z. Physiol.
Chem. 1972, 353, 759.
(10) Peypoux, F.; Bonmatin, J. M.; Wallach, J. Appl. Microbiol.
Biotechnol. 1999, 51, 553.
(11) de Crecy-Lagard, V.; Blanc, V.; Gil, P.; Naudin, L.; Lorenzon,
S.; Famechon, A.; Bamas-Jacques, N.; Crouzet, J.; Thibaut, D.
J. Bacteriol. 1997, 179, 705.
(12) Hubbard, B. K.; Walsh, C. T. Angew. Chem., Int. Ed. Engl. 2003,
42, 730.
(13) Steller, S.; Vollenbroich, D.; Leenders, F.; Stein, T.; Conrad, B.;
Hofemeister, J.; Jacques, P.; Thonart, P.; Vater, J. Chem. Biol.
1999, 6, 31.
(14) Steller, S.; Vater, J. J. Chromatogr., B: Biomed. Sci. Appl. 2000,
737, 267.
(15) Vanittanakom, N.; Loeffler, W.; Koch, U.; Jung, G. J. Antibiot.
(Tokyo) 1986, 39, 888.
(16) Rieber, M. T.; Imaeda, T.; Cesari, I. M. J. Gen. Microbiol. 1969,
55, 155.
(17) Konz, D.; Klens, A.; Schorgendorfer, K.; Marahiel, M. A. Chem.
Biol. 1997, 4, 927.
(18) Hori, K.; Yamamoto, Y.; Minetoki, T.; Kurotsu, T.; Kanda, M.;
Miura, S.; Okamura, K.; Furuyama, J.; Saito, Y. J. Biochem.
(Tokyo) 1989, 106, 639.
(19) Katsu, T.; Kobayashi, H.; Fujita, Y. Biochim. Biophys. Acta 1986,
860, 608.
(20) Stone, K. J.; Strominger, J. L. Proc. Natl. Acad. Sci. U.S.A. 1971,
68, 3223.
(21) Storm, D. R.; Strominger, J. L. J. Biol. Chem. 1973, 248, 3940.
(22) Maget-Dana, R.; Thimon, L.; Peypoux, F.; Ptak, M. Biochimie
1992, 74, 1047.
(23) Duitman, E. H.; Hamoen, L. W.; Rembold, M.; Venema, G.; Seitz,
H.; Saenger, W.; Bernhard, F.; Reinhardt, R.; Schmidt, M.;
Ullrich, C.; Stein, T.; Leenders, F.; Vater, J. Proc. Natl. Acad.
Sci. U.S.A. 1999, 96, 1329.
(24) Maget-Dana, R.; Peypoux, F. Toxicology 1994, 87, 151.
(25) Hutchison, M. L.; Gross, D. C. Mol. Plant Microbe Interact. 1997,
10, 347.
(26) Scholz-Schroeder, B. K.; Hutchison, M. L.; Grgurina, I.; Gross,
D. C. MPMI 2001, 14, 336.
(27) Hojati, Z.; Milne, C.; Harvey, B.; Gordon, L.; Borg, M.; Flett, F.;
Wilkinson, B.; Sidebottom, P. J.; Rudd, B. A. M.; Hayes, M. A.;
Smith, C. P.; Micklefield, J. Chem. Biol. 2002, 9, 1175.
(28) Alborn, W. E., Jr.; Allen, N. E.; Preston, D. A. Antimicrob. Agents
Chemother. 1991, 35, 2282.
(29) McHenney, M. A.; Hosted, T. J.; Dehoff, B. S.; Rosteck, P. R.,
Jr.; Baltz, R. H. J. Bacteriol. 1998, 180, 143.
(30) Canepari, P.; Boaretti, M. Microb. Drug Resist. 1996, 2, 85.
(31) Di Giambattista, M.; Engelborghs, Y.; Nyssen, E.; Clays, K.;
Cocito, C. Biochemistry 1991, 30, 7277.
(32) Weber, G.; Schorgendorfer, K.; Schneider-Scherzer, E.; Leitner,
E. Curr. Genet. 1994, 26, 120.
(33) Belshaw, P. J.; Schreiber, S. L. J. Am. Chem. Soc. 1997, 119,
1805.
(34) Liu, J.; Farmer, J. D., Jr.; Lane, W. S.; Friedman, J.; Weissman,
I.; Schreiber, S. L. Cell 1991, 66, 807.
(35) Thern, B.; Rudolph, J.; Jung, G. Angew. Chem., Int. Ed. 2002,
41, 2307.
(36) Quadri, L. E. Mol. Microbiol. 2000, 37, 1.
(37) Crosa, J. H.; Walsh, C. T. Microbiol. Mol. Biol. Rev. 2002, 66,
223.
(38) Sewald, N. Angew. Chem., Int. Ed. 2002, 41, 4661.
(39) Nicolaou, K. C.; Boddy, C. N. C.; Brase, S.; Winssinger, N. Angew.
Chem., Int. Ed. Engl. 1999, 38, 2096.
(40) Lipmann, F. Adv. Microb. Physiol. 1980, 21, 227.
(41) von Dohren, H.; Keller, U.; Vater, J.; Zocher, R. Chem. Rev. 1997,
97, 2675.
(42) Cosmina, P.; Rodriguez, F.; de Ferra, F.; Grandi, G.; Perego, M.;
Venema, G.; van Sinderen, D. Mol. Microbiol. 1993, 8, 821.

(43) Schwarzer, D.; Marahiel, M. A. Naturwissenschaften 2001, 88,
93.
(44) Cane, D. E.; Walsh, C. T. Chem Biol 1999, 6, R319.
(45) Stryer, L. Biochemistry, 4th ed.; Spektrum Akademischer Verlag
GmbH: Heidelberg-Berlin-Oxford, 1996.
(46) Linne, U.; Marahiel, M. A. Methods Enzymol. 2004, in press.
(47) Turgay, K.; Krause, M.; Marahiel, M. A. Mol. Microbiol. 1992,
6, 529.
(48) Conti, E.; Franks, N. P.; Brick, P. Structure 1996, 4, 287.
(49) Gulick, A. M.; Starai, V. J.; Horswill, A. R.; Homick, K. M.;
Escalante-Smemerena, J. C. Biochemistry 2003, 42, 2866.
(50) Conti, E.; Stachelhaus, T.; Marahiel, M. A.; Brick, P. EMBO J.
1997, 16, 4174.
(51) May, J.; Kessler, N.; Marahiel, M. A.; Stubbs, M. T. PNAS 2002,
99, 12120.
(52) Stachelhaus, T.; Mootz, H. D.; Marahiel, M. A. Chem. Biol. 1999,
6, 493.
(53) Gocht, M.; Marahiel, M. A. J. Bacteriol. 1994, 176, 2654.
(54) Saito, M.; Hori, K.; Kurotsu, T.; Kanda, M.; Saito, Y. J. Biochem.
(Tokyo) 1995, 117, 276.
(55) Starai, V. J.; Celic, I.; Cole, R. N.; Boeke, J. D.; EscalanteSemerena, J. C. Science 2002, 298, 2390.
(56) Du, L.; Sanchez, C.; Chen, M.; Edwards, D. J.; Shen, B. Chem.
Biol. 2000, 7, 623.
(57) Eppelmann, K.; Stachelhaus, T.; Marahiel, M. A. Biochemistry
2002, 41, 9718.
(58) Challis, G. L.; Ravel, J.; Townsend, C. A. Chem. Biol. 2000, 7,
211.
(59) Pavela-Vrancic, M.; Dieckmann, R.; von Dohren, H. Biochim.
Biophys. Acta 2004, 1696, 83.
(60) Ruttenberg, M. A.; Mach, B. Biochemistry 1966, 5, 2864.
(61) Bruckner, R. Reaktionsmechanismen; Spektrum Verlag: Heidelberg, 1996.
(62) Stein, T.; Vater, J.; Kruft, V.; Wittmann-Liebold, B.; Franke, P.;
Panico, M.; McDowell, R.; Morris, H. R. FEBS Lett. 1994, 340,
39.
(63) Stein, T.; Vater, J.; Kruft, V.; Otto, A.; Wittmann-Liebold, B.;
Franke, P.; Panico, M.; McDowell, R.; Morris, H. R. J. Biol.
Chem. 1996, 271, 15428.
(64) Stachelhaus, T.; Huser, A.; Marahiel, M. A. Chem. Biol. 1996,
3, 913.
(65) Schlumbohm, W.; Stein, T.; Ullrich, C.; Vater, J.; Krause, M.;
Marahiel, M. A.; Kruft, V.; Wittmann-Liebold, B. J. Biol. Chem.
1991, 266, 23135.
(66) Lambalot, R. H.; Gehring, A. M.; Flugel, R. S.; Zuber, P.; LaCelle,
M.; Marahiel, M. A.; Reid, R.; Khosla, C.; Walsh, C. T. Chem.
Biol. 1996, 3, 923.
(67) Walsh, C. T.; Gehring, A. M.; Weinreb, P. H.; Luis, E. N.; Flugel,
R. S. Curr. Opin. Chem. Biol. 1997, 1, 309.
(68) Quadri, L. E.; Weinreb, P. H.; Lei, M.; Nakano, M. M.; Zuber,
P.; Walsh, C. T. Biochemistry 1998, 37, 1585.
(69) Kealey, J. T.; Liu, L.; Santi, D. V.; Betlach, M. C.; Barr, P. J.
Proc. Natl. Acad. Sci. 1998, 95, 505.
(70) Grunewald, J.; Sieber, S. A.; Marahiel, M. A. Biochemistry 2004,
43, 2915.
(71) Clugston, S. L.; Sieber, S. A.; Marahiel, M. A.; Walsh, C. T.
(72) Reuter, K.; Mofid, M. R.; Marahiel, M. A.; Ficner, R. EMBO J.
1999, 18, 6823.
(73) Mofid, R. M.; Finking, R.; Essen, L. O.; Marahiel, M. A.
(74) Weber, T.; Baumgartner, R.; Renner, C.; Marahiel, M. A.; Holak,
T. A. Struct. Fold Des. 2000, 8, 407.
(75) Holak, T. A.; Kearsley, S. K.; Kim, Y.; Prestegard, J. H.
(76) Crump, M. P.; Crosby, J.; Dempsey, C. E.; Parkinson, J. A.;
Murray, M.; Hopwood, D. A.; Simpson, T. J. Biochemistry 1997,
36, 6000.
(77) Parris, K. D.; Lin, L.; Tam, A.; Mathew, R.; Hixon, J.; Stahl, M.;
Fritz, C. C.; Seehra, J.; Somers, W. S. Structure 2000, 8, 883.
(78) Mofid, M. R.; Finking, R.; Marahiel, M. A. J. Biol. Chem. 2002,
277, 50293.
(79) Schwarzer, D.; Mootz, H. D.; Linne, U.; Marahiel, M. A. Proc.
Natl. Acad. Sci. U.S.A. 2002, 99, 14083.
(80) Belshaw, P. J.; Walsh, C. T.; Stachelhaus, T. Science 1999, 284,
486.
(81) Ehmann, D. E.; Trauger, J. W.; Stachelhaus, T.; Walsh, C. T.
Chem. Biol. 2000, 7, 765.
(82) Keating, T. A.; Marshall, C. G.; Walsh, C. T.; Keating, A. E. Nat.
Struct. Biol. 2002, 9, 522.
(83) De Crecy-Lagard, V.; Marliere, P.; Saurin, W. C. R. Acad. Sci.
III 1995, 318, 927.
(84) Bergendahl, V.; Linne, U.; Marahiel, M. A. Eur. J. Biochem.
2002, 269, 620.
(85) Roche, E. D.; Walsh, C. T. Biochemistry 2002, 42, 1334.
(86) Marshall, C. G.; Hillson. N. J.; Walsh, C. T. Biochemistry 2002,
41, 244.
C. T. Nat. Prod. Rep. 1999, 16, 249.

(88) Duerfahrt, T.; Eppelmann, K.; Muller, R.; Marahiel, M. A. Chem.
Biol. 2004, 11, 261.
(89) Pfeifer, E.; Pavela-Vrancic, M.; von Dohren, H.; Kleinkauf, H.
(90) Stachelhaus, T.; Walsh, C. T. Biochemistry 2000, 39, 5775.
(91) Luo, L.; Walsh, C. T. Biochemistry 2001, 40, 5329.
(92) Hoffmann, K.; Schneider-Scherzer, E.; Kleinkauf, H.; Zocher, R.
J. Biol. Chem. 1994, 269, 12710.
(93) Linne, U.; Marahiel, M. A. Biochemistry 2000, 39, 10439.
(94) Linne, U.; Doekel, S.; Marahiel, M. A. Biochemistry 2001, 40,
15824.
(95) Linne, U.; Stein, D. B.; Mootz, H. D.; Marahiel, M. A. Biochemistry 2003, 42, 5114.
(96) Kessler, N.; Schuhmann, H.; Morneweg, S.; Linne, U.; Marahiel,
M. A. J. Biol. Chem. 2004, 279, 7413.
(97) Velkov, T.; Lawen, A. Biotechnol. Annu. Rev. 2003, 9, 151.
(98) Billich, A.; Zocher, R.; Kleinkauf, H.; Braun, D. G.; Lavanchy,
D.; Hochkeppel, H. K. Biol. Chem. Hoppe Seyler 1987, 368, 521.
(99) Schauwecker, F.; Pfennig, F.; Grammel, N.; Keller, U. Chem.
Biol. 2000, 7, 287.
(100) de Crecy-Lagard, V.; Saurin, W.; Thibaut, D.; Gil, P.; Naudin,
L.; Crouzet, J.; Blanc, V. Antimicrob. Agents Chemother. 1997,
41, 1904.
(101) Miller, D. A.; Walsh, C. T.; Luo, L. J. Am. Chem. Soc. 2001, 123,
8434.
(102) Weinig, S.; Hecht, H. J.; Mahmud, T.; Muller, R. Chem. Biol.
2003, 10, 939.
(103) Du, L.; Chen, M.; Sanchez, C.; Shen, B. FEMS Microbiol. Lett.
2000, 189, 171.
(104) Schneider, T. L.; Shen, B.; Walsh, C. T. Biochemistry 2003, 42,
9722.
(105) Silakowski, B.; Schairer, H. U.; Ehret, H.; Kunze, B.; Weinig,
S.; Nordsiek, G.; Brandt, P.; Blocker, H.; Hofle, G.; Beyer, S.;
Muller, R. J. Biol. Chem. 1999, 274, 37391.
(106) Shen, B.; Du, L.; Sabchez, C.; Edwards, D. J.; Chen, M.; Murell,
J. M. J. Nat. Prod. 2002, 65, 1262.
(107) Weinig, S.; Mahmud, T.; Muller, R. Chem. Biol. 2003, 10, 953.
(108) Reimmann, C.; Patel, H. M.; Serino, L.; Barone, M.; Walsh, C.
T.; Patel, H. M. J. Bacteriol. 2001, 813.
(109) Rouhiainen, L.; Paulin, L.; Suomalainen, S.; Hyytiainen, H.;
Buikema, W.; Haselkorn, R.; Sivonen, K. Mol. Microbiol. 2000,
37, 156.
(110) Tseng, C. C.; Bruner, S. D.; Kohli, R. M.; Marahiel, M. A.; Walsh,
C. T.; Sieber, S. A. Biochemistry 2002, 41, 13350.
(111) Ehmann, D. E.; Gehring, A. M.; Walsh, C. T. Biochemistry 1999,
38, 6171.
(112) Bruner, S. D.; Weber, T.; Kohli, R. M.; Schwarzer, D.; Marahiel,
M. A.; Walsh, C. T.; Stubbs, M. T. Structure (Cambridge) 2002,
10, 301.
Natl. Acad. Sci. U.S.A. 2001, 98, 14808.
(114) Kohli, R. M.; Takagi, J.; Walsh, C. T. PNAS 2002, 99, 1247.
(115) Kohli, R. M.; Trauger, J. W.; Schwarzer, D.; Marahiel, M. A.;
Walsh, C. T. Biochemistry 2001, 40, 7099.
(116) Sieber, S. A.; Tao, J.; Walsh, C. T.; Marahiel, M. A. Angew. Chem.
2004, 116, 499.
(117) Shaw-Reid, C. A.; Kelleher, N. L.; Losey, H. C.; Gehring, A. M.;
Berg, C.; Walsh, C. T. Chem. Biol. 1999, 6, 385.
(118) Weber, G.; Leitner, E. Curr. Genet. 1994, 26, 461.
(119) Haese, A.; Schubert, M.; Herrmann, M.; Zocher, R. Mol. Microbiol. 1993, 7, 905.
(120) Becker, J. E.; Moore, R. E.; Moore, B. S. Gene 2004, 325, 35.
(121) Witkowski, A.; Joshi, A.; Smith, S. Biochemistry 1996, 35, 10569.
(122) Kao, C. M.; Pieper, R.; Cane, D. E.; Khosla, C. Biochemistry 1996,
35, 12363.
(123) Staunton, J.; Caffrey, P.; Aparicio, J. F.; Roberts, G. A.; Bethell,
S. S.; Leadlay, P. F. Nat. Struct. Biol. 1996, 3, 188.
(124) Rangan, V. S.; Joshi, A. K.; Smith, S. J. Biol. Chem. 1998, 273,
34949.
(125) Witkowski, A.; Joshi, A. K.; Rangan, V. S.; Falick, A. M.;
Witkowska, H. E.; Smith, S. J. Biol. Chem. 1999, 274, 11557.
(126) Sieber, S. A.; Linne, U.; Hillson, N. J.; Roche, E.; Walsh, C. T.;
Marahiel, M. A. Chem. Biol. 2002, 9, 997.
Sieber and Marahiel

(127) Glinski, M.; Urbanke, C.; Hornbogen, T.; Zocher, R. Arch.
Microbiol. 2002, 178.
(128) Joshi, A. K.; Vangipuram, S. R.; Witkowski, A.; Smith, S. Chem.
Biol. 2003, 10, 169.
(129) Hillson, N. J.; Walsh, C. T. Biochemistry 2003, 42, 766.
(130) Richardt, A.; Kemme, T.; Wagner, S.; Schwarzer, D.; Marahiel,
M. A.; Hovemann, B. T. J. Biol. Chem. 2003, 278, 41160.
(131) Kasahara, T.; Kato, T. Nature 2003, 422, 832.
(132) Joshi, A. K.; Zhang, L.; Rangan, V. S.; Smith, S. J. Biol. Chem.
2003, 278, 33142.
(133) Bollag, D. M.; McQueney, P. A.; Zhu, J.; Hensens, O.; Koupal,
L.; Liesch, J.; Goetz, M.; Lazarides, E.; Woods, C. M. Cancer Res.
1995, 55, 2325.
(134) Molnar, I.; Schupp, T.; Ono, M.; Zirkle, R.; Milnamow, M.;
Nowak-Thompson, B.; Engel, N.; Toupet, C.; Stratmann, A.; Cyr,
D. D.; Gorlach, J.; Mayo, J. M.; Hu, A.; Goff, S.; Schmid, J.; Ligon,
J. M. Chem. Biol. 2000, 7, 97.
(135) Tang, L.; Shah, S.; Chung, L.; Carney, J.; Katz, L.; Khosla, C.;
Julien, B. Science 2000, 287, 640.
(136) Stachelhaus, T.; Schneider, A.; Marahiel, M. A. Science 1995,
269, 69.
(137) Schneider, A.; Stachelhaus, T.; Marahiel, M. A. Mol. Gen. Genet.
1998, 257, 308.
(138) Mootz, H. D.; Schwarzer, D.; Marahiel, M. A. Proc. Natl. Acad.
Sci. U.S.A. 2000, 97, 5848.
(139) Schwarzer, D.; Mootz, H. D.; Marahiel, M. A. Chem. Biol. 2001,
8, 997.
(140) Duerfahrt, T.; Doekel, S.; Sonke, T.; Quaedflieg, P. J. L. M.;
Marahiel, M. A. Eur. J. Biochem. 2003, 270, 4555.
(141) Mootz, H. D.; Kessler, N.; Linne, U.; Eppelmann, K.; Schwarzer,
D.; Marahiel, M. A. J. Am. Chem. Soc. 2002, 124, 10980.
(142) Davies, J. S. J. Pept. Sci. 2003, 9, 471.
(143) Sewald, N.; Jakubke, H.-D. Peptides: Chemistry and Biology;
Wiley-VCH: Weinheim, 2002.
(144) Sieber, S. A.; Marahiel, M. A. J. Bacteriol. 2003, 185, 7036.
(145) Kohli, R. M.; Walsh, C. T. Chem. Commun. (Cambridge) 2003,
297.
(146) Trauger, J.; Kohli, R.; Mootz, H.; Marahiel, M.; Walsh, C. Nature
2000, 407, 215.
(147) Trauger, J. W.; Kohli, R. M.; Walsh, C. T. Biochemistry 2001,
40, 7092.
(148) Bu, X.; Wu, X.; Xie G.; Guo, Z. Org. Lett. 2002, 4, 2893.
(149) Fernandez-Lopez, S.; Kim, H. S.; Choi, E. C.; Delgado, M.;
Granja, J. R.; Khasanov, A.; Kraehenbuehl, K.; Long, G.;
Weinberger, D. A.; Wilcoxen, K. M.; Ghadiri, M. R. Nature 2001,
412, 452.
(150) Kohli, R. M.; Walsh, C. T.; Burkart, M. D. Nature 2002, 418,
658.
(151) Kohli, R. M.; Burke, M. D.; Tao, J.; Walsh, C. T. J. Am. Chem.
Soc. 2003, 125, 7160.
(152) Garbe, D.; Sieber, S. A.; Bandur, N.; Koert, U.; Marahiel, M. A.
Chembiochem 2004, 5, 1000.
(153) Sieber, S. A.; Walsh, C. T.; Marahiel, M. A. J. Am. Chem. Soc.
2003, 125, 10862.
(154) Dawson, P. E.; Churchill, M. J.; Ghadiri, M. R.; Kent, S. B. H.
J. Am. Chem. Soc. 1996, 119, 4325.
(155) Muir, T. W.; Sondhi, D.; Cole, P. A. Proc. Natl. Acad. Sci. 1998,
95, 6705.
(156) Chan, W. C.; White P. D. Fmoc Solid-Phase Peptide Synthesis;
Oxford University Press: Oxford, 2000.
(157) Zhu, J.; Ma, D. Angew. Chem. 2003, 43, 5506.
(158) Schmidt, R.; Neubert, K. Int. J. Pept. Protein Res. 1991, 37, 502.
(159) Ehrlich, A.; Heyne, H.-U.; Winter, R.; Beyermann, M.; Haber,
H.; Carpino, L. A.; Bienert, M. J. Org. Chem. 1996, 61, 8831.
(160) Jackson, D. Y.; Burnier, J. P.; Wells, J. A. J. Am. Chem. Soc.
1995, 117, 819.
(161) Muir, T. W. Annu. Rev. Biochem. 2003, 72, 249.
(162) Nishino, N.; Xu, M.; Mihara, H.; Fujimoto, T.; Ueno, Y.; Kumagai, H. Tetrahedron Lett. 1992, 33, 1479.
(163) Klibanov, A. M. Nature 2001, 409, 241.
CR0301191
Chem. Rev. 2005, 105, 739758
739
Antitumor Antibiotics: Bleomycin, Enediynes, and Mitomycin

Ute Galm, Martin H. Hager, Steven G. Van Lanen, Jianhua Ju, Jon S. Thorson,*, and Ben Shen*,,
Division of Pharmaceutical Sciences and Department of Chemistry, University of Wisconsin, Madison, Wisconsin 53705
Contents
1. Introduction
2. Bleomycin
2.1. Discovery and Biological Activities
2.2. Clinical Resistance
2.2.1. Bleomycin Hydrolase
2.2.2. Enhanced DNA Repair
2.2.3. Bleomycin Binding Protein
2.2.4. Other Mechanisms
2.3. Resistance by the Producing Organisms
2.3.1. Bleomycin N-Acetyltranferase (BlmB)
2.3.2. Bleomycin Binding Protein (BlmA)
2.3.3. Transport Proteins
3. EnediynessNine-Membered Enediyne Core
Subfamily
3.2.1. Apo-Protein
3.2.2. DNA Repair
3.2.3. Transport
4. EnediynessTen-Membered Enediyne Core
Subfamily
5. Mitomycin
6. Perspective
7. Abbreviations
8. Acknowledgment
9. References
739
739
739
740
741
742
742
743
743
743
743
744
744
745
746
746
747
747
748
748
750
751
752
752
754
754
754
755
755
755
1. Introduction
Natural-product-derived cytotoxics remain a mainstay in current chemotherapy.1 This review focuses
on the current level of understanding and emerging
trends relevant to the DNA-damaging metabolite
families of the bleomycins, 9- and 10-membered
enediynes, and mitomycins. Within the context of
* To whom correspondence should be addressed. J.S.T.: phone,
608-262-3829;
fax,
608-262-5345;
e-mail,
jsthorson@
pharmacy.wisc.edu. B.S.: phone, (608) 263-2673; fax, (608) 2625345.; e-mail, bshen@pharmacy.wisc.edu.
Division of Pharmaceutical Sciences.
Department of Chemistry.
their clinical utilities and shortcomings, a comparison

of resistance mechanisms within producing organisms to those predominant among tumor cells reveals
remarkable potential for continued development of
these essential anticancer agents.
2. Bleomycin
The bleomycins (BLMs), such as bleomycinic acid
(1), BLM A2 (2), or BLM B2 (3), are a family of
glycopeptide-derived antibiotics originally isolated
from several Streptomyces species.2,3 Several structure variations of the naturally occurring BLMs have
been identified from fermentation broths, primarily
differing at the C-terminus of the glycopeptide. The
BLM structure was revised in 19784 and confirmed
by total synthesis in 1982.5,6 Structurally and biosynthetically related to the BLMs are the phleomycins (PLMs), such as PLM 12 (4) or PLM D1 (5),7-10
and tallysomycins (TLMs), such as TLM S2B (6) and
TLM S10B (7)11,12 (Figure 1).
BLMs are thought to exert their biological effects
through a sequence-selective, metal-dependent oxidative cleavage of DNA and RNA in the presence of
oxygen.13-16 The BLMs can be dissected into four
functional domains: (i) the pyrimidoblamic acid
(PBA) subunit along with the adjacent -hydroxyl
histidine constitutes the metal-binding domain that
provides the coordination sites required for Fe(II)
complexation and molecular oxygen activation responsible for DNA cleavage; (ii) the bithiazole and
C-terminal amine provide the majority of the BLMDNA affinity and may contribute to polynucleotide
recognition and the DNA cleavage selectivity; (iii) the
(2S,3S,4R)-4-amino-3-hydroxy-2-methylpentanoic acid
(AHM) subunit not only provides the connectivity
between the metal-binding and DNA-binding sites
but also plays an important role in the efficiency of
DNA cleavage by BLMs; (iv) the sugar moiety is
likely to participate in cell recognition by BLMs and
possibly in cellular uptake and metal-ion coordination. Consequently, there have been continuing attempts to develop new BLM congeners to define the
fundamental functional roles of the individual domains and search for anticancer drugs with better
clinical efficacy and lower toxicity. However, the
structural complexity of BLMs has limited most of
the modifications at either the C-terminal amine or
the N-terminal -aminoalaninamide moiety by either
directed biosynthesis or semisynthesis. Total chemi-
10.1021/cr030117g CCC: $53.50 2005 American Chemical Society

Ute Galm studied pharmacy at the University of Tuebingen, Germany. In

2003 she finished her doctoral research on aminocoumarin antibiotics
under Professor Lutz Heide and received her Ph.D. degree from the
University of Tuebingen. She was appointed as a postdoctoral associate
at the University of WisconsinsMadison in 2004, where she works with
Professor Ben Shen. Her current research interests are focused on
biosynthetic studies of the bleomycin family of compounds in different
Streptomyces strains.
Martin Hager, born in 1972 in Muenster, Germany, began his studies in

