You are on page 1of 41

ANNUAL

REVIEWS Further
Quick links to online content

Ann. Rev. Biochem. 1977. 46:723-63


Copyright © 1977 by Annual Reviews Inc. All rights reserved

POLYMYXIN AND RELATED -:-960

PEPTIDE ANTIBIOTICS!
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

Daniel R. Storm, Kenneth S. Rosenthal,


and Paul E Swanson

Department of Biochemistry, University of Illinois, Urbana, Illinois 61801

CONTENTS

PERSPECTIVES AND SUMMARY ................. ........................................................... 724


STRUCTURES OF THE PEPTIDES .......................................................................... 725
Native Antibiotics ......................................... .............................. ...........................
. . 725
Synthetic Derivatives ..................................... ......................................................... 727
ANTIMICROBIAL ACTIVITIES ................................................................................ 728
Native Antibiotics .................................................................................................... 728
Synthetic Derivatives .............................................................................................. 730
EFFECTS ON BACTERIAL METABOLISM ............................................................ 732
General Considerations .......................................................................................... 732
Membrane Permeability ........................................................................................ 733
Bacterial Respiration.............................................................................................. 735
INTERACTIONS WITH BACTERIAL MEMBRANES .......................................... 740
Outer Membranes .................................................................................................. 740
Cytoplasmic Membranes........................................................................................ 746
MODEL MEMBRANE STUDIES................................................................................ 748
Phospholipid and Lipopolysaccharide Aggregates ............................................ 749
Monolayers................................................................................................................ 751
Liposomes.................................................................................................................. 752
IMMOBILIZATION OF POLYMYXIN B AND OCTAPEPTIN............................ 753
MECHANISM OF ACTION ........................................................................................ 757
CONCLUSIONS.............................................................................................................. 759

lAbbreviations used: Dab, 2,4-diaminobutyric acid; aMG, methyl-a-D­


glucopyranoside; SDS, sodium dodecyl sulfate; DPH, 1,6-diphenyll , 3 , 5 -hexatriene;
DANSP, I-dimethylaminonaphthalene-5-sulfonamidopolymyxin.

723
724 STORM, ROSENTHAL & SWANSON

PERSPECTIVES AND SUMMARY

Various membrane�active antibiotics, including polymyxin and octapeptin,


have served as useful tools for studying the structure and function of biolog­
ical membranes ( 1-9). In addition, a number of the peptide antibiotics have
strong affinities for membrane lipids, and therefore can be used as well­
defined model systems for studying the chemistry of lipid-peptide interac­
tions (3, 1 0, 1 1). A prerequisite for the application of these antibiotics as
probes of membrane structure and function is a detailed description of the
mechanism for their antimicrobial activities. The objective of this chapter
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

is to review the research concerning the mechanism of action of the poly­


myxins and several related peptides. The peptides under consideration in­
clude the polymyxins, octapeptins, brevistin, stendomycin, polypeptin, and
cerexin, all of which have been isolated from various Bacillus strains. These
antibiotics are grouped together because they are all cyclic polycationic
peptides, containing a high percentage of diaminobutyric acid with a fatty
acid attached through an amide linkage. In addition, there is some evidence
that the polymyxins and octapeptins affect bacterial membrane structure
and function in a similar manner (9).
The first of these peptides to be isolated were the polymyxins, which were
obtained from a strain of Bacillus polymyxa in 1 947 ( 1 2-14). The circulins
and colistins were originally thought to be quite distinct from the polymyx­
ins but were later correctly classified as polymyxins ( 1 5 , 1 6). Octapeptin,
also referred to as EM 49 ( 1 7), brevistin ( 1 8-2 1), stendomycin (22), poly­
peptin (23), and cerexin (20) have all been isolated andlor characterized
within the last seven years. The polymyxins and octapeptins have been
studied most extensively; however, the other members of this family of
peptides may ultimately expand the range of compounds available for struc­
ture-function correlations. One of the objectives of this review is to compare
and contrast the effects of the polymyxins and octapeptins on bacterial
membranes.
It is well-established that the polymyxins and octapeptins alter bacterial
membrane structure (9, 24), although the detailed mechanisms for their
antimicrobial activities are still undefined. The amphipathic character of
these peptides might suggest that their disruptive influence on membrane
structure is completely analogous to simple cationic detergents. However,
these antibiotics exhibit biostatic or biocidal activities at concentrations
much lower than the common cationic detergents (9). Furthermore, rela­
tively minor structural modifications of these peptides can result in signifi­
cant changes in their biological activities. These general observations
suggest that interactions between these antibiotics and bacterial membranes
involve some degree of specificity beyond that expected for simple deter-
POLYMYXIN AND PEPTIDE ANTIBIOTICS 725

gents. There have been numerous attempts to identify the specific mem­
brane receptors for polymyxin. It is not clear, however, that association
between the peptide and the membrane lipid phase is ne,cessarily so specific
that the term receptor should be invoked for this interaction. It was
proposed in the early 1 950s that the primary effect of the polymyxins on
bacteria was a disorganization of membrane structure (24, 25). Twenty
years later we are still asking similar questions, but at a more detailed level.
How do these peptides perturb bacterial membrane structure, and why do
these changes lead to inhibition of bacterial growth? The answers to these
questions have been sought with a variety of microbiological, biochemical,
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

and physical techniques.


The timing of this review was dictated by the rapid proliferation of
research in this area and a more complete understanding of bacterial mem­
brane structure, which has accumulated since this topic was last reviewed
in depth. In addition, the recent discovery that polymyxin B and octapeptin
can be covalently attached to agarose beads with retention of antibiotic
activity (8) has made it possible to study interactions between these antibiot­
ics and the outer surface of intact bacterial cells. The polymyxins were first
extensively reviewed by Newton in 1956 (2), and several other relevant
reviews have been published since that time (26-32).

STRUCTURES OF THE PEPTIDES


Native Antibiotics
The polymyxins, circulins, and octapeptins are all related in structure and
general mode of action. Like most peptide antibiotics, they contain a mix­
ture of D- and L-amino acids. They are characterized by a heptapeptide ring,
a high percentage of 2,4-diaminobutyric acid (Dab), and a fatty acid at­
tached to the peptide through an amide bond (Figure 1). The peptides are
cyc1ized through the a-amino and carboxyl groups of the Dab residue at
position four, with. an additional peptide chain attached through the
y-amino group of this residue. The amino acid sequences of these antibio­
tics are all quite similar, with only two variable positions within the ring.
The polymyxins and circulins are decapeptides with molecular weights
in the range of 1 200 (29). They contain five to six Dab residues and variable
amounts of serine, threonine, D- or L-Ieucine, D-phenylalanine, and L­
isoleucine. In addition to the two variable amino acid positions of the
peptide ring, a third variable position is found in the peptide side chain
(Figure I). Either of two fatty acids can be found attached to the peptide,
6-methyl-octanoic acid or 6-methyl-heptanoic acid. These two fatty acyl
derivatives are usually distinguished by a subscript, e.g. polymyxin Bb B2,
and can be separated from one another by counter-current distribution
726 STORM, ROSENTHAL & SWANSON

OCTAPEPTIN ACYL ------.D-DAB-+ L-DAB � L-DAB � * � * 4 L-DAB -, L-DAB • L-LEU J


POLYMYXIN AC YL � L-DAB � L-THR +

1 2 3 4 5 6 7 8 9 10

ANTIBIOTIC COMPOSITION OF VARIABLE REGIONS

ACYL 3 6 7 REF,
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

POLYMYXIN A (M) MOA D-DAB D-LEU L-THR 15

POLYMYXIN Bl MOA L-DAB D-PHE L·LEU 46

lOA L-DAB D-PHE L-LEU 46


112
POLYMYXI tl Dl MOA D-SER D-LEU L-THR 165

POLYMYXIN El (COLISTIN A) MOA L-DAB D-LEU L-LEU 166

E2 (COLISTIN B) lOA L-DAB D-LEU L-LEU 166

CIRCULlN A MOA L-DAB D - LEU L·ILEU 167


OCTAPEPT! N Al EM49$ $·OH MD D-LEU L-LEU 47

A2 ....... {3 -OH MN D-LEU L-LEU 47


EM4Sa
A3/ {3 -OH ND D-LEU L-LE U 47

OCTAPEPTIN Bl EM 498 {3 -OH MD D-LEU L-PHE 47

B2 ....... {3-0H �1N D-LEU L-PHE 47


EM 49y
B3/ {3-0H ND D - LEU L-PHE 47
OCTAPEPT! N Cl 333-25 {3-0H MOA D-PHE L-LEU 18

BUL88 0 {3-0H MD 168


Y8495 168

Figure 1 Structures of the polymyxins and octapeptins. MOA, 6-methyloctanoic


acid; lOA, 6-methylheptanoic acid; MD, 8-methyldecanoic acid; MN, 8-methyl­
nonanoic acid; ND, n-decanoic acid. Structures of octapeptins C1 (Bu 1880 and
Y8495) have not been published.

(33). Differences in the amino acid and fatty acid compositions of the
polymyxins affect the biological activities of the peptides, and this is dis­
cussed in a later section.
The octapeptins differ from the polymyxins and circulins in that they
contain only eight amino acids, as the name implies, and have longer {3-
hydroxy fatty acids linked to the peptide (Figure 1). In addition, the only
amino acids found in the octapeptins are leucine, phenylalanine, and Dab.
The octapeptin fatty acids include the straight-chain J3-hydroxydecanoic
acid and the branched-chain 3-hydroxy-8-methyl nonanoic or 3-hydroxy-8-
methyldecanoic acids. Octapeptin classes A and B each contain a mixture
of these fatty acids. Although the octapeptins are similar in structure to the
POLYMYXIN AND PEPTIDE ANTIBIOTICS 727

polymyxins and circulins, distinct differences in their antimicrobial spec­


trum and biochemical actions have been reported (9, 34).
Other antibiotics containing a peptide ring with an acyl group attached
to the peptide have been isolated from Bacillus strains. They differ from the
polymyxins and octapeptins in their amino acid compositions, size of the
cyclic peptide portion, the means of cyclization (a lactone instead of an
amide bond), the nature of the fatty acid, as well as the bacterial source of
the compound. These compounds include brevistin ( 1 8, 1 9, 2 1 ), cerexin
(20), polypeptin (23), and stendomycin (22).
Attempts to correlate the secondary structure of the polymyxins in solu­
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

tion with antibiotic activities have had limited success. These studies have
utilized proton and l3C NMR (35, 36), tritium exchange (37), and thin-film
dialysis (38). Chapman & Golden (39) and Urry & Ohnishi (40) have
proposed secondary structures for polymyxin BJ, E, and circulin that con­
tain 2-5 amide-carbonyl hydrogen bonds. Deuterium exchange, measured
by NMR, and the temperature dependence of the amide chemical shifts
were reported to be consistent with this hydrogen-bonding scheme. Al­
though slow exchanging protons were detected with the chloride salts of
polymyxin Bl> E, and circulin, the corresponding sulfate salts did not
exhibit this phenomenon, nor did polymyxin A· HC l . Chapman & Golden
(39) concluded that a cross-beta conformation (41 ) for the polymyxins
would best explain the observed hydrogen bonding. In contrast, Craig has
proposed an extended structure for polymyxin (37) that does not contain
extensive intramolecular hydrogen bonding. Galardy, Craig & Printz (37)
have suggested that models involving extensive intramolecular hydrogen
bonding are invalid because the methods used by Chapman & Golden were
not reliable indications of hydrogen bonding. Considerably more informa­
tion concerning the secondary and tertiary structures of these pep tides is
required before meaningful correlations between the structures of the an­
tibiotics and their biological activities can be made.

Synthetic Derivatives
Numerous synthetic modifications of the polymyxins and octapeptins have
been made in order to alter their biological activities. These derivatives have
been valuable for identifying structural elements crucial for antibiotic activ­
ity. Structural modifications have included derivatization of the Dab "1-
amino groups, removal of the peptide side chain or fatty acid, replacement
of the fatty acid with other fatty acids, and changes in peptide sequence.
Attachment of 14C acetyl (42) or dimethylaminonaphthalene-5-sulfonyl
(24) groups to the "I-amino functions of Dab have provided radioactive- and
fluorescent-labeled polymyxins for studying interactions with whole cells or
membranes. Other analogues have been made by coupling single amino
728 STORM, ROSENTHAL & SWANSON

acids to the Dab 'Y-amino groups through amide linkages or Schiff base
formation with these same amino groups. The Schiff base derivatives are
extremely unstable and spontaneously hydrolyze in aqueous solutions (un­
published observations, Storm et al).
Total chemical synthesis of the polymyxins was used to aid in the deter­
mination of their structures. Vogler et al (43, 44) synthesized four cyclic
variants of polymyxin, which included the cyclic heptapeptide and octapep­
tide cyclized through either the a or '}' amino group of the appropriate Dab
residue. By similar methods, a polymyxin E analogue was synthesized with
lysines replacing all Dab residues (45). Another major class of polymyxin
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

or octapeptin derivatives includes those with the fatty acid either removed
or substituted with other fatty acids. Fatty acids have been removed from
the polymyxins (46) by digestion with subtilopeptidase A (EC 3.4.4. 1 6), and
deacylation of octapeptin has been done chemically by oxidation of the /3-
hydroxyl group of the fatty acid followed by treatment with hydroxylamine
(47). Reacylation with various fatty acids has provided a number of interest­
ing polymyxin and octapeptin derivatives that have been useful for defining
the role of the fatty acids for the biological activities of these peptides (48,
49).

