Review TRENDS in Biotechnology
1 January 2006
Directed evolution of metabolic
pathways
Ranjini Chatterjee1 and Ling Yuan2
1
Farasis Energy, Inc., 23575 Cabot Blvd., Hayward, CA 94545, USA
2
Department of Plant and Soil Sciences, Kentucky Tobacco and Research Center, University of Kentucky, Cooper and University
Drives, Lexington, KY 40546, USA
The modification of cellular metabolism is of biotechno-
logical and commercial significance because naturally
occurring metabolic pathways are the source of diverse
compounds used in fields ranging from medicine to
bioremediation. Directed evolution is the experimental
improvement of biocatalysts or cellular properties
through iterative genetic diversification and selection
procedures. The creation of novel metabolic functions
without disrupting the balanced intracellular pool of
metabolites is the primary challenge of pathway
manipulation. The introduction of coordinated changes
across multiple genetic elements, in conjunction with
functional selection, presents an integrated approach
for the modification of metabolism with benign physio-
logical consequences. Directed evolution formats take
advantage of the dynamic structures of genomes and
genomic sub-structures and their ability to evolve in
multiple directions in response to external stimuli. The
elucidation, design and application of genome-restruc-
turing mechanisms are key elements in the directed
evolution of cellular metabolic pathways.
Introduction
Darwin described the evolution of life as a competitive
process of ‘variation and natural selection’ arising from
successive cycles of mutation and adaptation; at the
molecular level, this results in the formation of genetic
diversity in organisms. Whereas selection to improve
phenotype has been practiced for centuries in plant and
animal breeding experiments, the concept of directed
evolution of nucleic acids was introduced as recently as
1967 [1]. Directed evolution incorporates Darwinian
principles of mutation and selection into experimental
strategies for improving biocatalyst or cellular properties.
In the laboratory, directed evolution comprises two
discrete components: first, genetic diversity is created
through the production of a library of genetic variants;
second, the library is evaluated by genetic selection and
high-throughput screening to identify variants with the
required function(s). Point mutation and recombination
are the two primary processes for the generation of genetic
diversity. Iteration of a directed-evolution cycle facilitates
Corresponding authors: Chatterjee, R. (rchatterjee@farasis.com), Yuan, L.
(Lyuan3@uky.edu).
Available online 18 November 2005
www.sciencedirect.com 0167-7799/$ - see front matter Q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.tibtech.2005.11.002
Vol.24 No.
the accumulation of beneficial mutations and the elimin-
ation of deleterious ones (Figure 1) [2,3].
Directed evolution differs from natural evolution in two
key aspects: (i) natural evolution occurs under multiple
and variable selection pressures, whereas directed evol-
ution is accomplished under controlled selection pressure
for predetermined functions; (ii) in directed evolution,
‘non-natural’ functions, of practical use, can be obtained
through the design of appropriate selection schemes,
whereas natural evolution favors functions advantageous
to the survival of the organism [4,5].
The development of practical methods for the evolution
of metabolic pathways is particularly pertinent because
most commercially valuable compounds are produced by,
or derived from, naturally occurring pathways. An array
of compounds, with varied structures and activities, is
present in the metabolome, which is defined as the full
complement of metabolites in an organism [6]. To be
commercially applicable, naturally occurring pathways
often require experimental manipulation. The combina-
torial aspect of directed evolution is a key characteristic
that distinguishes it from other successful strategies for
improving cellular properties, such as classical strain
improvement and metabolic engineering. Classical strain
improvement is a linear process in which repeated rounds
of mutagenesis and selection of individual parental
strains generate distinct progeny with improved pheno-
types. By contrast, directed evolution is a branched
process that allows the simultaneous incorporation of
mutations between genomic elements of multiple parents,
leading to more rapid improvements in phenotype. The
iterative genetic recombination events in directed evol-
ution formats can reduce the detrimental mutations that
frequently accumulate and can exert adverse effects on
strain physiology during classical strain improvement [7].