Biology in 1993 at the University of Tuebingen. He received his diploma
in 1998 and his Ph.D. degree in 2002 with Professor Ralph Bock at the
University of Freiburg for his work on the functional analyses of open
reading frames by reverse genetics. He then joined the laboratory of
Professor Albrecht E. Sippel to investigate EGF-receptor inhibitors and
new approaches to tumor targeting by exploiting antibody avidity. In 2003
he joined the laboratory of Professor Jon S. Thorson. His current research
objective is the directed evolution of glycosyltransferases and structural
analysis of resistance proteins. Martin Hager is a postdoctoral fellow of
the German Academy of Scientists Leopoldina. He also received grants
from the German Research Foundation and the German Ministry of
Education and Research.
cal synthesis of BLMs is expensive, thus limiting the

practicality in pharmaceutical applications. While
numerous BLM analogues have been synthesized in
the past two decades, none has improved properties.13,14,17 Therefore, the development of methods to
manufacture novel BLMs, particularly those unavailable or extremely difficult to prepare by chemical
synthesis, remains an important research goal.
BLMs exhibit strong antitumor activity and are
currently used clinically in combination with a number of other agents for the treatment of several types
of tumors, notably squamous cell carcinomas and
malignant lymphomas.13,14,18 The commercial product, Blenoxane, contains 2 and 3 as the principal
constituents. Unique to most anticancer drugs, BLMs
Galm et al.
Steven G. Van Lanen received his B.S. (1998) degree in Molecular Biology
from the University of WisconsinsMadison. In 2003 he received his Ph.D.
degree in Chemistry at Portland State University under the guidance of
Professor Dirk Iwata-Reuyl. He currently is a postdoctoral fellow in the
laboratory of Professor Ben Shen, studying the biosynthesis of C-1027
and other enediyne natural products.
Jianhua Ju was born in 1972 in Shandong, China. He received his B.S.

degree in Pharmacy from Shandong University in 1995 and obtained his
Ph.D. degree in Medicinal and Natural Product Chemistry at Peking Union
Medical College in 2000. Then he worked as a faculty member at the
Institute of Medicinal Plant Development, Chinese Academy of Medical
Sciences in Beijing, where he was engaged in new natural products
discovery research. He received the Servier Young Investigator Award in
Medicinal Chemistry in 2002 issued by the Chinese Pharmaceutical
Association & Institut de Recherches Internatinales Servier. In 2003 he
joined the research group of Professor Ben Shen as a postdoctoral
research associate at the School of Pharmacy, University of Wisconsins
Madison working in the field of natural product biosynthesis.
do not cause myelosuppression, promoting its wide

application in combination chemotherapy. Early development of drug resistance and cumulative pulmonary toxicity are the major limitations of BLMs in
chemotherapy.18

Thus far there is no proof of a mechanism that
explains the development of BLM resistance in some
tumor cells, although alteration of drug uptake and
efflux, enhanced repair of BLM-induced DNA lesions,
and increased inactivation of BLM might be possible
mechanisms.19-24 Several pathogenic microorganisms
have been found to exhibit BLM resistance, seemingly caused by the existence of a BLM-binding
protein in the respective organisms.25-28
Antitumor Antibiotics
Jon Thorson received his B.A. degree in Chemistry (1986) from Augsburg
College and his Ph.D. degree in Organic Chemistry (1993) from the
University of Minnesota with Professor Hung-wen (Ben) Liu. He held a
postdoctoral appointment as a Merck Postdoctoral Fellow of the Helen
Hay Whitney Foundation (19931996) at the University of California,
Berkeley, with Professor Peter Schultz. From 1996 to 2001 Jon held
appointments as an assistant member of the Memorial Sloan-Kettering
Cancer Center and assistant professor of Sloan-Kettering Division, Joan
and Sanford I. Weill Graduate School of Medical Sciences, Cornell
University, during which he was named a Rita Allen Foundation Scholar
(19982002) and Alfred P. Sloan Fellow (20002002). Professor Thorson
joined the School of Pharmacy in the summer of 2001, and since moving
to UW he has been awarded the American Society of Pharmacognosy
Matt Suffness Award (2004) and selected as a H. I. Romnes Fellow (2004).
His research interests include understanding and exploiting biosynthetic
pathways in various microorganisms, microbial pathway genomics,
mechanistic enzymology, mechanisms of resistance to highly reactive
metabolites, enzyme engineering and evolution to generate novel catalysts,
and development of chemoenzymatic and chemoselective ligation strategies for natural product glycorandomization.
2.2.1. Bleomycin Hydrolase

BLM can be metabolically inactivated in normal
and tumor tissues by an enzyme called BLM hydrolase as demonstrated by several earlier studies. BLM
hydrolase is a thiol protease that hydrolyzes the
C-terminus of BLM to generate the inactive deamido
metabolite.19,22,29 As evidence that this mechanism
may contribute to clinical BLM resistance,30-32 BLMresistant mammalian cells that were exposed to E64,
a specific thiol protease inhibitor that blocks BLM
hydrolase activity, became more sensitive to BLM.33
These experiments led to identification of the corresponding genes that encode BLM hydrolase from
yeast and mammalian cells.34-37
X-ray crystallographic studies of yeast38 and human37 BLM hydrolase revealed that both enzymes
share the same hexameric ring barrel structure with
the active sites embedded in a central cavity. The
central channel, which displays a very prominent net
positive charge in the yeast homologue and a slightly
negative net charge in the human enzyme, may
explain the differences in DNA binding among the
human and yeast BLM hydrolases.37 The threedimensional structures of these proteins have inspired mechanistic speculations regarding BLM hydrolysis. Notably, the primary amino group of the
metal-binding -aminoalanine moiety of BLM has
been proposed to serve as the amino terminus
anchoring to the C-terminal carboxylate of hydrolase
with the bulk of the BLM molecule protruding into
the large cavity in the center of the protein hex-
Ben Shen received his B.S. degree from Hangzhou University (1982),
M.S. degree from the Chinese Academy of Sciences (1984), and Ph.D.
degree from Oregon State University (1991, with Professor Steven J.
Gould) and carried out postdoctoral research in Professor C. Richard
Hutchinsons laboratory at the University of WisconsinsMadison (1991
1995). He started his independent career at the Department of Chemistry,
University of California, Davis, in 1995 and moved to the University of
WisconsinsMadison in 2001, where he currently holds the Charles M.
Johnson Chair in Pharmacy and Professor of Pharmaceutical Sciences
in the School of Pharmacy and Professor of Chemistry in the Department
of Chemistry. He was a Searle Scholar (1997) and recipient of a CAREER
award (19982003) from the National Science Foundation, the Matt
Suffness award (2000) and Jack L. Beal Award (2001) from the American
Society of Pharmacognosy, and an Independent Scientist Award (2001
2006) from the National Institutes of health. His current research interests
are the chemistry, biochemistry, and genetics of secondary metabolite
biosynthesis in Streptomyces and drug discovery and development by
combinatorial biosynthesis methods. Natural products currently under
investigation include aromatic polyketides, modular polyketides, the
enediynes, nonribosomal peptides, and hybrid peptidepolyketides.
amer.39 In this orientation the first BLM peptide bond

is cleaved via an aminopeptidase reaction. Such
mechanistic insights may ultimately be useful in
designing novel BLM analogues that are resistant to
BLM hydrolase.
Expression of the yeast BLM hydrolase gene blh1
in mammalian cells conferred a nearly 8-fold increase
in resistance to BLM, which could be antagonized by
the E64 inhibitor.40 This suggested that inactivation
of blh1 from yeast should result in a BLM-hypersensitive phenotype. However, independent studies revealed conflicting data regarding the role of Blh1 in
the detoxification of BLM.35,36 While two studies
suggested blh1 mutants were mildly sensitive to
BLM, two alternative studies indicated that these
mutants lacked BLM sensitivity.35,36,41,42 Moreover,
overexpression of blh1 in either parental or BLMsensitive yeast cells confers no additional resistance
to BLM,42,43 while the overproduced Blh1 clearly
inactivates BLM in vitro.42 Thus, the role of BLM
hydrolase in producing tumor resistance remains
controversial.
Furthermore, Blh1 binds specifically to the Gal4
transcription factor DNA-binding site and acts as a
repressor to negatively control the galactose metabolism pathway.38,44,45 Therefore, Blh1 could also play
a more general role in the cells to degrade proteins
or perhaps to regulate gene expression by degrading
certain transcription factors.45-47 If this is the case,
the BLM resistance observed by overexpression of
yeast blh1 in mammalian cells may be explained, for
Galm et al.
Figure 1. Structures of bleomycinic acid (1), bleomycin A2 (2), bleomycin B2 (3), phleomycin 12 (4), phleomycin D1 (5),
and tallysomycin S2B (6), and tallysomycin S10B (7).
example, by degrading pro-apoptotic factors, thus

preventing cell death.
2.2.2. Enhanced DNA Repair

To overcome BLM-induced genotoxicity, DNA repair may be the most important mechanism used by
cells. Thus, organisms exposed to BLM must recruit
a variety of enzymes and/or proteins to repair the
diverse types of BLM-induced DNA lesions. Apn1 was
the first yeast enzyme discovered to process BLMinduced DNA lesions in vitro.48,49 Apn1 is equipped
with an apurinic/apyrimidinic (AP) endonuclease
activity that cleaves the DNA backbone at AP sites
as well as a 3-diesterase activity that removes 3blocking groups (such as 3-phosphoglycolate) at
strand breaks. These enzymatic activities regenerate
3-hydroxyl groups that subsequently allow DNA
repair synthesis by DNA polymerase and ligase.50-52
Two additional yeast 3-diesterases, Apn2 and Tpp1,
also repair BLM-induced DNA lesions. Notably, only
the inactivation of all three 3-diesterase genes, apn1,
apn2, and tpp1, resulted in remarkable BLM sensitivity.53,54 In human cells only the Tpp1 and Apn2
homologues , hPNKP and Hap1, respectively, have
been identified to date.53,54 There is some evidence
that overproduction of Hap1 in mammalian cells can
lead to enhanced resistance to BLM,55 although the
contribution of hPNKP and Hap1 to the repair of
BLM-induced DNA lesions in human cells has not
been investigated.
Rad52 and Rad6 proteins, which are part of the
recombination and postreplication DNA repair path-
ways in yeast, respectively, are also involved in the

repair of BLM-induced DNA lesions. These findings
suggest that the repair of such lesions may not be
restricted to enzymes with the ability to cleave AP
sites or remove 3-blocking groups.56,57 The Rad52 and
Rad6 pathways also repair a wide spectrum of other
DNA lesions, including those generated by the alkylating agents, such as methyl methane sulfonate that
produces AP sites and 4-nitroquinoline-1-oxide that
forms bulky DNA adducts, and -rays.58-62 While the
exact contributions of these two proteins to DNA
repair of BLM-induced lesions has not been established, some evidence suggests that their contribution
depends on the extent of damage to the DNA.57
In Yeast expression of the DNA repair proteins
Apn1 and Rad6 resulted in restoration of wild-type
BLM resistance in the respective deletion mutants
but did not exceed the resistance level of the parental
cells.49,58 However, in mammalian cells one study
revealed the overproduction of the DNA repair enzyme Hap1 to enhance BLM resistance in normal or
tumor cells.55 This latter finding is in agreement with
one of the earlier predictions that tumor resistance
to BLM may be attributed to elevated DNA repair
activities.23 Therefore, attempts to promote the antitumor potential of BLM must take into consideration the importance of locally diminishing the DNA
repair capacity within the respective tumor cells.
2.2.3. Bleomycin Binding Protein

Although BLM has never been used as an antibacterial agent, many clinically isolated methicillin-
resistant Staphylococcus aureus (MRSA) strains were

found to be resistant to this drug at high levels.28 The
respective resistance gene blmS from one of these
strains was investigated in detail. Sequence analysis
revealed that it was identical with a gene located on
the staphylococcal plasmid pUB110,25 the gene product of which (BlmS) was determined to be a BLMbinding protein.28 Another gene, ble, located on the
transposon Tn5 (originally isolated from Klebsiella
pneumoniae),26,27 confers BLM resistance upon Escherichia coli with the gene product (BlmT) found to
also be a BLM-binding protein.62 While it could be
argued that pathogenic bacteria like K. pneumoniae
and S. aureus recruited their BLM-resistance genes
from BLM-producing organisms, BlmS and BlmT
share only 20% identity/40% similarity between
each other and to the BLM-binding proteins from
BLM producers. In contrast, the latter are highly
homologous with 60%identity/70% similarity
among each other (see section 2.3). Cumulatively, the
most common bacterial resistance mechanism against
BLM and its analogues appears to be development
and utilization of binding proteins via drug sequestering.
2.2.4. Other Mechanisms

Proper cell wall maintenance appears to play an
important role in protecting yeast against the lethal
effects of BLM. Certain mutants that were sensitive
to BLM were also identified to be defective in genes
encoding proteins that maintain proper cell wall
structure such as Fks1, a subunit of -1,3-glucan
synthetase.63 Defects in the cell wall integrity signaling pathway are therefore also able to cause sensitivity to BLM, likely due to improper cell wall structure
and enhanced permeability. Slg1, a plasma membrane sensor, detects cell wall perturbations and
signals activation of protein kinase C (Pkc1), and
varying sensitivities to BLM correlate to mutants
lacking either Slg1 or Pkc1.63
The Slg1 protein has been shown to be required to
attenuate the toxicity of BLM to yeast cells.63 Although mammalian cells do not have a cell wall, their
extracellular matrix is related to the yeast cell wall
and contains a class of protein receptors known as
integrins.64,65 The structural organization of integrins
is similar to yeast Slg1 protein.65,66 It is therefore
conceivable that a member of the integrin family
could function to sense BLM, in a manner comparable to yeast Slg1 protein, and activate a signal
transduction pathway leading to a defense mechanism against BLM.
Furthermore, a transcriptional activator, Imp2,
was found to be involved in detoxification of BLM.
Imp2 is a small protein that can activate transcription of a reporter gene by virtue of an acidic domain,43
and at least two genes, malT and malS, have been
identified to be positively regulated by Imp2.67 Deletion of imp2 resulted in mutants that displayed
hypersensitivity to BLM.43 Since Imp2 is a transcriptional activator, this protein may positively regulate
the expression of at least one gene encoding a protein
that repairs BLM-induced DNA lesions. The exact
mechanism of gene activation by Imp2, however,
remains to be elucidated.24
Interestingly, no drug transporters belonging to the

ATP-binding cassette (ABC) and major facilitator
superfamilies, such as Snq2, Yor1, Atr1, and Flr1,
have been found to be involved in BLM resistance,
as deficient mutants are not sensitive to the drug.68-74

The antibiotic-producing microorganisms must protect themselves from the lethal effects of their own
products, and multiple mechanisms of drug resistance are common for many antibiotic-producing
organisms. In most cases antibiotic production genes
have been found to be clustered in one region of the
bacterial chromosome, consisting of structural, resistance, and regulatory genes. Two resistance genes,
blmA and blmB, have been characterized whose
deduced products confer BLM resistance to the
producing organism by drug sequestering (BlmA) or
modification (BlmB), respectively.75-80
2.3.1. Bleomycin N-Acetyltranferase (BlmB)

The blmB gene from Streptomyces verticillus encodes an N-acetyltransferase that acetylates the
R-amine of the BLM -aminoalanine moiety in the
presence of CoA. The latter moiety is critical for metal
binding, and the acetylated BLM is no longer able to
chelate metal and, thus, lacks activity.75,79 BlmB has
been overproduced in E. coli, and the recombinant
protein has been purified and biochemically characterized.80 Surprisingly, neither the carboxylic acid
congener of BLM such as 1 nor PLMs such as 4 could
serve as substrates for the BlmB N-acetyltransferase.75,80 On the basis of this remarkable substrate
specificity it would now be of interest to determine
the three-dimensional structure of BlmB and thereby
shed light on the molecular interactions between
BLM and the BlmB N-acetyltransferase.
Within the TLM biosynthetic gene cluster from
Streptoalloteichus hindustanus a blmB homolog,
tlmB, was identified, the deduced product of which
showed 57% identity/64% similarity to BlmB (George,
N. P.; Wendt-Pienkowski, E.; Shen, B. Unpublished
data). On the basis of these results TlmB is presumed
to be the N-acetyltransferase responsible for inactivation of PLMs in S. hindustanus. In contrast, no
blmB homologue could be identified within the
sequenced PLM biosynthetic gene cluster from Streptomyces flavoviridis (Oh, T.-J.; George, N. P.; WendtPienkowski, E.; Yi, F.; Shen, B. Unpublished data).
From these findings two important questions arise:
(i) could the putative PLM N-acetyltransferase reside
outside the sequenced cluster or does it not exist
within the PLM producer and (ii) what is the relevance of the N-acetyltransferase for the BLM- and
TLM-producing organisms in order to maintain resistance against their own products while the PLM
producer does not seem to depend on this selfresistance mechanism? Both questions remain unresolved.
2.3.2. Bleomycin Binding Protein (BlmA)

In addition to the aforementioned BlmB N-acetyltransferase, S. verticillus has the blmA gene encoding
for a binding protein (BlmA) that displays a strong

affinity for BLM.28,76 The expression of both blmA and
blmB has been shown to increase simultaneously
with BLM production in the late exponential growth
phase of S. verticillus.80 This finding was not surprising since both gene products are likely to be responsible for self-resistance of the producer strain and
therefore need to be available in sufficient amounts
as BLM production increases. BlmA, an acidic protein
consisting of 122 amino acids with a calculated
molecular size of 13.2 kD, has been extensively
characterized biochemically.28 Determination of the
X-ray crystal structure revealed that BlmA forms a
dimer through N-terminal arm exchange.78 The
resulting concavity and groove may contribute to the
binding of two BLM molecules, and the formation of
a dimer may be necessary to retain an affinity for
BLM. The recent X-ray crystal structure of the
BlmA-BLM complex confirmed that two BLM molecules are indeed bound by the BlmA dimer.81 Additionally, these studies revealed the binding of the
first BLM molecule to support cooperative binding
of the second BLM.81 The interaction of BlmA and
BLM is assumed to result from an electrostatic
interaction between the basic antibiotic and acidic
protein, the affinity of which is enhanced upon Fe2+chelation. Moreover, the N-terminal Pro-9 of BlmA
may play a role as a hinge to support the dimer
structure. Replacement of Pro-9 in BlmA by Leu
abolished the binding affinity for BLM due to disruption of the quaternary structure.77
A BLM-binding protein was also isolated from S.
hindustanus, the producer of TLMs such as 6 and 7.
This protein was named Sh Ble and is able to form
1:1 protein-drug complex with high affinity to BLM.82
We recently completed the cloning and sequencing
of the TLM biosynthetic gene cluster from S. hindustanus and indeed localized the gene, tlmA, within
the sequenced TLM gene cluster that encodes the Sh
Ble protein (George, N. P.; Wendt-Pienkowski, E.;
Shen, B. Unpublished data). Finally, the PLM biosynthetic gene cluster from S. flavoviridis ATCC21892
has also been sequenced recently (Oh, T.-J.; George,
N. P.; Wendt-Pienkowski, E.; Yi, F.; Shen, B. Unpublished data). The sequenced PLM cluster also
contains a gene, plmA, whose deduced gene product
is highly homologous to BlmA and TlmA, supporting
its role as PLM-binding protein. The three binding
proteins, BlmA, PlmA, and TlmA, show high sequence homology (53-61% identities/63-73% similarities). Identification of these binding proteins in
all three producers implicates drug sequestering as
a main mechanism of self-resistance in these organisms.
Streptomyces lavendulae produces mitomycin (MTM)
C (19), a potent anticancer agent, and a resistance
protein, Mrd, was identified to act as a MTM C
binding protein83 (see section 4.3). Mrd belongs to a
larger family of proteins containing tandem R
motifs as representative structural elements. This
family also includes the BLM-resistance proteins of
S. hindustanus (Sh Ble), S. verticillus (BlmA), and
E. coli (BlmT on transposon Tn5). Although the BLMand Mrd-binding proteins show low sequence simi-
Galm et al.
larity, the X-ray crystal structure of Mrd is very

similar to that of the BLM-binding proteins.84 Both
structures share overall tertiary and quaternary
structural features and display a pair of symmetric
cavities that serve as binding sites. However, BLM
is a far larger molecule than MTM C, and these
compounds are chemically distinct. From a crystallographic study performed with the BlmT-BLM
complex85 it could be shown that the bithiazole moiety
of BLM is sequestered between two tryptophans of
the binding protein, a feature that has also been
observed for Mrd-MTM C complex. Indeed, Mrd was
recently confirmed to bind BLM and confer BLM
resistance.86 Thus, even though the nature of the
ligand is very different, the mode of drug binding by
Mrd and the BLM-binding proteins is very similar.
Taking these findings into consideration, it will be
interesting to explore whether the converse is true
and, more importantly, whether this drug-binding
motif serves as a general cross-resistance mechanism
for drug sequestering.
Since BLM and its homologues are excellent DNA
cleavage agents, the Sh ble gene also serves as an
invaluable resistance marker in commercial cloning
plasmids, particularly for prokaryotic-eukaryotic
shuttle plasmids. Although no naturally occurring
BLM-binding protein has been identified in eukaryotic cells so far, these proteins should also be considered as a possible emerging mechanism of resistance in eukaryotes in the future.
2.3.3. Transport Proteins

In the BLM-producer S. verticillus Blm-Orf7 has
been proposed to be a member of the ABC-transporter family of proteins.76 The latter proteins confer
resistance by transporting the drug out of the cells,87
although this function has not been confirmed for
Blm-Orf7.76 The gene product of Blm-Orf29 is
closely related to a family of transmembrane transporters and could also be involved in BLM resistance
by drug efflux.88 Further investigation of both transporters regarding their contribution to self-resistance
of the producer strain would be informative, although
BLM is not a substrate for the corresponding Pglycoprotein transporters (Pgp) involved in multidrug
resistance (MDR) in yeast or mammals.89
3. EnediynessNine-Membered Enediyne Core

Subfamily
The enediynes represent a steadily growing family
of natural products with unprecedented molecular
architecture. They have garnered much interest
because of their unique structure and mode of action
that confer clinically desirable attributes such as
antibiotic and antitumor activity.90,91 The enediynes
are structurally characterized by an unsaturated core
with two acetylenic groups conjugated to a double
bond or incipient double bond and have been categorized into two subfamilies: 9-membered ring chromophore cores or 10-membered rings (see section 4).
This portion of the review focuses on recent advances
in our understanding of the mechanism of the ninemembered ring enediynes and the potential applica-
Figure 2. Nine-membered enediyne chromophores whose structures have been elucidated: neocarzinostatin (8), kedarcidin
(9), C-1027 (10), maduropeptin (11), and N1999A2 (12).
tions of these natural products as therapeutic agents.

The discussion includes the general characteristics
of the family followed by specific information regarding individual enediynes with an emphasis on C-1027
and neocarzinostatin (NCS) as models for the ninemembered enediynes.

With the elucidation of the NCS chromophore (8)
in 198592 the nine-membered chromoprotein family
of enediyne has steadily grown to currently consist
of nine natural products: NCS from Streptomyces
carzinostaticus,93-95 kedarcidin (9) from Actinomycete
L585-6,96,97 C-1027 (10) from Streptomyces globisporus,98 maduropeptin (11) from Actinomadura
madurea,99 N1999A2 (12) from Streptomyces sp.
AJ9493,100 actinoxanthin from Actinomyces globisporus,101,102 largomycin from Streptomyces pluricolorescens,103 auromomycin from Streptomyces macromomyceticus,104-106 and sporamycin from Streptosporangium pseudovulgare.107,108 Although all of the
known nine-membered enediynes contain a common
bicylo[7.3.0]dodecdadiynene chromophore, only five
of the complete structures, 8-12, have been established (Figure 2). Recently, numerous cryptic gene
clusters encoding enediyne biosynthesis in a variety
of actinomycetes have been unveiled,109 suggesting
that these organisms have the potential to produce
uncharacterized enediynes. The latter finding may
significantly increase the pool of nine-membered
enediynes in the years to come.
As a group the nine-membered enediynes are the
most potent family of anticancer agents discovered,
some members of which are 5-8000 times more
potent than adriamycin, an antitumor antibiotic that
has been very effective in clinical use.110 For example,
C-1027 shows extremely potent cytotoxicity against
KB carcinoma cells (IC50 0.1 ng/mL) in vitro111 and
powerful antitumor activity toward tumor-bearing

mice in vivo.110 However, the enediynes have shown
delayed toxicity, limiting their use in clinical applications.110,112 To overcome this shortcoming a few
modified enediynes have been prepared for clinical
purposes and have shown great promise. Conjugation
of NCS with poly(styrene-co-maleic acid) (SMA) or
its various alkyl esters has dramatically improved
the uptake and overall toxicological profile, and as a
consequence, the polymer conjugated derivative of
NCS, SMANCS, has been used to treat hepatoma in
Japan since 1994.113 SMANCS in conjunction with
Lipiodol, a lipid contrast agent, has also shown great
promise for treatment of tumors in the lung, stomach,
pancreas, and gall bladder as well as lymphoma and
melanoma.113
An equally exciting development has been the
creation of monoclonal antibody (mAB)-enediyne
conjugates. The best example is the 10-membered
enediyne calicheamicin (CAL), which has been prepared as a mAB-CAL conjugate and is currently
approved by the FDA under the trade name of
Mylotarg to treat acute myeloid leukemia (AML) (see
section 4).114 Several mAB-C-1027 conjugates have
also been prepared and are characterized by increased tumor specificity and strong inhibition on the
growth of tumor xenografts.115-117 These conjugates
are currently being evaluated for clinical significance
as anticancer drugs118 and highlight the potential for
enediynes in therapeutic treatments.
The mode of action of nine-membered enediynes,
which is generally accepted for all enediynes, is the
ability to produce single-stranded or double-stranded
DNA lesions (and, in some cases, RNA lesions)119 by
a common mechanism shown for 10 in Scheme
1.120,121 The chromophore binds the minor groove of
DNA and undergoes an electronic rearrangement to
form a benzenoid diradical species (via a Bergman
Galm et al.
Scheme 1
or Myers rearrangement), which in turn abstracts

hydrogen atoms from the deoxyribose of DNA. Molecular oxygen can react with the newly formed
carbon-centered radicals, leading to site-specific DNA
breaks. The DNA damage in turn causes a significant
decrease in DNA replication competency and ultimately leads to cell death.
Similar to 10-membered enediynes, the 9-membered ring enediyne chromophores require activation
for biological activity, although they are remarkably
less stable overall. The most biologically relevant
mode of diradical initiation is thiol activation to
trigger radical formation, although other activators
such as acidic or basic pH and light have been shown
to initiate the electronic rearrangement.91 It is notable that 10 is the most labile enediyne studied
among the family as the free chromophore. Unlike
the others, DNA cleavage by C-1027 proceeds even
in the absence of thiol groups or nucleophiles.122,123
It has also been reported that maduropeptin, like
C-1027, is also capable of diradical formation without
activation.124

Unlike the 10-membered enediynes, the 9-membered enediyne cores are relatively unstable, and
therefore, it is of utmost importance for the producing
organism to control the production, transportation,
and export of the enediyne product. Similar to several
Streptomyces antibiotic-producing organisms, multiple mechanisms of self-resistance have been evolved
including drug sequestering, efflux transport, and
DNA repair.
3.2.1. Apo-Protein
The primary mechanism utilized by nine-membered enediyne-producing organisms is drug sequestering by the production of an apo-protein that
tightly, but noncovalently, binds and stabilizes the
chromophore.90,91 The observations that apo-proteins
for C-1027 and macromomycin are constitutively
produced and independent of the chromophore production suggest that the apo-protein function is
necessary for self-resistance.125,126 Furthermore, it
has been proposed that 10 is in equilibrium with its
p-benzyne form and is stabilized kinetically by the
CagA apo-protein.127 This hypothesis was tested and
supported by electron paramagnetic resonance analyses of the C-1027 chromoprotein complex and a spin-
trapping study of DNA cleavage induced by