ANTIMICROBIAL ACTIVITIES

Native Antibiotics

The polymyxins and circulins are broad-spectrum antibiotics exhibiting


activity against gram-positive and gram-negative bacteria, yeasts, and
protozoa (50). Characteristically, these peptides are biostatic at low concen­
trations and biocidal at higher concentrations and show complete cross­
resistance to each other. Unlike many other antibiotics, e.g. aminoglyco­
sides and penicillins (5 1 ), the bactericidal activity of the polymyxins has
changed little since their pharmaceutical introduction (52).
The polymyxins and circulins are quite active against most strains of
gram-negative bacteria. Because of their lower toxicities for eukaryotic
cells, polymyxins B and E have been studied most extensively. Polymyxins
B and E are very effective against strains of Escherichia coli, Pseudomonas
aeruginosa, Salmonella, Haemophilus, Shigella, Pasturella, Brucella,
Aerobacter aerogenes, and Bordatella bronchiseptica, with minimum inhibi­
tory concentrations less than 2 fLg/ml (50, 53). Naturally occurring poly­
myxin-resistant strains include Proteus mirabilis, Serratia marcescens (50),
Providencia (54), and Edwardsiella tarda (55), with minimum inhibitory
concentrations greater than 200 fLg/ml. Although polymyxin B resistance
has been induced in strains of P. aeruginosa by growth in increasing concen­
trations of the antibiotic, this was,an adaptive phenomenon, and resistance
POLYMYXIN AND PEPTIDE ANTIBIOTICS 729

was lost after one generation of growth in the absence of polymyxin B (56).
A number of polymyxin-resistant E. coli strains have been isolated that
include E coli SC 9252, SC 9253, and SC 8600 ( 1 7). Unfortunately, most
of the polymyxin-resistant mutants have not been characterized genetically.
Gram-positive bacteria are generally less sensitive to polymyxins than
gram-negative bacteria (28, 53). However, there are polymyxin-sensitive
strains of Staphylococcus, Bacillus, Streptococcus pyogenes, and Corynebac­
terium with minimum inhibitory concentrations less than 2 }Lg/ml. The
polymyxin- and circulin-producing strains, B. polymyxa and Bacillus circu­
lans, are resistant to these antibiotics; however, the basis for this resistance
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

has not been thoroughly investigated. The effects of the polymyxins and
circulins on the growth and metabolism of gram-positive bacteria have not
been studied as extensively as those for gram-negative bacteria. This is
unfortunate because the cell envelope of gram-positive organisms is much
less complicated than that of gram-negative bacteria (57).
The various polymyxins do differ somewhat in their antimicrobial activi­
ties. For example, Murray et al (58) examined the circulin and polymyxin
B sensitivity of 1 9 bacterial strains. Polymyxin B was more active than
circulin against all of these strains except Klebsiella pneumoniae and Mi­
crococcus pyogenes. Schwartz et al (50) and Wright & Welch (54) have
reported that polymyxin E is generally more active than polymyxin B
except against Staphylococcus aureus and Bacillus megaterium. Hayashi
& Suzuki have compared the antibiotic activities of polymyxins Band E
against several bacterial strains, and their results do not agree with the
studies described above (29). However, it should be emphasized that the
activities of these antibiotics are very dependent upon the growth media and
the method used by bioassaying the peptides. Therefore, the results from
different laboratories are sometimes contradictory because comparable con­
ditions were not used. Variation in the fatty acid present in analogous
polymyxins also affects antibiotic activity (59, 60). Polymyxins containing
the longer 6-methyloctanoic acid were generally more active than the 6-
methylheptanoic derivatives. Comparisons between the antibiotic activities
of the naturally occurring polymyxins and circulins have not provided a
great deal of insight regarding structure-activity relationships. However, it
is clear that the biological activities of the native peptides do differ consider­
ably, implying some degree of specificity for interactions between the antibi­
otics and the bacterial cells.
Because the octapeptins have only recently been isolated and character­
ized, most of the published research has been done with a mixture of
octapeptins A and B designated EM 49 (9, 34, 6 1 , 62). Although similar
structurally, the octapeptins and polymyxins differ significantly in antibiotic
activities. EM 49 is approximately three to ten times more active against
most gram-positive bacteria than is polymyxin B (34), and the two classes
730 STORM, ROSENTHAL & SWANSON

of peptides are not cross-resistant. The sensitivities of gram-negative bac­


teria to the octapeptins and polymyxins are comparable, except that the
polymyxin-resistant strains of Proteus are approximately four times more
sensitive to octapeptins. The most striking difference in the activities of the
polymyxins and octapeptins is illustrated by a comparison of their activities
against E coli SC 9251, 9252, 9253, 8599, and 8600. Whereas the wild-type
strains, SC 9251 and SC 8599, were equally sensitive to octapeptins and
polymyxins, sensitivity to octapeptins increased as resistance to polymyxin
increased in the series SC 9252, SC 9253, and SC 8600 (9, 1 7). For example,
the minimum inhibitory concentrations of octapeptin and polymyxin B for
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

E coli SC 9253 were 0.3 fLg/ml and> 200 fLg/ml, respectively. In one
study, the activities of purified octapeptins have been compared (34); the
greatest differences in activities were observed with S. aureus FDA 209p.
With this organism, minimum inhibitory concentrations for octapeptins
A2,3, AI> B2,3, and B, were 1 2.5, 6.3, 3.1, and 2.4 fLg/ml, respectively.

Synthetic Derivatives
The most definitive information concerning structure-function relationships
for the polymyxins and octapeptins has been obtained by systematic modifi­
cations of the antibiotic structures. These studies have shown that crucial
structural determinants for antibiotic activity include a positively charged
cycloheptapeptide and a fatty acid chain of intermediate length (C: 8-14) . .

Nakajima (60) and Teuber (63) have reported a number of observations


that emphasize the importance of the cyclic peptide moiety, and particu­
larly its positive charge for polymyxin antibiotic activity. For example,
polymyxin B was acetylated and assayed for antibiotic activity. The mono-,
di-, and tri- N-acetyl polymyxins were all active, whereas the tetra- and
penta-acetyl derivatives were inactive (63). Using polymyxins synthesized
chemically (43, 44), Volger et al have shown that expansion of the cyclic
peptide ring by one amino acid significantly reduces antibiotic activity. This
modification lowered activity against P. aeruginosa and B. bronchiseptica
sixfold and tenfold, respectively. Nakajima (60) has partially degraded
polymyxin E enzymatically and determined the sensitivity of E. coli B to
these derivatives. Deacylated polymyxin E, containing the cyclic peptide
and tripeptide side chain, was 30-fold less active than the native antibiotic.
The acylated tripeptide and the fatty acid were completely inactive. Simi­
larly, the activities of deacylated octapeptins against P. aeruginosa and S.
pyogenes were 20- and l Oa-fold less effective than the octapeptins (unpub­
lished results, K. S. Rosenthal and D. R. Storm).
The data discussed above indicate that the fatty acid side chain in the
polymyxins and octapeptins is required for full expression of activity. The
role of the fatty acids for the activities of the polymyxins has been explored
POLYMYXIN AND PEPTIDE ANTIBIOTICS 731

more thoroughly by systematically varying the length of the fatty acid side
chain (48, 49, 64). Chihara et al (48, 49) compared the activities of poly­
myxin E nonapeptide fatty acyl derivatives containing C ; 9 to C; 1 4 un­
branched fatty acids or iso C ; 9 and iso C ; 10 acyl groups. The longer fatty
acid derivatives were more effective against polymyxin E-resistant strains
of Micrococcus /uteus, S. aureus, Sarcina /utea, Bacillus subtilis, and Bacil­
lus cereus. However, the C; 10 and C; 12 derivatives were most effective
against the polymyxin E-sensitive strains of E. coli and P. aeruginosa. A
broader range of octapeptin unbranched fatty acid derivatives has been
examined (unpublished results, P. E. Swanson and D. R. Storm). As illus­
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

trated in Figure 2, there was an optimum chain length for the minimum
inhibitory concentrations against E. coli or B. subtilis. The greatest activity
against B. subtilis was observed with the native antibiotic and the C ; 12 or
C; 14 derivatives. The C : 12 derivative was 100 times more active against
B. subtilis than the C ; 2 analogue.

50

:=;
� 20

c:
0
.;:::
� 10
E
..
r:
u

0
I,) 5.0
>-
2
:a
:c
.5 2.0
- Octapeptin,
"6
E E. coli
'c
1.0
:i

0.5

� Octapeptin, B. subtilis
2 6 18

Lengfh of FOlly Acid Chain (No. of Carbons)

Figure 2 Minimal inhibitory concentrations of the C ; 2 to C : 18 acyl derivatives

of octapeptin. B. subtilis GSY 201,0--0. E. coli SC 9251,-.


732 STORM, ROSENTHAL & SWANSON

As a first approximation, one might expect antibiotic activity to increase


with the length of the fatty acid, since this would lead to greater hydro.
phobic interactions with the membrane phase. However, this does not
occur. The C : 18 derivative was five times less active against B. subtilis than
the C: 12 or c: 14 analogues. A similar chain-length optimum was ob·
served with E. coli, although it was shifted to shorter fatty acids. The
decrease in activities with longer fatty acids was not due to micelle forma­
tion, since the concentrations of the peptides used in this study were well
below critical micelle concentrations. In the case of E. coli, it could be
argued that the more hydrophobic peptides have greater difficulty overcom­
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

ing the outer membrane permeability barrier, since Nikaido (65) has
proposed that the outer membranes have a tendency to exclude hydro­
phobic molecules. However, this would not apply to B. subtilis. The signifi­
cance of these observations in terms of membrane interactions is discussed
in greater detail in a later section.

EFFECTS ON BACTERIAL METABOLISM


General Considerations
The polymyxins or octapeptins affect a wide variety of biochemical pro­
cesses in bacteria, which include selective membrane permeability (9, 60,
66-70), transport phenomena (70), respiration (70-73), ATP pool size (un­
published observations, this laboratory), nucleic acid synthesis (70), protein
synthesis (70, 74, 75), lipopolysaccharide and peptidoglycan synthesis (70),
and specific enzyme activities (69, 76, 77). The major problem has been to
separate secondary effects of the antibiotics from the primary lesion. This
is complicated by the fact that killing of bacteria by various agents often
results in rapid autolysis leading to the degradation of macromolecules and
inhibition of their synthesis (78-82). For example. inhibition of protein
synthesis in gram-negative bacteria by chloramphenicol resulted in the
release of outer-membrane material, even though chloramphenicol has no
direct effect on outer membranes (83). In addition, the biosynthesis of
macromolecules is dependent upon the energy charge of the cell (84);
therefore, inhibition of electron transport and/or oxidative phosphorylation
can lead to general inhibition of biosynthetic processes. Definition of the
primary site of action for an antibiotic requires a detailed examination of
the kinetics and concentration dependence for inhibition of various bio­
chemical phenomena by the antibiotic. Such an analysis for the polymyxins
and octapeptins leads to the conclusion that these peptides first bind to
bacterial membranes, resulting in rapid permeability changes of the cyto­
plasmic membrane (70). All other effects on bacterial metabolism are most
likely secondary to changes in the membrane permeability barrier.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 733

This general conclusion, which was first advocated by Few (85) and
Newton (66) and extensively developed by Teuber (70), has been strength­
ened considerably by recent studies with polymyxin B and octapeptin cova­
lently attached to agarose (8). These studies have shown that both
polymyxin B and octapeptin can inhibit the growth and respiration of
gram-negative bacteria and spheroplasts without entering the cell. Contact
between the immobilized peptides and the outer surface of the membrane
was sufficient to alter membrane permeability and inhibit respiration.

Membrane Permeability
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

At concentrations of the polymyxins comparable to minimum biocidal


concentrations, the permeability of the membranes for small-molecular­
weight compounds is rapidly increased and the efflux of cytoplasmic compo­
nents into the media occurs. For example, it was originally reported by
Newton (66) that treatment of bacteria with polymyxin stimulated the
release of material absorbing at 260 nm. Polymyxins A, B, D, and E were
equally effective in this respect, and this phenomenon has since been con­
firmed by several other investigators (60,69). The released compounds have
been partially characterized and shown to include adenine, adenosine, gua­
nine, and guanosine (60). Efflux of the nucleosides and bases occurred as
early as I to 2 min following treatment, and permeability changes were
detected at concentrations of the peptide as low as 3 JLg/ml. Furthermore,
the rate of release of these compounds was accelerated with increasing
concentrations of the peptide. The polymyxin-induced leakage of A260 ab­
sorbing compounds was inhibited by divalent cations (86), which is consis­
tent with the observations that magnesium and calcium ions inhibit the
antibiotic activity of polymyxin B and its binding to bacteria (9,68,87,88).
Polymyxin-stimulated permeability changes have also been observed with
E. coli Y-lO grown in the presence of H 3 32P04 (9). Treatment with poly­
myxin B at concentrations as low as 4 JLg/ml stimulated the release of
32P-Iabeled material into the media. The 32P-Iabeled compounds released by
polymyxin B were not thoroughly characterized; however, they were water­
soluble, of low molecular weight, and dialyzable.
The antibiotic activities of several polymyxin E derivatives have been
compared to their effects on the membrane permeability of E. coli B (69).
Deacylated polymyxin E, which contains the cyclic peptide and the tripep­
tide side chain, released approximately three times less A260 absorbing
material compared with polymyxin E at comparable concentrations, and it
was 30 times less active as an antibiotic. The acylated tripeptide moiety
(methyloctanyl-L-Dab-L-Thr-L-Dab) had no effect on membrane permea­
bility and exhibited no antibiotic activity. These data indicate that the major
structural requirement for disruption of membrane permeability is the cy-
734 STORM, ROSENTHAL & SWANSON

clic peptide domain; however, the fatty acid side chain does contribute
significantly to this activity.
Teuber (70) has examined the kinetics for breakdown of the membrane
permeability barrier of Salmonella typhimurium by polymyxin B using cells
preloaded with [l4C]methyl-a-D-glucopyranoside (aMG). This sugar is
taken up by the cells through the phosphotransferase system (89) and is not
further metabolized. The efflux of the accumulated aMG was examined at
several time intervals following addition of polymyxin B. The effect of
polymyxin B on the permeability of the cells was extremely rapid and could
be detected within 15 to 30 sec following introduction of the peptide. When
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

the kinetics for leakage were compared to the rate of killing, it was clear
that permeability changes either preceded or occurred simultaneously with
cell death. Uptake of aMG was also inhibited by polymyxin B; however,
this effect was subsequent to breakdown of the membrane permeability
barrier. Nucleic acid and protein synthesis were inhibited by polymyxin B;
however, these effects must be secondary to membrane damage because
polymyxin B can inhibit the growth and respiration of gram-negative bac­
teria without entering the cell (8). In addition, in vitro protein synthesis was
not inhibited by polymyxin B at concentrations several orders of magnitude
higher than its minimal biocidal concentrations (90). It can be concluded
that one of the earliest actions of polymyxin B is perturbation of the cyto­
plasmic membrane structure, increasing its permeability with respect to
charged or polar molecules. It should be emphasized that the rapid permea­
bility changes caused by the polymyxins were effective only for low-molecu­
lar-weight compounds. Efflux of cytoplasmic macromolecules, such as
specific enzymes, did not occur during this time scale (9 1). If indeed the
peptide destroys the normal packing of the membrane phospholipids, the
"holes" formed initially are large enough for small-molecular-weight com­
pounds but not for proteins.
The octapeptins and polymyxins share the common property of inducing
membrane permeability changes [(9); unpublished results, K. S. Rosenthal
and D. R. Storm]. Both E. coli and B. subtilis released low-molecular­
weight phosphate containing compounds when treated with octapeptin in
the concentration range of 4 to 10 f-tg/ml. The release of these compounds
was extremely rapid and could be detected within one minute or less upon
the addition of octapeptin. Octapeptin also stimulated the release of potas­
sium ions from B. subtilis, and this occurred within 30 sec following addi­
tion of the antibiotic. Furthermore, treatment of E. coli or B. subtilis with
octapeptin resulted in proton leakage, which was detected within 5 sec
following addition of octapeptin. Proton leakage seems to be one of the
earliest permeability changes affected by octapeptin.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 735