Metabolic engineering is a rational approach that relies
extensively on information about cellular and enzymatic
function [7–10]. Although strategies for the rational
engineering of metabolism can be guided by the inte-
gration of differential gene expression, characterization of
metabolic flux and enzyme activities, our ability to
accurately predict the outcome of a metabolic engineering
experiment is often impeded by the network of metabolic
activities that might not be fully elucidated and the
stringent regulatory mechanisms controlling individual
pathways. Thus, unforeseen cellular responses to a
 Review TRENDS in Biotechnology
Parental gene(s)
x x
x
Selection
Desired variants
Figure 1. A typical scheme for directed evolution. Individual variants in a mutant library (right) differ from the parental genes by containing mutations that are functionally
neutral (green circle), deleterious (X) or beneficial (red triangle). Progeny expressing these mutant genes are screened for the function of interest. The gene variants isolated
from the improved progeny (red) can be used as parents for the subsequent cycle of evolution.
well-controlled genetic alteration can complicate the
practice of rational methods [8,10–12]. Directed evolution
has the unique advantage of adapting the biocatalyst to
imposed conditions through the design and implemen-
tation of genetic selections and screens for function
without requiring extensive prior knowledge of gene-to-
function relationships.
In this review, we describe the approaches available for
directed evolution and the complementary technologies
that are most relevant to the coordinated manipulation of
metabolic pathways and cellular behavior. We focus on two
technological areas: (i) the construction of nucleotide
sequence diversity, in a concerted manner, across multiple
genetic elements; (ii) the integration of multiple metabolic
segments towards a specific cellular phenotype. The
advances in screening technologies for identification of
useful biological functions have been reviewed elsewhere
[4,13].
Established technologies for directed evolution
In the laboratory, genetic diversity can be introduced by
point mutagenesis, recombination or a combination of the
two methods. Computational methods can combine
rational design with directed evolution through the
incorporation of elements of protein structure and
stability into the design of the gene library. Some
established strategies for directed evolution are briefly
outlined in this section (Table 1). These methods, which
have been primarily demonstrated on single enzymes,
have been extensively reviewed elsewhere [4,5,14,15].
All of the above formats are applicable towards the
directed evolution of metabolic pathways in instances
where limiting steps, including specific enzyme reactions,
transport, or regulatory functions, have been identified,
and the relevant genes are contiguous in the nucleotide
sequence to be evolved. Introduction of an enzyme with
novel specificity can redirect the flux of metabolites in a
www.sciencedirect.com
Vol.24 No.1 January 2006 29
Mutation and/or recombination
xx x Mutant library
Progeny
TRENDS in Biotechnology
host and result in new products. These recombinant
enzymes can be obtained from other organisms known to
produce the compounds, or by directed evolution to create
the desired specificity from an enzyme that normally
catalyzes other reactions. Pathway engineering often
requires alteration of the substrate pools for the key
steps. Thus, directed evolution of the enzymes responsible
for the production of these substrates can enhance or even
redirect biosynthetic pathways. In practice, the selection
of a specific technique is determined by the size and
configuration of the target genetic element, availability of
the nucleotide sequence and protein structure infor-
mation, and access to genetic manipulation techniques
for the particular host organism. The primary limitation of
a majority of the directed evolution methods developed for
single genes is that they cannot target the multiple,
isolated genetic elements that comprise or control
metabolic pathways simultaneously. Whole genome shuf-
fling and combinatorial libraries enhanced by recombina-
tion in yeast (CLERY) are particularly useful for directed
pathway evolution. These are in vivo techniques that
use cellular recombination systems and can access
multiple, distributed genetic elements to redirect cellular
metabolites.
Cellular mechanisms for genome alteration
Cellular functions leading to mutation and genome
restructuring are central to evolutionary change.