C-1027.128,129
The apo-protein sequences for 8 (NcsA),130,131 9
(Ked),124,132 10 (CagA),133 actinoxanthin (AxnA),102
and auromomycin (McmA)134 have been reported and
show 40% identity among them. The apo-protein
sequence for 11 is presumably known, but the
authors reported size ambiguities ranging from 13
kD by mass spectroscopy to 29 kD or 31 kD by SDSPAGE.124,135 We recently sequenced the maduropeptin biosynthetic gene cluster from A. madurea. While
sequence analysis failed to identify an apo-protein
candidate that is homologous to known apo-proteins,
a single ORF, encoding a hypothetical protein with
unknown function that is small (16 kD) and acidic
(pI 3.5), was identified. The latter could potentially
function as the chromophore-binding protein, although our current study falls short of excluding the
possibility that the apo-protein may reside outside
of the sequenced cluster (Liu, W.; Wendt-Pienkowski,
E.; Oh, T.-J.; Van Lanen, S. G.; Shen, B. Unpublished
data).
The structures for both apo-proteins (CagA and
NcsA) and chromoprotein complexes (C-1027 and
NCS) have been established by NMR spectroscopy.136-140 Preliminary X-ray diffraction data for
actinoxanthin, macromycin, and C-1027 apo-proteins
or chromoproteins have also been reported,141-143
although detailed descriptions of the crystal structures are lacking. The apo-proteins share several
characteristics as expected from high sequence homology, including being small, acidic peptides that
are rich in -sheet secondary structure.
The apo-protein for C-1027, CagA, was revealed to
consist of three antiparallel -sheets: a four-stranded
-sheet, a three-stranded -sheet, and a two-stranded
-sheet. A hydrophobic pocket is formed by the fourstranded -sheet and three loops. Modeling of the
C-1027 chromoprotein complex demonstrated that
the chromophore is packaged compactly by folding
its benzoxazolinate and aminosugar moieties in a
manner to interact with several hydrophobic side
chains of CagA, which include Tyr-32, Ala-34, Pro47, Ala-50, and Pro-76. Electrostatic interactions are
also plausible within the holo-protein, which include
a salt bridge and two hydrogen bonds.
In the case of the aromatized C-1027 chromophoreCagA complex, the benzodihydropentalene core is
located in the center of the pocket and its molecular
plane is nearly perpendicular to the bottom of the

pocket. The benzene ring of the core, where the C3
and C6 of the carbon-centered diradicals are formed,
faces toward the bottom of the pocket and is masked
by the -tyrosine, benzoxazolinate, and aminosugar
moieties. A mechanism for diradical stabilization was
proposed in which the H-R of Gly-96 serves as a
candidate for hydrogen abstraction by the C6 radical
while H-1 and H-2 of Pro-76 present to the C3
radical.128,129 The former interaction has recently been
tested by observing kinetic isotope effects which
revealed that the [U-2H]Gly-96 CagA exhibited a
better chromophore-stabilizing ability.144
The structure for apo-protein of NCS, NcsA, has
been solved by NMR spectroscopy.136,137,145 Similar to
CagA, NcsA is comprised primarily of -sheets. In
addition to structural analysis of NcsA, binding data
was obtained using fluorescence and NMR spectroscopy. Binding of 8 to NcsA, analogous to 10 to CagA,
was attributed mainly to hydrophobic interactions
between the naphthoic acid moiety and several
hydrophobic side chains of NcsA including Gly-35,
Leu-45, Phe-78, Val-95, and Trp-39. NcsA binds the
naphthoic acid moiety deep in the pocket, and the
enediyne core is located above a Cys-47-Cys-37 disulfide bond and surrounded by the aromatic rings
of Phe-52 and Phe-78, the former of which, in
conjunction with Asp-33, Ser-98, and the protonated
methylamino group, infringe upon nucleophilic attack
and thereby inhibit diradical formation. The carbonate carbonyl and aminosugar group provide some
binding energy but to a much lesser extent.
Although the primary function of the apo-protein
is to sequester and stabilize the enediyne for proper
delivery, the apo-proteins have been proposed to
exhibit protease activity. As demonstrated for kedarcidin and NCS, Ked or NcsA isolated from the native
producer had specific endopeptidase activity, preferentially cleaving histone H1 compared to other histones and various proteins.124,135 This putative activity was very attractive as it enabled the specific
delivery of the chromoprotein complex to the DNA
target. However, the protease activity has recently
been reexamined for NCS by production and isolation
of a recombinant NcsA from E. coli, and results from
two separate groups confirmed that protease activity
can be separated from NcsA, suggesting this activity
is due to minor contaminating proteases.146,147 The
recombinant NcsA maintained its structural integrity
based on far-UV, CD, and NMR spectroscopy and
functioned in binding 8, although with a much
greater Kd (>200-fold). It is unclear whether the
differences in 8 binding were due to experimental
variations or if the recombinant NcsA protein had
lost some functionality during production in a heterologous host. In general, the native apo-proteins
have >1000-fold lower specific activities and no
primary sequence or tertiary structure similarities
to known proteases,135,148-151 supporting the conclusion that the protease activity is indeed an artifact
of the purification and assay conditions. The possibility still remains that the apo-protein could
require a cofactor for protease activity that is lost
upon recombinant expression and purification (per-
haps the chromophore itself) or could recruit a

specific protease by noncovalent interactions. The
availability of numerous genes and the strategies to
heterologously express them for apo-protein production should facilitate further experiments to ascertain
the nature of the protease activity, such as sitedirected mutagenesis and chemical trapping of potential substrate-enzyme intermediates.
Regardless of the protease function, the binding of
the apo-protein to the enediyne chromophore affords
protection for the producer but also provides a
blueprint as a natural drug delivery system. Having
the wealth of structural data will now allow opportunities to genetically engineer and chemically
manipulate the chromophore complex for specific
targeting and optimized drug delivery during future
clinical trials.
3.2.2. DNA Repair

A pivotal discovery in our understanding the
resistance mechanisms for the enediyne family was
the identification and cloning of the gene clusters for
C-1027 and CAL.152,153 Primary sequence analysis of
the gene products revealed several possible candidates responsible for self-resistance in conjunction
with drug binding by the apo-protein and the putative involvement of encoded DNA repair enzymes.
The upstream boundary of the C-1027 gene cluster
contains an ORF, sgcB2, whose deduced product
showed high sequence similarity to E. coli UvrA, a
protein involved in excision repair,154 and Streptomyces peucetius DrrC, a UvrA-like drug-resistance
protein.156,157 UvrA is part of a three-component
protein assembly that recognizes, unwinds, and
excises damaged DNA, and UvrA has been shown to
strongly bind DNA and release DNA-bound anthramycin.155 Similar to UvrA, DrrC was shown to have
DNA-binding activity that was mediated by ATP and
enhanced in the presence of the antibiotic daunorubicin.156,157 Therefore, SgcB2, which has 27% identity/37% similar to DrrC, is speculated to bind to
DNA regions as a general mechanism to inhibit
C-1027 binding or could hinder the activation of the
enediyne core while inside the producer.
3.2.3. Transport
Sequencing of the gene clusters for NCS and
C-1027 and BLAST searches of the gene products
revealed homologues to efflux pumps and other
candidates for transport, one of which is conserved
between the two clusters: SgcB for C-1027 and
NcsA1 for NCS. These proteins are putative efflux
transporters that have several homologues within
Streptomyces including Pur8 in the puromycin gene
cluster from Streptomyces alboniger.158 Pur8, when
expressed in S. lividans, induced specific antibiotic
resistance and was implicated in the excretion of the
last intermediate in the puromycin biosynthetic
pathway, N-acetylpuromycin.159 Sequence analysis
revealed Pur8 to contain 14 transmembrane-spanning regions, and as a result, Pur8 is believed to be
necessary for puromycin efflux energized by a protondependent electrochemical gradient. From the high
sequence homology (SgcB/Pur8, 36% identity/56%
Galm et al.
Figure 3. Structures of 10-membered enediynes: calicheamicin (13), esperamicin (14), namenamicin (15), shishijimicin
(16), and dynemicin (17).
similarity and NcsA1/Pur8, 34% identity/53% similarity), it is reasonable to assume that SgcB and
NcsA1 have similar activity and provide the means
for enediyne efflux transport.
Interestingly, the C-1027 gene cluster also contains
an unshared antibiotic transporter homolog, SgcB4,
which consists of conserved domains from a family
of predicted drug exporters of the resistance-nodulation-cell division permease superfamily,160 and
AcrB, a cation/multidrug efflux pump utilized as a
defense mechanism. The latter family consists of
proteins that have been biochemically confirmed to
be involved in multidrug efflux with wide substrate
specificity as demonstrated in E. coli,161 the stressinduced efflux system of E. coli,162 and the secretion
of the siderophore pyoverdine in Pseudomonas aeruginosa.163,164 SgcB4, therefore, represents an additional
candidate for C-1027 efflux that is not shared in the
nine-membered enediyne core subfamily and could
be a general mechanism for C-1027 resistance.
4. EnediynessTen-Membered Enediyne Core

Subfamily
The second set of enediynes is very similar to those
previously mentioned in that they also contain three
critical functional domainssthe enediyne moiety (or
warhead), a recognition unit which delivers the
enediyne moiety to its DNA target, and a trigger
device that initiates the generation of the reactive
chemical species. The notable difference, however, is
that the enediyne is found within a 10-membered
ring system, presenting metabolites that do not
require sequestration by an apo-protein. To date, this
subfamily encompasses the natural products CAL

(13) from Micromonospora echinospora ssp. calichensis,165,166 esperamicin (14) from Actinomadura verrucosospora,167 namenamicin (15) from Polysyncraton
lithostrotum,168 shishijimicin (16) from Didemnum
proliferum,169 and dynemicin (17) from Micromonospora chersina (Figure 3).170,171 Like the other members of the enediyne antibiotic family, the 10membered enediynes also function in vitro and in
vivo as DNA-damaging agents.172,173

Discovery of the 10-membered enediynes began
with 14 in 1985 and sparked a great interest in this
new class of antitumor antibiotics.167 In the following
years the structures of 13, 15, and 17 were reported
with the discovery of 16 in 2003 representing the
most recent family member.169 In a manner similar
to that described for the 9-membered enediynes, the
10-membered enediynes bind DNA with high affinity,
and some reported sequence specificity, culminating
in sequence-selective oxidative strand cleavage. DNA
cleavage results from quenching of the benzenoid
diradical formed upon reductive activation via DNA
backbone hydrogen abstraction (Scheme 2).174-176
The staggering cytotoxicity of the enediynes represents their greatest strength. However, the relatively unspecific mode of action poses the problem of
general toxicity and subsequent systemic side effects.
Therefore, current enediyne research is focused upon
the development of enediyne analogues with enhanced selectivity toward cancerous cells. One promising approach to compensate for this limitation has
Scheme 2
been to conjugate 10-membered enediynes to tumordirected mAbs.177 Such antibody-targeted chemotherapy is heavily dependent upon the specific delivery of the enediyne to tumor cells via the tumorassociated antigen-mAb recognition to provide a
localized exposure to the cytotoxic agent.178 The high
toxicity of the 10-membered enediynes (they are
capable of triggering cell cycle arrest and apoptosis
in the picomolar range) favors this approach as only
a small number of immunoconjugates bind to the cell
surface and are internalized.179 Two different mABdirected strategies have proven successful in the
application of 10-membered enediynes. An alternative strategy employs the specific tissue-localized
enzymatic activation of enediynes. Both strategies
ultimately limit overall general toxicity and are
described in more detail below.180,181
As the pioneering example of localized delivery, the
first mAB-CAL conjugate strategically focused upon
the treatment of AML,112,182 a disease for which 13
had already displayed notable antileukemic potency.
To reduce general 13 cytotoxicity a semisynthetic
derivative of the drug was covalently coupled to a
humanized mAb (HuM195) specific for the antigen
CD33.183 The combination of mAb and immunoconjugate linker turned out to be the key to the success
of this new approach. The antibody HuM195 binds
the antigen CD33, a glycosylated transmembrane
protein with an expression pattern that is basically

confined to the hematopoietic system. In 90% of
patients with AML this 67 kD glycoprotein is absent
from healthy hematopoietic stem cells and nonhematopoietic tissues.184 Furthermore, binding of CD33
results in endocytosis of the antibody-drug-complex, further enhancing specific drug delivery. The
linker was designed as a metabolically stable but
acid-labile hydrazone conjugate such that lysosomal
acid hydrolysis of the mAb-CAL conjugate efficiently
releases the 10-membered enediyne.185 This antiCD33 antibody-CAL conjugate was named gemtuzumab ozogamicin (CMA-676) and later given FDA
approval under the name Mylotarg for the treatment
of first-relapse AML in patients >60 years of age.
Due to its reduced toxicity, antibody-targeted chemotherapy with Mylotarg is today a clinically validated therapeutic option for CD33-positive AML.186
Mylotarg represents the first and is presently the
only approved antibody-targeted cytotoxic small molecule agent for clinical use in the United States.
Several attempts have been made to expand this
strategy. CD22 emerged in these studies as an
attractive candidate for immunotoxin-based therapeutic strategies involving the treatment of B-lymphoid malignancies.187 As a sialic-acid-binding lectin,
CD22 combines two desirable qualitiessits expression is restricted to B-lymphocytes188,189 and CD22
is rapidly internalized when bound by ligand or
antibody.190,191 In an approach paralleling the treatment of AML with Mylotarg, a CD22-targeted immunoconjugate of CAL (CMC-544 or inotuzumab
ozogamicin) bound CD22 with subnanomolar affinity
and exhibited a potent cytotoxicity against CD22+
B-lymphoma cells. These results bode well for the
future use of CMC-544 in the treatment of patients
with non-Hodgkin B-cell lymphoma.192
The versatility of CAL-immunoconjugates has recently been demonstrated by two more applications
that have been advanced to the stage of clinical
evaluation. The compound CMB-401 employs a mAb
that is directed against polymorphic epithelial mucin
(PEM), a glycoprotein that shows aberrant expression
levels in malignancies of epithelial origin and is
implicated in the increased metastatic potential of
ovarian cancers. This new derivative of a CALimmunoconjugate showed potent selective cytotoxicity toward PEM-positive cell lines in tissue culture.
Therefore, CMB-401 is now being evaluated as a
monotherapy for the treatment of epithelial ovarian
carcinoma.193 The oligosaccharide Lewisy was also
considered as a promising antigen for antibodytargeted chemotherapy. Following the concept established for Mylotarg, CMC-544 and CMB-401, CAL
has also been conjugated to the humanized IgG1
antibody Hu3S193 which recognizes the Lewisy antigen. This specific antigen is highly expressed on
carcinomas of the colon, breast, lung, ovary, and
prostate, while expression of Lewisy in normal tissues
is restricted to the gastric mucosa, small intestine,
and pancreas. This conjugate may therefore prove
useful for selectively targeting tumors that express
Lewisy.194
Galm et al.
Scheme 3
In a related strategy, a simplified analogue of

another naturally occurring 10-membered enediyne,
17, has been employed in mAb-directed enzyme
prodrug therapy (ADEPT). This approach relies upon
directing an enzyme specifically to tumor cells via a
tumor-specific mAb. Subsequently, the enediyne prodrug is administered and locally activated at the
tumor site by the tumor-associated mAb-enzyme
conjugate, thereby greatly reducing toxicity to normal
tissue.195 Among the 10-membered enediynes 17 is
unique in that it combines structures that are
characteristic of both anthracycline and enediyne
antibiotics.180,196 As a DNA-damaging agent, 17 induces single- and double-strand breaks after reductive activation. Interaction with the DNA is established by intercalation of the anthraquinone into
DNA presenting the enediyne in the minor groove.197
Reductive activation of the anthraquinone, mediated
by physiological reductants, initiates the cascade of
events toward Bergman-cycloaromatization and ultimately DNA cleavage.196,198,199
The enediyne-ADEPT strategy employed a simplified analogue (18; Scheme 3) of 17 and the catalytic
mAb 38C2. Antibody 38C2 is a monoclonal aldolase
antibody with dual function, providing both tumor
specificity through an integrin-targeting RGD peptidomimetic and catalytic activity through aldolase
catalysis.200 As illustrated in Scheme 3, 18 serves as
a substrate for 38C2 and was specifically equipped
with a consecutive aldol and oxa-Michael-dependent
linker for enediyne activation.201
In this elegant design only the carbon-carbon
bond-cleaving retro-aldol reaction catalyzed by 38C2
was able to reveal the hidden retro-Michael substrate. Once exposed the oxa-Michael motif underwent retro-aldol and -elimination to result in the
activated, cycloaromatized drug (Scheme 3).195 While
38C2 was able to trigger prodrug activation by a
reaction cascade and the free drug inhibited tumor
cell growth in vitro,201 the immunogenicity of 38C2
remains a major limitation for the transition of this
strategy from preclinical to clinical evaluation. Yet,

the demonstration of enediyne-ADEPT technology
is highly versatile and opens new therapeutic perspectives.

The demonstrated potency of enediynes coupled
with their unique architecture and mechanism attracted great attention, which ultimately led to their
expanding clinical utility. However, clinical enediyne
use has also been accompanied by growing reports
of clinical drug resistance in enediyne-treated patients.202 Notably, the Mylotarg treatment of elderly
relapsed and secondary AML patients revealed clinical resistance to correlate with both disease resurgence and the expression of functional MDR.203
Although there has been great progress in the treatment of AML and most adult de novo AML patients
achieve complete remission (CR) with conventional
chemotherapies, the majority of responding patients
relapse and ultimately die with treatment-refractory
disease. In these cases, MDR was the leading cause
of this therapeutic failure and accounts for a poor
prognosis in AML.204,205
MDR characterizes a series of events by which
leukemia cells become resistant to various chemotherapeutic drugs that are structurally and functionally unrelated. It can result from the overexpression
of ATP-dependent efflux pumps that are members of
the ABC transporter superfamily.206 The Pgp is the
most prominent member of this family and responsible for the efflux of several chemotherapeutic drugs
currently employed for AML, including anthracyclines and etoposide. Pgp is expressed in 70% of older
patients with de novo AML, whereas only 40% of
younger subjects are typically affected. Pgp was also
shown to have a high frequency of expression in
patients with secondary and relapsed adult AML.207,208
Since the drugs that are used for chemotherapy in
AML patients are substrates of Pgp, it is assumed
that treatment with various chemotherapy regimens
induces and upregulates Pgp expression. This overexpression of Pgp in leukemia cell lines was shown
to result from drug-induced changes in mRNA stability and transcriptional activation.206 Therefore, Pgp
expression accounts for increased efflux of CAL and
is linked to lower CR, higher rates of refractory
disease, and shorter overall survival after treatment
with these chemotherapeutics. Multidrug resistance
and its impact on the treatment of AML with CALimmunoconjugates has prompted the development of
MDR modifiers that inhibit efflux of antileukemia
agents and restore the effect of chemotherapeutic
agents in resistant cell lines.208,213,214 Cyclosporine
and PSC 833 are Pgp antagonists that are used as
chemosensitizing agents in AML treatment trials.
These MDR modifiers are able to restore the cytocidal
effect of Mylotarg in Pgp-expressing cell lines.
In the context of this review it is interesting to note
that MTM C (19, vide infra) is able to suppress the
activity of Pgp. The underlying mechanism by which
this occurs, however, remains elusive.209 Taken together, these observations indicate that avoiding the
induction of Pgp expression is of paramount importance when it comes to choosing an appropriate and
successful therapeutic regimen. It also points to
reconsidering the role of mAB-CAL conjugates in the
treatment of relapsed and refractory AML since the
higher specificity and lower side effects cannot compensate for developed resistance mechanisms. Given
the lower expected frequency of MDR expression in
de novo AML, Mylotarg may show greater efficacy
in this group of patients.
Recent observations suggest that non-Pgp transporters and mechanisms other than drug efflux may
also contribute to clinical 13 resistance. These resistance mechanisms involve the multidrug-resistanceassociated proteins (MRPs) that also belong to the
ABC-transporter family and the major vault protein
LRP, a ribonucleoprotein found to be overexpressed
in many chemoresistant cancer cell lines and implicated in the sequestration of drugs.206,210 Vaults are
large-sized complexes that have an estimated molecular mass of 13 MDa. Their barrel-like structures
indicate a function in intracellular drug transport,
and they have been linked to drug resistance. However, they share no similarity with the apo-proteins
of nine-membered enediynes, the MTM C-resistance
protein Mrd (see section 5.3) or BlmA (see section
2.3).206,210 Recently, the breast-cancer-resistant protein (BCRP), the equivalent of mitoxantrone-resistant
protein or placental ABC transporter, was described
in AML and shown to play an important role in the
development of MDR.211 This unique transporter
belongs to the family of ABC transporters yet is
evolutionarily distinct from Pgp or MRPs as it
requires dimerization. The AML chemotherapeutics
mitoxantrone, daunorubicin, and etoposide are all
substrates for this transporter. Notably, BCRP mRNA
levels in patients resistant to Mylotarg that did not
achieve remission after the first chemotherapy were
found to be 10-times higher as compared to patients
who did achieve remission.212

Similar to the nine-membered enediynes, the members of this group also require activation to become
biologically active, yet unlike most nine-membered
enediynes, the 10-membered analogues do not spontaneously cycloaromatize.176 Thus, all reported 10membered enediynes lack an apo-protein, and until
recently, the predominant mechanism of self-resistance among 10-membered enediyne-producing organisms was lacking. This issue was recently addressed by using one of the best characterized 10membered enediynes, 13, in a screen as a resistance
marker. From a genomic cosmid library of the CALproducing M. echinospora expressed in E. coli, six
colonies showed growth on CAL-impregnated media
and eventually led to the discovery of a single gene
(calC) that was able to confer resistance to 13.152,215
Expression of this gene in E. coli rendered the
organism resistant to CAL at concentrations that
were 3 orders of magnitude higher than the lethal
dose for wild-type E. coli.216 This finding demonstrated the protective potential of the CalC protein
and prompted studies to explore the underlying
mechanism. The early in vivo studies also revealed
the CalC-CAL interaction to be specific since the
calC-expressing E. coli failed to grow in the presence
of either 14 or 17 yet could thrive in the presence of
15 and 16.
The key in vitro evidence pointing to an unprecedented mechanism came from gel electrophoresis
experiments that revealed cleavage of CalC by reductively activated 13 with proteolytic-like precision
to give two specific peptide fragments. Further
analyses suggested a CAL-catalyzed proteolysis
through hydrogen abstraction from Gly-113 with
direct R-hydrogen transfer from Gly-113 to the cycloaromatized 13 (CLM) observed via isotopic labeling.216 These key experiments further corroborated
the hypothesis that CalC was acting as a quencher
for the enediyne diradical in a mechanism parallel
to hydrogen abstraction from the deoxyribose of DNA
(Scheme 4). Amino acid point mutagenesis of CalC
at position 113 reduced resistance, thereby providing
additional support for the proposed CalC mechanism
and the importance of Gly-113.216
The elucidation of the CalC-mediated resistance to
13 represents a truly unprecedented and unusually
extravagant paradigm in antibiotic self-resistance.
Although it seems counterintuitive that M. echinospora would destroy its own biosynthetic product
while sacrificing the resistance protein, the catastrophic reactivity of the enediynes may dictate such
an excessive response. CalC has, to date, no significant homologues, and it is also not known if the other
producers of 10-membered enediynes employ a similar self-sacrifice mechanism. However, given CalC
was able to confer resistance to the 10-membered
enediynes 15 and 16, it is reasonable to assume the
gene clusters from Polysyncraton lithostrotum and
Didemnum proliferum could harbor calC homologues.
It also remains to be determined whether CalC
expression correlates directly with the production of
13 in M. echinospora. Finally, it is unclear if CalC
(or a CalC variant) can, at some stage, act as a
Galm et al.
Scheme 4
binding protein for 13 prior to hydrogen abstraction,

i.e., whether CalC is in fact derived from an apoprotein-like progenitor. Intriguingly, the secondary
structure prediction for CalC shows a symmetrical
pattern with a R-RR fold for the putative
monomeric protein. In light of the recently unraveled
R-R monomeric signatures of the MTM
C-resistance protein Mrd and the BLM-resistance
protein BlmA (see section 2.3), the secondary structure prediction of CalC might indicate a possible
functional conservation of the CalC structure. This
is further supported by recent CalC NMR experiments that reveal CalC contains predominantly a
-sheet secondary structure (Hager, M. H.; Hallenga,
K.; Thorson, J. S. Unpublished data).
In view of the recently established resistance
mechanism of M. echinospora, it will be interesting
to see if the producers of 10-membered enediynes all
rely solely on the strategy of providing a surrogate
substrate for detoxification or if there are other
parameters that play a role in self-resistance.152 In
the CAL producer M. echinospora the biosynthetic
gene cluster harbors several genes of putative efflux
proteins that belong to the family of ABC transporters. The gene calT5, for example, encodes a protein
with high homology to DrrA, an ATP-binding protein
which confers resistance to daunorubicin. A counterpart to calT5, orf 42, is found in the 15-producer M.
chersina (Gao, Q.; Thorson, J. S. Unpublished data),
which points to a possible role of this gene in the

efflux of 13 and 17. This is consistent with the role
of CalC homologues as a last-ditch emergency brake
and the predominant mechanism of self-resistance
in 10-membered enediyne producers via efflux transporters.
5. Mitomycin
The mitomycins (MTMs) are potent antibiotics that
belong to the family of antitumor quinones. In
contrast to BLM and the enediyne antibiotics, MTMs
do not cause DNA-backbone cleavage but rather form
covalent linkages with DNA and function as alkylating agents.217 A unique hallmark of the MTMs is
their conversion to the active drug through an
enzymatic reduction process that preferentially proceeds in the absence of oxygen.218

In 1956 mitomycin A (20) and B (21) were isolated
from Streptomyces caespitosus,219-221 and shortly after
mitomycin C (19) was found from the same strain
(Figure 4).222 The N-methyl derivative of 19, porfiromycin, was isolated in 1960 from Streptomyces ardus,
followed by the discovery of mitiromycin from Streptomyces verticillatus.223,224 Among all these different
Figure 4. Structures of mitomycin C (19) and analogues

mitomycin A (20) and mitomycin B (21).
MTMs, 19 led to early widespread clinical use given

its uniquely superior activity against solid tumors
and reduced toxicity as compared to the natural
counterparts 20 and 21.225,226
The MTMs are comprised of aziridine, quinone, and
carbamate moieties arranged in a compact pyrrolo[1,2-a] indole structure,227,228 which presents the
extraordinary ability to cross-link DNA with high
efficiency and specificity for the sequence CpG.229
Similar to the other DNA-damaging agents such as
CAL, 19 is reductively activated, which converts the
molecule from a noncyctotoxic prodrug into a shortlived and highly reactive quinone methide.230 In this
activated form the compound lacks the methoxide
substituent and aziridine ring opening exposes an
electrophilic C1, a highly reactive species termed
mitosene. Covalent DNA cross-links are then rapidly
formed via attack at both the C1 and C10 positions
(Scheme 5).231
Activation of 19 can occur chemically or via enzymatic activation by intracellular flavin reductases
and can proceed by either one- or two-electron reduction (Scheme 5).232 The one-electron reduction of 19
Scheme 5
produces the semiquinone anion radical intermediate

(23), which is regenerated to the parental compound
19 under aerobic conditions by electron transfer to
molecular oxygen, generating a superoxide radical
anion. Alternatively, a second electron can be transferred to 23 to form the aforementioned mitosene
species. Due to its potential to undergo activation
through a variety of different enzymes in aerobic and
hypoxic conditions, 19 is a natural agent for the
approach termed bioreductive chemotherapy.233 This
treatment is dependent upon tumors rich in reductive
activation proteins and, thus, is most successful in
certain cancer lines including colon, breast, lung, and
head cancer. Additionally, bioreductive chemotherapy
can be employed to selectively attack carcinomas
since it capitalizes on the differences in oxygen
content and cellular pH between normal tissue and
tumor tissue.234 In particular, hypoxic cells of solid
tumors that are resistant to most chemotherapeutic
agents create an environment which favors the
reductive processes that are needed for the activation
of 19. Therefore, 19 displays greater toxicity to
oxygen-deficient cells in comparison to their oxygenated counterpartssa quality that makes 19 a valuable antitumor drug able to attack hypoxic regions
of solid tumors.
The discovery of the alkaloid FR900482 from
Streptomyces sandaensis in 1987 revealed a structurally related mitomycinoid with a mode of action
analogous to 19.235 This representative of a new class
of antitumor agents displays markedly lower hematotoxicity,236 and its semisynthetic derivative FK317
has shown promising activity in human clinical trials.
Due to their greatly reduced toxicity and superior
DNA cross-linking activity, these compounds may
have the potential to replace 19 in the clinic.237

Despite the high efficacy of 19 in the treatment of
solid tumors, acquired or intrinsic drug resistance of
tumor-cell populations is also responsible for refractory malignant target tissue and the limited utility
of this antineoplastic drug. Several factors have been
implicated to be involved in the acquisition of MTM
C resistance in mammalian cells. These include the
deficiency of activating enzymes such as an NAD(P)H oxidoreductase, DNA repair processes, and
increased drug efflux.238 Whereas these resistance
mechanisms have also been reported for other drugs,
cells that become specifically insensitive to 19 might
also employ a distinct detoxification mechanism. In
recent years it has been hypothesized that tumor cells
become resistant to 19 by reoxidizing the reduced,
cytotoxic intermediate of 19 to the parent drug
through a redox mechanism. This assumption was
prompted by the observation that a 54-kDa flavoprotein from S. lavendulae is able to confer 19-resistance
to Chinese hamster ovary (CHO) cells.239 This protein
is called MCRA (mitomycin C resistance associated)
and renders the 19 producer insensitive to high
concentrations of its own toxic product.240 In S.
lavendulae MCRA acts as a hydroquinone oxidase
and protects DNA from cross-linking by oxidizing the
toxic 19 hydroquinone (22, Scheme 5). That this
mechanism is oxygen dependent in S. lavendulae is
in accordance with the observation that expression
of MCRA in CHO confers profound resistance only
under aerobic conditions but not under hypoxia.239
Nontransfected 19-resistant mammalian cell lines
that were developed by stepwise exposure to increasing 19 concentrations also showed 19 resistance only
in the presence of oxygen. Furthermore, overexpression of 19-activating enzymes such as the NAD(P)H
oxidoreductase restored 19 sensitivity in both transfected and drug-selected cell lines.241 All these observations lend further support to the idea that 19
resistance in tumor cells not only is derived from
increased drug efflux, DNA repair, or differential
rates of reduction/activation, but also involves an
oxygen-dependent resistance mechanism possibly
analogous to that established by MCRA.242 Unlike the
MTMs and, in particular, 19, clinical trials for the
FR900482 class are in process, and specific biological
implications with respect to resistance to these
antitumor drugs should be available soon.