Bacterial Respiration

The rapid effects of the polymyxins and octapeptins on bacterial membrane


permeability suggest that these antibiotics should affect bacterial respiration
and oxidative phosphorylation, since these processes require an intact,
sealed cytoplasmic membrane (92, 93). Inhibition of bacterial respiration
has been observed with both antibiotics (69, 70, 7 1 , 73). Wahn et al (7 1)
measured oxygen consumption by E coli B manometrically in the presence
of polymyxin B. Within the limits of detection, there was an immediate
decrease in the rate of oxygen uptake following addition of the antibiotic.
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

Significant inhibition of respiration occurred at concentrations as low as 1


p.g/ml of polymyxin B, and the extent of inhibition increased proportion­
ally with higher concentrations of the peptide. This phenomenon was ob­
served with resting or proliferating bacterial cells. Similar observations have
been made with S. typhimurium, although the kinetics of the response were
somewhat slower (70). At biocidal concentrations of polymyxin B, oxygen
uptake by S. typhimurium was inhibited within 1 to 2 min after injection
of the antibiotic (70).
A comparison between the kinetics for changes' in membrane permeabil­
ity and inhibition of respiration indicated that the effects on respiration
occurred after alterations in membrane permeability. In contrast to poly­
myxin B, polymyxin E stimulated autorespiration of P. aeruginosa by 17%
and the oxidation of succinate and pyruvate by 38% and 41 %, respectively
(69). However, oxidation of 2-ketogluconate, oxaloacetate, and acetate were
inhibited 50%, 74%, and 80%, respectively. Stimulation of oxygen con­
sumption has not been observed at any concentration of polymyxin B (70,
7 1 , 73). This difference between polymyxins B and E is quite interesting,
since octapeptin does stimulate oxygen uptake at low concentrations, and
polymyxin E resembles the octapeptins structurally more so than any of the
other polymyxins (73).
The effects of octapeptin and polymyxin B on the respiration of E. coli,
B. subtilis, and E. coli spheroplasts have been compared (unpublished
results, K. S. Rosenthal and D. R. Storm). Both antibiotics caused a rapid
change in the rates of oxygen uptake (10 to 1 5 sec); however, the effects of
the two peptides were qualitatively and quantitatively distinct. At concen­
trations of octapeptin less than the minimum biocidal concentration, the
rates of oxygen consumption by E. coli were stimulated (Figure 3). How­
ever, at concentrations equivalent to or greater than minimum biocidal
concentrations, octapeptin inhibited respiration. The maximum stimulation
of respiration was 15 to 20%, with 100% inhibition occurring at higher
concentrations of the peptide. The respiration of B. subtilis was also affected
736 STORM, ROSENTHAL & SWANSON

by octapeptin (Table 1). At concentrations of octapeptin less than 0.7


Il-g/ml, B. subtilis respiration was stimulated; at concentrations > 0.7
Il-g/ml, respiration was inhibited. The kinetics for inhibition of B. subtilis
respiration by octapeptin were first-order (Figure 4). In contrast to octapep­
tin, polymyxin B did not stimulate oxygen uptake at any concentration.
Treatment of E. coli SC 925 1 or B. subtilis with concentrations of poly­
myxin B greater than 0.3 Il-g/ml inhibited respiration within 1 0 to 1 5 sec
(Table 1). The respiration of two polymyxin-resistant strains, E. coli SC
9252 and 9253, was unaffected by polymyxin B up to 200 Il-g/ml.
Stimulation of respiration by low concentrations of octapeptin is similar
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

to the actions of uncouplers of oxidative phosphorylation (94). The observa-

A. Enhancement of Respiration Rate


0.80
,
'"

o. 70 '� EM 49 (2fLg/ml)

0.60

E
'- 0.50
V>
Q)


.:!- 0.40
c:
o 0�4-�-4--+-��-4--+-��--�
e B. Inhibition of Respiration Rate
i: 0.80
Q)
u
c:
o
o
c: 0.70
Ql
c>

x
o
0.60

0.50

0
0 1 2 3 4 5 6 7 8 9 10 11
Time (minutes)

Figure 3 Effects of EM 49 on the respiration of E. coli SC 9251. E coli was treated


with 2 p.g/ml EM 49 (A) or 10 p.glml EM 49 (B). The change in respiration rate
after treatment with 2 fLg/ml EM 49 was an enhancement of 20%. EM 49 is a
mixture of octapeptins A and B.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 737

Table 1 Effects of octapeptin and polymyxin B on bacterial respiration

Optimal Minimum
Minimum concentration Minimum concentration
inhibitory for stimulation biocidal for inhibition
concentrationa . of respirationb concentrationc of respirationd

Octapeptin
E. coli SC 9251 1.7 ± O.le 2.0 ± 1.0e 4.5 ± 0.2e 4.6 ± 1.0e
E. coli SC 9252 0.8 ± 0.2 0.5 ± 0.1 1.3 ± 0.3 1.5 ± 0.2
E. coli SC 9253 0.3 t 0.1 0.6 t 0.1 1.0 t 0.2 1.2 :t 0.2
B. subtilis GSY 201 0.2:t 0.1 0.2:t 0.1 1.2 :t 0.1 0.7:t 0.2
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

Polymyxin B
E. coli SC 9251 1.2 ± 0.1 3.0 ± 1.0 0.3 ± 0.1
E. coli SC 9252 >100 >200 >200
E. coli SC 9253 >200 >200 >200
B. subtilis GSY 201 0.6 ± 0.1 0.9:t 0.1 0.3 ± 0.1

aMinimal inhibitory concentration of the peptide that completely inhibited bacterial


growth, �g/ml.
bOptimal concentration of octapeptin for stimulation of the rate of respiration, �g/ml.
cMinimal biocidal concentration in IJ.g/ml.
dMinimal concentration of the peptide to inhibit the rate of respiration by 10% follow­
ing 0.2 min treatment, IJ.g/ml.
eThese data are the average of three or more determinations.

tion that the peptide increased the permeability of the membrane with
respect to protons suggests that octapeptin, at low concentrations, may
relax the membrane proton gradient. Inhibition of respiration at higher
concentrations of octapeptin or polymyxin B may reflect extensive mem­
brane damage resulting in the efHux of metabolites used for reducing poten­
tial in electron transport, or disorganization of the electron-transport chain.
In order to study directly the interactions between these peptides and the
cytoplasmic membrane of gram-negative bacteria, the effects of octapeptin
and polymyxin B on E. coli spheroplast respiration were examined. In
addition, the role of the outer membrane as a determinant for polymyxin
resistance was evaluated by comparing spheroplasts prepared from E. coli
SC 9251, SC 9252, and SC 9253. The latter two strains are resistant to
polymyxin B but quite sensitive to octapeptin (Table 1). Treatment of E.
coli spheroplasts with EM 49 affected their rates of oxygen uptake in a way
analogous to whole cells. Spheroplast respiration was stimulated by oc­
tapeptin in the range of 1 to 2 fLg/ml and inhibited at higher concentrations.
Concentrations of polymyxin B greater than 5 ,...g/ml inhibited the respira­
tion of spheroplasts prepared from E. coli SC 9251, SC 9252, and SC 9253,
even though intact E. coli SC 9252 and SC 925 3 respiration was unaffected
by concentrations of polymyxin B up to 200 fLg/ml (Table 1). These data
738 STORM, ROSENTHAL & SWANSON

� 0.8
(/l

� 0.7
::l...
0.6

..-
0

-E
Q)
0.5
u
c:
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org

80.4
Access provided by University of Washington on 02/03/16. For personal use only.

c:
(])
gO.3
x
0

°0 1 2 3 4 5 6 7 8 9 10
Time (minutes)

Figure 4 Inhibition of B. subtilis respiration by EM 49. EM 49 (4 j..Lg/ml) was


added as indicated and the rate of oxygen utilization was continually monitored.
Complete inhibition of respiration occurred after 2 min. EM 49 is a mixture of
octapeptins A and B.

indicate that resistance of E coli SC 9252 and SC 9253 to polymyxin B was


due, at least in part, to their outer membranes.
The influence of polymyxin B and octapeptin on membrane permeability
and bacterial respiration suggests that changes in respiration rate may be
secondary to alterations in membrane permeability. In order to test this
hypothesis, the minimum inhibitory concentration of octapeptin for E coli
SC 925 1 was determined under anaerobic and aerobic conditions. The
minimum inhibitory concentration was identical under aerobic and anaero­
bic conditions, 1.7 fLg/ml. Thus, the effects of octapeptin on bacterial
respiration cannot be the primary killing mechanism, which is consistent
with the proposal that the primary effect of these peptides is disruption of
selective membrane permeability. Even under anaerobic conditions, an in­
tact membrane permeability barrier is required for bacterial growth, and the
membrane potential is maintained by the CaH, MgH-ATPase (95).
Inhibition of bacterial respiration by octapeptin and polymyxin B should
lead to depressed levels of intracellular ATP. Both peptides lowered cellular
ATP levels (unpublished results, this laboratory) measured by the firefly
luciferase assay (96). The absolute decline and the rate of this decrease in
ATP concentration was dependent upon the concentration of the antibiotic
(Figure 5). For example, octapeptin concentrations of 10 and 20 fLg/ml
decreased E coli ATP levels 25% and 70%, respect ively . A decline in ATP
POLYMYXIN AND PEPTIDE ANTIBIOTICS 739

2 • A
7 0

II
....
=
:;
u

-
o --�------�--B
E
"-
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org

0
Access provided by University of Washington on 02/03/16. For personal use only.

II
"0
E
0.

180
---�O>--___-2..0
C
o

°OL-----�4��---8����1�2 --������2�
O��

Time (minutes)

Figure 5 ATP pool size of B. subtilis following treatment with octapeptin. ATP
concentrations of D. subtilis samples were determined by the firefly luciferase assay
(96) at various time intervals following addition of octapeptin. A, 0 J.Lg/ml octapep­
tin; D, 10 J.Lg/ml; C. 17 J.Lg/ml.

pool size was detectable within 1 min following treatment. Under compara­
ble conditions, the earliest detectable changes in proton permeability, rates
of oxygen consumption, and ATP concentration were observed at 5 sec, 10
sec, and 60 sec, respectively. However, the sensitivity of the assays used to
follow these various parameters varied considerably, and this variation
imposes serious limitations on comparative kinetic studies.
If the chemical potential across the inner membrane is relaxed because
of permeability changes induced by octapeptin, the bacterial cells could
strive to maintain this gradient through the action of the membrane-bound
Ca2+, Mg2+-ATPase. Indirect evidence supporting this hypothesis was ob­
tained by comparing the effects of octapeptin, dicyclohexylcarbodiimide,
and a combination of the two on the viability of E. coli cells. Dicyclohexyl­
carbodiimide is a potent inhibitor of the Ca2+, Mg2+-ATPase complex (97,
98). Octapeptin at 2 p.g/ml and dicyclohexylcarbodiimide at I mM were
both biostatic but not biocidal. However, a combination of the two at these
concentrations was biocidal (unpublished results, this laboratory). The ap­
parent synergism between these two compounds supports the hypothesis
under consideration but by no means proves this proposal. It is clear,
740 STORM, ROSENTHAL & SWANSON

however, that alterations in membrane permeability are one of the earliest


changes caused by the polymyxins and octapeptins, and that effects on
respiration and ATP levels are probably secondary to membrane damage.

INTERACTIONS WITH BACTERIAL MEMBRANES

The actions of the polymyxins and octapeptins on metabolism have strongly


implicated bacterial membranes as the primary site of action for these
peptides. Polymyxins and octapeptins are generally more active against
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org

gram-negative bacteria than against gram-positive bacteria (28, 34, 50, 54).
Access provided by University of Washington on 02/03/16. For personal use only.

Activity against gram-negative bacteria requires that the antibiotic over­


come the outer membrane permeability barrier to reach the inner mem­
brane, if indeed the latter is the primary site of action. There is considerable
evidence that the structure of gram-negative outer membranes is disrupted
by the polymyxins and octapeptins, and this disruption either indirectly
alters inner membrane permeability or allows direct interactions between
the peptides and inner membranes.