Endogenous mechanisms of genetic change include
mutation, recombination, amplification, dispersal, inser-
tion and fusion of sequence elements. These are mediated
through DNA repair enzymes and cellular recombination
systems, conjugal plasmids, transducing phage and
transposons or mobile genetic elements [16]. To devise
effective experimental approaches for the evolution of
metabolic pathways, it will be important to dissect the
molecular processes that affect coordinated changes in the
30 Review TRENDS in Biotechnology Vol.24 No.1 January 2006

Table 1. Examples of established directed evolution methods

Directed evolution method
Random point mutagenesis
Homologous recombination-based methods
DNA shuffling
Staggered extension process
(StEP)
Gene site saturation mutagen-
esis (GSSM)
Random chimeragenesis on
transient templates (RACHITT)
Combinatorial libraries
enhanced by recombination in
yeast (CLERY)
Assembly of designed oligo-
nucleotides (ADO)
Mutagenic and unidirectional
reassembly (MURA)
Y-ligation based block shuf-
fling (YLBS)
Whole genome shuffling
(WGS)
Non-homologous recombination-based methods
Incremental truncation for the
creation of hybrid enzymes
(ITCHY)
Sequence homology indepen-
dent protein recombination
(SHIPREC)
Exon shuffling
Nonhomologous random
recombination (NRR)
Structure-based combinatorial methods
Structure-based combinatorial
protein engineering (SCOPE)
Synthesis of rational design and directed evolution
Computational design
Description Refs
Genetic diversity introduced by random point mutagenesis using error-prone PCR (EP-PCR). This [2]
method is generally applicable to both single genes as well as to multiple contiguous genes
constituting a metabolic pathway.
DNA shuffling methods rapidly recombine mutations that arise from point mutagenesis through [3,54,55]
a fragmentation and PCR-mediated re-assembly process. ‘Family shuffling’ is a format in which
nucleotide sequences from multiple homologous genes are recombined to form highly diverse
genetic variants. ‘Synthetic shuffling’ is a variation of the original family shuffling format that
does not require naturally occurring segments of homology for recombination. DNA shuffling
methods have been applied to single genes, metabolic pathways, groups of homologous genes
in family shuffling formats, and to entire genomes of organisms.
StEP generates recombination events through PCR amplification of a template with very short [56]
extension times, whereby partially elongated oligonucleotides are able to anneal.
GSSM uses sets of degenerate oligonucleotides to introduce all 19 amino acid substitutions at [57]
every position of the gene.
RATCHITT is similar to DNA shuffling. It increases recombination frequencies through the use of [58]
small oligonucleotide linkers between low-homology regions.
A PCR- and in vivo recombination-based strategy for recombining multiple parental genes in [59]
yeast.
ADO is similar to synthetic shuffling in that designed oligonucleotides are used in the formation [60]
of full-length genes with defined crossover points and mutations.
MURA utilizes elements of both DNA shuffling and incremental truncation to generate highly [61]
diverse gene libraries.
YLBS is a general method that allows modules of variable size to be incorporated into sequences [62]
through repeated cycles of ligation of sequence blocks with a stem and two branches, formed by
two unique single-stranded DNA molecules.
The WGS process generates recombination events across entire genomes without requiring [63]
knowledge of the genome sequence. In a manner analogous to family shuffling, WGS enables
recombination between genomes of multiple parental strains. WGS is carried out by recursive,
multiparental protoplast fusions. The fusion of protoplasts offers the advantage of a high
frequency of recombination across complete genomes and mimics the crossing over of entire
genomes that occur in conventional breeding experiments.
ITCHY is a method for generating chimeric gene libraries by non-homologous recombination [64]
through a process of nested gene deletions and fusions.
The SHIPREC method generates variants with a single crossover without requiring sequence [65]
homology between the input sequences.
In vitro exon shuffling is performed using a mixture of oligonucleotides to allow combinations of [66]
exons to be spliced together to generate mosaic proteins.
NRR is based on DNase I fragmentation, blunt-end ligation and/or extension, and capping using [67]
two asymmetrical DNA hairpins to stop the extension. This method potentially provides higher
flexibility in modulating fragment size and crossover frequency, as well as the number of
parental genes.
Multiple PCR primers are designed, based on protein primary, secondary and tertiary structure, [68]
and used to create crossover gene libraries from distantly related proteins. SCOPE provides an
effective means for the creation of functional gene libraries from gene families that share both
low and high sequence homology.
Incorporation of mutations determined to be functionally relevant based on protein structure, [69]
fitness predictions of sequence, and ‘packing’ algorithms.

genome in response to environmental signals. Investi-
gations into the cellular mechanisms of genome re-
assortment date back to the discovery of transposable
elements in maize and the study of episomes in microbes
[17]. In this section, we describe some natural cellular-
mechanisms of genetic change, and their metabolic
consequences, as a preface to experimental strategies for
metabolic pathway evolution. Examples of selected
genome-altering processes are summarized in Table 2.