To protect itself from the harmful effects of 19, S.
lavendulae developed several mechanisms to ensure
self-resistance. Three genes have been identified so
far to be involved in the cellular protection against
19 in this MTM-producing organism. The previously
mentioned mcrA gene is located outside of the MTM
C biosynthetic gene cluster and encodes a flavoprotein (MCRA) which is able to convey MTM C resistance upon the heterologous host S. lividans.240 In
addition to MCRA-based MTM C regeneration, a
second gene (mrd) was located within the MTM C
gene cluster.83 It encodes a small soluble protein
(Mrd)84 that is able to confer 30-fold enhanced MTM
C resistance if heterologously expressed in E. coli.
Galm et al.
Further characterization revealed that Mrd was able

to reversibly bind and sequester 19 with no observable antibiotic modification.243 However, Mrd was
shown to require NADH to exert its protective drugbinding function. This NADH requirement led to the
discovery that Mrd can, in fact, act as a weak
activator of 19 through the Mrd-dependent generation of 1,2-cis-1-hydroxy-2,7-diaminomitosene, a compound that is produced in the reductive 19 activation
cascade.243 The finding of a 19-activating activity
appeared to be conflicting with the previously identified protective effect of Mrd. However, the reductive
reaction catalyzed by Mrd is slow and results in a
prolonged association of 19 and its corresponding
reduced product with the protein. Therefore, reduction represents a prerequisite for binding of 19 to Mrd
and rapid removal of the drug through a specific
transport protein Mct (encoded by the third resistance gene). Mct displays extensive amino acid
sequence similar to several antibiotic-exporting proteins. Heterologous expression of both mct and mrd
in E. coli indicated that both proteins together form
an efficient drug-binding export system.244
As it was mentioned in section 2.3, the Mrdresistance protein features on the structural level an
interesting monomeric tandem R motif.84 Tandem R motifs are characteristic of a large family
comprising several other proteins with diverse functions such as the BLM-resistance protein from S.
verticillus (BlmA) and the methymalonyl-coenzyme
A epimerase from Propionibacterium shermanii. In
the context of this review, it is important to note that
despite the remarkable structural differences between their ligands, Mrd and BlmA share substantial
structural similarity and may implicate the R
motif as a signature for the general drug-sequestering-resistance paradigm. It remains to be determined
whether there is a relationship between this structural family of sequestering proteins and the -sheetrich enediyne apo-proteins and/or CalC.
6. Perspective
In the context of the DNA-damaging agents discussed (BLMs, enediynes, and MTMs), one can derive
a fairly unique distinction among the mammalian
mechanisms of resistance in contrast to the mechanisms employed by the drug-producing prokaryotic
counterparts. For example, eukaryotic BLM resistance relies upon BLM hydrolase and DNA repair
enzymes, while these higher organisms are devoid
of the predominant prokaryotic-resistance components (BLM-binding proteins, acetyltransferases, and
BLM-specific transporters). In a similar fashion,
MDR appears to be the predominant mammalian
resistance mechanism for enediynes, while sequestration, self-sacrifice, and, to a lesser extent, efflux
lend to the self-preservation of enediyne-producing
prokaryotes. The case for the MTMs is perhaps less
clear as the often observed phenomenon of aerobic
drug resistance in human cancer cell lines could be
attributed to an MCRA-analogous reoxidization process. Such correlations may lead to new perspectives
in terms of drug development. For example, naturally
inactivated acetyl-BLMs from the producing organ-
ism may serve as reasonable prodrugs which, upon

tumor-cell-dependent deacetylation, present the active DNA-cleavage agents. Alternatively, the naturally occurring prokaryotic cytotoxin-binding proteins
(e.g., BlmA, Mrd, enediyne apo-proteins) could be
rationally engineered to display tumor-targeting elements, and these complexes subsequently could be
produced via direct fermentation, thereby enhancing
tumor specificity of these highly reactive drugs and
eliminating the current need for tedious mAbconjugation strategies. In addition, as we continue
to learn more about the specific mechanisms of tumor
resistance to these agents, sensitizing agents (e.g.,
BLM hydrolase inhibitors, DNA repair inhibitors,
MDR inhibitors, MTM oxidative regeneration inhibitors) can be specifically incorporated into chemotherapeutic regimens to enhance efficacy. Armed
with this information, physicians may also begin to
profile patients to individually tailor chemotherapeutic treatment. For example, although the exact
mechanism is not clear, a receptor protein appears
to exist on the plasma membrane of mammalian and
yeast cells that may mediate BLM internalization.245-247 Increasing the receptor production in
tumor cells or patient profiling for overproduction of
this putative BLM-receptor gene may assist in a
predetermination for patients most amenable to BLM
chemotherapy.24 As gene therapy comes of age, one
might even imagine delivering specific resistance
proteins (e.g., CalC) to normal tissues to serve as
chemoprotective agents.
Within the drug-producing organisms, understanding the mechanisms of resistance, efflux, and regulation is critical to the continued success of these
reagents. For example, overexpression of genes encoding drug binding and/or efflux proteins and/or
simple affinity labeling of the inherent binding
proteins may radically enhance production levels and/
or purification strategies. Pathway engineering toward novel variants is also at the mercy of resistance
and efflux elements. Moreover, a molecular-level
understanding of the drug-binding protein interactions is essential to (i) fine tune intracellular release
of the parent drugs within a tumor (e.g., as in the
case of chromoprotein enediynes), (ii) potentially
engineer de novo drug carriers (e.g., proteins capable
of accommodating drugs other than the natural
metabolite),248,249 (iii) rationally incorporate tumordirecting elements, and (iv) possibly understand the
subtle distinctions and/or evolution between catalysis
and binding (e.g., as in the case of Mrd or possibly
even CalC). Such projected research areas will continue to present exciting challenges of significant
therapeutic value for years to come.
7. Abbreviations
ABC
ADEPT
AHM
AML
AP
BCRP
BLM
CAL
CHO
ATP-binding cassette
mAb-directed enzyme prodrug therapy
4-amino-3-hydroxy-2-methylpentanoic acid
acute myeloid leukemia
apurinic/apyrimidinic
breast-cancer-resistant protein
bleomycin
calicheamicin
Chinese hamster ovary

CR
mAB
MCRA
MDR
MRP
MRSA
MTM
NCS
PBA
PEM
Pgp
PLM
TLM
complete remission
monoclonal antibody
mitomycin C resistance associated protein
multi-drug resistance
multi-drug-resistance-associated proteins
methicillin-resistant S. aureus
mitomycin
neocarzinostatin
pyrimidoblamic acid
polymorphic epithelial mucin
P-glycoprotein transporters
phleomycin
tallysomycin
8. Acknowledgment
Bleomycin investigations were supported in part
by the Searle Scholars Program/Chicago Community
Trust and by the National Institutes of Health
(AI40475 and CA94426). The nine-membered enediyne work is supported in part by the National
Institutes of Health (CA78747). The 10-membered
enediyne work was supported in part by the National
Institutes of Health (CA84374 and AI52218), the
Wisconsin Alumni Research Foundation Romnes
Fellowship, a Rita Allen Scholar Award (J.S.T.), and
an Alfred P. Sloan Foundation Fellowship (J.S.T.).
We gratefully acknowledge a postdoctoral fellowship
to U.G. from the Deutsche Forschungsgemeinschaft.
M.H.H. is a Postdoctoral Fellow of the German
Academy of Scientists Leopoldina (BMBF-LPD 9901/
8-82). S.V.L is supported by National Institutes of
Health postdoctoral fellowship F32 CA105984. B.S.
is the recipient of the National Science Foundation
CAREER Award (MCB9733938) and National Institutes of Health Independent Scientist Award
(AI51687).
9. References
(1) Newman, D. J.; Cragg, G. M.; Snader, K. M. J. Nat. Prod. 2003,
66, 1022.
(2) Umezawa, H.; Maeda, K.; Takeuchi, T.; Okami, Y. J. Antibiot.
1966, 19, 200.
(3) Takita, T.; Muraoka, Y.; Yoshioka, T.; Fujii, A.; Maeda, K. J.
Antibiot. 1972, 25, 755.
(4) Takita, T.; Muraoka, Y.; Nakatani, T.; Fujii, A.; Umezawa, Y.;
Naganawa, H.; Umezawa, H. J. Antibiot. 1978, 31, 801.
(5) Takita, T.; Umezawa, Y.; Saito, S. I.; Morishima, H.; Naganawa,
H.; Umezawa, H.; Tsuchiya, T.; Miyake, T.; Kageyama, S.;
Umezawa, S.; Muraoka, Y.; Suzuki, M.; Otsuka, M.; Narita, M.;
Kobayashi, S.; Ohno, M. Tetrahedron Lett. 1982, 23, 521.
(6) Aoyagi, Y.; Katano, K.; Suguna, H.; Primeau, J.; Chang, L. H.;
Hecht, S. M. J. Am. Chem. Soc. 1982, 104, 5537.
(7) Maeda, K.; Kosaka, H.; Yagashita, K.; Umezawa, H. J. Antibiot.
1956, 9, 82.
(8) Takita, T.; Maeda, K.; Umezawa, H. J. Antibiot. 1959, 12, 111.
(9) Takita, T. J. Antibiot. 1959, 111, 285.
(10) Hamamichi, N.; Hecht, S. M. J. Am. Chem. Soc. 1993, 115,
12605.
(11) Kawaguchi, H.; Tsukiura, H.; Tomita, K.; Konishi, M.; Saito, K.;
Kobaru, S.; Numata, K.; Fujisawa, K.; Miyaki, T.; Hatori, M.;
Koshiyama, H. J. Antibiot. 1977, 30, 779.
(12) Konishi, M.; Saito, K.; Numata, K.; Tsuno, T.; Asama, K.;
Tsukiura, H.; Naito, T.; Kawaguchi, H. J. Antibiot. 1977, 30,
789.
(13) Boger, D. L.; Cai, H. Angew. Chem., Int. Ed. Engl. 1999, 38, 448.
(14) Hecht, S. M. J. Nat. Prod. 2000, 63, 158.
(15) Stubbe, J.; Kozarich, J. W. Chem. Rev. 1987, 87, 1107.
(16) Burger, R. M. Chem. Rev. 1998, 98, 1153.
(17) Leitheiser, C. J.; Rishel, M. J.; Wu, X.; Hecht, S. M. Org. Lett.
2000, 2, 3397.
(18) Sikic, B. I.; Rosenzweig, M.; Chater, S. K. Bleomycin chemotherapy; Academic Press: New York, 1985.
(19) Akiyama, S.; Ikezaki, K.; Kuramochi, H.; Takahashi, K.; Kuwano, M. Biochem. Biophys. Res. Commun. 1981, 101, 55.

(20) Miyaki, M.; Ono, T.; Hori, S.; Umezawa, H. Cancer Res. 1975,
35, 2015.
(21) Morris, G.; Mistry, J. S.; Jani, J. P.; Mignano, J. E.; Sebti, S.
M.; Lazo, J. S. Mol. Pharmacol. 1992, 42, 57.
(22) Sebti, S. M.; Jani, J. P.; Mistry, J. S.; Gorelik, E.; Lazo, J. S.
Cancer Res. 1991, 51, 227.
(23) Urade, M.; Ogura, T.; Mima, T.; Matsuya, T. Cancer 1992, 69,
2589.
(24) Ramotar, D.; Wang, H. Curr. Genet. 2003, 43, 213.
(25) Semon, D.; Movva, N. R.; Smith, T. F.; el Alama, M.; Davies, J.
Plasmid 1987, 17, 46.
(26) Genilloud, O.; Garrido, M. C.; Moreno, F. Gene 1984, 32, 225.
(27) Mazodier, P.; Cossart, P.; Giraud, E.; Gasser, F. Nucleic Acids
Res. 1985, 13, 195.
(28) Sugiyama, M.; Kumagai, T.; Matsuo, H.; Bhuiyan, M. Z.; Ueda,
K.; Mochizuki, H.; Nakamura, N.; Davies, J. E. FEBS Lett. 1995,
362, 80.
(29) Lazo, J. S.; Humphreys, C. J. Proc. Natl. Acad. Sci. U.S.A. 1983,
80, 3064.
(30) Nishimura, C.; Suzuki, H.; Tanaka, N.; Yamaguchi, H. Biochem.
Biophys. Res. Commun. 1989, 163, 788.
(31) Sebti, S. M.; Lazo, J. S. Pharmacol. Ther. 1988, 38, 321.
(32) Umezawa, H.; Hori, S.; Sawa, T.; Yoshioka, T.; Takeuchi, T. J.
Antibiot. 1974, 27, 419.
(33) Jani, J. P.; Mistry, J. S.; Morris, G.; Lazo, J. S.; Sebti, S. M.
Oncol. Res. 1992, 4, 59.
(34) Bromme, D.; Rossi, A. B.; Smeekens, S. P.; Anderson, D. C.;
Payan, D. G. Biochemistry 1996, 35, 6706.
(35) Enenkel, C.; Wolf, D. H. J. Biol. Chem. 1993, 268, 7036.
(36) Magdolen, U.; Muller, G.; Magdolen, V.; Bandlow, W. Biochim.
Biophys. Acta 1993, 1171, 299.
(37) OFarrell, P. A.; Gonzalez, F.; Zheng, W.; Johnston, S. A.; JoshuaTor, L. Struct. Fold. Des. 1999, 7, 619.
(38) Joshua-Tor, L.; Xu, H. E.; Johnston, S. A.; Rees, D. C. Science
1995, 269, 945.
(39) Zheng, W.; Johnston, S. A.; Joshua-Tor, L. Cell 1998, 93, 103.
(40) Pei, Z.; Calmels, T. P.; Creutz, C. E.; Sebti, S. M. Mol. Pharmacol.
1995, 48, 676.
(41) Kambouris, N. G.; Burke, D. J.; Creutz, C. E. J. Biol. Chem.
1992, 267, 21570.
(42) Wang, H.; Ramotar, D. Biochem. Cell Biol. 2002, 80, 789.
(43) Masson, J. Y.; Ramotar, D. Mol. Cell Biol. 1996, 16, 2091.
(44) Xu, H. E.; Johnston, S. A. J. Biol. Chem. 1994, 269, 21177.
(45) Zheng, W.; Xu, H. E.; Johnston, S. A. J. Biol. Chem. 1997, 272,
30350.
(46) Koldamova, R. P.; Lefterov, I. M.; DiSabella, M. T.; Almonte,
C.; Watkins, S. C.; Lazo, J. S. Biochemistry 1999, 38, 7111.
(47) Schwartz, D. R.; Homanics, G. E.; Hoyt, D. G.; Klein, E.;
Abernethy, J.; Lazo, J. S. Proc. Natl. Acad. Sci. U.S.A. 1999,
96, 4680.
(48) Popoff, S. C.; Spira, A. I.; Johnson, A. W.; Demple, B. Proc. Natl.
Acad. Sci. U.S.A. 1990, 87, 4193.
(49) Ramotar, D.; Popoff, S. C.; Gralla, E. B.; Demple, B. Mol. Cell
Biol. 1991, 11, 4537.
(50) Demple, B.; Johnson, A.; Fung, D. Proc. Natl. Acad. Sci. U.S.A.
1986, 83, 7731.
(51) Levin, J. D.; Johnson, A. W.; Demple, B. J. Biol. Chem. 1988,
263, 8066.
(52) Ramotar, D.; Popoff, S. C.; Demple, B. Mol. Microbiol. 1991, 5,
149.
(53) Jilani, A.; Ramotar, D.; Slack, C.; Ong, C.; Yang, X. M.; Scherer,
S. W.; Lasko, D. D. J. Biol. Chem. 1999, 274, 24176.
(54) Vance, J. R.; Wilson, T. E. J. Biol. Chem. 2001, 276, 15073.
(55) Robertson, K. A.; Bullock, H. A.; Xu, Y.; Tritt, R.; Zimmerman,
E.; Ulbright, T. M.; Foster, R. S.; Einhorn, L. H.; Kelley, M. R.
Cancer Res. 2001, 61, 2220.
(56) Abe, H.; Wada, M.; Kohno, K.; Kuwano, M. Anticancer Res. 1994,
14, 1807.
(57) Keszenman, D. J.; Salvo, V. A.; Nunes, E. J. Bacteriol. 1992,
174, 3125.
(58) He, C. H.; Masson, J. Y.; Ramotar, D. Can. J. Microbiol. 1996,
42, 1263.
(59) Jentsch, S.; McGrath, J. P.; Varshavsky, A. Nature 1987, 329,
131.
(60) Madura, K.; Prakash, S.; Prakash, L. Nucleic Acids Res. 1990,
18, 771.
(61) Resnick, M. A.; Martin, P. Mol. Gen. Genet. 1976, 143, 119.
(62) Kumagai, T.; Nakano, T.; Maruyama, M.; Mochizuki, H.; Sugiyama, M. FEBS Lett. 1999, 442, 34.
(63) Leduc, A.; He, C. H.; Ramotar, D. Mol. Genet. Genomics 2003,
269, 78.
(64) Faik, A.; Laboure, A. M.; Gulino, D.; Mandaron, P.; Falconet,
D. Eur. J. Biochem. 1998, 253, 552.
(65) Mayer, U. J. Biol. Chem. 2003, 278, 14587.
(66) Philip, B.; Levin, D. E. Mol. Cell Biol. 2001, 21, 271.
(67) Lodi, T.; Goffrini, P.; Ferrero, I.; Donnini, C. Microbiology 1995,
141 (Pt 9), 2201.
(68) Alarco, A. M.; Balan, I.; Talibi, D.; Mainville, N.; Raymond, M.
J. Biol. Chem. 1997, 272, 19304.
Galm et al.
(69) Coleman, S. T.; Tseng, E.; Moye-Rowley, W. S. J. Biol. Chem.
1997, 272, 23224.
(70) Decottignies, A.; Lambert, L.; Catty, P.; Degand, H.; Epping, E.
A.; Moye-Rowley, W. S.; Balzi, E.; Goffeau, A. J. Biol. Chem.
1995, 270, 18150.
(71) Katzmann, D. J.; Hallstrom, T. C.; Mahe, Y.; Moye-Rowley, W.
S. J. Biol. Chem. 1996, 271, 23049.
(72) Katzmann, D. J.; Epping, E. A.; Moye-Rowley, W. S. Mol. Cell
Biol. 1999, 19, 2998.
(73) Oskouian, B.; Saba, J. D. Mol. Gen. Genet. 1999, 261, 346.
(74) Wemmie, J. A.; Moye-Rowley, W. S. Mol. Microbiol. 1997, 25,
683.
(75) Sugiyama, M.; Thompson, C. J.; Kumagai, T.; Suzuki, K.;
Deblaere, R.; Villarroel, R.; Davies, J. Gene 1994, 151, 11.
(76) Calcutt, M. J.; Schmidt, F. J. Gene 1994, 151, 17.
(77) Kumagai, T.; Hibino, R.; Kawano, Y.; Sugiyama, M. FEBS Lett.
1999, 450, 227.
(78) Kawano, Y.; Kumagai, T.; Muta, K.; Matoba, Y.; Davies, J.;
Sugiyama, M. J. Mol. Biol. 2000, 295, 915.
(79) Sugiyama, M.; Kumagai, T.; Shionoya, M.; Kimura, E.; Davies,
J. E. FEMS Microbiol. Lett. 1994, 121, 81.
(80) Matsuo, H.; Mochizuki, H.; Davies, J.; Sugiyama, M. FEMS
(81) Sugiyama, M.; Kumagai, T.; Hayashida, M.; Maruyama, M.;
Matoba, Y. J. Biol. Chem. 2002, 277, 2311.
(82) Gatignol, A.; Durand, H.; Tiraby, G. FEBS Lett. 1988, 230, 171.
(83) Sheldon, P. J.; Johnson, D. A.; August, P. R.; Liu, H. W.;
Sherman, D. H. J. Bacteriol. 1997, 179, 1796.
(84) Martin, T. W.; Dauter, Z.; Devedjiev, Y.; Sheffield, P.; Jelen, F.;
He, M.; Sherman, D. H.; Otlewski, J.; Derewenda, Z. S.;
Derewenda, U. Structure (Cambridge) 2002, 10, 933.
(85) Maruyama, M.; Kumagai, T.; Matoba, Y.; Hayashida, M.; Fujii,
T.; Hata, Y.; Sugiyama, M. J. Biol. Chem. 2001, 276, 9992.
(86) Matoba, Y.; Danshiitsoodol, N.; Kumagai, T.; Sugiyama, M. 13th
International Symposium on the Biology of Actinomycetes, Melbourne, Australia, Dec 1-5, 2003.
(87) Mendez, C.; Salas, J. A. FEMS Microbiol. Lett. 1998, 158, 1.
(88) Du, L.; Sanchez, C.; Chen, M.; Edwards, D. J.; Shen, B. Chem.
Biol. 2000, 7, 623.
(89) Sidorova, T. A.; Nigmatov, A. G.; Kakpakova, E. S.; Stavrovskaya,
A. A.; Gerassimova, G. K.; Shtil, A. A.; Serebryakov, E. P. J.
Med. Chem. 2002, 45, 5330.
(90) Thorson, J. S.; Shen, B.; Whitwam, R. E.; Liu, W.; Li, Y.; Ahlert,
J. Bioorg. Chem. 1999, 27, 172.
(91) Xi, Z.; Goldberg, I. H. Comprehensive Natural Products Chemistry; Elsevier: New York, 1999; Chapter 7, p 533.
(92) Edo, K.; Mizugaki, M.; Koide, Y.; Seto, H.; Furihata, K.; Otake,
N.; Ishida, N. Tetrahedron Lett. 1985, 26, 331.
(93) Edo, K.; Saito, K.; Matsuda, Y.; Akiyamamurai, Y.; Mizugaki,
M.; Koide, Y.; Ishida, N. Chem. Pharm. Bull. 1991, 39, 170.
(94) Myers, A. G.; Proteau, P. J.; Handel, T. M. J. Am. Chem. Soc.
1988, 110, 7212.
(95) Koide, Y.; Ishii, F.; Hasuda, K.; Koyama, Y.; Edo, K.; Katamine,
S.; Kitame, F.; Ishida, N. J. Antibiot. 1980, 33, 342.
(96) Kawata, S.; Ashizawa, S.; Hirama, M. J. Am. Chem. Soc. 1997,
119, 12012.
(97) Leet, J. E.; Schroeder, D. R.; Hofstead, S. J.; Golik, J.; Colson,
K. L.; Huang, S.; Klohr, S. E.; Doyle, T. W.; Matson, J. A. J.
Am. Chem. Soc. 1992, 114, 7946.
(98) Otani, T.; Minami, Y.; Sakawa, K.; Yoshida, K. J. Antibiot. 1991,
44, 564.
(99) Schroeder, D. R.; Colson, K. L.; Klohr, S. E.; Zein, N.; Langley,
D. R.; Lee, M. S.; Matson, J. A.; Doyle, T. W. J. Am. Chem. Soc.
1994, 116, 9351.
(100) Ando, T.; Ishii, M.; Kajiura, T.; Kameyama, T.; Miwa, K.;
Sugiura, Y. Tetrahedron Lett. 1998, 39, 6495.
(101) Khokhlov, A. S.; Cherches, B. Z.; Reshetov, P. D.; Smirnova, G.
M.; Sorokina, I. B. J. Antibiot. 1969, 22, 541.
(102) Khokhlov, A. S.; Reshetov, P. D.; Chupova, L. A.; Cherches, B.
Z.; Zhigis, L. S.; Stoyachenko, I. A. J. Antibiot. 1976, 29, 1026.
(103) Yamaguch, T.; Furumai, T.; Sato, M.; Okuda, T.; Ishida, N. J.
Antibiot. 1970, 23, 369.
(104) Chimura, H.; Ishizuka, M.; Hamada, M.; Hori, S.; Kimura, K.
J. Antibiot. 1968, 21, 44.
(105) Hidaka, T.; Yano, Y.; Yamashita, T.; Watanabe, K. J. Antibiot.
1979, 32, 340.
(106) Yamashita, T.; Naoi, N.; Watanabe, K.; Takeuchi, T.; Umezawa,
H. J. Antibiot. 1976, 29, 415.
(107) Komiyama, K.; Umezawa, I. J. Antibiot. 1978, 31, 473.
(108) Okamoto, M.; Komiyama, K.; Takeshima, H.; Yamamoto, H.;
Umezawa, I. J. Antibiot. 1979, 32, 386.
(109) Zazopoulos, E.; Huang, K. X.; Staffa, A.; Liu, W.; Bachmann, B.
O.; Nonaka, K.; Ahlert, J.; Thorson, J. S.; Shen, B.; Farnet, C.
M. Nat. Biotechnol. 2003, 21, 187.
(110) Zhen, Y.; Ming, X.; Yu, B.; Otani, T.; Saito, H.; Yamada, Y. J.
Antibiot. 1989, 42, 1294.
(111) Sugimoto, Y.; Otani, T.; Oie, S.; Wierzba, K.; Yamada, Y. J.
Antibiot. 1990, 43, 417.
(112) Thorson, J. S.; Sievers, E. L.; Ahlert, J.; Shepard, E.; Whitwam,
R. E.; Onwueme, K. C.; Ruppen, M. Curr. Pharm. Des. 2000, 6,
1841.
(113) Maeda, H. Enediyne Antibiotics as Antitumor Agents; Marcel
Dekker: New York, 1995; p 363.
(114) Sievers, E. L.; Appelbaum, F. R.; Spielberger, R. T.; Forman, S.
J.; Flowers, D.; Smith, F. O.; Shannon-Dorcy, K.; Berger, M. S.;
Bernstein, I. D. Blood 1999, 93, 3678.
(115) Li, J. Z.; Jiang, M.; Xue, Y. C.; Zhen, Y. S. Yao Xue. Xue. Bao.
1993, 28, 260.
(116) Li, J.; Zhen, Y.; Yang, Z. Zhongguo Yi. Xue. Ke. Xue. Yuan Xue.
Bao. 1994, 16, 328.
(117) Shao, R. G.; Zhen, Y. S. Yao Xue. Xue. Bao. 1992, 27, 486.
(118) Brukner, I. Curr. Opin. Oncol. Endocr. Met. Invest. Drugs 2000,
2, 344.
(119) Totsuka, R.; Aizawa, Y.; Uesugi, M.; Okuno, Y.; Matsumoto, T.;
Sugiura, Y. Biochem. Biophys. Res. Commun. 1995, 208, 168.
(120) Dedon, P. C.; Goldberg, I. H. J. Biol. Chem. 1990, 265, 14713.
(121) Dedon, P. C.; Goldberg, I. H. Chem. Res. Toxicol. 1992, 5, 311.
(122) Xu, Y. J.; Zhen, Y. S.; Goldberg, I. H. Biochemistry 1994, 33,
5947.
(123) Yoshida, K.; Minami, Y.; Azuma, R.; Saeki, M.; Otani, T.
(124) Zein, N.; Schroeder, D. R. Adv. DNA Seq.-Spec. Agents 1998, 3,
201.
(125) Sakata, N.; Kanbe, T.; Tanabe, M.; Hayashi, H.; Hori, M.; Hotta,
K.; Hamada, M. J. Antibiot. 1989, 42, 1704.
(126) Sakata, N.; Ikeno, S.; Hori, M.; Hamada, M.; Otani, T. Biosci.,
Biotechnol., Biochem. 1992, 56, 1592.
(127) Iida, K.; Hirama, M. J. Am. Chem. Soc. 1995, 117, 8875.
(128) Hirama, M.; Akiyama, K.; Tanaka, T.; Noda, T.; Iida, K.; Sato,
I.; Hanaishi, R.; Fukuda-Ishisaka, S.; Ishiguro, M.; Otani, T.;
Leet, J. E. J. Am. Chem. Soc. 2000, 122, 720.
(129) Usuki, T.; Inoue, M.; Akiyama, K.; Hirama, M. Chem. Lett. 2002,
1148.
(130) Gibson, B. W.; Herlihy, W. C.; Samy, T. S.; Hahm, K. S.; Maeda,
H.; Meienhofer, J.; Biemann, K. J. Biol. Chem. 1984, 259, 10801.
(131) Maeda, H.; Glaser, C. B.; Kuromizu, K.; Meienhofer, J. Arch.
Biochem. Biophys. 1974, 164, 379.
(132) Hofstead, S. J.; Matson, J. A.; Malacko, A. R.; Marquardt, H. J.
Antibiot. 1992, 45, 1250.
(133) Otani, T.; Yasuhara, T.; Minami, Y.; Shimazu, T.; Zhang, R.;
Xie, M. Y. Agric. Biol. Chem. 1991, 55, 407.
(134) Samy, T. S.; Hahm, K. S.; Modest, E. J.; Lampman, G. W.;
Keutmann, H. T.; Umezawa, H.; Herlihy, W. C.; Gibson, B. W.;
Carr, S. A.; Biemann, K. J. Biol. Chem. 1983, 258, 183.
(135) Zein, N.; Solomon, W.; Colson, K. L.; Schroeder, D. R. Biochemistry 1995, 34, 11591.
(136) Izadi-Pruneyre, N.; Quiniou, E.; Blouquit, Y.; Perez, J.; Minard,
P.; Desmadril, M.; Mispelter, J.; Adjadj, E. Protein Sci. 2001,
10, 2228.
(137) Kwon, Y.; Xi, Z.; Kappen, L. S.; Goldberg, I. H.; Gao, X.
(138) Okuno, Y.; Iwashita, T.; Sugiura, Y. J. Am. Chem. Soc. 2000,
122, 6848.
(139) Okuno, Y.; Otsuka, M.; Sugiura, Y. J. Med. Chem. 1994, 37,
2266.
(140) Urbaniak, M. D.; Muskett, F. W.; Finucane, M. D.; Caddick, S.;
Woolfson, D. N. Biochemistry 2002, 41, 11731.
(141) Chen, Y.; Shao, R.; Bartlam, M.; Li, J.; Jin, L.; Gao, Y.; Liu, Y.;
Tang, H.; Zhen, Y.; Rao, Z. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 2002, 58, 173.
(142) Pletnev, V. Z.; Kuzin, A. P.; Malinina, L. V. Bioorg. Khim. 1982,
8, 1637.
(143) Sieker, L. C.; Samy, T. S. A. J. Mol. Biol. 1984, 174, 739.
(144) Usuki, T.; Inoue, M.; Hirama, M.; Tanaka, T. J. Am. Chem. Soc.
2004, 126, 3022.
(145) Valerio-Lepiniec, M.; Nicaise, M.; Adjadj, E.; Minard, P.; Desmadril, M. Protein Eng. 2002, 15, 861.
(146) Heyd, B.; Lerat, G.; Adjadj, E.; Minard, P.; Desmadril, M. J.
Bacteriol. 2000, 182, 1812.
(147) Nozaki, S.; Tomioka, Y.; Hishinuma, T.; Inoue, M.; Nagumo, Y.;
Tsuruta, L.; Hayashi, K.; Matsumoto, T.; Kato, Y.; Ishiwata, S.;
Itoh, K.; Suzuki, T.; Hirama, M.; Mizugaki, M. J. Biochem.
(Tokyo) 2002, 131, 729.
(148) Sakata, N.; Tsuchiya, K. S.; Moriya, Y.; Hayashi, H.; Hori, M.;
Otani, T.; Nagai, M.; Aoyagi, T. J. Antibiot. 1992, 45, 113.
(149) Zaheer, A.; Zaheer, S.; Montgomery, R. J. Biol. Chem. 1985, 260,
1787.
(150) Zein, N.; Casazza, A. M.; Doyle, T. W.; Leet, J. E.; Schroeder, D.
R.; Solomon, W.; Nadler, S. G. Proc. Natl. Acad. Sci. U.S.A. 1993,
90, 8009.
(151) Zein, N.; Reiss, P.; Bernatowicz, M.; Bolgar, M. Chem. Biol. 1995,
2, 451.
(152) Ahlert, J.; Shepard, E.; Lomovskaya, N.; Zazopoulos, E.; Staffa,
A.; Bachmann, B. O.; Huang, K.; Fonstein, L.; Czisny, A.;
Whitwam, R. E.; Farnet, C. M.; Thorson, J. S. Science 2002, 297,
1173.