Outer Membranes
The cell envelopes of gram-negative and gram-positive bacteria differ in a
number of important ways (57). Notable in this respect is the outer mem­
brane of gram-negative bacteria, which is external to the peptidoglycan
layer and inner membrane. The structure of the outer membrane has not
been completely defined, although there is considerable evidence that it is
a bilayer structure composed of lipopolysaccharide, phospholipids, and
proteins (99-106). The fatty acid chains of the phospholipids and lipopoly­
saccharides are thought to extend into the hydrophobic interior with the
phospholipid polar head-groups and the lipopolysaccharide polysaccharide
chains at the aqueous-membrane interface. In addition, magnesium and
calcium ions are important for the maintenance of outer-membrane struc­
tural integrity, and Leive (107) has proposed that these divalent cations act
as metal-ion bridges between phosphate groups of phospholipids and/or
lipopolysaccharides. Nikaido and co-workers (106) have proposed an asym­
metric model for the outer membranes of S. typhimurium in which the
outer half of the bilayer is exclusively composed of lipopolysaccharide and
protein, with the inner half containing all of the phospholipid. This model
predicts that the polymyxins and octapeptins must be able to interact with
the lipopolysaccharide or proteins of outer membranes, since the external
monolayer of the outer membrane is the first structure encountered by the
antibiotic interacting with whole cells. Strong interactions between the
polymyxins and lipopolysaccharide have been reported (108, 109).
POLYMYXIN AND PEPTIDE ANTIBIOTICS 74 1

The first evidence for outer-membrane damage by the polymyxins was


changes in the morphology of the cells detected by electron microscopy (71 ,
1 10-1 13). For example, Koike, Iida & Matsuo ( 1 1 2) have observed changes
on the surface of E. coli and P. aeruginosa upon treatment with polymyxin
B or E. Electron micrographs of treated bacteria revealed numerous protru­
sions or blebs extending from the outside surface of the cells. The number
of blebs formed by the polymyxins increased with antibiotic concentration
and was inhibited by the presence of magnesium ions. Blebs are occasionally
seen on the surface of gram-negative bacteria during normal growth; how­
ever, the number of protrusions seen in the presence of the polymyxins far
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

exceeded that observed in the absence of the antibiotic. Similar observations


have been reported by Wahn et al (71). Bleb formation on the outside
surface of E. coli occurred at concentrations of polymyxin B up to 10
JA-g/ml. At higher concentrations, additional effects were observed includ­
ing cell autolysis and brightening of the nuclear area.
It might be argued that the blebs seen in electron micrographs of poly­
myxin-treated E. coli or P. aeruginosa were artifacts resulting from fixation
and thin-sectioning of antibiotic-weakened membranes. However, the re­
sults from several other studies examining this phenomenon by alternative
techniques tend to argue against this possibility. Schindler & Teuber ( 1 13)
examined polymyxin B-treated S. typhimurium by freeze-etching, a tech­
nique that is less destructive to the cell envelope than fixation and thin­
sectioning ( 1 14). Freeze-etching revealed three layers, which were
interpreted as being the outer and inner monolayer of outer membranes, and
the outer surface of the cytoplasmic membrane. Treatment with polymyxin
B caused the rapid formation of numerous blebs only on the outermost
surface, which corresponds to the external monolayer of the outer mem­
branes. These protrusions varied in size from 10 to 30 nm and were ran­
domly distributed across the surface of the membrane at a density of 200
to 500 globules/ JA-m2. Schindler & Teuber estimated that the membrane
surface would be expanded approximately 5% by simple intercalation of
polymyxin B ( 1 13). The observed surface area increase was 1 1 %, indicating
that the packing of membrane components must have been changed by
polymyxin. The formation of blebs by polymyxin is not limited to bacteria.
Polymyxin B caused the formation of bulges on the surface of rat peritoneal
mast cells. These protrusions occurred within seconds after exposure to the
peptide and were detected by freeze-fracture methods (115). Blebs have also
been observed on the outside surface of Chlamydia psittaci treated with
polymyxin B (1 1 6).
Changes in the surface morphology of E. coli have also been observed
following treatment with octapeptin (9, 17). Transmission electron micro-
742 STORM, ROSENTHAL & SWANSON

graphs of E coli Y- 1 0 treated with octapeptin or polymyxin B showed the


formation of numerous blebs on the outside surface that were absent in
untreated controls ( 1 7). There was, however, an important difference be­
tween octapeptin- and polymyxin-treated cells. Octapeptin, in contrast to
polymyxin B, stimulated the release of vesicular material into the surround­
ing media. This phenomenon has recently been examined by scanning
electron microscopy to visualize octapeptin-induced morphological changes
(9). A number of previous studies have used this technique to study the
surface morphology of microorganisms ( 1 1 7, 1 1 8) as well as antibiotic­
induced changes on their surface ( 1 1 9). The critical-point drying technique
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

was used because it provides the most artifact-free method for sample
preparation with maintenance of cell size and dimensions. Particulate mate­
rial was seen on the surface of both controls and octapeptin-treated bacteria
(Figure 6). However, the untreated cells had a much smoother surface
compared with those treated with octapeptin. The morphological changes
caused by the antibiotic were extensive and covered a significant fraction
of the cell's surface.

Figure 6 Scanning electron microscopy of octapeptin-treated E. coli cells. E. coli


SC 9252 was treated with 50 p.g/ml of octapeptin and allowed to grow for another
30 min (B). Controls (A) were treated identically in the absence of added antibiotic.
The bacteria were fixed in 2% glutaraldehyde added directly to the media at 4°C
for 4 hr and then postfixed in 0.6% KMn04 for 6 hr. The fixed cells were dehydrated
in ethanol and critical-point-dried onto glass cover slips, coated with Au/Pd (60%/
40%), and viewed with a JEOL U-3 scanning electron microscope.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 743

The origin of the blebs caused by polymyxins and octapeptin, or the


identity of the particles released by octapeptin, could not be unambiguously
determined by electron microscopy. Therefore, the material released by
octapeptin was isolated and characterized (9). On the basis of a number of
criteria, these particles were shown to be outer-membrane fragments. They
had a phospholipid composition quite similar to outer membranes and a
density of 1 .22 g/cc, which was identical to isolated outer membranes. The
ratio of phospholipid to lipopolysaccharide to protein was comparable to
outer membranes, and SDS gel profiles of the released membrane fragments
indicated the presence of the major outer-membrane proteins. Release of
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

outer-membrane particles occurred within 2 min or less following addition


of octapeptin, which indicated that this phenomenon was not a secondary
event resulting from cell death and lysis. Approximately 10 to 1 5% of the
E. coli lipopolysaccharide was released upon treatment with octapeptin. A
combination of 1 mM MgCl2 and CaCl2 completely inhibited the release of
outer-membrane vesicles, which is consistent with the observations that
divalent cations inhibited the antibiotic activities of the polymyxins and
octapeptin (68, 87), binding of the peptides to bacteria (9, 87, 88), and
formation of blebs in the presence of polymyxin B ( 1 1 2). These observa­
tions, coupled with the proposed structural role of divalent cations in outer
membranes ( 1 07), suggested that these peptides competed with magnesium
and/or calcium ions for anionic sites in the membrane.
Interactions between octapeptin or polymyxin B and isolated membranes
have been studied by fluorescence polarization (9). Changes in fluorescence
polarization and fluorescence intensity of a hydrophobic fluorescent-dye
molecule incorporated into membranes can give information concerning the
hydrophobicity and fluidity of the dye environment ( 1 20). The fluorescent
dye 1 ,6-diphenyl- l ,3,5-hexatriene (DPH) was used as a fluorescence probe
of outer-membrane structure ( 1 2 1 ). Upon treatment of DPH-containing
outer membranes with octapeptin or polymyxin B (5 JLg/ml), the fluores­
cence polarization and intensity increased. After correction for fluorescent
lifetime changes, these data indicated that octapeptin and polymyxin B, in
the concentration range of 0 to 5 JLg/ml, caused a decrease in membrane
fluidity. These data cannot be rigorously interpreted in terms of specific
molecular alterations of outer-membrane structure, but they do illustrate
that octapeptin and polymyxin B caused structural changes in outer mem­
branes.
Interactions between the polymyxins and outer membranes have been
detected by a number of indirect methods. The O-antigenic lipopolysaccha­
ride of gram-negative bacteria is synthesized in the cytoplasmic membrane
( 1 0 1 , 1 03) but is found in the outer membrane with the O-antigen-specific
744 STORM, ROSENTHAL & SWANSON

carbohydrate chain extending into the aqueous environment. The O-antigen


is the source of the pyrogenic immune reaction to gram-negative bacteria.
Polymyxins B and E neutralized the lethality of E. coli endotoxin for chick
embryos, rabbits, and adrenalectomized mice (122, 1 23). Similar observa­
tions were made with endotoxins of Aerobacter aerogenes ( 1 24). Ritkind
(124) also reported that the antibiotic activity of polymyxin B was inhibited
by endotoxin preparations. Endotoxins derived from polymyxin B-sensitive
strains were approximately five times more effective in neutralizing the
antibiotic activity of polymyxin B than were those obtained from polymyx­
in-resistant strains. The reciprocity between the biological activities of the
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

endotoxins and polymyxins suggested a direct interaction between those


two molecules, which has since been confirmed (108). The free y amino
groups of the polymyxins were important for interactions with endotoxins,
since modification of these functions lowered the potency of polymyxin E
endotoxin neutralization 13-fold (122). Cooperstock (125) has also reported
inactivation of endotoxins from a number of aerobic gram-negative bacteria
by polymyxin B. The susceptibility of various endotoxins to polymyxin B
varied over a 1O,OOO-fold range. In contrast to Rifkind's results, the bacteri­
cidal and endotoxin-detoxifying properties of polymyxin B were not directly
related because the endotoxin most sensitive to polymyxin B was obtained
from a polymyxin-resistant strain. Similarly, absorption of phages T3, T4,
and T7 to E coli B was inhibited by polymyxin B ( 126). Since the receptors
for these phages are located in the outer-membrane lipopolysaccharide
these studies also indicate interaction between polymyxin B and outer­
membrane lipopolysaccharide.
The outer membrane of gram-negative bacteria is a permeability barrier
prohibiting certain molecules from reaching the inner layers of the cell
envelope (107). If the polymyxins damage the outer-membrane structure,
then the barrier function of the outer membrane should be altered. Changes
in outer-membrane permeability have been detected by the selective release
of periplasmic proteins in the presence of polymyxins (91, 127-129). Peri­
plasmic proteins are located between the inner and outer membranes of
gram-negative bacteria and are released from E coli by lysozyme-EDT A
treatment ( 1 30) or by osmotic shock ( 1 3 1 , 1 32). A number of enzyme
activities, including alkaline phosphatase, 5'-nucleotidase and ribonuclease
I, are classified as periplasmic proteins, and their general location between
inner and outer membranes has been confirmed by histochemical methods
( 1 33, 1 34). Cerny & Teuber (9 1 ) have reported the selective release of
periplasmic proteins from E coli B after 1 min of treatment with polymyxin
B. The proteins obtained from the supernatant of polymyxin-treated cells
were compared to those obtained by osmotic shock using SDS gel electro­
phoresis. The two SDS gel profiles were identical. Longer incubations with
POLYMYXIN AND PEPTIDE ANTIBIOTICS 745

polymyxin B, up to 60 min, resulted in the release of 100% of the cytoplas­


mic proteins. The periplasmic enzymes released by polymyxin B included
5'-nucleotidase, 3'-nucleotidase, ribonuclease I, acid phosphatase, alkaline
phosphatase (91), an aminopeptidase activity (128), and a phosphate-bind­
ing protein (129). Magnesium at 10 mM inhibited periplasmic protein
release by polymyxin B ( 1 27).
There have been a number of other studies demonstrating destruction of
the outer-membrane permeability barrier by the polymyxins. For example,
lysozyme cannot reach the peptidoglycan layer of gram-negative bacteria
unless the outer-membrane structure is perturbed by EDTA complexation
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

of divalent cations. Polymyxin B can substitute for EDTA, and a combina­


tion of the antibiotic and lysozyme has been used to make spheroplasts ( 1 35,
136). Many antibiotics active against gram-positive bacteria are much less
active against gram-negative bacteria because they are excluded by the outer
membrane. The activities of actinomycin D ( 1 37), penicillin ( 1 38), rifampi­
cin (139), and streptogramin (140) against gram-negative bacteria are ac­
centuated by EDTA. Synergistic effects between polymyxin B and
tetracyclines, amphotericin B, or chloramphenicols have also been reported
( 1 4 1 , 142).
The polymyxin resistance of some gram-negative strains is apparently
attributable to their outer membranes. For example, Proteus mirabilis is
normally quite resistant to polymyxin B. Conversion of a polymyxin-resist­
ant strain of P. mirabilis to L-forms or spheroplasts increased sensitivity to
polymyxin B 400-fold (143). The susceptibility to polymyxin B was lost
when L-forms or spheroplasts were allowed to convert back into the bacil­
lary form. Teuber proposed that the resistance of some Proteus strains to
polymyxins is due to the impermeability of their outer-cell envelopes to
polymyxin. A similar conclusion was made by Sud & Feingold (144) on the
basis of liposome studies that showed equivalent polymyxin sensitivities for
liposomes prepared from lipids extracted from several P. mirabilis strains
differing in polymyxin sensitivity. The resistance of some gram-negative
strains to polymyxin B, which was not observed with the corresponding
spheroplast preparations, could be due either to the peptidoglycan layer or
to the outer membrane. Protoplasts formed from Streptococcus faecalis
were five times as sensitive to polymyxin as were whole cells (145). How­
ever, this difference in sensitivity was not comparable to that observed when
polymyxin-resistant gram-negative bacteria were compared to their corre­
sponding spheroplasts.
Comparison between the antibiotic sensitivity of protoplasts or sphero­
plasts and whole cells is also complicated by the possibility that the cyto­
plasmic membrane is made more sensitive to the antibiotic by the removal
of outer layers. Therefore, it is not known whether the polymyxin resistance
746 STORM, ROSENTHAL & SWANSON

of the gram-negative bacteria discussed above was due to impermeability of


the peptidoglycan layer or outer membranes, although the latter seems
much more likely. This conclusion is supported by the studies of Suling &
O'Leary (146), which showed that polymyxin B-resistant strains of P.
mirabilis were sensitized to the antibiotic by several detergents, including
cetyltrimethylammonium bromide and n -alkyldimethyl betaines. In addi­
tion, Brown & Richards (147) have reported that the activity of polymyxin
B against P. aeruginosa was potentiated by the sodium salt of EDTA.
A wide range of experiments has established that the structure of gram­
negative outer membranes is affected by polymyxins and octapeptins. The
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

effects of the peptides on inner-membrane functions are due either to direct


interactions with the antibiotics or to indirect effects resulting from damage
to outer membranes. In fact, results reported in a later section show that
both phenomena can occur.