In prokaryotes, co-transcribed genes encoding proteins
performing coordinated functions are frequently orga-
nized into cluster structures called operons. The gene
www.sciencedirect.com
content, order, and linked or unlinked regulatory elements
of an operon define a specific metabolic function. The trp
operon and the aromatic catabolic operons contain
some interesting genetic rearrangements, which lead to
altered metabolic functions [18,19]. The trp pathway has
evolved within the context of cellular metabolism and is
integrated with other metabolic pathways. Rearrange-
ments of the trp operon, such as gene fusion events, can
result in the incorporation of multiple catalytic functions
into a single polypeptide, gene dispersal (physical separ-
ation of trp operon genes) leading to deregulation, altered
regulatory modes, or participation of the genes in the trp
 Review TRENDS in Biotechnology Vol.24 No.1 January 2006 31

Table 2. Select examples of natural genome altering processes [26,70]

System DNA substrate DNA rearrangement Functional consequence
SOS mutagenesis Non-specific, UV-damaged DNA. Point mutations. Alteration, loss or gain of function.
Homologous F plasmid and chromosomal insertion Reciprocal crossing over. Hfr formation, F’ excision.
recombination sequence (IS) elements.
Plasmid IS elements or tandem r-det Unequal crossing over. Amplification of resistance
repeats. determinants.
Site-specific Antibiotic resistance cassettes, Integration. Generation of multiple resistance
recombination integrons. operons.
Salmonella typhimurium H2 flagellar Promoter inversion. Flagellar-phase variation.
region.
Phage Mu G region. Partial coding sequence Host-range variation.
inversion.
Transposition Plasmid IS1 elements. Replicon fusion. Formation of co-integrate R
plasmids.
IS1 or IS5 termini, bgl operon Transposition and/or insertion. Silent promoter activation.
promoter region.
Bacterial gene transfer Ti-plasmid. Ti-plasmid conjugation, opine Agrobacterium and other bacteria
to plants metabolism and transferred-DNA are capable of horizontally
(T-DNA). transmitting genes into plants.

operon in pathways other than primary amino acid
biosynthesis [18].
The evolution of catabolic pathways for the degradation
of aromatic compounds has been shown to involve gene
transfer, recombination and transposition events [19]. The
role of insertion sequence (IS) elements and plasmids in
aromatic degradation is summarized in Table 3. Certain
catabolic operons are part of IS elements, which are
capable of random insertion into chromosomal and
plasmid DNA, mobilization of adjacent sequences and
insertional activation or inactivation of adjacent (formerly
silent) genes. These IS elements are capable of conferring
extensive catabolic diversity to microbial strains.
The horizontal transfer of genes between bacterial taxa
has played a significant role in genome diversification.
The transfer of genes between the closely related enterics
Escherichia coli and Salmonella enterica includes genes
conferring functions for the biosynthesis of polysacchar-
ides, pilins and coenzyme B12, the transport of iron and
phosphate, and the catabolism of lactose, propanediol and
phosphonates [20]. Examples of inter-domain horizontal
gene-transfer have been observed between genes of the
tryptophan (trp) operon and genes responsible for the
biosynthesis of isopentenyl diphosphate (IPP), the uni-
versal precursor of isoprenoid compounds [18,21].
At the level of cellular metabolism, two prevalent
models explain the evolution of enzymes in the context
of a pathway. The retro-evolution model proposes that
enzymes catalyzing early reactions in a pathway evolved
by duplication and mutation of enzymes catalyzing the
later reactions. As substrates for later reactions were
depleted, selection pressure on preceding enzymes
necessitated the synthesis of the required substrates
from precursor molecules: examples of homology between
enzymes catalyzing sequential steps exist in the histidine,
tryptophan and methionine biosynthetic pathways. The
recruitment model, however, proposes that enzymes with
similar reaction mechanisms, but in different pathways,
evolved by duplication and mutation; the retention of
catalytic mechanisms is the basis of this hypothesis.