(153) Liu, W.; Christenson, S. D.; Standage, S.; Shen, B. Science 2002,
297, 1170.
(154) Lage, C.; de Padula, M.; de Alencar, T. A. M.; Goncalves, S. R.
D.; Vidal, L. D.; Cabral-Neto, J.; Leitao, A. C. Mutat. Res. 2003,
544, 143.
(155) Kacinski, B. M.; Rupp, W. D. Cancer Res. 1984, 44, 3489.
(156) Furuya, K.; Hutchinson, C. R. FEMS Microbiol. Lett. 1998, 168,
243.
(157) Lomovskaya, N.; Hong, S. K.; Kim, S. U.; Fonstein, L.; Furuya,
K.; Hutchinson, R. J. Bacteriol. 1996, 178, 3238.
(158) Tercero, J. A.; Espinosa, J. C.; Lacalle, R. A.; Jimenez, A. J. Biol.
Chem. 1996, 271, 1579.
(159) Tercero, J. A.; Lacalle, R. A.; Jimenez, A. Eur. J. Biochem. 1993,
218, 963.
(160) Tseng, T. T.; Gratwick, K. S.; Kollman, J.; Park, D.; Nies, D. H.;
Goffeau, A.; Saier, M. H., Jr. J. Mol. Microbiol. Biotechnol. 1999,
1, 107.
(161) Elkins, C. A.; Nikaido, H. J. Bacteriol. 2002, 184, 6490.
(162) Aono, R. Extremophiles 1998, 2, 239.
(163) Lamont, I. L.; Beare, P. A.; Ochsner, U.; Vasil, A. I.; Vasil, M.
(164) McMorran, B. J.; Merriman, M. E.; Rombel, I. T.; Lamont, I. L.
Gene 1996, 176, 55.
(165) Lee, M. D.; Dunne, T. S.; Siegel, M. M.; Chang, C. C.; Morton,
G. O.; Borders, D. B. J. Am. Chem. Soc. 1987, 109, 3464.
(166) Lee, M. D.; Dunne, T. S.; Chang, C. C.; Ellestad, G. A.; Siegel,
M. M.; Morton, G. O.; Mcgahren, W. J.; Borders, D. B. J. Am.
Chem. Soc. 1987, 109, 3466.
(167) Konishi, M.; Ohkuma, H.; Saitoh, K.; Kawaguchi, H.; Golik, J.;
Dubay, G.; Groenewold, G.; Krishnan, B.; Doyle, T. W. J.
Antibiot. 1985, 38, 1605.
(168) McDonald, L. A.; Capson, T. L.; Krishnamurthy, G.; Ding, W.
D.; Ellestad, G. A.; Bernan, V. S.; Maiese, W. M.; Lassota, P.;
Discafani, C.; Kramer, R. A.; Ireland, C. M. J. Am. Chem. Soc.
1996, 118, 10898.
(169) Oku, N.; Matsunaga, S.; Fusetani, N. J. Am. Chem. Soc. 2003,
125, 2044.
(170) Konishi, M.; Ohkuma, H.; Tsuno, T.; Oki, T.; Vanduyne, G. D.;
Clardy, J. J. Am. Chem. Soc. 1990, 112, 3715.
(171) Konishi, M.; Ohkuma, H.; Matsumoto, K.; Saitoh, K.; Miyaki,
T.; Oki, T.; Kawaguchi, H. J. Antibiot. 1991, 44, 1300.
(172) Shiraki, T.; Sugiura, Y. Nucleic Acids Symp. Ser. 1989, 21, 53.
(173) Devoss, J. J.; Hangeland, J. J.; Townsend, C. A. J. Am. Chem.
Soc. 1990, 112, 4554.
(174) Myers, A. G.; Cohen, S. B.; Kwon, B. M. J. Am. Chem. Soc. 1994,
116, 1255.
(175) Myers, A. G.; Kort, M. E.; Cohen, S. B.; Tom, N. J. Biochemistry
1997, 36, 3903.
(176) Nicolaou, K. C.; Smith, A. L.; Yue, E. W. Proc. Natl. Acad. Sci.
U.S.A. 1993, 90, 5881.
(177) Trail, P. A.; King, H. D.; Dubowchik, G. M. Cancer Immunol.
Immunother. 2003, 52, 328.
(178) Goldenberg, D. M. CA Cancer J. Clin. 1994, 44, 43.
(179) Jackson, S. P. Biochem. Soc. Trans. 2001, 29, 655.
(180) Nicolaou, K. C.; Dai, W. M.; Tsay, S. C.; Estevez, V. A.; Wrasidlo,
W. Science 1992, 256, 1172.
(181) Countouriotis, A.; Moore, T. B.; Sakamoto, K. M. Stem Cells
2002, 20, 215.
(182) Giles, F.; Estey, E.; OBrien, S. Cancer 2003, 98, 2095.
(183) Giles, F. J. Expert Rev. Anticancer Ther. 2002, 2, 630.
(184) Lo-Coco, F.; Cimino, G.; Breccia, M.; Noguera, N. I.; Diverio, D.;
Finolezzi, E.; Pogliani, E. M.; Di Bona, E.; Micalizzi, C.; Kropp,
M.; Venditti, A.; Tafuri, A.; Mandelli, F. Blood 2004, Epub ahead
of print.
(185) Jedema, I.; Barge, R. M.; van, d. V., V.; Nijmeijer, B. A.; Van
Dongen, J. J.; Willemze, R.; Falkenburg, J. H. Leukemia 2004,
18, 316.
(186) Larson, R. A.; Boogaerts, M.; Estey, E.; Karanes, C.; Stadtmauer,
E. A.; Sievers, E. L.; Mineur, P.; Bennett, J. M.; Berger, M. S.;
Eten, C. B.; Munteanu, M.; Loken, M. R.; Van Dongen, J. J.;
Bernstein, I. D.; Appelbaum, F. R. Leukemia 2002, 16, 1627.
(187) Coleman, M.; Goldenberg, D. M.; Siegel, A. B.; Ketas, J. C.; Ashe,
M.; Fiore, J. M.; Leonard, J. P. Clin. Cancer Res. 2003, 9, 3991S.
(188) Engel, P.; Nojima, Y.; Rothstein, D.; Zhou, L. J.; Wilson, G. L.;
Kehrl, J. H.; Tedder, T. F. J. Immunol. 1993, 150, 4719.
(189) Powell, L. D.; Varki, A. J. Biol. Chem. 1995, 270, 14243.
(190) Shan, D.; Press: O. W. J. Immunol. 1995, 154, 4466.
(191) Shih, L. B.; Goldenberg, D. M.; Xuan, H.; Lu, H. W.; Mattes, M.
J.; Hall, T. C. Cancer Immunol. Immunother. 1994, 38, 92.
(192) DiJoseph, J. F.; Armellino, D. C.; Boghaert, E. R.; Khandke, K.;
Dougher, M. M.; Sridharan, L.; Kunz, A.; Hamann, P. R.;
Gorovits, B.; Udata, C.; Moran, J. K.; Popplewell, A. G.;
Stephens, S.; Frost, P.; Damle, N. K. Blood 2004, 103, 1807.
(193) Chan, S. Y.; Gordon, A. N.; Coleman, R. E.; Hall, J. B.; Berger,
M. S.; Sherman, M. L.; Eten, C. B.; Finkler, N. J. Cancer
Immunol. Immunother. 2003, 52, 243.
(194) Boghaert, E. R.; Sridharan, L.; Armellino, D. C.; Khandke, K.
M.; DiJoseph, J. F.; Kunz, A.; Dougher, M. M.; Jiang, F.;
(195)
(196)
(197)
(198)
(199)
(200)
(201)
(202)
(203)
(204)
(205)
(206)
(207)
(208)
(209)
(210)
(211)
(212)
(213)
(214)
(215)
(216)
(217)
(218)
(219)
(220)
(221)
(222)
Kalyandrug, L. B.; Hamann, P. R.; Frost, P.; Damle, N. K. Clin.

Cancer Res. 2004, 10, 4538.
Shabat, D.; Rader, C.; List, B.; Lerner, R. A.; Barbas, C. F., III
Sugiura, Y.; Shiraki, T.; Konishi, M.; Oki, T. Proc. Natl. Acad.
Sci. U.S.A. 1990, 87, 3831.
Wender, P. A.; Kelly, R. C.; Beckham, S.; Miller, B. L. Proc. Natl.
Acad. Sci. U.S.A. 1991, 88, 8835.
Kusakabe, T.; Uesugi, M.; Sugiura, Y. Biochemistry 1995, 34,
9944.
Unno, R.; Michishita, H.; Inagaki, H.; Suzuki, Y.; Baba, Y.;
Jomori, T.; Nishikawa, T.; Isobe, M. Bioorg. Med. Chem. 1997,
5, 987.
Rader, C.; Turner, J. M.; Heine, A.; Shabat, D.; Sinha, S. C.;
Wilson, I. A.; Lerner, R. A.; Barbas, C. F. J. Mol. Biol. 2003,
332, 889.
Sinha, S. C.; Li, L. S.; Miller, G. P.; Dutta, S.; Rader, C.; Lerner,
R. A. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 3095.
Linenberger, M. L.; Hong, T.; Flowers, D.; Sievers, E. L.; Gooley,
T. A.; Bennett, J. M.; Berger, M. S.; Leopold, L. H.; Appelbaum,
F. R.; Bernstein, I. D. Blood 2001, 98, 988.
Rathnasabapathy, R.; Lancet, J. E. Cancer Control 2003, 10, 469.
Tsimberidou, A.; Cortes, J.; Thomas, D.; Garcia-Manero, G.;
Verstovsek, S.; Faderl, S.; Albitar, M.; Kantarjian, H.; Estey,
E.; Giles, F. J. Leuk. Res. 2003, 27, 893.
Naito, K.; Takeshita, A.; Shigeno, K.; Nakamura, S.; Fujisawa,
S.; Shinjo, K.; Yoshida, H.; Ohnishi, K.; Mori, M.; Terakawa, S.;
Ohno, R. Leukemia 2000, 14, 1436.
Scotto, K. W. Oncogene 2003, 22, 7496.
Walter, R. B.; Raden, B. W.; Hong, T. C.; Flowers, D. A.;
Bernstein, I. D.; Linenberger, M. L. Blood 2003, 102, 1466.
Matsui, H.; Takeshita, A.; Naito, K.; Shinjo, K.; Shigeno, K.;
Maekawa, M.; Yamakawa, Y.; Tanimoto, M.; Kobayashi, M.;
Ohnishi, K.; Ohno, R. Leukemia 2002, 16, 813.
Ihnat, M. A.; Lariviere, J. P.; Warren, A. J.; La Ronde, N.;
Blaxall, J. R.; Pierre, K. M.; Turpie, B. W.; Hamilton, J. W. Clin.
Cancer Res. 1997, 3, 11339.
Mossink, M. H.; van Zon, A.; Scheper, R. J.; Sonneveld, P.;
Wiemer, E. A. Oncogene 2003, 22, 7458.
Plasschaert, S. L.; Van Der Kolk, D. M.; De Bont, E. S.; Vellenga,
E.; Kamps, W. A.; De Vries, E. G. Leuk. Lymphoma 2004, 45,
649.
van den Heuvel-Eibrink, M. M.; Wiemer, E. A.; Prins, A.;
Meijerink, J. P.; Vossebeld, P. J.; van der Holt, B.; Pieters, R.;
Sonneveld, P. Leukemia 2002, 16, 833.
Walter, R. B.; Raden, B. W.; Cronk, M. R.; Bernstein, I. D.;
Appelbaum, F. R.; Banker, D. E. Blood 2004, 103, 4276.
Tsimberidou, A.; Estey, E.; Cortes, J.; Thomas, D.; Faderl, S.;
Verstovsek, S.; Garcia-Manero, G.; Keating, M.; Albitar, M.;
OBrien, S.; Kantarjian, H.; Giles, F. Cancer 2003, 97, 1481.
Whitwam, R. E.; Ahlert, J.; Holman, T. R.; Ruppen, M.; Thorson,
J. S. J. Am. Chem. Soc. 2000, 122, 1556.
Biggins, J. B.; Onwueme, K. C.; Thorson, J. S. Science 2003, 301,
1537.
Sastry, M.; Fiala, R.; Lipman, R.; Tomasz, M.; Patel, D. J. J.
Mol. Biol. 1995, 247, 338.
Sartorelli, A. C.; Hodnick, W. F.; Belcourt, M. F.; Tomasz, M.;
Haffty, B.; Fischer, J. J.; Rockwell, S. Oncol. Res. 1994, 6, 501.
Hata, T.; Hoshi, T.; Kanamori, K.; Matsumae, A.; Sano, Y.;
Shima, T.; Sugawara, R. J. Antibiot. 1956, 9, 141.
Hata, T.; Sugawara, R. J. Antibiot. 1956, 9, 147.
Powis, G. Drug Metab. Rev. 1989, 20, 379.
Wakaki, S.; Marumo, T.; Tomioka, T.; Shimizu, E.; Kato, H.;
Kamada, S.; Kudo, S.; Fujimoto, Y. Antibiot. Chemother. 1958,
8, 228.
Galm et al.
(223) DeBoer, C.; Dietz, A.; Lummus, N. E.; Savage, G. M. Antimicrob.
Agents Ann. 1960, 1960, 17.
(224) Lefemine, D. V.; Dann, M.; Barbatsch, F.; Hausmann, W. K.;
Zbinovsky, V.; Monnikendam, P.; Adam, J.; Bohonos, N. J. Am.
Chem. Soc. 1962, 84, 3184.
(225) Planting, A. S.; Schellens, J. H.; van der Burg, M. E.; BoerDennert, M.; Winograd, B.; Stoter, G.; Verweij, J. Anticancer
Drugs 1999, 10, 821.
(226) Verweij, J.; Schellens, J. H. M.; Loo, T. L.; Pinedo, H. M. Cancer
Chemotherapy and Biotherapy: Principles and Practice; 1996;
p 409.
(227) Tomasz, M.; Lipman, R.; Chowdary, D.; Pawlak, J.; Verdine, G.
L.; Nakanishi, K. Science 1987, 235, 1204.
(228) Tomasz, M.; Chawla, A. K.; Lipman, R. Biochemistry 1988, 27,
3182.
(229) Begleiter, A. Front. Biosci. 2000, 5, E153.
(230) Suresh, K. G.; Lipman, R.; Cummings, J.; Tomasz, M. Biochemistry 1997, 36, 14128.
(231) Kumar, G. S.; He, Q. Y.; Behr-Ventura, D.; Tomasz, M. Biochemistry 1995, 34, 2662.
(232) Celli, C. M.; Jaiswal, A. K. Cancer Res. 2003, 63, 6016.
(233) Workman, P. Oncol. Res. 1994, 6, 461.
(234) Seow, H. A.; Penketh, P. G.; Baumann, R. P.; Sartorelli, A. C.
Methods Enzymol. 2004, 382, 221.
(235) Kiyoto, S.; Shibata, T.; Yamashita, M.; Komori, T.; Okuhara, M.;
Terano, H.; Kohsaka, M.; Aoki, H.; Imanaka, H. J. Antibiot.
1987, 40, 594.
(236) Hirai, O.; Shimomura, K.; Mizota, T.; Matsumoto, S.; Mori, J.;
Kikuchi, H. J. Antibiot. 1987, 40, 607.
(237) Naoe, Y.; Inami, M.; Matsumoto, S.; Nishigaki, F.; Tsujimoto,
S.; Kawamura, I.; Miyayasu, K.; Manda, T.; Shimomura, K.
Cancer Chemother. Pharmacol. 1998, 42, 31.
(238) Wu, H. I.; Brown, J. A.; Dorie, M. J.; Lazzeroni, L.; Brown, J.
M. Cancer Res. 2004, 64, 3940.
(239) Belcourt, M. F.; Penketh, P. G.; Hodnick, W. F.; Johnson, D. A.;
Sherman, D. H.; Rockwell, S.; Sartorelli, A. C. Proc. Natl. Acad.
Sci. U.S.A. 1999, 96, 10489.
(240) August, P. R.; Flickinger, M. C.; Sherman, D. H. J. Bacteriol.
1994, 176, 4448.
(241) Baumann, R. P.; Hodnick, W. F.; Seow, H. A.; Belcourt, M. F.;
Rockwell, S.; Sherman, D. H.; Sartorelli, A. C. Cancer Res. 2001,
61, 7770.
(242) Cummings, J. Drug Resist. Update 2000, 3, 143.
(243) He, M.; Sheldon, P. J.; Sherman, D. H. Proc. Natl. Acad. Sci.
U.S.A. 2001, 98, 926.
(244) Sheldon, P. J.; Mao, Y.; He, M.; Sherman, D. H. J. Bacteriol.
1999, 181, 2507.
(245) Aouida, M.; Tounekti, O.; Belhadj, O.; Mir, L. M. Antimicrob.
(246) Pron, G.; Belehradek, J., Jr.; Mir, L. M. Biochem. Biophys. Res.
Commun. 1993, 194, 333.
(247) Pron, G.; Mahrour, N.; Orlowski, S.; Tounekti, O.; Poddevin, B.;
Belehradek, J., Jr.; Mir, L. M. Biochem. Pharmacol. 1999, 57,
45.
(248) Jones, G. B.; Hynd, G.; Wright, J. M.; Purohit, A.; Plourde, G.
W.; Huber, R. S.; Mathews, J. E.; Li, A.; Kilgore, M. W.; Bubley,
G. J.; Yancisin, M.; Brown, M. A. J. Org. Chem 2001, 66, 3688.
(249) Heyd, B.; Pecorari, F.; Collinet, B.; Adjadj, E.; Desmadril, M.;
Minard, P. Biochemistry 2003, 42, 5674.
CR030117G
Chem. Rev. 2005, 105, 759774
759
New Targets and Screening Approaches in Antimicrobial Drug Discovery

Eric D. Brown* and Gerard D. Wright*
Antimicrobial Research Centre, Department of Biochemistry and Biomedical Sciences, McMaster University, 1200 Main Street West,
Hamilton, Ontario, Canada L8N 3Z5
Received August 19, 2004
Contents
1. Introduction
2. Historical View of Target-Based Antibiotic
Discovery
3. Genomics and New Antimicrobial Target
Discovery
4. Features of Antimicrobial Targets
5. New Directions in Celebrated and Emerging
Antimicrobial Targets
5.1. Peptidoglycan Biosynthesis
5.2. DOXP Pathway
5.3. Wall Teichoic Acid
5.4. Folate Biosynthesis
5.5. Fatty Acid Biosynthesis
5.6. Protein Secretion
5.7. Peptide Deformylase
5.8. Proteins of Unknown Function
5.9. Fungal Targets
6. Genomic and Systems Methods in Antimicrobial
Research
6.1. Transcriptomics
6.2. Chemical Genomics
6.3. Phenotype-Based Screening Followed by
Target Discovery
7. Conclusions
8. Acknowledgments
9. References
759
760
760
761
762
762
762
764
765
765
765
766
767
767
769
769
770
770
771
771
771
1. Introduction
There is no doubt that the existing arsenal of
antimicrobial agents we have in hand for the treatment of infectious diseases is insufficient to protect
us over the long term.1-3 The primary reason for this
state of affairs is the inexorable drive of evolution
that leads to antimicrobial resistance. At the same
time, we must also acknowledge our inability to
predict with any accuracy the nature of new emerging infections. Witness the stunning impact of HIV
over the past 20 years, the unexpected causative link
between peptic ulcer disease and Helicobacter pylori,
the recent emergence of SARS and avian influenza,
and the ensuing dramatic impact these epidemics
have had on human and animal health as well as the
international economy. Our continued vulnerability
* To whom correspondence should be addressed. E.D.B. e-mail:
ebrown@mcmaster.ca. G.D.W. e-mail: wrightge@mcmaster.ca.
Eric Brown is an Associate Professor in the Department of Biochemistry

He is also Director of the McMaster High Throughput Screening Laboratory.
He received his Ph.D. degree in Biochemistry in 1992 for studies of proline
utilization in bacteria and spent 2 years as a Natural Sciences and
Engineering Research Council of Canada Postdoctoral Fellow in Professor
Christopher Walshs laboratory at Harvard Medical School researching
bacterial cell wall biosynthesis. He worked for 3 years in the Boston
pharmaceutical sector with Myco Pharmaceuticals and Astra Research
Center Boston before joining the Department of Biochemistry at McMaster
in 1998 and receiving a Medical Research Council of Canada Scholarship.
Erics research program focuses on emerging targets in antibacterial
research. He presently holds a Canada Research Chair in Microbial
Biochemistry.
For a photograph and biography of Gerry Wright, see his previous paper
in this issue.
to the effects of microbes should be humbling. Paradoxically, large pharmaceutical companies are retreating from the field of antimicrobial agents, preferring to concentrate on chronic diseases that promise
longer term profits.4,5
Resistance to antibiotics is an unavoidable side
effect of their use. The time scale of the life cycle of
microbes and the adapt or die paradigm that is
imposed with the current arsenal of agents that
either stop growth or cause cell death conspire
against the indefinite longevity of antibiotics. We
cannot avoid resistance nor can we predict with any
accuracy the emergence of new infectious agents, but
we can work to mitigate these issues with research
that will yield new agents of novel mechanism and
chemical class. Such research will include studies to
validate and characterize novel targets for efforts in
discovering and developing new leads for new antimicrobials. The availability of complete genome
10.1021/cr030116o CCC: $53.50 2005 American Chemical Society

sequences of dozens of pathogenic bacteria and an

ever-increasing number of fungal species has enabled
the development of systematic genome-wide approaches to target discovery and validation. Indeed,
new screening approaches are beginning to employ
systems approaches and parallelization that have
been enabled by genomics. In this review, we have
attempted to summarize, without being comprehensive, many of these developments that span new
screens and targets for the next generation of antimicrobials.
2. Historical View of Target-Based Antibiotic