Cytoplasmic Membranes

Interactions between outer membranes and polymyxins or octapeptins have


been extensively studied. Unfortunately, direct interactions between the
antibiotics and cytoplasmic membranes have not been characterized to the
same extent. Binding between polymyxin and bacterial membranes was first
shown by Newton (24, 87), using fluorescent-labeled polymyxin, l -dime­
thylaminonaphthalene-5-sulphonamidopolymyxin (DANSP), which re­
tained its bactericidal activity. Whole P. aeruginosa cells were treated with
DANSP, broken, and fractionated. The labeled polymyxin was found asso­
ciated with membrane particles, although no distinction was made between
inner and outer membranes. Binding to P. aeruginosa was completely inhib­
ited by divalent cations, and it was proposed that the antibiotic binds to
phosphate groups on the cell surface. DANSP was also bound by Sarcina
lutea and protoplasts prepared from Micrococcus lysodeikticus and Bacillus
megaterium, indicating that polymyxin has affinity for the cytoplasmic
membrane.
Teuber & Bader (42, 1 48) have measured binding of 1 4C-Iabeled mono­
acetyl polymyxin B to inner- and outer-membrane preparations made from
S. typhimurium. Inner and outer membranes bound 30 and 60 nmole of
peptide per milligram of membrane, respectively. When these values were
corrected for the amount of phospholipid or lipopolysaccharide phosphate
present, the two membrane preparations bound equivalent amounts of poly­
myxin per mole of lipid phosphate. Saturation of the membranes with
polymyxin B occurred at 0.1 mg of antibiotic per milligram of membrane
protein. It is important to note that binding of polymyxins to bacterial
membranes shows saturation only at concentrations one to two orders of
magnitude higher than minimum inhibitory or minimum biocidal concen-
POLYMYXIN AND PEPTIDE ANTIBIOTICS 747

trations. For examples, absorption of approximately 2 X 105 molecules of


polymyxin B per S. typhimurium cell was sufficient to kill the bacteria
within 30 sec ( 1 49). Nonsaturable binding at physiologically meaningful
concentrations of the antibiotic, coupled with the correlation between bind­
ing capacity and membrane lipid phosphate composition, strongly suggests
that there is not a specific protein receptor for polymyxin in the membranes.
Teuber & Bader (42, 1 48) have proposed that the polymyxins bind preferen­
tially to negatively charged phospholipids or lipopolysaccharide in mem­
branes.
There have been a limited number of studies examining the interactions
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

of these peptides with membranes using biophysical probes to monitor


membrane structural changes. Other than the fluorescence polarization
study of outer membranes (9), described in a previous section, the only
other attempts to measure perturbation of membrane structure by physical
techniques have been reported by Barrett-Bee et al (35) and Pache, Chap­
man & Hillaby ( 1 50). In the latter study, it was concluded on the basis of
differential scanning calorimetry, ESR, and NMR data that polymyxin B
associated with both the polar and hydrophobic domains of phosphatidyl­
choline model membranes. On the other hand, gramicidin, a cyclic peptide
containing no fatty acid, interacted only with the phospholipid polar head­
groups. Binding of polymyxin B to whole E. coli cells has been examined
by high-resolution NMR (35). Addition of E. coli suspensions to solutions
of polymyxin B caused line broadening of the peptide NMR spectra, in­
dicating restricted motion of polymyxin B upon binding to E. coli cells.
Line broadening of the peptide resonances was eliminated in the presence
of magnesium ions, which confirmed other reports that binding of poly­
myxin B to bacteria is inhibited by divalent cations (9, 87). Although
broadening of all peptide resonances occurred, the largest effects were
observed with the hydrophobic residues of 6-methyloctanoic acid,
phenylalanine, and leucine. The spin-lattice relaxation times indicated that
these hydrophobic residues were most extensively immobilized upon bind­
ing to cells. It was concluded that the nonpolar residues of polymyxin B
penetrate the hydrophobic domain of the membrane with the polar groups
situated at the membrane/aqueous interface. In this same study, alterations
in the structure of the bacterial membranes were detected by using fluores­
cent probes incorporated into the membrane. The fluorescence intensity of
l -anilino-8-naphthalene sulfonate and N-phenyl- l -naphthylamine was in­
creased by addition of polymyxin B. It was suggested that polymyxin B
opened up the membrane structure, allowing for more binding of the
fluorescent dye to the membrane. However, these fluorescence-intensity
changes could have been due to redistribution of the dyes within the cell
or membranes, and do not necessarily indicate that the membrane structure
was expanded by polymyxin B.
748 STORM. ROSENTHAL & SWANSON

The original proposal by Few (25) that binding of the polymyxins to


bacterial membranes involves electrostatic interactions with the negatively
charged phospholipids has been supported by a number of different studies
(42, 148, 151, 152). Feingold has suggested an alternative mechanism invok­
ing proton exchange between polymyxin and the amino group of phos­
phatidylethanolamine (153, 154). Teuber & Bader (148) have attempted to
resolve this question by modification of the composition of Acholeplasma
laidlawii B membrane phospholipid. A. laidlawii B was fused with phos­
pholipid vesicles and the polymyxin B sensitivity of the phospholipid-sup­
plemented bacteria was examined. This organism, which has no cell wall,
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

can be fused with phospholipid vesicles, giving incorporation of the exoge­


nous lipids into the bacterial cytoplasmic membrane (155). Using spin­
labeled dipalmitoyl phosphatidylcholine Grand & McConnell have shown
that incorporated lipid diffuses laterally within the membrane; however,
some of the incorporated lipid remained concentrated in patches ( 1 55).
A. laidlawii B is normally resistant to polymyxin B. However, fusion
with phosphatidylglycerol cardiolipin, or mixed vesicles of phosphatidyl­
ethanolamine and phosphatidylglycerol increased sensitivity to polymyxin
B by 1 0- to 30-fold. Phosphatidylcholine or mixed vesicles of phosphatidyl­
choline and phosphatidylethanolamine were without effect. Incorporation
of phospholipids was measured with tritium-labeled lipids, and all phos­
pholipids were shown to be incorporated at significant levels. A. laidlawii
B cells treated with phosphatidylglycerol bound 12 times more polymyxin
than the naturally resistant cells. On the basis of these results, it was
concluded that the presence of phosphatidylethanolamine was not a crucial
determinant for polymyxin susceptibility. It should be noted, however, that
the presence of phosphatidylcholine depressed the sensitivity of liposomes
to polymyxin B (152), and it was not possible to study the effects of phos­
phatidylethanolamine incorporation alone because it does not swell and
form sealed vesicles. This study supports the general proposal that poly­
myxin B disrupts membrane structure by electrostatic interactions with
acidic phospholipids. This conclusion is based on the assumption that in­
teraction of the phospholipid vesicles with the bacterial membranes did not
make them generally more sensitive to surface active agents. In this respect,
Teuber & Bader noted that the fusion process did make these cells more
sensitive to mechanical stress encountered in centrifugation and resuspen­
sion manipulations.

MODEL MEMBRANE STUDIES

The evidence discussed thus far indicates that the polymyxins and octapep­
tins inhibit the growth of bacteria by disrupting the selective permeability
POLYMYXIN AND PEPTIDE ANTIBIOTICS 749

of bacterial membranes. Perturbation of membrane structure could conceiv­


ably result from association of the peptides with membrane lipids, proteins,
or both. Interactions with membrane proteins have not been strongly ad­
vocated, and there is, as of yet, no evidence for such associations. In addi­
tion, the binding curves for interaction between these peptides and whole
cells or membranes were not indicative of specific protein receptors in the
membranes. These observations and the results discussed in this section
have led to the prejudice that the polymyxins and octapeptin bind to and
disrupt the structure of membranes by virtue of interactions with membrane
lipids. The data discussed below prove that these peptides have strong
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

affinities for bacterial phospholipids and lipopolysaccharide, and that the


structures of lipid aggregates, monolayers, and liposomes are affected by
polymyxins or octapeptins. Changes in the structure of lipid bilayers, in the
absence of membrane proteins, are sufficient to increase the permeability of
the lipid barrier with respect to charged or polar molecules.

Phospholipid and Lipopolysaccharide Aggregates


Complex formation between polymyxin and bacterial lipids has been de­
tected by a number of different methods. The first indication that such
associations occur was inhibition of polymyxin antibiotic activity by exoge­
nously added phospholipids. Bliss, Chandler & Schoenback ( 1 56) first noted
that the addition of soaps or soya bean phospholipids inhibited the bacteri­
cidal activity of polymyxins. Similarly, Few (25) has shown that the activity
of polymyxin E against Pseudomonas denitrificans was inhibited by bac­
terial phospholipids. Cardiolipin was the most effective, while phos­
phatidylcholine had no effect on polymyxin E activity. The minimum
bactericidal concentration of polymyxin E was elevated from 2 ,u.g/ml to
12 ,u.g/ml in the presence of 25 ,u.g/ml cardiolipin. It was estimated that
the biological activity of polymyxin E was completely neutralized at a ratio
of three cardiolipin phosphates per four polymyxin E amino groups. The
antagonism of polymyxin antibacterial action by phospholipids does not
necessarily directly prove complex formation between the peptide and phos­
pholipids. For example, it has been proposed by Baker, Harrison & Miller
( 1 57) that protection of bacteria from detergents by phospholipids was due
to alteration of membrane structure rather than complex formation between
lipids and detergent.
Direct interactions between the polymyxins or octapeptins and bacterial
phospholipids have been detected by several independent techniques (9,
151, 1 58). Phosphatidic acid or phosphatidylglycerol were shown by Teuber
( 1 5 1 ) to form relatively stable complexes with polymyxin B. The peptide did
not migrate when applied to a paper chromatography system using chloro­
form/methanol/water for development. However, in the presence of phos-
750 STORM, ROSENTHAL & SWANSON

phatidic acid or phosphatidylglycerol, polymyxin B did migrate in this


system. In addition, the extraction of polymyxin B from aqueous solutions
into chloroform/methanol was facilitated by the presence of phosphatidic
acid or phosphatidylglycerol. Interaction between polymyxin E and phos­
phatidylethanolamine, phosphatidylserine, or cardiolipin was not detected
by these techniques.
Disruption of phospholipid aggregates by octapeptin and polymyxin B
(9) has been studied with the Hummel & Dreyer technique ( 1 59). It was
discovered that both peptides extracted phospholipids from phospholipid
micelles. In the absence of polymyxin B or octapeptin, the phospholipid
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

aggregates were excluded from Sephadex G-25. In the presence of either


peptide, a fraction of the total phospholipid was included in the gel as a
peptide-lipid complex. The apparent association constants for interaction of
octapeptin or polymyxin B with E. coli phospholipids are reported in Table
2. These values are not true binding constants for the interaction of mono­
meric phospholipid and the antibiotics, since interactions between phos­
pholipids in the micelle had to be overcome during the extraction process.
Clearly, octapeptin and polymyxin B interacted very strongly with phos­
phatidylethanolamine, phosphatidylglycerol, and cardiolipin. There was no
indication, however, that the antibiotics interacted exclusively with any
particular phospholipid class. Bacitracin, SDS, and quaternary ammonium
salts did not show this phenomenon at concentrations comparable to poly­
myxin B or octapeptin.
The results of several different studies, discussed in an earlier section,
have indirectly suggested that polymyxin binds to lipopolysaccharide in the
outer membrane of gram-negative bacteria. Lopes & Inniss ( 1 08) have
studied the effects of polymyxin B on the structure of lipopolysaccharide
using electron microscopy. Lipopolysaccharide isolated from E. coli has a

Table 2 Binding constants for the interaction of octapeptin and polymyxin B with E.
coli phospholipids

Phosphatidylethanoiamine Phosphatidylglycerol Cardiolipin

Octapeptina 1 .2 X 1 0 5 2.1 X 1 0 5 5 .3 X 1 0 5
Polymyxin B 2.2 X 1 0 5 3.9 X 1 0 5 1 .8 X 1 0 5
Bacitracin A NAb NA NA
CTABc NA NA NA
SDS NA NA NA

a Binding constants (M- 1 ) determined by the method of Hummel & Dreyer (1 1 , 159).
b
No interaction detected using the Hummel & Dreyer technique in the concentration
range examined (10-6 to 1 0- 3 M).
c CTAB, cetyltrimethyiammonium bromide.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 751

membranous, ribbon-like structure with periodic branching. Treatment


with polymyxin B at 25 J.Lg/ml resulted in breakdown of these structure to
give shorter fragments or completely disaggregated material. Similar obser­
vations have also been made by Koike & !ida (1 26). Bader & Teuber ( 1 09)
have reported the formation of stable non covalent complexes between S.
typhimurium lipopolysaccharide and polymyxin B. Dissociation of these
complexes resulted upon treatment with cetylpyridinium chloride or upon
lowering the pH from 7.0 to 1 .0. On the basis of comparative binding studies
using various lipopolysaccharide derivatives, it was concluded that the
primary interaction was with the negatively charged ketodeoxyoctanoate­
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

lipid A portion of lipopolysaccharide. Binding of mono-N-dimethylamino­


naphthalenesulfonyl polymyxin to lipopolysaccharide gave a blue shift
from 550 to 5 1 5 nM, suggesting that the cyclic peptide moiety was at a
polar-apolar interface. The peptide-lipid complex contained approximately
3 moles of polymyxin per mole of lipopolysaccharide.