Studies comparing sequence, structure and reaction
mechanisms of enzymes have examined the evolution of
enzymes in pathways. To date, the majority of the studies
indicate that it is more probable that new enzymes evolved
from enzymes within the same class (with similar
chemical function) rather than from enzymes in the
same pathway [22–24]. The assignment of orthologs –
genes derived from a common ancestor that retain similar
functions but are separated by speciation – and paralogs –
genes that have evolved distinct functions following
duplication within a genome – in sequenced genomes
has been informative in tracing the formation of metabolic
pathways. An interesting illustration of the recruitment
theory is seen in the recent evolution of a pathway for

Table 3. Evolutionary mechanisms of an aromatic degradation pathway [19]

Mechanism of sequence alteration
Conjugative transmission of TOL
plasmids carrying xyl operons for
degradation of methylbenzenes
Transposition of insertion sequence
(IS) elements into genomes:
i) TOL catabolic operons (carrying xyl
genes) belong to transposable element
Tn4651, a Tn3-type transposon
ii) IS193 element
iii) Tn5280
Functional consequence
Transfer of TOL plasmid pWW0 to Pseudomonas sp. strain B13 extended its degradative capability
from 3-chlorobenzoate to 4-, and 3, 5-chlorobenzoate by acquisition of the plasmid-borne toluate
dioxygense xylXYZ genes. Conjugative transmission ensures coordinate transfer and regulation of
operon genes.
XylF- and XylJ-like genes flanking todC1C2BAD operon form novel toluene degradation pathway as
a result of transposition.
Multiple copies of IS193 present in chq locus for degradation of 2,4,5-trichlorophenoxyacetic acid in
P. cepacia; role in development of the pathway by activating expression of adjacent genes through
transposition.
Tn5280 consists of chlorobenzene dioxygenase genes of Pseudomonas sp. strain P51, flanked by
IS1066 and IS1067, forming a composite transposon capable of inserting randomly into the genome.

www.sciencedirect.com
32 Review TRENDS in Biotechnology
the degradation of pentachlorophenol (PCP) in Sphingo-
monas chlorophenolica [25]. Here, the new pathway
appears to have been assembled, and evolved, from a
combination of genes for the degradation of dichlorophenol
compounds and an aromatic-amino-acid catabolism
pathway.
Strategies for the directed evolution of metabolic
pathways
In contrast to the numerous studies on the directed
evolution of single enzyme catalysts, fewer examples exist
concerning the experimental evolution of metabolic path-
ways. This is not unexpected because the coordinated
improvement of contiguous or dispersed genes towards a
cellular product or phenotype involves complexities not
encountered in the optimization of a single gene. Although
several directed evolution formats can be used to improve
an isolated enzyme in a pathway, the concerted evolution
of pathway components, in the context of cellular
metabolism, is dependent on multiple factors, including
the configuration of the target sequence elements (linked
or dispersed), localization (plasmid- or chromosome-
borne), regulation, and host genetics and physiology.
Here, we discuss some strategies of practical use for the
experimental evolution of metabolic pathways.
Adaptive mutation
Molecular genetics suggest that microbes have a low
tolerance for random variability because they possess a
range of sophisticated DNA-repair systems. Mutations that
arise in slowly dividing or non-dividing cells under
prolonged exposure to non-lethal selection conditions are
generally referred to as adaptive mutations. These
mutations are non-random, specific for the selection
conditions and provide a growth advantage to the cell [26,
27]. The study of evolution through adaptive mutagenesis
has been based on the ability of strains to acquire mutations
that initiate new pathways for catabolism of a carbon source
not metabolized by wild-type strains. Commonly observed
adaptive mechanisms include constitutive expression of
previously inducible enzymes, altered substrate specificities
and improved transport of the limiting nutrient.
One example of adaptive mutation was demonstrated
experimentally by the araB–lacZ fusion system [26,28]. A
pre-fusion strain of E. coli, containing the Mu prophage
sequence, which separates araB from a promoter-less lacZ
gene, lacked the ability to grow on lactose because of the
absence of a functional promoter; to recover this ability
requires the formation of a hybrid araB–lacZ operon.
Under prolonged selection on lactose, Mu-mediated
excision resulted in the creation of a functional araB–
lacZ operon, leading to growth of the mutant on lactose.
Subsequent molecular mechanistic studies have eluci-
dated the roles of multiple cellular factors in catalyzing
the DNA rearrangement processes.