Discovery
The renaissance in antimicrobial research of the
1990s was marked by a paradigm shift from phenotype-based to target-based screening and from a
reliance on natural product libraries to an emphasis
on libraries of synthetic small molecules. Precipitated
by steady erosion in the activity of conventional
antibiotics, an urgent need was recognized for research into new targets for new antibacterial drugs.
Organisms such as Staphylococcus aureus continue
to be a major cause of hospital-acquired infection and
have become increasingly difficult to treat due to
resistance to multiple antibiotics including methicillin6 and vancomycin.7 In fact, in the past 40 years
only two new chemical classes of antibiotics, the
oxazolidinone linezolid8 and the lipopeptide daptomycin,9 have reached the clinic and existing antibiotics are directed at a small number of targets, principally cell wall, DNA, and protein biosynthesis. For
example, fewer than 30 proteins have been exploited
commercially as targets for antibacterial drugs.10
Thus multidrug resistance among bacterial pathogens is thought to be due in large part to the limited
repertoire of antibacterial drugs that eradicate bacteria using a narrow range of mechanisms. The
answers then lie in new targets and in a genes to
drugs research approach that begins with the sequencing of bacterial genomes.
In recent years, genomics has provided staggering
amounts of sequence information and simultaneously
expanded the target base for new antimicrobials.
Microbial genomics arrived with much fanfare and
promise to provide a blueprint for the next generation
of drugs. Given the time required to discover and
develop new chemical entities, it is clearly too soon
to judge the contribution of genomics to antimicrobial
drug discovery. It has nevertheless delivered far more
questions about protein biochemistry and cell physiology than were probably anticipated. Since the first
report of a complete genome sequence for a free-living
organism (Haemophilus influenzae) in 199511 researchers have been faced with the daunting realization that about one-third of the genes in any microbe
encode biochemically uncharacterized proteins.12 Similar estimates have been made for the human genome,13,14 and there is now widespread agreement
that the next challenge facing biomedical researchers
in a variety of fields is the assignment of function in
the uncharted complement of genes. Perhaps not
surprisingly, microbial genome sequencing, informatics, and functional genomics have also brought to
Brown and Wright
light how thin our understanding is of even fairly well

studied areas of bacterial physiology. Examples of
very significant recent and genome-inspired advances
include discoveries in folate,15 isoprenoid,16 thiamin,17
and adenylate18 metabolism.
3. Genomics and New Antimicrobial Target

Discovery
Mutagenesis experiments in a variety of microbes
have suggested that literally hundreds of proteins are
indispensable for the growth of any given microorganism. Among the first investigations was an
effort to isolate some 440 mutants in Salmonella
typhimurium that were temperature sensitive for
growth on rich media.19 This early work was pregenomic and used the power of genetics to make a
determination of the scope of the essential bacterial
gene set. Several years later another genetic methodology, footprinting by random transposon mutagenesis, began to dominate but relied on genomic
information. Footprinting was first applied to chromosome V of yeast Saccharomyces cerevisiae and
revealed 53 putatively essential genes.20 The footprinting approach relies on high-density transposon
mutagenesis followed by analysis of the cells carrying
mutations. Genes for which no insertions are recovered are inferred to be essential. This approach led
Hutchison et al. to conclude that 265-350 of the 517
genes in Mycoplasma genitalium were indispensable
for growth.21 Similar studies of H. influenzae22 and
Escherichia coli23 identified 259 and 620 genes,
respectively, as putatively essential. Inhibition of
gene function by antisense RNA has been applied to
S. aureus to reveal about 150 putatively essential
genes.24 Working with a short list of 347 candidate
open reading frames from Streptococcus pneumoniae,
Thanassi et al. used a targeted approach to determine
that 113 of these genes could not be disrupted with
a suicide vector.25
Comprehensive gene knockout projects have been
initiated in the model Gram-negative bacterium E.
coli26 and reported on for the model Gram-positive
Bacillus subtilis27 as well as the model fungus S.
cerevisiae.28 The latter works involved a systematic
construction of mutants in virtually all genes in the
genomes of B. subtilis and S. cerevisiae and concluded
that 271 and 1105 genes, respectively, were essential
for growth on rich media. Interestingly, although
perhaps not surprisingly, construction of conditional
mutants in the pathogenic yeast Candida albicans
revealed that only 61% of the essential genes in S.
cerevisiae were also required for C. albicans growth.29
Genome-wide studies in other important filamentous
fungi have lagged behind the yeasts largely as a
result of their larger genome size (30-40 Mb) and
genetic intractability; however, a recent systematic
transposon integration method in cosmids has the
potential to bring these organisms into the comprehensive gene knockout era.30
In general, genome-scale dispensability projects
have revealed the magnitude of essential gene sets
and are valuable contributions to microbial functional
genomics. They are, however, cursory investigations
that should not be over-interpreted. Footprinting, for
Antimicrobial Drug Discovery
Table 1. Features of Antimicrobial Targets

features
pros/cons
ideal target
function well characterized

essential on a Petri dish
amenable to target-based drug discovery toolbox
conserved in significant pathogens
pros
tractable
large market
proven track record
cons
indiscriminate activity and resistance pressure
narrow spectrum target

virulence target
targets a narrow spectrum of pathogens

targets a virulence determinant in a pathogen
example, suffers from a variety of potential artifacts

including difficulties with smaller genes, cold spots
for recombination, and polar effects in operons.31 Not
surprisingly, therefore, many instances of disagreement can be found among the studies that are
presumably not all due, for example, to differences
in the dispensability of a given orthologue in each
bacterial system. A further qualification on the limits
of gene dispensability data is the potential for
dispensable genes to be part of essential gene families. Penicillin-binding proteins (PBPs), for example,
are not typically essential; however, inhibition of
multiple PBPs can be lethal.32
4. Features of Antimicrobial Targets

While there are a plethora of new targets with
potential for new drug discovery, the worth of any
one of these remains the subject of considerable
debate. Those that have seen the most intense drug
discovery activity in the pharmaceutical sector have
had properties considered to be ideal.3 The ideal
target typically is an enzyme with fully elucidated
function and mechanism (Table 1). In general, the
target can be highly purified in quantity as a recombinant protein such that it is amenable to highthroughput assay and any of a number of downstream biochemical investigations to sort out mechanism of inhibition including co-structural examinations. Most often the target is absolutely required for
survival of the organism in vitro on a Petri dish. As
such, targets are normally validated with in-vitro
genetic methods to determine gene essentiality.33,34
The ideal target is typically present in a number of
species of clinical and economic significance where
the development of an effective agent will, in all
probability, lead to a broad-spectrum drug. In the
case of an antibacterial, the ideal target is absent in
the human host. While there have been efforts on
antibacterial targets that defy the latter criterion,
such as dihydrofolate reductase targeted by trimethoprim, it is generally accepted that avoiding host
physiology will be a crucial step in avoiding host
toxicity.
Narrow-spectrum antibiotics have been called for
by infectious disease specialists as a means to
selectively target particular pathogens.35 The use of
broad-spectrum agents has the effect of indiscriminately denuding the indigenous and more often
benign (and perhaps even helpful) microbial flora as
pros
benign (beneficial) flora potentially unharmed
resistance selection from a smaller population
cons
need for diagnostics
smaller market
a result of their lack of specificity. This results in the

eradication of all susceptible bacteria and subsequent
superinfection by other potentially problematic species such as Candida albicans and Clostridium
difficile. The challenge in such a smart bomb approach is the accurate and rapid identification of
infection-causing organisms. Selectivity inherent in
narrow-spectrum therapeutics, it could be argued,
might also curtail the spread of resistance determinants. With a selective antimicrobial agent the
pressure for drug resistance is, by definition, on a
smaller population of organisms. Continuing improvements in molecular diagnostics including rapid
PCR-based strategies are a step in the right direction;
however, in many cases it is difficult to pinpoint the
specific cause of infection as multiple genera and
species may be detected at the site of infection, e.g.,
in pneumonia.36 Nonetheless, in some cases such as
Pseudomonas colonization and subsequent infection
in patients with cystic fibrosis and infections specifically caused by one organism or genus such as
Mycobacterium tuberculosis and Chlamydia spp.,
organism-specific antibiotics make good sense.
Another strategy is to target molecular factors that
are essential to microbial virulence. In this approach,
proteins that are specifically associated with infection
rather than ubiquitously required for growth are
targeted. Various technologies have been developed
to identify virulence factors that are expressed only
upon host infection, e.g., in vivo expression technology (IVET).37-39 In principle, this is an ingenious
approach as it targets only those factors that play a
role in infection, and therefore, one would predict that
this strategy would be highly selective for diseasecausing microbes.40 A potential downside of the
approach is that it does not distinguish between
targets involved in bacterial colonization or persistence. Drugs targeting the former category would
presumably have a requirement for prophylactic use.
In any case, this approach will require a paradigm
shift in the anti-infective drug discovery field where
the traditional in-vitro minimal inhibitory concentration (MIC) results hold primacy in drug development
decision making processes, for example, in lead
selection. Another axiom of the anti-infective drug
discovery field that is violated by targeting virulence
is that of dead bugs dont mutate.41 Again, a favorable drug-resistance argument can be made for
virulence targets since agents that target such functions are likely to be particularly selective for a given
pathogen. While vital housekeeping functions are

commonly shared across diverse microbial species,
those functions associated with virulence are often
rather peculiar to a given pathogen (or group of
pathogens).
Despite appeals for strategies focusing on narrow
spectra of bacteria and on virulence targets, the lions
share of the efforts in the pharmaceutical and biotech
sectors has been on targeting universal and essential
housekeeping functions. A challenge to successful
execution of more selective approaches is to identify
an economic model that will provide sufficient financial return to companies who pursue this approach
as market share will be smaller than for broadspectrum agents.
5. New Directions in Celebrated and Emerging

Antimicrobial Targets
For well-validated targets, there have been numerous efforts over the years to identify new classes of
molecules that block activity. For emerging targets
the emphasis has been on validation and characterization of the structure and function of newly discovered proteins. Here, we summarize some of what
is been done on novel antimicrobial targets and
progress in new directions on more established
targets.
5.1. Peptidoglycan Biosynthesis

The biosynthesis of murein (peptidoglycan) is a
well-validated antibacterial target. Numerous antibiotics and a number of blockbuster drugs are known
to interact with the enzymes (and their substrates)
that catalyze both the early-stage (intracellular) and
late-stage (extracellular) events that comprise this
pathway (Figure 1). Screening approaches directed
toward identifying cell wall active agents have therefore been pursued extensively. A cell-based reporter
system using the ampR-ampC loci of Citrobacter
freundi cloned into E. coli has been reported.42
Expression of the AmpC -lactamase through the
regulatory protein AmpR is highly sensitive to cell
wall degradation products. These arise as a result of
exposure of the cell to peptidoglycan biosynthesis
inhibitors that trigger the activity of lytic transglycosylases. The screen consists of monitoring the
expression of AmpC through its -lactamase activity
in 96-well format using the colorimetric -lactamase
reagent nitrocefin. The assay effectively detected
known inhibitors of both early and late stages of
peptidoglycan assembly and was demonstrated to be
useful in the screening of compound libraries and
crude extracts for inhibitors of cell wall biosynthesis.42,43 Another cell-based approach for screening
molecules for late-stage murein synthesis has been
reported that exploits the promoter regulating the
sigE operon of Streptomyces coelicolor.44 This promoter was shown to be sensitive to inhibitors of
peptidoglycan synthesis such as glycopeptides, some
-lactams, and ramoplanin. Fusion of the promoter
to a kanamycin-resistance reporter enabled construction of a system with the capacity to be used in a
screen for agents that disrupt cell wall synthesis.
Brown and Wright
In-vitro screening assays suitable for high throughput have been developed for a number of the individual murein biosynthesis enzymes including
MurA,45 MurG,46 and MraY.47 Since many of the
Mur enzymes use ATP as a substrate with concomitant product inorganic phosphate formation
(Figure 1), the sensitive malachite green/phosphomolybdic acid assay48 has proven to be a method of
choice for facile assay in high throughput.49
A challenge in this one-enzyme-at-a-time in-vitro
approach is commercial access to substrates, especially for the MurB through MurF enzymes. To
alleviate this problem, an in-vitro pathway assay that
links the six enzymes MurA, MurB, MurC, MurD,
MurE, and MurF has been described and optimized
for screening.50,51 This clever approach uses only
purified enzymes and the commercially available
substrates for MurA, NADPH (for MurB), and ATP
and the readily available amino acid substrates for
MurC, MurD, MurE, and MurF to monitor flux
through the reconstituted metabolic pathway. A
permeabilized whole-cell assay that monitors incorporation of 14C-labeled UDP-GlcNAc into peptidoglycan has been reported that expands the number of
biosynthetic steps assayed at once.52
5.2. DOXP Pathway

Isoprenoids comprise a large family of natural
products from virtually all living organisms and
include a variety of polyisoprenoids in microorganisms including ubiquinone, menaquinone, and undecaprenyl pyrophosphate. For many years it was
thought that isopentenyl pyrophosphate, the fundamental unit of polyisoprenoid biosynthesis, was the
product of a mevalonate-dependent biosynthetic pathway in all organisms.53 A large and recent body of
work from several laboratories has established that
a mevalonate-independent pathway, the DOXP pathway, exists for the synthesis of isopentenyl-pyrophosphate in many bacteria, protozoa, and plants (reviewed in ref 16). The six committed steps of the
DOXP pathway and their respective enzymes have
been identified (Figure 2). They involve successive
action of newly discovered enzymes on 1-deoxy-Dxylulose-5-phosphate (1) to yield isopentenyl-diphosphate (7) and its isomer dimethylallyl-diphosphate
(8), the building blocks of polyisoprenoids.
Gene distribution in microbial genomes indicates
that the mevalonate-dependent pathway predominates in archea and low G+C Gram-positive cocci
while the DOXP route is used by the majority of
eubacteria, particularly Gram-negative organisms.16
The conservation of the DOXP genes among significant human pathogens (e.g., M. tuberculosis, P.
aeruginosa, H. pylori) and their absence in humans
is a compelling finding. Indeed, growing evidence
suggests that the nonmevalonate pathway is essential in bacteria.54-57 Known functions of isoprenoids in bacteria include the modification of
tRNA,58 dolichol production, and the formation of
respiratory quinones.59 To probe the dominant mechanism of lethality associated with lesions in this
pathway, Campbell and Brown60 characterized conditional mutants in ispF in E. coli and B. subtilis and
Figure 1. Biosynthesis of bacterial peptidoglycan. (A) Early-stage intracellular assembly of the N-acetylmuramylpentapeptide subunit. The antibiotic phosphomycin blocks MurA in some bacterial species, while D-cycloserine inhibits
both the D-Ala-D-Ala ligase (Ddl) and Ala racemase (Dad/Aln) steps. (B) Late-stage polymer assembly. MurG is inhibited
by ramoplanin, and MraY is inhibited by tunicamycin and the mureidomycins. Bacitracin blocks the membrane-associated
pyrophosphatase (not shown) required for recycling of the undecaprenyl carrier lipid once released by the transglycosylases.
The -lactam antibiotics target the penicillin-binding proteins (transpeptidases) that are involved in peptidoglycan crosslinking. The glycopeptide antibiotics such as vancomycin block both the transpeptidases and transglycosylases.
Brown and Wright
ism.63,64 Exploitation of this pathway in modern

target-based anti-infective discovery efforts currently
suffers from the practical obstacle of procuring the
substrates for each of the reactions. Recent published
efforts in inhibitor discovery have understandably
focused on small molecule screening for inhibitors of
the first committed step, the DOXP reductase65,66 for
which the substrates are commercially available.
5.3. Wall Teichoic Acid
Figure 2. Committed steps of the DOXP pathway. Compound numbers refer to the following: 1-deoxy-D-xylulose5-phosphate (1), methyl-D-erythritol-4-phosphate (2), 4-diphosphocytidyl-2C-methyl-D-erythritol (3), diphosphocytidyl-2C-methyl-D-erythritol 2-phosphate (4), 2C-methyl-Derythritol 2,4-cyclodiphosphate (5), 1-hydroxy-2-methyl-2(E)-butenyl (6), isopentyl diphosphate (7), dimethylallyl
diphosphate (8), and fosmidomycin (9). Protein names are
those from E. coli. Several of these were renamed recently,
and the previous names are indicated in brackets.
revealed rod to filament and rod to sphere transitions

upon depletion of the respective proteins. These
findings pointed to a primary impact on cell wall
synthesis and were supported by experiments showing that inhibitors of peptidoglycan biogenesis, in
particular, showed synergy with the depletion of IspF
in both E. coli and B. subtilis. While recent genetic
efforts on this novel bacterial pathway have focused
on validation of the enzymes as therapeutic targets,
it is interesting to note that the DOXP pathway is
the target of a natural product antibiotic, fosmidomycin (9), isolated in the 1970s and evaluated in
clinical trials (but abandoned) for bacterial infections.61 It was not until the discovery of the DOXP
pathway that the DOXP reductase (IspC) was understood to be the target of fosmidomycin.62 Indeed,
fosmidomycin has been shown to be effective in
controlling the malaria parasite P. falciparum through
inhibition of the DOXP reductase in that organ-
While the cell walls of Gram-negative bacteria are

composed primarily of peptidoglycan, Gram-positive
bacteria have a more substantial cell wall and
produce, in addition to peptidoglycan, large amounts
of the polymer teichoic acid. Teichoic acids are a
chemically diverse group of linear, hydrophilic, anionic polymers of polyol phosphates some 30-50
residues long. Of the Gram positives so far studied,
a large number produce either poly(ribitol phosphate)
or poly(glycerol phosphate) as the major cell wall
teichoic acid;67 however, a number of bacteria are
known to produce polymers having alternate repeating units, for example, -glycerol-phosphate-glycosylphosphate- or -N-acetylgalactosamine-1-P-.68
B. subtilis has been the model Gram positive for
understanding the genetics and biochemistry of wall
teichoic acid biogenesis. The major teichoic acids of
B. subtilis strains 168 and W23 are linear 1,3- and
1,5-linked poly(glycerol phosphate) and poly(ribitol
phosphate), respectively, which are modified with
glucose or alanine (Figure 3). The polymer is anchored in the cell wall by covalent attachment via a
phosphodiester bond between the linkage unit of
teichoic acid and the 6-hydroxyl of N-acetylmuramic
acid of peptidoglycan.69,70 In contrast to the polymers,
the linkage units have a less variant composition
among many Gram-positive bacteria, consisting of a
disaccharide-phosphate component, most often Nacetylmannosamine(1-4)N-acetylglucosaminephosphate and a series of glycerol phosphate units,
numbering between one and three.71-73
Although considered a dispensable and auxiliary
polymer for decades, isolation of temperature-sensitive mutants derived from random mutagenesis of
strain 16874 suggested that wall teichoic acid may be
essential. That work led to the isolation of the tag
genes for poly(glycerol phosphate) synthesis in B.
subtilis 168 and ultimately the tar genes for poly(ribitol phosphate) synthesis in strain W23.75 The
indispensability of tagD, encoding glycerol 3-phosphate cytidylyltransferase, was unequivocally demonstrated with the construction of a deletion strain
where tagD was precisely replaced by an antibiotic
resistance cassette while a complementing copy of the
gene, under xylose-inducible control, was inserted at
the amyE locus.76 In the absence of xylose, this
deletion strain displayed extremely poor growth,
abnormal cocoid morphology, and lysis comparable
to lesions in peptidoglycan synthesis. The gene tarD,
coding for glycerol 3-phosphate cytidylyltransferase,
in a hybrid W23/168 strain, was similarly shown to
be essential to poly(ribitol phosphate) synthesis.77
While the function of teichoic acid in the Grampositive cell wall has been the subject of speculation,
Figure 3. Structure of wall teichoic acid in B. subtilis, strains 168 and W23. The structure can be divided into the linkage
unit and polymer portion, where the former has an apparently conserved role in linking diverse polymers to peptidoglycan
through a phosphate ester with the 6-hydroxyl of muramic acid (not shown).
convention has held that it can be replaced by a

phosphate-free, glucuronic-acid-containing polymer
(teichuronic acid) under phosphate-limiting conditions. Most recently, teichoic acid biosynthesis was
shown to be indispensable to B. subtilis under
phosphate-limiting conditions.77 These findings suggest that teichoic acid has an essential function that
cannot be replaced by teichuronic acid.
The functions of most of the teichoic acid biosynthetic genes are predictable from sequence similarity.75 Biochemical characterization of several of these
functions has been undertaken: glycerol 3-phosphate
cytidylyltransferase,78,79 CDP-ribitol synthase,80,81 poly(glycerol phosphate) polymerase,82 N-acetylglucosamine 2-epimerase,83 and N-acetylglucosamine-1phosphate transferase.84 Thus, while significant gaps
remain in our understanding of the synthesis, export,
and linkage of this unusual polymer to peptidoglycan,
there is a reasonable foundation in target validation
and enzyme characterization to support drug discovery initiatives in wall teichoic acid synthesis.
5.4. Folate Biosynthesis

There remain a number of examples of emerging
targets and new work on well-studied targets that
are too numerous to detail here. Some, however, are
particularly worthy of mention, including folic acid
synthesis. Folate biosynthesis is a well-validated
route to antibiotics. Tetrahydrofolate and other reduced folates are required in bacteria for the biosynthesis of purines, thymidylate, panthothenate, and
some amino acids. The sulfa drugs (e.g., sulfamethoxazole) and trimethoprim target dihydropteroate synthase and dihydrofolate reductase, respectively, and
are commonly used in a combination therapy to treat
infections as disparate as respiratory and urinary
tract infections. There continues to be a significant
and ongoing effort on new and more potent inhibitors
for dihydrofolate reductase including two recent
screens of the E. coli enzyme.85,86 Nevertheless,
untapped targets remain in earlier steps in this
pathway that are uncommon to the host but where
the genetic and biochemical literature has been
slower to develop.
5.5. Fatty Acid Biosynthesis

The bacterial fatty acid synthase system is a
dissociated type II system that is distinct from its
eukaryotic counterpart. FabI, the enoyl-acyl carrier
protein reductase, catalyzes the last step in each cycle
of elongation87 and has been identified as the target
of hydroxyl-diphenyl ethers such as triclosan88 and
a metabolite of isoniazid (Figure 4).89 Beta-ketoacylacyl carrier protein (ACP) synthases have likewise
been identified as targets of the natural products
thiolactomycin90 and cerulenin.91 Recently, a fattyacid-pathway-specific reporter assay has been described.92 With an abundance of validating antibacterials and a depth of basic understanding of the lipid
biosynthesis93 the area has attracted considerable
attention and remains very promising.
5.6. Protein Secretion

Essential components of the bacterial protein secretion pathway provide attractive points of intervention in processing and translocation steps that are
required for proteins bound for extracytoplasmic
destinations. Significant challenges remain to exploiting these targets because they are by and large
membrane bound and not all of the steps of the socalled general secretory pathway have so far been
identified, much less understood.94 The proteolytic
processing of signal peptides is a particularly appealing point for intervention as this occurs on the
extracellular side of the cytoplasmic membrane,
where permeability might be less of an obstacle to
getting a drug on target (Figure 5). A soluble,
catalytically active, periplasmic fragment of signal
peptidase, a large transmembrane Ser proteinase of
the general secretory pathway, has proven to be
highly tractable to biochemical characterization.95,96
A robust, genetic, cell-based screen for secretion
inhibitors has been reported97 as has a recent cellbased assay for evaluation of signal peptidase inhibitors.98 The crystal structure of the periplasmic fragment has been solved in complex with a -lactam99
and natural product inhibitor.100 The latter molecule,
arylomycin A2, is a member of a recently described
class of lipopeptide antibiotics (Figure 5).101 Interest-
Brown and Wright
Figure 4. Fatty acid biosynthesis is a target for antibiotics. The synthetic antibacterials triclosan and isoniazid inhibit
the 2,3-enoylreductase FabI, while the natural products cerulenin and thiolactamycin inhibit -keto-ACP synthase (FabB
in E. coli). During fatty acid synthesis the acyl chains are immobilized on acyl carrier protein (ACP) via a thioester linkage.
Figure 5. Signal peptidase cleaves peptides on the extracellular side of the cytoplasmic membrane and is inhibited by
arylomycin A2.
ingly, the signal peptidase of the lipoprotein secretory

pathway appears also to be a reasonable target as
globomycin, a cyclic peptide antibiotic effective against
Gram-negative enteric bacteria, is understood to be
a specific inhibitor of prolipoprotein signal peptidase.102 Thus, the essential steps of protein secretion
appear underexploited as targets in strategies for
new antibacterials.
5.7. Peptide Deformylase

The last several years have seen considerable
activity around the antibacterial target peptide deformylase. The metalloproteinase peptide deformylase (PDF) has the critical role in eubacteria of
removing the formyl group in formyl-methioneinitiated protein synthesis (Figure 6). This is necessary because the methionine aminopeptidase (MAP)
cannot hydrolyze N-blocked polypeptides. Thus, both
PDF and MAP are essential, for example, in E. coli,
though the former becomes dispensable in mutants
lacking formyltransferase, the enzyme that Nformylates methionyl-tRNA.103,104 The crystal struc-
Figure 6. Reaction catalyzed by peptide deformylase, and

structure of the natural product inhibitor actinonin.
ture of PDF bound to a potent natural product

antibiotic actinonin105 revealed a metal-chelating role
for a hydroxamate moiety in that inhibitor.106 Indeed,
many laboratories have had considerable success in
designing potent enzyme inhibitors against this metalloprotease, and nearly all of these efforts have
relied on a strategy using a chelating pharmacophore.105,107 Unfortunately, potent enzyme inhibition
and relatively weak antibacterial activity have been

hallmarks of these efforts. PDF nevertheless represents a well-validated antibacterial target with tangible prospects for yielding a novel therapeutic.
5.8. Proteins of Unknown Function