Monolayers
The studies with phospholipid or lipopolysaccharide dispersions indicated
that the polymyxins not only bind to these lipids but also significantly alter
their structures. Few (25) has directly demonstrated changes in the packing
of phospholipids by polymyxin E using monolayers prepared from P. deni­
trificans or S. aureus phospholipids. Addition of polymyxin E to the subso­
lution below the monolayer caused an increase in the surface pressure of the
monolayer. Monolayers prepared from a crude mixture of phosphatidylse­
rine and phosphatidylethanolamine or lipid extracted from P. denitrificans
were most sensitive to polymyxin E. Total lipids from S. aureus or cardio­
lipin were somewhat less sensitive to polymyxin E. Changes in surface
pressure were detectable at polymyxin E concentrations as low as 0.5
J.Lg/ml, and the maximum response was generally attainable at I JLg/ml of
peptide. The surface pressure of phosphatidylcholine monolayers was only
:narginally increased by polymyxin E. The low affinity of the polymyxins
for phosphatidylcholine is the most consistent observation that has been
reported in various studies of polymyxin-phospholipid associations.
The monolayer studies of Few (25) illustrate that polymyxin can affect
the structure of phospholipid monolayers with some specificity for particu­
lar phospholipid classes. Increases in the surface pressure of monolayers
could be due to insertion of the fatty acid tail into the hydrophobic core of
the membrane, electrostatic interactions between the peptide amino groups
and phospholipid phosphate groups, or a combination of both. Comparative
monolayer studies using the polymyxins and deacylpolymyxin would have
been useful for distinguishing between these various possibilities; however,
such comparisons have not yet been made.
752 STORM, ROSENTHAL & SWANSON

Liposomes

The influence of polymyxin B on the glucose permeability of phospholipid


liposomes has been examined by two different laboratories (152-154). Al­
though their conclusions regarding specificity for interactions with particu­
lar phospholipid classes differ, both studies have demonstrated that the
permeability of lipid bilayers for polar molecules is rapidly increased upon
treatment with physiologically significant levels of polymyxin B. In both
studies, liposomes were prepared by the general method of Kinsky et al
(160), and permeability changes were monitored by measuring efflux of
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

trapped glucose. In neither case were the liposome preparations extensively


characterized, except for their ability to trap glucose. The size of the lipo­
somes was not reported, and it was not specified as to whether the prepara­
tions contained multi-lamellar structures or single membrane units. These
specifications are crucial because comparisons of polymyxin sensitivity for
liposomes of varying phospholipid classes were made, and it is not entirely
clear that comparable physical structures were obtained when the lipid
composition was varied.
Feingold and co-workers (153 , 154) have reported that liposomes con­
taining phosphatidylethanolamine are extremely sensitive to polymyxin B,
whereas those prepared from N-methylphosphatidylethanolamine, N,N­
dimethylphosphatidylethanolamine, and phosphatidylcholine were insensi­
tive to polymyxin B. On the basis of these and other observations, Feingold
concluded that the sensitivity of biological membranes to polymyxin re­
quires some threshold density of phosphatidylethanolamine. This specificity
for phosphatidylethanolamine was proposed to explain the sensitivity of
certain bacteria to polymyxin B as opposed to cells such as erythrocytes,
which are rich in phosphatidylcholine. The liposomes used in this study
contained 50% cholesterol and, in some cases, dicetylphosphate. Therefore,
conclusions regarding specific interactions between polymyxin B and phos­
phatidylethanolamine are not justified by these data.
Imai, Inoue & Nojima (152) arrived at somewhat different conclusions
regarding the lipid requirements for polymyxin sensitivity. Liposomes were
prepared either from E. coli phospholipids or from mixtures of purified
phospholipids in varying ratios. Permeability changes caused by polymyxin
B were quite rapid. For example, glucose release from liposomes prepared
from E. coli lipids was detected within I to 2 min following treatment with
2.2 ILg/ml of polymyxin B. It was not possible to study pure phosphatidyl­
ethanolamine because it does not swell in aqueous solutions to give sealed
vesicles (161). In this study, phosphatidylcholine liposomes were sensitive
to polymyxin B as long as they contained a negatively charged amphipathic
molecule such as cardiolipin.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 753

The apparent discrepancy between the results of these two studies was
probably due to the high levels of cholesterol incorporated into the lipo­
somes used by Feingold. Incorporation of cholesterol into liposomes pre­
pared from E. coli phospholipids decreased their sensitivity to polymyxin
B ( 1 52). Liposomes containing varying ratios of phosphatidylcholine and
cardiolipin or phosphatidylglycerol were also compared. Susceptibility to
polymyxin was not significantly altered as the percentage of phosphatidyl­
glycerol or cardiolipin was increased. These studies indicate that polymyx­
in-induced permeability changes are not dependent on the presence of
phosphatidylethanolamine, but do require the presence of a negatively
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

charged amphipathic molecule. The literature concerning the specificity for


polymyxin phospholipid interactions is quite contradictory and reflects, in
part, the different techniques used and, in some cases, the utilization of
impure phospholipid preparations. There is no evidence that the polymyx­
ins or octapeptins interact exclusively with any particular phospholipid,
although they do show some preference for the acidic phospholipids.

IMMOBILIZATION OF POLYMYXIN B
AND OCTAPEPTIN

It has been generally assumed that the effects of the polymyxins or octapep­
tin on the growth and respiration of gram-negative bacteria are due to a
direct interaction between the peptides and the inner membrane (9, 143,
1 54). The possibility that structural damage to the outer membrane by these
peptides may indirectly affect the functions of the inner membrane had not
been examined. In order to examine this question more directly, polymyxin
B and octapeptin were covalently attached to agarose beads to limit contact
to the outer surface of gram-negative bacteria (8). Polymyxin-agarose and
octapeptin-agarose inhibited the growth and respiration of two gram-nega­
tive bacteria, E coli and P. aeruginosa, but not that of B. subtilis.
The antibiotics were attached to agarose by stable covalent bonds. The
structure of the chemical linkage between agarose and the peptides is shown
here for polymyxin B.

� �
e
y
Agarose O - C H2 - C H2- N - C -CH2-CH2- C - N - CH2-CH2- Polymyxin B
I I
H H
Both peptides were covalently attached to the spacer arm through an amide
bond between the 'Y amino group of a Dab residue and the carboxyl group
at the end of the spacer arm. The ether bond is extremely stable, and
hydrolysis of the amide bonds would not be expected. Although the amide
bonds used for covalent coupling of the peptides to agarose should not be
754 STORM, ROSENTHAL & SWANSON

susceptible to attack by bacterial proteases, it was important to prove that


free peptide, in an active form, was not released in the presence of bacteria.
The experiment summarized in Table 3 was designed to detect the presence
of free, active polymyxin in the presence of E. coli. Dialysis tubing contain­
ing nutrient media was suspended in a flask containing nutrient media,
stoppered, and autoclaved. The sterilized media inside and outside of the
bag were inoculated with E. coli, and polymyxin-agarose was introduced
inside the dialysis bag. After 1 8 hr at 37°C, there was no growth in the
dialysis bag, but growth occurred normally outside. When polymyxin­
agarose was placed outside the dialysis bag, bacteria grew inside but not
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

outside. Free polymyxin B readily diffused through the dialysis tubing. This
experiment illustrated that free polymyxin B was not released from the
agarose beads in the presence of E. coli. In addition, 3H-Iabeled octapeptin
was coupled to agarose by an analogous procedure and incubated with E.
coli. Octapeptin, either bound to bacteria or free in solution, was not
released from octapeptin-agarose. In addition, the growth and respiration
of B. subtilis GSY 201 was unaffected by polymyxin-agarose. This strain
of B. subtilis was quite sensitive to free polymyxin. These results, taken
collectively, indicate that the peptides were not released from the peptide­
agarose derivatives in the presence of bacteria.

Table 3 Demonstration that free polymyxin B was not released from polymyxin-agarose
beads in the presence of H.
coli

Number of viable H. coli cells b


after 1 8 hr (bacteria/ml)

Sample descriptiona in dialysis bag outside dialysis bag

1 .0 mg/ml polymyxin-agarose
i n dialysis bag 0.0 1 .9 X 1 09
4 /Jog/ml free polymyxin
in dialysis bag 0.0 0.0
1 .0 mg/ml u n derivatized agarose
in dialysis bag 2.0 X 1 09 2.0 X 1 09
1 .0 mg/ml polymyxin-agarose
outside dialysis bag 2 .0 X 1 09 0.0

a Dialysis tubing containing 5 0 ml of nutrient media was suspended i n 1 00 ml of nutri­


ent media in a n Erlenmeyer flask, stoppered with a cotton plug, and autoclaved. An
inoculum of E. coli SC 9 2 5 1 ( 1 .0 X 1 0 6 cells/ml) was in troduced both inside and outside
the dialysis bag; polymyxin-agarose beads, underivatized agarose beads, or free poly­
myxin B were placed either inside or outside the bag as indicated. The flask containing
the dialysis bag was then vigorously shaken at 3 7° C for 1 8 hr.
b Yiable counts were determined by serial dilution and growth on nutrient agar plates.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 755

The growth of E. coli SC 9251 was inhibited by polymyxin-agarose


(Figure 7). This was manifested as a lag in the growth curves, which
increased with concentration of immobilized polymyxin. At a concentration
of 1 mg/ml, no growth occurred even after 96 hr, and the growth lag period
was inversely proportional to inoculum size (Table 4). Underivatized aga­
rose beads, bacitracin-agarose, and vancomycin-agarose had no effect on the
growth of gram-negative bacteria. Octapeptin-agarose also inhibited the
growth of E coli and P. aeruginosa but not B. subtilis. The concentrations
of the immobilized peptides used in these experiments might seem rather
high. However, the polymyxin-agarose beads used contained 1 50 J-Lg of
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

polymyxin/mg of polymyxin-agarose. If it is assumed that the bacteria can


only interact with the surface of the beads, then only a minor percentage
(no more than 1 %) of the polymyxin was available for interaction with the
bacteria. A concentration of 1 mg/ml of polymyxin-agarose, which com­
pletely inhibited the growth of E. coli, would correspond to 1 .5 J-Lg/ml or
less of accessible polymyxin.

240
A
8 0
-;;; 2 00 C
-
'c
=:J
_ 160
Q;

.r::.
120

e
(!) 80
.2

'"
ti
0
co

4 8 12 16 20 24 28 32 36 40
T i me ( hours)

Figure 7 Inhibition o f E. coli growth i n the presence o f polymyxin B covalently


attached to agarose beads. An inoculum of E. coli SC 9251 (2.3 X 106 bacteria/ml)
was grown at 3 7°C in the absence or presence of polymyxin B covalently attached
no polymyxin-agarose or 0.5 mg/ml of underivatized agarose; B, 0. 1
to agarose. A,
mg/ml polymyxin-agarose; C, 0.25 mg/ml polymyxin-agarose; and D, 0.50 mg/ml
polymyxin-agarose. A sample containing 1.0 mg/ml of polymyxin-agarose did not
show any growth when monitored up to 96 hr.
756 STORM, ROSENTHAL & SWANSON

The respiration of E. colt: E coli spheroplasts, and B. subtilis protoplasts


was rapidly inhibited by polymyxin-agarose; however, the respiration of B.
subtilis was unaffected by immobilized polymyxin. These experiments allow
a number of important conclusions to be made concerning the mechanism
of polymyxin action against gram-negative bacteria and the functional sig­
nificance of outer membranes. The polymyxins and octapeptins do not have
to enter a bacterial cell to inhibit growth, nor are they required to move back
and forth across the membrane. The effects of these peptides on biochemical
processes such as nucleic acid or protein biosynthesis must be secondary to
membrane damage. In the case of gram-positive bacteria, such as B. subtilis,
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

the intact cell was insensitive to immobilized polymyxin; however, proto­


plast respiration was inhibited. The sensitivity of the gram-negative bacteria
to polymyxin-agarose or octapeptin-agarose must have been due to interac­
tions with the outer membrane. This is consistent with the observations that
both peptides caused significant changes in the structure of outer mem­
branes. Inhibition of spheroplast or whole-cell respiration by polymyxin­
agarose indicates that the peptide can alter this function either directly by
interactions with the inner membrane, which is the site for electron trans­
port and oxidative phosphorylation, or indirectly by perturbation of the
outer-membrane structure. This latter point is quite interesting, since low

Table 4 Inhibition of growth of E. coli and P. aeruginosa by immobilized polymyxin B

b
Inoculum sizea Polymyxin-agarose Growth lag
Bacterial strain (bacteria/ml) (mg/ml) (hr)

E. coli SC 925 1 2.3 X 1 06 0.00 0.0


2.3 X 1 06 0.10 1 .8
2.3 X 1 06 0.25 6.0
2.3 X 1 06 0 .5 0 3 1 .0
2.3 X 1 06 1 .00 > 96
2.3 X 1 05 0.10 10.6
2.3 X 1 04 0.10 13.6
P. aeruginosa PaG 1 3 9.1 X l OS 0.00 0.0
9.1 X l OS 0.10 0.0
9.1 X l OS 0.25 2.5
9.1 X 1 05 0.50 6.6
9.1 X l OS 1 .00 1 6 .8
B. subtilis GSY 201 3 .0 X l OS 0.10 0.0
3.0 X 1 04 0.10 0.0
3.0 X 1 04 0.25 0.0
3.0 X 1 04 0.50 0.0
3.0 X 1 04 1 .00 0.0

a
Bacteria were grown in nutrient media at 37°C with shaking.
b
Growth lag is the difference in time required for the control and a sample treated
with po!ymyxin-agarose to reach 50% of their maximum growth.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 757

levels of EDTA, which were just sufficient to complex Mg2+ and CaH
crucial for outer-membrane structure (107), made E coli osmotically frag­
ile (1 62) and caused the lysis of P. aeruginosa (78). It is possible that the
outer membrane of E coli, which is attached to the peptidoglycan through
the Braun lipoprotein (104, 105, 1 63), functions with the peptidoglycan as
a mechanical barrier to prevent lysis. If the latter is true, then damage to
the outer membrane by the polymyxins or octapeptins could indirectly
affect the selective permeability ofthe inner membrane and inhibit bacterial
respiration.
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

MECHANISM OF ACTION

Any model proposed for the effects of these peptides on bacterial metabo­
lism and growth must take into account the following observations. Al­
though polymyxin inhibits many biochemical processes, one of the earliest
detectable changes is an increase in the permeability of the cytoplasmic
membrane with respect to polar or charged molecules. Inhibition of biosyn­
thetic pathways that occur within the cell must be secondary to membrane
damage because the peptides do not have to enter the cell to inhibit growth.
Changes in membrane permeability are sufficient to account for all other
biochemical effects, including alteration in the rates of respiration. The
studies with immobilized polymyxin or octapeptin indicate that these pep­
tides are probably not ionophores facilitating the transport of ions by diffus­
ing back and forth across the membrane. Binding of the peptides to the
outer surface of membranes is sufficient to alter their structure and permea­
bility. This general model has been developed through the efforts of several
different research laboratories over the last 25 years (8, 9, 24, 25, 42, 67,
68, 85, 1 36, 1 50- 1 52, 1 54).
One of the primary questions raised in the introduction to this chapter
concerned the detailed molecular mechanism for polymyxin- or octapeptin­
induced permeability changes. A complete description of this phenomenon
does not exist, although certain aspects of this process are now more clearly
understood. These peptides bind to and disrupt the structure of phos­
pholipid and lipopolysaccharide aggregates. The permeability oflipid bilay­
ers for polar molecules is increased by the actions of these peptides.
Inhibition of bacterial growth, binding of the peptides to membranes and
lipids, and perturbation of membrane structure are all specifically inhibited
by divalent cations, which are crucial structural elements of biological
membranes. The influence of these divalent cations on the activities of these
antibiotics suggests that polymyxin and octapeptin may competitively dis­
place Mg2+ or Ca2+ from negatively charged phosphate groups on mem­
brane lipids. Electrostatic interactions between the peptide and lipid
phosphates are crucial determinants of activity; however, the fatty acid side
758 STORM, ROSENTHAL & SWANSON

chain also seems to play an important role. The affinity of the antibiotics
for membrane lipids is undoubtedly increased by hydrophobic interactions
between the peptide fatty acid and the hydrophobic domain of the mem­
brane. Insertion of the antibiotic's fatty acid into the core of the membrane
may perturb the normal packing of membrane phospholipid fatty acids,
since the antibiotic's fatty acid is shorter (C : 8 to C : 1 1 ) than the average
fatty acid chain of membrane phospholipids (C : 1 6 to c : 1 8). This possibil­
ity was suggested by structure-function correlations in which antibiotic
activity was optimal for derivatives having a fatty acid chain of C ; 8 to
C : 1 2. Longer chain fatty acids, such as C : 1 6 or C ; 1 8, actually dimin­
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

ished antibiotic activity.


A tentative model for the effects of polymyxin and octapeptin on lipid­
bilayer permeability can be proposed on the basis of the conclusions cited
above. Diffusion of charged or polar molecules through a lipid bilayer
occurs at an extremely low rate because of the low dielectric constant of the
hydrophobic domain of the membrane. The structure of the bilayer is
dictated by specific packing of the phospholipid or lipopolysaccharide polar
head-groups and hydrophobic interactions between fatty acid chains oriented
parallel to each other. Disruption of phospholipid packing could occur by
a combination of interactions with the antibiotics (Figure 8). Electrostatic
interactions between the peptide amino groups and lipid phosphates could

OUTSIDE

+2 +2
Mg Mg

Figure 8 Hypothetical model for interaction of polymyxin or octapeptin with a


phospholipid bilayer. It is proposed that the fatty acid tail of the peptide penetrates
the hydrophobic domain of the bilayer, with the peptide amino groups interacting
electrostatically with phospholipid phosphates.
POLYMYXIN AND PEPTIDE ANTIBIOTICS 759

displace divalent cations that normally function as cationic bridges between


adjacent phosphates. Since the polycationic cyclopeptide is much larger
than Mg2+ or Ca2+, normal packing between lipid polar head-groups would
necessarily be perturbed, probably leading to an expansion of the outer
monolayer. Insertion of the fatty acid tail into the center of the membrane
would also weaken hydrophobic interactions between adjacent phos­
pholipid fatty acids. This would be analogous to the effects exhibited when
short-chain fatty acids or cis unsaturated fatty acids are incorporated into
phospholipid bilayers (1 64). The combination of these interactions could
lead to disruption of membrane lipid packing, increasing the permeability
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

of the lipid bilayer for polar or charged molecules. The release of outer­
membrane particles by octapeptin may reflect extensive alteration of lipid
packing when a sufficient amount of the peptide is absorbed at a particular
site on the surface of the membrane. This general model, although specula­
tive, is consistent with all available data and can be directly tested by further
experimentation.

CONCLUSIONS

The general mechanism for the antibiotic activities of the polymyxins and
octapeptins has been elucidated by research using a broad range of experi­
mental techniques. However, this phenomenon has not been described in
detailed molecular terms, and this must be one of the major goals for future
research in this area. Since 1 947, when polymyxin was first isolated, there
have been tremendous advances in our knowledge of membrane structure.
The application of biophysical technology such as NMR, ESR, fluorescence
spectroscopy, differential scanning calorimetry, and electron microscopy
has been particularly valuable for studying model and biological membrane
structures. It is these techniques which will provide a detailed molecular
mechanism for the effects of these peptide antibiotics on membrane struc­
ture. In addition, the large number of antibiotic derivatives available should
be exploited more extensively for structure-function correlations. The ulti­
mate goal is to correlate the biological properties of these peptides with their
effects on the physical properties of membranes and to rationalize these
events in terms of lipid-peptide interactions.
760 STORM, ROSENTHAL & SWANSON

Literature Cited

1 . Lardy, H. A., Ferguson, S. M. 1 969. 27. Gale, E. F. 1963. Pharmacol. Rev.


Ann. Rev. Biochem. 38:99 1 - 1 034 1 5 :481-530
2. Kinsky, S. C. 1970. Ann. Rev. Phar­ 28. Sebek, O. K. 1 967. In A ntibiotics I:
macol. 10: 1 1 9-42 Mechanisms ofAction, ed. D. Gottlieb,
3. Storm, D. R. 1 974. Ann. NY Acad. Sci. P. Shaw, pp. 1 42-52. New York:
235:387-98 Springer. 785 pp.
4. Andreoli, T. E. 1974. Ann. NY Acad. 29. Hayashi, K., Suzuki, T. 1967. Bull. Inst.
Sci. 235:448-68 Chem. Res. Kyoto Univ. 43:259-77
5. Lau, A. L., Chan, S. I. 1 975. Proc. Natl. 30. Paulus, H. 1967. In Antibiotics fl· Bio­
Acad. Sci. USA 72:21 70-74 synthesis, ed. D. Gottlieb, P. Shaw, pp.
·6. Ryabova, I. D ., Gorneva, G. A., Ov­ 254-67. New York: Springer. 760 pp.
chinnikov, Yu. A. 1 975. Biochim. Bio­ 3 1 . Korzybski, T., Kowszyk-Gindifer, Z.,
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org

phys. Acta 40 1 : 109-1 8 Kurylowicz, W. 1 967. Antibiotics: Ori­


Access provided by University of Washington on 02/03/16. For personal use only.

7. Pressman, B. C. 1 976. Ann. Rev. Bio­ gin, Nature and Properties, 1 :98- 1 1 8.
chem. 45:501-30 New York: Pergamon. ·1 144 pp.
8. LaPorte, D., Rosenthal, K. S., Storm, 32. Studer, R. O. 1 967. Prog. Med. Chem.
D . R. 1977. Biochemistry. In press 5 : 1 -5 5
9. Rosenthal, K. S., Swanson, P. E., 33. Hausmann, W . , Craig, L. C. 1 954. J.
Storm , D. R. 1976. Biochemistry 1 5 : Am. Chem. Soc. 76:4892-96
573-76 34. Meyers, E., Pansy, F. E., Basch, H. I.,
10. Hauser, H., Finer, E. G., Chapman, D. McRipley, R. J., Slusarchyk, D. S.,
1 970. J. Mol. Bioi. 53:4 1 9-33 Graham, S. F., Trejo, W. H. 1973. J.
11. Storm, D . R., Strominger, J. L. 1 974. Antibiot. 26:457-62
J. Bioi. Chem. 249 : 1 823-27 35. Barrett-Dee, K., Radda, G. K.,
12. Ainsworth, G. C., Brown, A. M., Thomas, N. A. 1972. In Mitochondria:
Brownlee, G. 1947. Nature 1 60:263 Biogenesis and Bioenergetics. Biomem­
13. Stansly, P. G., Shepherd, R. G., White, branes: Molecular Arrangements and
J. 1947. Bull. Johns Hopkins Hosp. Transport Mechanisms. FEBS Proc.
8 1 :43-54 Meet. 8th, 28:231-52. New York: El­
14. Benedict, R. G., Langlykke, A. F. 1 947. sevier. 4 1 4 pp.
J. Bacteriol. 54:24--2 5 36. Gibbons, W. A., Alms, H., Bockman,
1 5. Wilkinson, S., Lowe, L. A. 1 966. Na­ R. S., Wyssbrod, H. R. 1972. Biochem­
ture 2 1 2:3 1 1 istry 1 1 : 1 72 1-25
1 6. Fujikawa, K., Suketa, Y., Hayashi, K., 37. Galardy, R. E., Craig, L. C., Printz, M.
Suzuki, T. 1965. Experientia 2 1 :307-8 P. 1974. Biochemistry 1 3 : 1 674--77
1 7. Meyers, E., Parker, W. L., Brown, W. 38. Craig, L. C. 1 964. Science 144: 1093-99
E. 1974. Ann. NY Acad. Sci. 235 :493- 39. Chapman, T. M., Golden, M. R. 1 972.
501 Biochem. Biophys. Res. Commun.
18. Shoji, J., Hinoo, H., Wakisaka, Y., 46:2040--47
Koizumi, K., Mayama, M., Matsuura, 40. Vrry, D. W., Ohnishi, M. 1 970. In Spec­
S., Matsumoto, K. 1 976. J. Antibiot. troscopic Approaches to Biomolecular
29 : 5 1 6-20 Configurations, ed. D. W. Vrry, pp.
1 9. Shoji, J., Hinoo, H., Sakazaki, R. 1 976. 263-300. Chicago: Am. Med. Assoc.
J. Antibiot. 29:521-25 Press
20. Shoji, J., Hinoo, H. 1975. J. A ntibiot. 41. Venkatachalam, C. M. 1 968. Biopo/y­
28:60-63 mers 6: 1 425-36
21. Shoji, J., Kato, T. 1 976. J. Antibiot. 42. Teuber, M., Bader, J. 1976. Arch. Mi­
29:3 80-89 crobiol. 109:51-58
22. Bodanszky, M., Izdebski, J., Mura­ 43. Vogler, K., Studer, R. 0., Lergier, W.,
matsu, I. 1969. J. Am. Chem. Soc. Lanz, P. 1 960. Helv. Chim. Acta
9 1 :235 1-58 43: 1 7 5 1-60
23. Sogn, J. A. 1976. J. Med. Chem. 44. Vogler, K., Studer, R. 0., Lanz, P., Ler­
1 9: 1 228-32 gier , W., Boehni, E. 1 964. Experientia
24. Newton, B. A. 1 956. Bacteriol. Rev. 20:365--66
20: 1 4-27 45. Kurihara, T., Takeda, H., Ito, H. 1972.
25. Few, A. V. 1955. Biochim. Biophys. Yakugaku Zasshi 92: 1 29-34
Acta 1 6 : 1 37-45 46. Suzuki, T., Hayashi, K., Fijikawa, K.,
26. Davis, B. D ., Feingold, D . S. 1 962. In Tsukamoto, K. 1 9 64 . J. Biochem.
The Bacteria. ed. I. C. Gunsalus, E. Y. Tokyo 56:335-43
Stanier, 4:343-97. New York: Aca­ 47. Parker, W. L., Rathnum, M. L. 1975.
demic J. Antibiot. 28:379-89
POLYMYXIN AND PEPTIDE ANTIBIOTICS 761

48. Chihara, S., Ito, A . , Yahata, M . , Tobita, 72. Pruul, H., Reynolds, B. L. 1972. In/ect.
T., Koyama, Y. 1974. Agric. Bioi. Immun. 6:709- 1 7
Chem. 38:521-29 7 3 . Rosenthal, K. S . , Storm, D. R . 1975.
49. Chihara, S., Ito, A., Yahata, M., Tobita, Uncoupling of Oxidative Phosphoryla­
T., Koyama, Y. 1974. Agric. Bioi. tion by EM 49. Presented at Intersci.
Chem. 3 8 : 1 767-77 Conf. Antimicrob. Agents Chemother.,
50. Schwartz, B. S., Warren, M. R., Bark� 1 5th, Washington DC, Abstr. No. 422
ley, F. A., Landis, L. 1960. Antibiotic 74. Ennis, H. L. 1965. I. Bacteriol 90:
Ann. 7:41-60, 1959-1 960 1 102-8
5 1 . Dulaney, E. L., Laskin, A. I. 197 1 . Ann. 75. Ennis, H. L. 1965. 1. Bacterial 90:
NY Acad. Sci. 82: 1--4 1 5 1 1 09- 1 9
52. Weinstein, L . 1970. In The Pharmaco­ 7 6 . Cohen, S . , Purdy, C. V . , Kushnick, J. B .
logical Basis of Therapeutics, ed . L. S. 1952. Fed. Proc. II :463-64
Goodman, A. Gilman, pp. 1 287-90. 77. Sai to, K., Akada, Y., Kawasaki, N.
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org

1972. I. Biochem. Tokyo 72:2 1 3- 1 4


Access provided by University of Washington on 02/03/16. For personal use only.