The antibiotic-resistance problem has driven the
premise that new antimicrobial drugs of novel chemical and mechanistic classes will come from nontraditional therapeutic targets. It is tempting to
conclude that microbial proteins of unknown function
provide the best prospects in terms of novelty. This
class of proteins is, however, plagued with a very
significant obstacle. In the absence of a demonstrable
biochemical function, small molecule drug discovery
is an impractical activity no matter how interesting
the target. Indeed, as discussed above, genetic methods for validation of the essentiality of a potential
target in, for example, model bacteria are relatively
straightforward even for a gene of unknown function.
It is the downstream efforts that are awkward. Those
studies are in cell physiology to understand the
impact and worth of a given lesion to cell function
and in protein function to develop assays for the
discovery of lead inhibitors of biochemical activity.
It is unclear how large the vital unknown fraction
of bacterial genomes is. To date, the only nearly
comprehensive analysis of gene dispensability in
bacteria is that done in B. subtilis, where 271 where
deemed essential. About 10% of these encode proteins
whose functions are completely uncharacterized or
only predicted in a general way. Nonetheless, interest
in this class of targets has been considerable. Among
the first works in this area was that of Arigoni and
co-workers,108 who shortly after the release of the first
bacterial genome sequences, short listed some 26
conserved, uncharacterized open reading frames for
rigorous dispensability testing in E. coli. Through
conditional complementation experiments these researchers demonstrated that six genes were essential
to the viability of both E. coli and B. subtilis. Other
efforts in identifying novel targets in the unknown
class include a 27-gene study in E. coli by Freiberg
et al.56 yielding six essential genes and a 144-gene
analysis in S. pneumoniae by Zalacain et al.109 The
latter revealed 36 essential genes, many of which are
conserved and appear also to be essential in other
pathogens.
Interestingly, a number of conserved essential and
uncharacterized bacterial genes encode proteins with
signature GTP- or ATP-binding motifs and have been
the subject of ongoing study. Here, we discuss a
subset of these: yjeQ, era, der, obg, and yjeE (E. coli
gene names). Gene yjeQ was shown to encode a
GTPase with unusual burst kinetics110 whose structure is circularly permuted relative to prototype
GTPases such as Ras.111 Biochemical studies indicated that YjeQ participates in a guanine nucleotidedependent interaction with the ribosome.112 Work
using protein translation inhibitors as biochemical
and genetic probes have further implicated YjeQ in
ribosome function,113 however, the exact role of this
putative translation factor remains elusive. Era has
been implicated in ribosomal RNA processing.114-116
Der was so named because it has a slow GTPase with

two GTP-binding domains related to Era (double
Era).117 Little is known of the function of Der;
however, a recent high-resolution crystal structure
has revealed a C-terminal domain that resembles an
RNA-recognition element.118 Obg has been under
study in a variety of organisms and implicated in
cellular functions including GTP-GDP sensing,119
sporulation initiation,120,121 and translation.122 The
structure of Obg was recently solved,123 and the
protein was shown to be associated predominantly
with the 50S ribosomal subunit and a stress response
protein SpoT, implicating Obg in a SpoT-mediated
stress response.124 Structural genomics efforts revealed that the YjeE protein binds ADP by means of
an unusual kinase-like fold.125 Most recently, AllaliHassani et al.126 characterized the extremely slow
ATPase activity catalyzed by YjeE and demonstrated
that mutants in the Walker motifs of the ATPbinding region were incapable of complementing an
yjeE deletion strain in E. coli.
Presented here is just a sampling of the efforts on
individual targets in the area of conserved and
essential proteins of unknown function. It is, however, clear that the idiosyncratic nature of physiology
and biochemistry that is associated with any given
gene product profoundly limits genome-scale approaches to understanding function. As such, an
understanding of the functions of essential uncharacterized proteins will be most significantly advanced
one gene/protein at a time through concerted investigations of phenotype and biochemistry.
5.9. Fungal Targets

Unlike bacteria, the number of well-validated molecular targets in antifungal drug discovery is relatively small. This is not the result of a dearth in
antifungal agents; rather, it is a measure of the state
of target-based fungicide and antifungal drug development. As noted above, there is a paucity of tractable fungal organisms amenable to molecular biological approaches in target discovery, let alone
genome-scale methods, and consequently, research
has focused on the yeasts. These organisms nonetheless do provide good models for fungal biochemistry
and have been the mainstay of antifungal target
validation. An additional difficulty in antifungal drug
development is the potential of toxicity associated
with small molecule inhibitors that can arise as a
result of the similarity of many pathways and proteins between fungi and humans.
One of the best validated pathways in antifungal
therapy is the biosynthesis of the major fungal sterol
ergosterol (Figure 7). Sterols, in particular, the
planar ergosterol, are essential for fungal physiology.
Consequently, disruption of their biosynthesis at any
of several points results in impairment of growth that
can result in cell death.127,128 Ergosterol biosynthesis
is the target of some of the most clinically useful
antifungal drugs including the triazoles such as
fluconazole, which target the product of the ERG11
gene, the P450 lanosterol 14R-demethylase (Figure
8). Furthermore, this pathway is the target of the
allylamine antibiotics such as terbinafine, which
Brown and Wright
Figure 7. Biosynthesis of the fungal sterol ergosterol in yeast. All of the gene products shown to be essential are shown
in bold. The HMG CoA synthase genes HMG1 and HMG2 are redundant and can be dispensed with individually; however,
the double mutant is inviable. Mutations or deletion of the genes shown in italics give viable organisms but with growth
defects.
inhibits squalene epoxidase (ERG1), and the morpholine antifungals such as tridemorph that block the
8-7 isomerase ERG2 and the 14-reductase,
ERG24.
Another important target, which parallels the case
in bacteria, is cell wall biosynthesis. Fungal cell walls
are heterogeneous among genera, but many contain
structural saccharide components including mannans
(manose polymers), -glucans (glucose polymers), and
chitin (polymer of N-acetylglucosamine).129 Inhibition
of the biosynthesis of these components is a good
target for antifungal compounds. For example, the
echinocandins such as caspofungin (Figure 8), which
was approved by the FDA in 2001, block the synthesis of -(1,3)-glucan.130 A 384-well cell-based assay
suitable for the discovery of fungal cell wall inhibitors
has been reported.131
Other targets have also been explored. Inhibition

of protein synthesis, a very well validated target of
many antibacterial drugs, has been underexploited
in antifungal research. One exception is the sordarins
(Figure 8) that inhibit protein synthesis by binding
to elongation factor 2.132,133 Amino acid biosynthesis
is another potential target given the essential requirements for amino acids in cell growth and the
generally low availability of these even in serum. Met
and Thr biosynthesis, in particular, have emerged as
exploitable pathways. The antifungal natural products azoxybacillin,134 rhizocticin,135 and 2-amino-5hydroxy-4-oxopentanoic acid136 have been shown to
target homocysteine synthase, threonine synthase,
and homoserine dehydrogenase, respectively (Figure
8). The potent inhibition of the latter enzyme by
2-amino-5-hydroxy-4-oxopentanoic acid by a slow
Figure 8. Structures of representative antifungal agents that block fungal sterol, cell wall and amino acid biosynthesis,
and translation.
Table 2. Genomic and Systems Methods in Antimicrobial Research
features
outcomes
transcriptomics
impact of antibiotics on microbial

gene transcription
mechanism of action
molecular targets
promoter identification for reporter strains
new screens
chemical genomics
emphasis on parallelization
universal assays for target validation and
lead identification are hallmarks
high target throughput

small molecule probes for target validation
access to unknown fraction of targets
phenotype-based screens
and target discovery
not target-focused
cell-based screening for phenotype
(growth inhibition)
leads with appropriate permeability characteristics

mechanism and molecular target if successful
binding mechanism that results in antifungal activity137 has spurred in-vitro screen development that
has identified new small molecule inhibitors of the
enzyme138 and the establishment of an assay suitable
for high-throughput screening that links four enzymes required for Met and Thr biosynthesis, including homoserine dehydrogenase.139
6. Genomic and Systems Methods in

Antimicrobial Research
6.1. Transcriptomics
DNA microarrays that cover the genomic landscape
of a wide spectrum of bacteria and yeast have
presented a unique opportunity to probe the effects
of exposure to antibiotics on microbial gene tran-
scription. This information can be used to help

identify the molecular targets of antimicrobial agents
and simultaneously inform on the impact of antibiotics on the regulation of biochemical pathways, stress
response, resistance mechanisms, and pleiotropic
impact of antibiotics throughout the organism (Table
2). The approach was first used in profiling the effect
of the anti-tubercular drug, isoniazid, on gene transcription in M. tuberculosis.140 This antibiotic is a
prodrug that is activated by an intracellular peroxidase, KatG, and inhibits InhA, an enoyl acyl carrier
protein reductase essential for the biosynthesis of
mycolic acids that form a protective lipid barrier that
surrounds the organism in the macrophage.141 Microarray analysis of M. tuberculosis transcription in the
presence of isoniazid demonstrated that exposure to
the antibiotic increased levels of expression of genes
required for mycolic acid biosynthesis consistent with

the mode of action. These genes were not induced in
a KatG-deficient strain, supporting the hypothesis
that it is the oxidatively activated form of isoniazid
that is the antibiotic.140 In another study researchers
examined the effect of low levels of antibiotics that
inhibit bacterial translation on gene transcription in
Streptococcus pneumoniae.142 Levels of translationassociated genes were increased upon exposure to
these antibiotics as well as heat shock and purine
biosynthesis genes. Statistical analysis of the levels
and patterns of gene expression were informative,
demonstrating that antibiotics that blocked translation at different steps resulted in different patterns
of impact on gene expression and that antibiotics that
blocked the same steps produced similar patterns of
gene expression. These studies therefore have the
potential to be used to identify the targets of translation inhibitors of unknown mechanism. Most recently, a comprehensive analysis was published on
the transcriptional profile of B. subtilis on exposure
to 37 antibacterials of diverse mechanism and chemical class.143,144 This enabled the prediction of the
mechanism of action of various classes of antibiotics
by scrutiny of impact of sublethal doses of antibiotic
on the transcriptome143 and identification of promoters that lead to the development of 12 reporter
strains that informed on a variety of pathways.144 A
similar approach has also been reported by Fischer
and co-workers,92 who demonstrated proof of principle for the concept in a 900 000 compound screen
using a luciferase-fused promoter (fabHB) that was
identified to be responsive to fatty acid synthesis
inhibition.
Similarly, gene expression profiling has been used
to probe the impact of antifungal agents. Exposure
of S. cerevisiae to 5-flurocytosine, a widely used
antifungal drug that interferes with nucleic acid
metabolism, induced genes involved in nucleic acid
synthesis and repair and cell cycle control.145,146
Incubation of azole antifungals with either S. cerevisiae146,147 or C. albicans148 increased the expression
levels of ergosterol biosynthesis and sterol uptake
genes. Caspofungin, an echinocandin that blocks
fungal cell wall biosynthesis, induces expression of
genes required for maintenance of the integrity of the
wall.146 On the other hand, amphotericin, which
physically disrupts the fungal cell membrane, has
broader effects in cell membrane biosynthesis, transport, cell stress, metabolism, and cell wall integrity.146
Genome-wide transcript analysis can therefore greatly
assist in target identification and validation and
serve as a powerful tool in antimicrobial development.
6.2. Chemical Genomics

With the accumulation of antimicrobial targets in
recent years, including many that are incompletely
validated, large-scale approaches that facilitate prioritization and provide probes of poorly understood
cell physiology and biochemistry will be a welcome
addition. Methods that facilitate systematic target
validation and small molecule screening will be
Brown and Wright
amenable to emerging drug discovery formats that

emphasize parallelization and throughput. The growing emphasis on parallelization implies that new
approaches are needed that facilitate small molecule
screening efforts on large numbers of incompletely
characterized targets. In this so-called chemical
genomics strategy screening is put up front in the
discovery process so that a large number of small
molecule inhibitors are generated for a multitude of
targets. These provide decision points and impetus
for in-depth target characterization and validation.149
Target-selective inhibitors generated in this manner
are also a resource for the determination of protein
function and for validation and optimization of leads.
While the chemical genomics strategy has potential
for transforming drug discovery, especially antimicrobial drug discovery where so many targets are
readily exploited, its application awaits some further
technological advancement. Among these hurdles are
assay technologies that are suitable for parallelization. Perhaps the best example of a universal assay
is the use of peptides as surrogate ligands for protein
targets.150-152 In this approach a peptide which shows
affinity for a given protein target is selected from a
combinatorial peptide library. Such ligands can be
used in competitive binding assays involving small
molecule libraries. In this approach a small molecule
of interest displaces the surrogate peptide ligand, and
this is assessed using spectral measurements, most
often fluorescence.
Recently Liu et al.153 reported a method where
strategies used by bacteriophages to disable bacteria
provide starting points for a surrogate ligand approach. The concept begins with bacteriophage genomics and a search for key phage proteins that
disable vulnerable targets in bacteria. These targets
represent validated and novel points of weakness in
bacterial physiology. As such, the phage proteins
directed at these bacterial targets could serve as
novel probes to validate and characterize the bacterial targets. Furthermore, as binding partners for the
target proteins, the phage-encoded probes provide a
route to screen for small molecules that bind to the
bacterial target and displace the phage protein.
6.3. Phenotype-Based Screening Followed by

Target Discovery
Phenotype-based screens for small molecules with
biological activity, for example, growth inhibition in
the case of antimicrobials, have accounted for an
overwhelming majority of pharmaceutical drugs in
use today and are finding increasing use in a chemical genetics research paradigm emerging in academic
circles.154,155 One of the most significant hurdles in
the use of phenotype-based small molecule screens
is identification of the macromolecular target. In
contrast, the discovery and optimization of inhibitors
of protein function using target-based screens is
plagued with the dilemma of creating molecules that
penetrate and persist in cells in order that they have
biological activity. Indeed, there is growing appreciation that small molecule permeability in bacteria and
fungi is governed to an extent by ubiquitous multi-
drug efflux pumps,156 whose substrate specificities

are very broad and hard to define.157 Hence, phenotype-based screens have a particular advantage in
that bioactive molecules identified have physicalchemical properties that are compatible with microbial cell permeation. What are needed, however, are
robust genetic approaches for the identification of
protein targets of small molecules identified in phenotype-based small molecule screens of microbes.
Classically, identification of protein targets for
phenotype-perturbing small molecules has been accomplished biochemically using labeled or immobilized molecules. Recent advances in target identification have included DNA microarrays92 and threehybrid systems.158 A competitive growth assay using
a pool of bar-coded genome-wide heterozygous yeast
strains has been used to identify mutants that
fail to grow in the presence of growth-inhibitory
drugs.159-161 The procedure, termed drug-induced
haploinsufficiency, has been shown to be effective in
identifying the targets of well-characterized clinical
and agricultural agents. Most recently, Li et al.162
used an approach well known to microbial geneticists
as multicopy suppression. The technique involves
creation of a plasmid-encoded random genomic library followed by a selection or screen for clones that
have a suppressor phenotype. Proof of principle for
this system was achieved with the isolation of genes
encoding the targets of known antibiotics and of novel
antibacterial leads targeting dihydrofolate reductase.
Another cell-based approach to take advantage of the
importance of target copy number to growth inhibition by antibacterial small molecules is that reported
by DeVito et al.163 These researchers constructed an
array of E. coli strains where each strain was
engineered for low-level expression of an essential
gene product. The hypersensitivity of these engineered strains to inhibitors specific for the targeted
protein permitted the identification of new inhibitors
of bacterial growth that targeted MurA, catalyzing
the first committed step of peptidoglycan biosynthesis.
7. Conclusions
Filling the antimicrobial drug discovery pipeline
has never been as challenging as it is now. The evermounting problem of resistance fuels the need for
new agents; however, the retreat of many large
pharmaceutical and biotechnology companies from
the anti-infective arena guarantees that there will
be fewer therapeutic options in the future.5,164 Paradoxically, on the scientific front, we have perhaps
never been better equipped to discover new targets
and pathways suitable for antimicrobial drug development. The stunning progress in omics-scale research has now been available to the antimicrobial
community for several years. The screening methods
and targets discussed in this review present simply
the first pass at exploitation of these new opportunities. Nonetheless, the challenges of leveraging this
information into profitable drugs remain significant.
However, research in this area must continue if we
are to be able to address the real needs we will face
over the next several years.
8. Acknowledgments
We thank Jeff Schertzer for critical comments on
the manuscript. E.D.B is supported by a Canada
Research Chair in Microbial Biochemistry, and G.D.W.
is supported by a Canada Research Chair in Antibiotic Biochemistry.
9. References
(1) Bax, R.; Mullan, N.; Verhoef, J. Int. J. Antimicrob. Agents 2000,
16, 51.
(2) Coates, A.; Hu, Y.; Bax, R.; Page, C. Nat. Rev. Drug Discov. 2002,
1, 895.
(3) Barrett, C. T.; Barrett, J. F. Curr. Opin. Biotechnol. 2003, 14,
621.
(4) Projan, S. J. Curr. Opin. Microbiol. 2003, 6, 427.
(5) Spellberg, B.; Powers, J. H.; Brass, E. P.; Miller, L. G.; Edwards,
J. E., Jr. Clin. Infect. Dis. 2004, 38, 1279.
(6) Salgado, C. D.; Farr, B. M.; Calfee, D. P. Clin. Infect. Dis. 2003,
36, 131.
(8) Ford, C. W.; Zurenko, G. E.; Barbachyn, M. R. Curr. Drug
Targets Infect. Disord. 2001, 1, 181.
(9) LaPlante, K. L.; Rybak, M. J. Expert Opin. Pharmacother. 2004,
5, 2321.
(10) Haselbeck, R.; Wall, D.; Jiang, B.; Ketela, T.; Zyskind, J.; Bussey,
H.; Foulkes, J. G.; Roemer, T. Curr. Pharm. Des. 2002, 8, 1155.
(11) Fleischmann, R. D.; Adams, M. D.; White, O.; Clayton, R. A.;
Kirkness, E. F.; Kerlavage, A. R.; Bult, C. J.; Tomb, J. F.;
Dougherty, B. A.; Merrick, J. M.; et al. Science 1995, 269, 496.
(12) Tatusov, R. L.; Galperin, M. Y.; Natale, D. A.; Koonin, E. V.
Nucleic Acids Res. 2000, 28, 33.
(13) Lander, E. S.; Linton, L. M.; Birren, B.; Nusbaum, C.; Zody, M.
C.; Baldwin, J.; Devon, K.; Dewar, K.; Doyle, M.; FitzHugh, W.;
Funke, R.; Gage, D.; Harris, K.; Heaford, A.; Howland, J.; Kann,
L.; Lehoczky, J.; LeVine, R.; McEwan, P.; McKernan, K.;
Meldrim, J.; Mesirov, J. P.; Miranda, C.; Morris, W.; Naylor, J.;
Raymond, C.; Rosetti, M.; Santos, R.; Sheridan, A.; Sougnez, C.;
Stange-Thomann, N.; Stojanovic, N.; Subramanian, A.; Wyman,
D.; Rogers, J.; Sulston, J.; Ainscough, R.; Beck, S.; Bentley, D.;
Burton, J.; Clee, C.; Carter, N.; Coulson, A.; Deadman, R.;
Deloukas, P.; Dunham, A.; Dunham, I.; Durbin, R.; French, L.;
Grafham, D.; Gregory, S.; Hubbard, T.; Humphray, S.; Hunt,
A.; Jones, M.; Lloyd, C.; McMurray, A.; Matthews, L.; Mercer,
S.; Milne, S.; Mullikin, J. C.; Mungall, A.; Plumb, R.; Ross, M.;
Shownkeen, R.; Sims, S.; Waterston, R. H.; Wilson, R. K.; Hillier,
L. W.; McPherson, J. D.; Marra, M. A.; Mardis, E. R.; Fulton, L.
A.; Chinwalla, A. T.; Pepin, K. H.; Gish, W. R.; Chissoe, S. L.;
Wendl, M. C.; Delehaunty, K. D.; Miner, T. L.; Delehaunty, A.;
Kramer, J. B.; Cook, L. L.; Fulton, R. S.; Johnson, D. L.; Minx,
P. J.; Clifton, S. W.; Hawkins, T.; Branscomb, E.; Predki, P.;
Richardson, P.; Wenning, S.; Slezak, T.; Doggett, N.; Cheng, J.
F.; Olsen, A.; Lucas, S.; Elkin, C.; Uberbacher, E.; Frazier, M.
Nature 2001, 409, 860.
(14) Venter, J. C.; Adams, M. D.; Myers, E. W.; Li, P. W.; Mural, R.
J.; Sutton, G. G.; Smith, H. O.; Yandell, M.; Evans, C. A.; Holt,
R. A.; Gocayne, J. D.; Amanatides, P.; Ballew, R. M.; Huson, D.
H.; Wortman, J. R.; Zhang, Q.; Kodira, C. D.; Zheng, X. H.; Chen,
L.; Skupski, M.; Subramanian, G.; Thomas, P. D.; Zhang, J.;
Gabor Miklos, G. L.; Nelson, C.; Broder, S.; Clark, A. G.; Nadeau,
J.; McKusick, V. A.; Zinder, N.; Levine, A. J.; Roberts, R. J.;
Simon, M.; Slayman, C.; Hunkapiller, M.; Bolanos, R.; Delcher,
A.; Dew, I.; Fasulo, D.; Flanigan, M.; Florea, L.; Halpern, A.;
Hannenhalli, S.; Kravitz, S.; Levy, S.; Mobarry, C.; Reinert, K.;
Remington, K.; Abu-Threideh, J.; Beasley, E.; Biddick, K.;
Bonazzi, V.; Brandon, R.; Cargill, M.; Chandramouliswaran, I.;
Charlab, R.; Chaturvedi, K.; Deng, Z.; Di Francesco, V.; Dunn,
P.; Eilbeck, K.; Evangelista, C.; Gabrielian, A. E.; Gan, W.; Ge,
W.; Gong, F.; Gu, Z.; Guan, P.; Heiman, T. J.; Higgins, M. E.;
Ji, R. R.; Ke, Z.; Ketchum, K. A.; Lai, Z.; Lei, Y.; Li, Z.; Li, J.;
Liang, Y.; Lin, X.; Lu, F.; Merkulov, G. V.; Milshina, N.; Moore,
H. M.; Naik, A. K.; Narayan, V. A.; Neelam, B.; Nusskern, D.;
Rusch, D. B.; Salzberg, S.; Shao, W.; Shue, B.; Sun, J.; Wang,
Z.; Wang, A.; Wang, X.; Wang, J.; Wei, M.; Wides, R.; Xiao, C.;
Yan, C. Science 2001, 291, 1304.
(15) Myllykallio, H.; Leduc, D.; Filee, J.; Liebl, U. Trends Microbiol.
2003, 11, 220.
(16) Rohdich, F.; Kis, K.; Bacher, A.; Eisenreich, W. Curr. Opin.
Chem. Biol. 2001, 5, 535.
(17) Settembre, E.; Begley, T. P.; Ealick, S. E. Curr. Opin. Struct.
Biol. 2003, 13, 739.

(18) Gerdes, S. Y.; Scholle, M. D.; DSouza, M.; Bernal, A.; Baev, M.
V.; Farrell, M.; Kurnasov, O. V.; Daugherty, M. D.; Mseeh, F.;
Polanuyer, B. M.; Campbell, J. W.; Anantha, S.; Shatalin, K.
Y.; Chowdhury, S. A.; Fonstein, M. Y.; Osterman, A. L. J.
Bacteriol. 2002, 184, 4555.
(19) Schmid, M. B.; Kapur, N.; Isaacson, D. R.; Lindroos, P.; Sharpe,
C. Genetics 1989, 123, 625.
(20) Smith, V.; Chou, K. N.; Lashkari, D.; Botstein, D.; Brown, P. O.
Science 1996, 274, 2069.
(21) Hutchison, C. A.; Peterson, S. N.; Gill, S. R.; Cline, R. T.; White,
O.; Fraser, C. M.; Smith, H. O.; Venter, J. C. Science 1999, 286,
2165.
(22) Akerley, B. J.; Rubin, E. J.; Novick, V. L.; Amaya, K.; Judson,
N.; Mekalanos, J. J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 966.
(23) Gerdes, S. Y.; Scholle, M. D.; Campbell, J. W.; Balazsi, G.;
Ravasz, E.; Daugherty, M. D.; Somera, A. L.; Kyrpides, N. C.;
Anderson, I.; Gelfand, M. S.; Bhattacharya, A.; Kapatral, V.;
DSouza, M.; Baev, M. V.; Grechkin, Y.; Mseeh, F.; Fonstein, M.
Y.; Overbeek, R.; Barabasi, A. L.; Oltvai, Z. N.; Osterman, A. L.
J. Bacteriol 2003, 185, 5673.
(24) Forsyth, R. A.; Haselbeck, R. J.; Ohlsen, K. L.; Yamamoto, R.
T.; Xu, H.; Trawick, J. D.; Wall, D.; Wang, L.; Brown-Driver,
V.; Froelich, J. M.; C, K. G.; King, P.; McCarthy, M.; Malone,
C.; Misiner, B.; Robbins, D.; Tan, Z.; Zhu, Z. Y.; Carr, G.; Mosca,
D. A.; Zamudio, C.; Foulkes, J. G.; Zyskind, J. W. Mol. Microbiol.
2002, 43, 1387.
(25) Thanassi, J. A.; Hartman-Neumann, S. L.; Dougherty, T. J.;
Dougherty, B. A.; Pucci, M. J. Nucleic Acids Res. 2002, 30, 3152.
(26) Mori, H.; Isono, K.; Horiuchi, T.; Miki, T. Res. Microbiol. 2000,
151, 121.
(27) Kobayashi, K.; Ehrlich, S. D.; Albertini, A.; Amati, G.; Andersen,
K. K.; Arnaud, M.; Asai, K.; Ashikaga, S.; Aymerich, S.;
Bessieres, P.; Boland, F.; Brignell, S. C.; Bron, S.; Bunai, K.;
Chapuis, J.; Christiansen, L. C.; Danchin, A.; Debarbouille, M.;
Dervyn, E.; Deuerling, E.; Devine, K.; Devine, S. K.; Dreesen,
O.; Errington, J.; Fillinger, S.; Foster, S. J.; Fujita, Y.; Galizzi,
A.; Gardan, R.; Eschevins, C.; Fukushima, T.; Haga, K.; Harwood, C. R.; Hecker, M.; Hosoya, D.; Hullo, M. F.; Kakeshita,
H.; Karamata, D.; Kasahara, Y.; Kawamura, F.; Koga, K.; Koski,
P.; Kuwana, R.; Imamura, D.; Ishimaru, M.; Ishikawa, S.; Ishio,
I.; Le Coq, D.; Masson, A.; Mauel, C.; Meima, R.; Mellado, R. P.;
Moir, A.; Moriya, S.; Nagakawa, E.; Nanamiya, H.; Nakai, S.;
Nygaard, P.; Ogura, M.; Ohanan, T.; OReilly, M.; ORourke, M.;
Pragai, Z.; Pooley, H. M.; Rapoport, G.; Rawlins, J. P.; Rivas, L.
A.; Rivolta, C.; Sadaie, A.; Sadaie, Y.; Sarvas, M.; Sato, T.; Saxild,
H. H.; Scanlan, E.; Schumann, W.; Seegers, J. F.; Sekiguchi, J.;
Sekowska, A.; Seror, S. J.; Simon, M.; Stragier, P.; Studer, R.;
Takamatsu, H.; Tanaka, T.; Takeuchi, M.; Thomaides, H. B.;
Vagner, V.; van Dijl, J. M.; Watabe, K.; Wipat, A.; Yamamoto,
H.; Yamamoto, M.; Yamamoto, Y.; Yamane, K.; Yata, K.;
Yoshida, K.; Yoshikawa, H.; Zuber, U.; Ogasawara, N. Proc. Natl.
Acad. Sci. U.S.A. 2003, 100, 4678.
(28) Giaever, G.; Chu, A. M.; Ni, L.; Connelly, C.; Riles, L.; Veronneau, S.; Dow, S.; Lucau-Danila, A.; Anderson, K.; Andre, B.;
Arkin, A. P.; Astromoff, A.; El-Bakkoury, M.; Bangham, R.;
Benito, R.; Brachat, S.; Campanaro, S.; Curtiss, M.; Davis, K.;
Deutschbauer, A.; Entian, K. D.; Flaherty, P.; Foury, F.;
Garfinkel, D. J.; Gerstein, M.; Gotte, D.; Guldener, U.; Hegemann, J. H.; Hempel, S.; Herman, Z.; Jaramillo, D. F.; Kelly, D.
E.; Kelly, S. L.; Kotter, P.; LaBonte, D.; Lamb, D. C.; Lan, N.;
Liang, H.; Liao, H.; Liu, L.; Luo, C.; Lussier, M.; Mao, R.;
Menard, P.; Ooi, S. L.; Revuelta, J. L.; Roberts, C. J.; Rose, M.;
Ross-Macdonald, P.; Scherens, B.; Schimmack, G.; Shafer, B.;
Shoemaker, D. D.; Sookhai-Mahadeo, S.; Storms, R. K.; Strathern, J. N.; Valle, G.; Voet, M.; Volckaert, G.; Wang, C. Y.; Ward,
T. R.; Wilhelmy, J.; Winzeler, E. A.; Yang, Y.; Yen, G.; Youngman, E.; Yu, K.; Bussey, H.; Boeke, J. D.; Snyder, M.; Philippsen,
P.; Davis, R. W.; Johnston, M. Nature 2002, 418, 387.
(29) Roemer, T.; Jiang, B.; Davison, J.; Ketela, T.; Veillette, K.;
Breton, A.; Tandia, F.; Linteau, A.; Sillaots, S.; Marta, C.; Martel,
N.; Veronneau, S.; Lemieux, S.; Kauffman, S.; Becker, J.; Storms,
R.; Boone, C.; Bussey, H. Mol. Microbiol. 2003, 50, 167.
(30) Hamer, L.; Adachi, K.; Montenegro-Chamorro, M. V.; Tanzer,
M. M.; Mahanty, S. K.; Lo, C.; Tarpey, R. W.; Skalchunes, A.
R.; Heiniger, R. W.; Frank, S. A.; Darveaux, B. A.; Lampe, D.
J.; Slater, T. M.; Ramamurthy, L.; DeZwaan, T. M.; Nelson, G.
H.; Shuster, J. R.; Woessner, J.; Hamer, J. E. Proc. Natl. Acad.
Sci. U.S.A. 2001, 98, 5110.
(31) Hare, R. S.; Walker, S. S.; Dorman, T. E.; Greene, J. R.; Guzman,
L. M.; Kenney, T. J.; Sulavik, M. C.; Baradaran, K.; Houseweart,
C.; Yu, H.; Foldes, Z.; Motzer, A.; Walbridge, M.; Shimer, G. H.,
Jr.; Shaw, K. J. J. Bacteriol. 2001, 183, 1694.
(32) Denome, S. A.; Elf, P. K.; Henderson, T. A.; Nelson, D. E.; Young,
K. D. J. Bacteriol. 1999, 181, 3981.
(33) Bhavsar, A. P.; Zhao, X.; Brown, E. D. Appl. Environ. Microbiol.
2001, 67, 403.
(34) Guzman, L. M.; Belin, D.; Carson, M. J.; Beckwith, J. J.
Bacteriol. 1995, 177, 4121.
Brown and Wright

(35) Nathan, C. Nature 2004, 431, 899.
(36) Murdoch, D. R. Clin. Infect. Dis. 2003, 36, 1162.
(37) Mahan, M. J.; Tobias, J. W.; Slauch, J. M.; Hanna, P. C.; Collier,
R. J.; Mekalanos, J. J. Proc. Natl. Acad. Sci. U.S.A. 1995, 92,
669.
(38) Wang, J.; Mushegian, A.; Lory, S.; Jin, S. Proc. Natl. Acad. Sci.
U.S.A. 1996, 93, 10434.
(39) Retallack, D. M.; Deepe, G. S., Jr.; Woods, J. P. Microb. Pathog.
2000, 28, 169.
(40) Lee, Y. M.; Almqvist, F.; Hultgren, S. J. Curr. Opin. Pharmacol.
2003, 3, 513.
(41) Stratton, C. W. Emerg. Infect. Dis. 2003, 9, 10.
(42) Sun, D.; Cohen, S.; Mani, N.; Murphy, C.; Rothstein, D. M. J.
(43) DeCenzo, M.; Kuranda, M.; Cohen, S.; Babiak, J.; Jiang, Z. D.;
Su, D.; Hickey, M.; Sancheti, P.; Bradford, P. A.; Youngman, P.;
Projan, S.; Rothstein, D. M. J. Antibiot. (Tokyo) 2002, 55, 288.
(44) Hong, H. J.; Paget, M. S.; Buttner, M. J. Mol. Microbiol. 2002,
44, 1199.
(45) Baum, E. Z.; Montenegro, D. A.; Licata, L.; Turchi, I.; Webb, G.
C.; Foleno, B. D.; Bush, K. Antimicrob. Agents Chemother. 2001,
45, 3182.
(46) Hu, Y.; Helm, J. S.; Chen, L.; Ginsberg, C.; Gross, B.; Kraybill,
B.; Tiyanont, K.; Fang, X.; Wu, T.; Walker, S. Chem. Biol. 2004,
11, 703.
(47) Stachyra, T.; Dini, C.; Ferrari, P.; Bouhss, A.; van Heijenoort,
J.; Mengin-Lecreulx, D.; Blanot, D.; Biton, J.; Le Beller, D.
(48) Hohenwallner, W.; Wimmer, E. Clin. Chim. Acta 1973, 45, 169.
(49) Cogan, E. B.; Birrell, G. B.; Griffith, O. H. Anal. Biochem. 1999,
271, 29.
(50) Wong, K. K.; Kuo, D. W.; Chabin, R. M.; Fournier, C.; Gegnas,
L. D.; Waddell, S. T.; Marsilio, F.; Leiting, B.; Pompliano, D. L.
J. Am. Chem. Soc. 1998, 120, 13527.
(51) Wong, K. K.; Pompliano, D. L. Adv. Exp. Med. Biol. 1998, 456,
197.
(52) Barbosa, M. D.; Yang, G.; Fang, J.; Kurilla, M. G.; Pompliano,
D. L. Antimicrob. Agents Chemother. 2002, 46, 943.
(53) Beytia, E. D.; Porter, J. W. Annu. Rev. Biochem. 1976, 45, 113.
(54) Takahashi, S.; Kuzuyama, T.; Watanabe, H.; Seto, H. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, 9879.
(55) Campos, N.; Rodriguez-Concepcion, M.; Seemann, M.; Rohmer,
M.; Boronat, A. FEBS Lett. 2001, 488, 170.
(56) Freiberg, C.; Wieland, B.; Spaltmann, F.; Ehlert, K.; Brotz, H.;
Labischinski, H. J. Mol. Microbiol. Biotechnol. 2001, 3, 483.
(57) McAteer, S.; Coulson, A.; McLennan, N.; Masters, M. J. Bacteriol.
2001, 183, 7403.
(58) Bjork, G. R.; Ericson, J. U.; Gustafsson, C. E.; Hagervall, T. G.;
Jonsson, Y. H.; Wikstrom, P. M. Annu. Rev. Biochem. 1987, 56,
263.
(59) Sherman, M. M.; Petersen, L. A.; Poulter, C. D. J. Bacteriol.
1989, 171, 3619.
(60) Campbell, T. L.; Brown, E. D. J. Bacteriol. 2002, 184, 5609.
(61) Kuemmerle, H. P.; Murakawa, T.; Sakamoto, H.; Sato, N.;
Konishi, T.; De Santis, F. Int. J. Clin. Pharmacol. Ther. Toxicol.
1985, 23, 521.
(62) Zeidler, J.; Schwender, J.; Muller, C.; Weisner, J.; Weidemeyer,
C.; Beck, E.; Jomaa, H.; Lichtenthaler, H. K. 1998, 53, 980.
(63) Lell, B.; Ruangweerayut, R.; Wiesner, J.; Missinou, M. A.;
Schindler, A.; Baranek, T.; Hintz, M.; Hutchinson, D.; Jomaa,
H.; Kremsner, P. G. Antimicrob. Agents Chemother. 2003, 47,
735.
(64) Jomaa, H.; Wiesner, J.; Sanderbrand, S.; Altincicek, B.; Weidemeyer, C.; Hintz, M.; Turbachova, I.; Eberl, M.; Zeidler, J.;
Lichtenthaler, H. K.; Soldati, D.; Beck, E. Science 1999, 285,
1573.
(65) Altincicek, B.; Hintz, M.; Sanderbrand, S.; Wiesner, J.; Beck,
E.; Jomaa, H. FEMS Microbiol. Lett. 2000, 190, 329.
(66) Gottlin, E. B.; Benson, R. E.; Conary, S.; Antonio, B.; Duke, K.;
Payne, E. S.; Ashraf, S. S.; Christensen, D. J. J. Biomol. Screen
2003, 8, 332.
(67) Ward, J. B. Microbiol. Rev. 1981, 45, 211.
(68) Neuhaus, F. C.; Baddiley, J. Microbiol. Mol. Biol. Rev. 2003, 67,
686.
(69) Araki, Y.; Ito, E. Crit. Rev. Microbiol. 1989, 17, 121.
(70) Yokoyama, K.; Miyashita, T.; Araki, Y.; Ito, E. Eur. J. Biochem.
1986, 161, 479.
(71) Kaya, S.; Yokoyama, K.; Araki, Y.; Ito, E. Biochem. Biophys. Res.
Commun. 1983, 111, 312.
(72) Kaya, S.; Yokoyama, K.; Araki, Y.; Ito, E. J. Bacteriol. 1984,
158, 990.
(73) Kojima, N.; Araki, Y.; Ito, E. Eur. J. Biochem. 1985, 148, 29.
(74) Briehl, M.; Pooley, H. M.; Karamata, D. J. Gen. Microbiol. 1989,
135, 1325.
(75) Lazarevic, V.; Abellan, F. X.; Moller, S. B.; Karamata, D.; Mauel,
C. Microbiology 2002, 148, 815.
(76) Bhavsar, A. P.; Beveridge, T. J.; Brown, E. D. J. Bacteriol. 2001,
183, 6688.

(77) Bhavsar, A. P.; Erdman, L. K.; Schertzer, J. W.; Brown, E. D.
J. Bacteriol. 2004, 186, 7865.
(78) Park, Y. S.; Sweitzer, T. D.; Dixon, J. E.; Kent, C. J. Biol. Chem.
1993, 268, 16648.
(79) Badurina, D. S.; Zolli-Juran, M.; Brown, E. D. Biochim. Biophys.
Acta 2002, 1646, 196.
(80) Zolli, M.; Kobric, D. J.; Brown, E. D. Biochemistry 2001, 40, 5041.
(81) Pereira, M. P.; Brown, E. D. Biochemistry 2004, 43, 11802.
(82) Schertzer, J. W.; Brown, E. D. J. Biol. Chem. 2003, 278, 18002.
(83) Soldo, B.; Lazarevic, V.; Pooley, H. M.; Karamata, D. J. Bacteriol.
2002, 184, 4316.
(84) Soldo, B.; Lazarevic, V.; Karamata, D. Microbiology 2002, 148,
2079.
(85) Kay, B. K.; Hamilton, P. T. Comb. Chem. High Throughput
Screen 2001, 4, 535.
(86) Zolli-Juran, M.; Cechetto, J. D.; Hartlen, R.; Daigle, D. M.;
Brown, E. D. Bioorg. Med. Chem. Lett. 2003, 13, 2493.
(87) Heath, R. J.; Rock, C. O. J. Biol. Chem. 1995, 270, 26538.
(88) Heath, R. J.; Yu, Y. T.; Shapiro, M. A.; Olson, E.; Rock, C. O. J.
Biol. Chem. 1998, 273, 30316.
(89) Banerjee, A.; Dubnau, E.; Quemard, A.; Balasubramanian, V.;
Um, K. S.; Wilson, T.; Collins, D.; de Lisle, G.; Jacobs, W. R.,
Jr. Science 1994, 263, 227.
(90) Tsay, J. T.; Rock, C. O.; Jackowski, S. J. Bacteriol. 1992, 174,
508.
(91) Price, A. C.; Choi, K. H.; Heath, R. J.; Li, Z.; White, S. W.; Rock,
C. O. J. Biol. Chem. 2001, 276, 6551.
(92) Fischer, H. P.; Brunner, N. A.; Wieland, B.; Paquette, J.; Macko,
L.; Ziegelbauer, K.; Freiberg, C. Genome Res. 2004, 14, 90.
(93) Heath, R. J.; White, S. W.; Rock, C. O. Prog. Lipid Res. 2001,
40, 467.
(94) Economou, A. Mol. Membr. Biol. 2002, 19, 159.
(95) Kuo, D.; Weidner, J.; Griffin, P.; Shah, S. K.; Knight, W. B.
(96) Tschantz, W. R.; Paetzel, M.; Cao, G.; Suciu, D.; Inouye, M.;
Dalbey, R. E. Biochemistry 1995, 34, 3935.
(97) Alksne, L. E.; Burgio, P.; Hu, W.; Feld, B.; Singh, M. P.;
Tuckman, M.; Petersen, P. J.; Labthavikul, P.; McGlynn, M.;
Barbieri, L.; McDonald, L.; Bradford, P.; Dushin, R. G.; Rothstein, D.; Projan, S. J. Antimicrob. Agents Chemother. 2000, 44,
1418.
(98) Barbosa, M. D.; Lin, S.; Markwalder, J. A.; Mills, J. A.; DeVito,
J. A.; Teleha, C. A.; Garlapati, V.; Liu, C.; Thompson, A.; Trainor,
G. L.; Kurilla, M. G.; Pompliano, D. L. Antimicrob. Agents
Chemother. 2002, 46, 3549.
(99) Paetzel, M.; Dalbey, R. E.; Strynadka, N. C. Nature 1998, 396,
186.
(100) Paetzel, M.; Goodall, J. J.; Kania, M.; Dalbey, R. E.; Page, M.
G. J. Biol. Chem. 2004, 279, 30781.
(101) Schimana, J.; Gebhardt, K.; Holtzel, A.; Schmid, D. G.;
Sussmuth, R.; Muller, J.; Pukall, R.; Fiedler, H. P. J. Antibiot.
(Tokyo) 2002, 55, 565.
(102) Dev, I. K.; Harvey, R. J.; Ray, P. H. J. Biol. Chem. 1985, 260,
5891.
(103) Mazel, D.; Pochet, S.; Marliere, P. EMBO J. 1994, 13, 914.
(104) Chang, S. Y.; McGary, E. C.; Chang, S. J. Bacteriol. 1989, 171,
4071.
(105) Chen, D. Z.; Patel, D. V.; Hackbarth, C. J.; Wang, W.; Dreyer,
G.; Young, D. C.; Margolis, P. S.; Wu, C.; Ni, Z. J.; Trias, J.;
White, R. J.; Yuan, Z. Biochemistry 2000, 39, 1256.
(106) Guilloteau, J. P.; Mathieu, M.; Giglione, C.; Blanc, V.; Dupuy,
A.; Chevrier, M.; Gil, P.; Famechon, A.; Meinnel, T.; Mikol, V.
J. Mol. Biol. 2002, 320, 951.
(107) Yuan, Z.; Trias, J.; White, R. J. Drug Discov. Today 2001, 6,
954.
(108) Arigoni, F.; Talabot, F.; Peitsch, M.; Edgerton, M. D.; Meldrum,
E.; Allet, E.; Fish, R.; Jamotte, T.; Curchod, M. L.; Loferer, H.
Nat. Biotechnol. 1998, 16, 851.
(109) Zalacain, M.; Biswas, S.; Ingraham, K. A.; Ambrad, J.; Bryant,
A.; Chalker, A. F.; Iordanescu, S.; Fan, J.; Fan, F.; Lunsford, R.
D.; ODwyer, K.; Palmer, L. M.; So, C.; Sylvester, D.; Volker, C.;
Warren, P.; McDevitt, D.; Brown, J. R.; Holmes, D. J.; Burnham,
M. K. J. Mol. Microbiol. Biotechnol. 2003, 6, 109.
(110) Daigle, D. M.; Rossi, L.; Berghuis, A. M.; Koonin, E. V.; Brown,
E. D. Biochemistry 2002, 41, 11109.
(111) Levdikov, V. M.; Blagova, E. V.; Brannigan, J. A.; Cladiere, L.;
Antson, A. A.; Isupov, M. N.; Seror, S. J.; Wilkinson, A. J. J.
Mol. Biol. 2004, 340, 767.
(112) Daigle, D. M.; Brown, E. D. J. Bacteriol. 2004, 186, 1381.
(113) Campbell, T. L.; Daigle, D. M.; Brown, E. D. Unpublished results.
(114) Hang, J. Q.; Meier, T. I.; Zhao, G. Eur. J. Biochem. 2001, 268,
5570.
(115) Inoue, K.; Alsina, J.; Chen, J.; Inouye, M. Mol. Microbiol. 2003,
48, 1005.
(116) Hang, J. Q.; Zhao, G. Eur. J. Biochem. 2003, 270, 4164.
(117) Hwang, J.; Inouye, M. J. Biol. Chem. 2001, 276, 31415.
(118) Robinson, V. L.; Hwang, J.; Fox, E.; Inouye, M.; Stock, A. M.
Structure (Cambridge) 2002, 10, 1649.

(119) Lin, B.; Covalle, K. L.; Maddock, J. R. J. Bacteriol. 1999, 181,
5825.
(120) Okamoto, S.; Ochi, K. Mol. Microbiol. 1998, 30, 107.
(121) Vidwans, S. J.; Ireton, K.; Grossman, A. D. J. Bacteriol. 1995,
177, 3308.
(122) Lin, B.; Thayer, D. A.; Maddock, J. R. J. Bacteriol. 2004, 186,
481.
(123) Kukimoto-Niino, M.; Murayama, K.; Inoue, M.; Terada, T.; Tame,
J. R.; Kuramitsu, S.; Shirouzu, M.; Yokoyama, S. J. Mol. Biol.
2004, 337, 761.
(124) Wout, P.; Pu, K.; Sullivan, S. M.; Reese, V.; Zhou, S.; Lin, B.;
Maddock, J. R. J. Bacteriol. 2004, 186, 5249.
(125) Teplyakov, A.; Obmolova, G.; Tordova, M.; Thanki, N.; Bonander,
N.; Eisenstein, E.; Howard, A. J.; Gilliland, G. L. Proteins 2002,
48, 220.
(126) Allali-Hassani, A.; Campbell, T. L.; A., H.; Schertzer, J. W.;
Brown, E. D. Biochem. J. 2004, 384, 577.
(127) Parks, L. W. CRC Crit. Rev. Microbiol. 1978, 6, 301.
(128) Parks, L. W.; Casey, W. M. Annu. Rev. Microbiol. 1995, 49, 95.
(129) Masuoka, J. Clin. Microbiol. Rev. 2004, 17, 281.
(130) Denning, D. W. Lancet 2003, 362, 1142.
(131) Evans, J. M.; Zaworski, P. G.; Parker, C. N. J. Biomol. Screen
2002, 7, 359.
(132) Gomez-Lorenzo, M. G.; Garcia-Bustos, J. F. J. Biol. Chem. 1998,
273, 25041.
(133) Dominguez, J. M.; Kelly, V. A.; Kinsman, O. S.; Marriott, M. S.;
Gomez de las Heras, F.; Martin, J. J. Antimicrob. Agents
Chemother. 1998, 42, 2274.
(134) Fujiu, M.; Sawairi, S.; Shimada, H.; Takaya, H.; Aoki, Y.; Okuda,
T.; Yokose, K. J. Antibiot. (Tokyo) 1994, 47, 833.
(135) Kugler, M.; Loeffler, W.; Rapp, C.; Kern, A.; Jung, G. Arch.
Microbiol. 1990, 153, 276.
(136) Yamaki, H.; Yamaguchi, M.; Tsuruo, T.; Yamaguchi, H. J.
(137) Jacques, S. L.; Mirza, I. A.; Ejim, L.; Koteva, K.; Hughes, D. W.;
Green, K.; Kinach, R.; Honek, J. F.; Lai, H. K.; Berghuis, A. M.;
Wright, G. D. Chem. Biol. 2003, 10, 989.
(138) Ejim, L.; Mirza, I. A.; Capone, C.; Nazi, I.; Jenkins, S.; Chee, G.
L.; Berghuis, A. M.; Wright, G. D. Bioorg. Med. Chem. 2004, 12,
3825.
(139) Bareich, D. C.; Nazi, I.; Wright, G. D. Chem. Biol. 2003, 10, 967.
(140) Wilson, M.; DeRisi, J.; Kristensen, H. H.; Imboden, P.; Rane,
S.; Brown, P. O.; Schoolnik, G. K. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 12833.
(141) Rozwarski, D. A.; Grant, G. A.; Barton, D. H.; Jacobs, W. R.,
Jr.; Sacchettini, J. C. Science 1998, 279, 98.
(142) Ng, W. L.; Kazmierczak, K. M.; Robertson, G. T.; Gilmour, R.;
Winkler, M. E. J. Bacteriol. 2003, 185, 359.
(143) Hutter, B.; Schaab, C.; Albrecht, S.; Borgmann, M.; Brunner,
N. A.; Freiberg, C.; Ziegelbauer, K.; Rock, C. O.; Ivanov, I.;
Loferer, H. Antimicrob. Agents Chemother. 2004, 48, 2838.
(144) Hutter, B.; Fischer, C.; Jacobi, A.; Schaab, C.; Loferer, H.
(145) Zhang, L.; Zhang, Y.; Zhou, Y.; Zhao, Y.; Cheng, J. Int. J.
Antimicrob. Agents 2002, 20, 444.
(146) Agarwal, A. K.; Rogers, P. D.; Baerson, S. R.; Jacob, M. R.;
Barker, K. S.; Cleary, J. D.; Walker, L. A.; Nagle, D. G.; Clark,
A. M. J. Biol. Chem. 2003, 278, 34998.
(147) Bammert, G. F.; Fostel, J. M. Antimicrob. Agents Chemother.
2000, 44, 1255.
(148) De Backer, M. D.; Ilyina, T.; Ma, X. J.; Vandoninck, S.; Luyten,
W. H.; Vanden Bossche, H. Antimicrob. Agents Chemother. 2001,
45, 1660.
(149) Croston, G. E. Trends Biotechnol. 2002, 20, 110.
(150) Hyde-DeRuyscher, R.; Paige, L. A.; Christensen, D. J.; HydeDeRuyscher, N.; Lim, A.; Fredericks, Z. L.; Kranz, J.; Gallant,
P.; Zhang, J.; Rocklage, S. M.; Fowlkes, D. M.; Wendler, P. A.;
Hamilton, P. T. Chem. Biol. 2000, 7, 17.
(151) Tao, J.; Wendler, P.; Connelly, G.; Lim, A.; Zhang, J.; King, M.;
Li, T.; Silverman, J. A.; Schimmel, P. R.; Tally, F. P. Proc. Natl.
Acad. Sci. U.S.A. 2000, 97, 783.
(152) Christensen, D. J.; Gottlin, E. B.; Benson, R. E.; Hamilton, P.
T. Drug Discov. Today 2001, 6, 721.
(153) Liu, J.; Dehbi, M.; Moeck, G.; Arhin, F.; Bauda, P.; Bergeron,
D.; Callejo, M.; Ferretti, V.; Ha, N.; Kwan, T.; McCarty, J.;
Srikumar, R.; Williams, D.; Wu, J. J.; Gros, P.; Pelletier, J.;
DuBow, M. Nat. Biotechnol. 2004, 22, 185.
(154) Strausberg, R. L.; Schreiber, S. L. Science 2003, 300, 294.
(155) Brown, E. D. J. Biomol. Screen 2003, 8, 377.
(156) Li, X. Z.; Nikaido, H. Drugs 2004, 64, 159.
(157) Yu, E. W.; Aires, J. R.; Nikaido, H. J. Bacteriol. 2003, 185, 5657.
(158) Licitra, E. J.; Liu, J. O. Proc. Natl. Acad. Sci. U.S.A. 1996, 93,
12817.
(159) Giaever, G.; Shoemaker, D. D.; Jones, T. W.; Liang, H.; Winzeler,
E. A.; Astromoff, A.; Davis, R. W. Nat. Genet. 1999, 21, 278.
(160) Baetz, K.; McHardy, L.; Gable, K.; Tarling, T.; Reberioux, D.;
Bryan, J.; Andersen, R. J.; Dunn, T.; Hieter, P.; Roberge, M. Proc.
Natl. Acad. Sci. U.S.A. 2004, 101, 4525.

(161) Lum, P. Y.; Armour, C. D.; Stepaniants, S. B.; Cavet, G.; Wolf,
M. K.; Butler, J. S.; Hinshaw, J. C.; Garnier, P.; Prestwich, G.
D.; Leonardson, A.; Garrett-Engele, P.; Rush, C. M.; Bard, M.;
Schimmack, G.; Phillips, J. W.; Roberts, C. J.; Shoemaker, D.
D. Cell 2004, 116, 121.
(162) Li, X.; Zolli-Juran, M.; Cechetto, J. D.; Daigle, D. M.; Wright,
G. D.; Brown, E. D. Chem. Biol. 2004, 11, 1423.
Brown and Wright

(163) DeVito, J. A.; Mills, J. A.; Liu, V. G.; Agarwal, A.; Sizemore, C.
F.; Yao, Z.; Stoughton, D. M.; Cappiello, M. G.; Barbosa, M. D.;
Foster, L. A.; Pompliano, D. L. Nat. Biotechnol. 2002, 20,
478.
(164) Wenzel, R. P. N Engl. J. Med. 2004, 351, 523.
CR030116O

Antibiotic Resistance

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Antibiotic Resistance

Uploaded by

Copyright:

Available Formats

Volume 105, Number 2

Introduction: Antibiotic Resistance

Christopher T. Walsh is the Hamilton Kuhn Professor of Biological

Gerry Wright is Professor and Chair of the Department of Biochemistry

One of the central themes of success in human

In the course of these discovery efforts, the active

10.1021/cr030100y CCC: $53.50 2005 American Chemical Society

392 Chemical Reviews, 2005, Vol. 105, No. 2

Table 1. History of Antibiotic Classes in Clinical Use

waves of resistant bacteria. In the macrolide lineage,

transmembrane efflux pumps, keeping antibiotic

cells, discussing fluoroquinolones that target type II

Chemical Reviews, 2005, Vol. 105, No. 2 393

new targets for antibiotics are discovered in bacteria

Chem. Rev. 2005, 105, 395424

Bacterial Resistance to -Lactam Antibiotics: Compelling Opportunism,

cephems) -lactams but now includes natural and

10.1021/cr030102i CCC: $53.50 2005 American Chemical Society

396 Chemical Reviews, 2005, Vol. 105, No. 2

Jed F. Fisher is an alumnus of S.U.N.Y. Stony Brook (B.S., under Bill

Samy Meroueh studied chemistry at Wayne State University, where he

-lactams, the (ancient) transformation of these

Shahriar Mobashery received his undergraduate (1981) and doctoral (1985)

primary literature of the past 4 years with citation

Bacterial Resistance to -Lactam Antibiotics

then the enzyme will fail to complete its catalytic

Chemical Reviews, 2005, Vol. 105, No. 2 397

may be achieved by alteration of the transpeptidase

398 Chemical Reviews, 2005, Vol. 105, No. 2

recognized as a peptidomimetic by both enzyme

the divisome when as part of cell division) to bacterial

wall into its own structure. As anticipated, compound

Bacterial Resistance to -Lactam Antibiotics

strand of the peptidoglycan (red portions in structure

Chemical Reviews, 2005, Vol. 105, No. 2 399

400 Chemical Reviews, 2005, Vol. 105, No. 2

recognized by this ancestral enzyme as a mimic of

Figure 2. (A) Stereoview of the active site of the PBP2x

conserved motifs (Figure 2C). The first motif is the

Bacterial Resistance to -Lactam Antibiotics

thus facilitating its activation as a nucleophile and

Chemical Reviews, 2005, Vol. 105, No. 2 401

The expansion of resistant PBPs is a medical

402 Chemical Reviews, 2005, Vol. 105, No. 2

during chemotherapy is insufficient to inactivate the

high molecular weight PBP, and its structure was

2.2. -Lactam-Sensing Proteins

Bacterial Resistance to -Lactam Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 403

Figure 3. (A) Schematic of the various proteins that are

licheniformis has been recently solved by X-ray

404 Chemical Reviews, 2005, Vol. 105, No. 2

active serine residue where a valine residue would

Bacterial Resistance to -Lactam Antibiotics

these developments have been outlined by Bush and

Chemical Reviews, 2005, Vol. 105, No. 2 405

A summary of the mechanistic expectations for

406 Chemical Reviews, 2005, Vol. 105, No. 2

already on this face and makes the self-annihilation

3.2. Class A -Lactamases

diffusion controlled, where the apparent second-order

Bacterial Resistance to -Lactam Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 407

structure of the Streptomyces R61 D,D-peptidase/