New York: Macmillan. 1794 pp.


53. Wright, W. W., Welch, H. 1960. Antibi­ 7 8 . Eagon, R. G., Carson, K. J. 1965. Can.
otic Ann. 7:61-74, 1959-1 960 I. Microbial. 1 1 : 1 93-201
54. von Graevenitz, A., Nourbakhsh, M. 79. Neu, H. c., Ashman, D. F., Price, T. D.
1972. Med. Microbial Immunol. 1 57: 1967. I. Bacterial. 93:1 360-68
1 42-48 80. Ghuysen, J. M. 1968. Bacterial Rev.
55. Muyembe, T., Vandepitte, J., De­ 32:425-64
smyter, J. 1973. Antimicrob. Agents 8 1 . Schwarz, U., Asmus, A., Frank, H.
Chemother. 4:521-24
1969. 1. Mol. Bioi. 4 1 :419-29
56. Gilleland, H. E. Jr . , Murray, R. E.
82. Elmros, T., Burman, L. G., Bloom, G.
1976. 1. Bacterial 125:267-8 1
D. 1 976. I. Bacterial. 1 26:969-76
57. Bayer, M. E. 1974. Ann. NY A cad. Sci.
83. Rothfield, L., Pearlman-Kothencz, M.
235:6--29
1969. 1. Mol. Bioi. 44:477-92
58. Murray, F. J., Tetrault, P. A., Kauf­
84. Boyer, P. D., Cross, R. L., Momsen, W.
mann, O. W., Koffler, H., Peterson, D.
1973. Proc. Natl. Acad. Sci. USA
H., Colingsworth, D. R. 1949. 1. Bac­
70:2837-39
terial. 57:305-12
85. Few, A. V., Schulman, J. H. 1953. I.
59. Suzuki, T., Hayashi, K., Fujikawa, K.,
Gen. Microbial. 9:454-66
Tsukamoto, K. 1963. 1. Biochem.
86. Nakajima, K., Kawamata, J. 1965.
Tokyo 54:555-56
Biken I. 8:233-39
60. Nakajima, K. 1967. Chem. Pharm.
87. Newton, B. A. 1955. 1. Gen. Microbial
Bull. 1 5 : 1 2 1 9-24
1 2:226--3 6
6 1 . Meyers, E., Brown, W. E., Principe, P.
A., Rathnum, M. L., Parker, W. L. 88. Chen, C. H., Feingold, D. S. 1972. An ­
1973. 1. Antibiot. 26:444--4 8 timicrob. Agents Chemother. 2:33 1-35
62. Parker, W. L., Rathnum, M. L. 1973. 89. Kundig, W., Kundig, F. D., Anderson,
1. Antibiot. 26:449-56 B., Roseman, S. 1966. J. Bioi. Chem.
63. Teuber, M. 1 970. Z Naturforsch. Teil B 241 :3243-46
25: 1 1 7 90. Tanaka, N., Igusa, S. 1968. J. Antibiot.
64. Chihara, S., Tobita, T., Yahata, M., Ito, 2 1 :239-40
A., Koyama, Y. 1973. Agric. Bioi. 9 1 . Cerny, G., Teuber, M. 1 9 7 1 . Arch. Mik­
Chem. 37:2455-62 robio/. 7 8 : 1 66-79
65. Nikaido, H. 1 976. Biochim. Biophys. 92. Mitchell, P. 1970. Symp. Soc. Gen. Mi­
Acta 433 : 1 1 8-32 crobial. 20th. pp. 1 2 1-66
66. Newton, B. A. 1953. 1. Gen. Microbial 93. Simoni, R. D., Postma, P. W. 1975.
9:54-64 Ann. Rev. Biochem. 44:523-54
67. Newton, B. A. 1953. 1. Gen. Microbial. 94. Wilson, D. F., Ting, H. P., Koppelman,
8:6-7 M. S. 1 9 7 1 . Biochemistry 10:2897-902
68. Newton, B. A. 1953. Nature 1 72: 95. Pavlasova, E., Harold, F. M. 1969. I.
1 60-6 1 Bacterial. 98:1 98-204
69. Mohan, R. R., Pianotti, R. S., Leverett, 96. Kimmich, G. A., Randles, J., Brand, J.
R., Schwartz, B. S. 1 962. Antimicrob. S. 1975. Anal Biochem. 69: 1 87-206
Agents Chemother. 2:801-14 97. Raisin, M. P., Kepes, A. 1973. Biochim.
70. Teuber, M. 1974. A rch. Microbial. Biophys. Acta 305:249-59
100: 1 3 1-44 98. Patel, L., Kaback, H. R. 1976. Bio­
7 1 . Wahn, K., Lutsch, G., Rockstroh, T., chemistry 1 5:2741-46
Zapf, K. 1968. Arch. Mikrobiologiya 99. De Petris, S. 1967. I. Ultrastruct. Res.
63: 103-16 1 8 :45-53
762 STORM, ROSENTHAL & SWANSON

100. Romeo, D., Hinckley, A., Rothfield, L. 1 27. Cerny, G., Teuber, M. 1 97 1 . Arch. Mik­
1 970. J. Mal. Bial. 53:491-501 robiol. 82:36 1-70
1 0 1 . Osborn, M. J., Gander, J. E., Parisi, E., 128. Teuber, M., Cerny, G. 1973. Arch. Mik­
Carson, J. 1972. J. Bial. Chem. robiol. 9 1 :235-40
247:3962-72 1 29. Willsky, G. R., Malamy, M. H. 1 976.
102. Inoue, K. 1 974. Biachim. Biaphys. Acta J. Bacteriol. 1 27:595-609
339:390-402 1 30. Malamy, M. H., Horecker, B. L. 1 964.
103. Osborn, M. J., Rick, P. D., Lehmann, Biochemistry 3 : 1 889-93
Y., Rupprecht, E., Singh, M. 1974. Ann. 1 3 1 . Neu, H. C., Heppel, L. A. 1 964. Bio­
NY Acad. Sci. 235:52-65 chem. Biophys. Res. Commun. 14:
104. Braun, Y., Bosch, Y., HantJ<e, K., 109- 1 2
Schaller, K. 1974. Ann. NY Acad. Sci. 1 32. Neu, H. C. 1 970. J. Bacteriol. 102:
235:66-82 537-39
105. Braun, Y. 1975. Biachim. Biaphys. Acta 1 33. Nisonson, I., Tannenbaum, M., Neu, H.
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

4 1 5 :335-77 C. 1 9 69. J. Bacteriol. 100: 1083-90


1 06. Smit, J., Kamio, Y., Nikaido, H. 1 975. 1 34. Wetzel, B. K., Spicer, S. S., Dvorak, H.
J. Bacterial. 124:942-58 F., Heppel, L. A. 1 970. J. Bacteriol.
107. Leive, L. 1 974. Ann. NY Acad. Sci. 1 04:529-42
235 : 1 09-29 1 35. Warren, G. H., Gray, J., Yurchenco, J.
108. Lopes, J., Inniss, W. E. 1969. J. Bac­ A. 1 957. J. Bacterial. 74:788-93
terial. 100: 1 1 28-30 1 36. Teuber, M. 1 970. Arch. Mikrobiol. 70:
109. Bader, J., Teuber, M. 1973. Z. Natur­
1 3 9-46
1 37. Leive, L. 1 968. J. BioI. Chem. 243:
forsch. Teil C 28:422-30
2373-80
1 10. Kaye, J. J., Chapman, G. B. 1 963. J.
1 3 8. Weiser, R., Asscher, A. W., Wimpenny,
Bacteriol. 86:536-43
J. 1 968. Nature 2 1 9 : 1 365-66
1 1 1 . Suganuma, A., Hara, K., Kishida, T.,
Nakajima, K., Kawamata, J. 1968. 1 39. Reid, P., Speyer, J. 1 970. J. Bacterial.
1 04:376-89
Biken J. 1 1 : 1 49-55
140. Ennis, H. L. 1967. J. Bacteriol. 93:
1 1 2. Koike, M., Iida, K., Matsuo, T. 1 969.
1 88 1-87
J. Bacterial. 97:448-52
1 4 1 . Jawetz, E. 1 96 1 . Pediatr. Clin. North
1 1 3. Schindler, P. G., Teuber, M. 1 975. An­
Am. 8 : 1 057-7 1
timicrab. Agents Chemather. 8:95-104
142. Collins, M. S., Pappagianis, D. 197:5.
1 14. Bayer, M. E., Remsen, C. C. 1 970. J.
A ntimicrob. Agents Chemother. 7:
Bacterio!. 1 0 1 :304- 1 3
781-87
1 1 5. Chi, E . Y., Lagunotf, D . , Koehler, J . K. 143. Teuber, M. 1 969. J. Bacteriol. 98:
1 976. Prac. Natl. A cad. Sci. USA
347-50
73:2823-27 144. Sud, I. J., Feingold, D. S. 1 970. J. Bac­
1 1 6. Matsumoto, A., Higashi, N., Tamura, teriol. 104:289-94
A. 1973. J. Bacterial. 1 1 3:357-64 145. Shockman, G. D., Lampen, J. O. 1 962.
1 1 7. Barnes, W. G. A., Flesher, A. E., J. Bacterial. 84:508-1 2
Berger, A. E., Arnold, 1. D. 197 1 . J. 146. Suling, W. J., O'Leary, W. M. 1 975.
Bacterial 1 06:276-80 A ntimicrob. Agents Chemother. 8:
1 1 8. Klainer, A. 5., Perkins, R. L. 1 97 1 . J. 334-43
Am. Med. Assac. 2 1 5 : 1 655-57 147. Brown, M. R. W., Richards, R. M. E.
1 1 9. Greenwood, D., O'Grady, F. 1969. 1965. Nature 207 : 1 39 1-93
Science 163: 1076-78 148. Teuber, M., Bader, J. 1 976. Antimicrob.
1 20. Weber, G. 1 966. In Fluarescence and Agents Chemother. 9:26-35
Phosphorescence AnalYSis, ed. D . M. 149. Teuber, M., Bader, J. 1 97 1 . FEDS Lett.
Hercules. New York: Interscience 1 6 : 1 95-97
1 2 1 . Shinitzky, M., Inbar, M. 1974. J. Mal. 1 50. Pache, W., Chapman, D., Hillaby, R.
BioI. 85 :603- 1 5 1 972. Biochim. Diophys. ACla 255:
1 22. Rifkind, D ., Palmer, J. D. 1966. J. Bac­ 358-64
teriol 92:8 1 5- 1 9 1 5 1 . Teuber, M. 1973. Z. Naturforsch. 28:
1 23. Rifkind, D. 1967. J. Bacteriol. 93: 476-77
1 463-64 1 52. Imai, M . , Inoue, K., Nojima, S. 1975.
1 24. Rifkind, D. 1 967. J. Infect. Dis. 1 1 7: Diochim. Diophys. Acta 375: 1 30-37
433-38 1 53. Hsu Chen, c.-C., Feingold, D. S. 1 973.
1 25 . Cooperstock, M. S. 1 974. A ntim icrob. Biochemistry 1 2:2105- 1 1
Agents Chemother. 6:422-25 1 54. Feingold, D. S., Hsu Chen, C.-c., Sud,
1 26. Koike, M., !ida, K. 1 97 1 . J. Bacterial. I. J. 1 974. Ann. NY Acad. Sci. 235:
108: 1402- 1 1 480-92
POLYMYXIN AND PEPTIDE ANTIBIOTICS 763

1 5 5. Grant, C. W. M., McConnell, H. M. 1 62. Birdsell, D. c., Cota-Robles, E. H.


1973. Proc. Natl. A cad. Sci. USA 1 967. J. Bacteriol. 93:427-37
70: 1 238-40 1 63. Inouye, M., Hirashima, A., Lee, N.
1 56. Bliss, E. A., Chandler, C. A., Schoen­ 1974. Ann. NY Acad. Sci. 235:83-90
bach, E. B. 1 949. Ann. NY Acad. Sci. 1 64. Shimshick, E. I., McConnell, H. M.
5 1 :944-5 1 1973. Biochemistry 1 2:23 5 1 -60
1 57. Baker, Z., Harrison, R. W., Miller, B. 1 65. Hayashi, K., Suketa, Y., Tsukamoto,
F. 1 94 1 . J. Exp. Med. 74:621-37 K., Suzuki, T. 1966. Experientia 22:
1 58. Latterade, C., Macheboeuf, M. 1 950. 354--5 5
Ann. Inst. Pasteur Paris 78:753-57 1 66. Suzuki, T., Hayashi, K., Fujikawa, K.,
1 59. Hummel, I. P., Dreyer, W. I. 1962. Bio­ Tsukamoto, K. 1965. 1. Biochem.
chim. Biophys. Acta 63:530-32 Tokyo 57:226--2 7
1 60. Kinsky, S. c., Haxby, I. A., Zopf, D. 1 67. StUder, R. 0., Lergier, W., Vogler, K.
A., A1ving, C. R., Kinsky, C. B. 1969. 1966. Helv. Chim. Acta 49:974--8 5
Biochemistry 8:4 149-58 168. Meyers, E., Parker, W. L., Brown, W.
Annu. Rev. Biochem. 1977.46:723-763. Downloaded from www.annualreviews.org
Access provided by University of Washington on 02/03/16. For personal use only.

1 6 1 . Papahadjopoulos, D., Miller, N. 1 967. E., Shoji, 1., Wakisaka, Y. 1976. J. An­
Biochim. Biophys. Acta 1 35:624--52 tibiot. 29: 1 24 1-42

You might also like