The ebg operon was also experimentally evolved for
lactose utilization by adaptive mutagenesis [29]. The ebg
operon is paralogous to the lac operon, and was identified
as encoding a second b-galactosidase in a DlacZ strain of
E. coli. The ebg operon includes ebgR, which encodes a
repressor that regulates the ebgABC genes, and ebgA and
www.sciencedirect.com
Vol.24 No.1 January 2006
ebgC, which encode the a and b subunits of the Ebg
b-galactosidase enzyme, respectively; the function of the
ebgB product is unknown. The wild-type EBG enzyme is
unable to support growth on lactose, even when the ebg
operon is constitutively expressed. When selected for
growth on lactose, single point-mutations accumulated
in the ebgA gene of a DlacZebgRK strain, which conferred
the ability to grow on lactose and other b-galactoside
sugars [30]. Kinetic analyses of functional, mutant EBG
enzymes indicated that substrate specificities increased
by at least an order of magnitude. The experimental
evolution of the ebg operon, and characterization of the
resulting mutants, demonstrates the creation of a novel
metabolic pathway and represents a model for the
examination of evolutionary potential within genomes
[29,31]. A strain carrying deletions in both lacZ and ebgA
was unable to produce spontaneous mutants capable of
using lactose under various mutagenesis protocols,
indicating that the potential for the evolution of lactose
use in E. coli is limited to the ebg operon.
Adaptive mutagenesis is a practical strategy for
directed evolution of pathways because of the ease of
execution and the fact that the selection conditions can
both ‘direct’ useful mutations to genes encoding a specific
metabolic function and lead to the identification of latent
metabolic capabilities encoded in the genome. Following
selection, rigorous phenotype-to-genotype correlations are
essential to ensure capture and stabilization of beneficial
mutations. The activation of natural genome-altering
mechanisms under severe stress is a related approach to
directed mutagenesis. Mismatch repair-deficient mutator
strains, under stringent selection conditions, have been
shown to increase the frequency of non-random, beneficial
mutations [27]. The underlying mechanisms of adaptive
mutation are complex, requiring interactions between the
systems altering the cellular genome and the signal
transduction and regulatory machineries. The activities
of insertion elements and transposons are known to
require multiple cellular components and functions,
including integration host factor, DnaA, Dam methyl-
ation, programmed frame-shifting and differential utiliz-
ation of internal promoters [26]. As the biochemistry of
DNA rearrangement is further elucidated, the application
of these naturally occurring molecular mechanisms will
become increasingly practical for experimental
pathway evolution.
Transposition and conjugation
Metabolic modifications achieved through the use of mobile
genetic elements have been illustrated experimentally by
the activation of cryptic genes, through transposition and by
the conjugal transfer of genes [32] – a transposition-based
method for the detection of cryptic antibiotic-resistance
genes has been described [33,34]. Combining a synthetic
transposon with a variant Tn5 transposase forms a stable
transposon–transposase complex that can insert randomly
into target DNA, in vivo or in vitro; following insertion, the
expression of downstream genes is initiated from a strong
transposon-encoded promoter. The transposition method
was used to investigate the S. enterica strain, LT2, for the
presence of cryptic antibiotic-resistance genes. Following
 Review TRENDS in Biotechnology
transposition, the library was selected on 16 commonly used
antibiotics. Three classes of antibiotic resistance were
detected: two with increased resistance to aminoglycosides,
and one with resistance to fluoroquinolones and tetra-
cyclines. Isolation and characterization of the resistance-
encoding genes in a separate strain revealed that activation
of the cryptic genes aac(6 0 )-Iaa and rma conferred
aminoglycoside and fluroquinolone resistance, respectively,
in the previously antibiotic-sensitive strain of S. enterica
LT2. The transposition approach to pathway evolution is
applicable to diverse operations, such as activation (or
deactivation) of selected metabolic functions, targeting of
entire metabolic pathways to random genomic locations to
confer favorable expression or regulation properties, and for
predicting and identifying novel metabolic functions.
Transposition provides a rapid and systematic method to
examine the genome for novel metabolic functions, and for
manipulating metabolism at the genome level. Routine use
of transposition technologies is enabled by the availability of
commercial products for in vivo and in vitro transposition,
and for the systematic analysis of transposition libraries.
Intensive parallel screening of a transposon library at the
genetic and functional levels is required for accurate
genotype-to-phenotype assignments.
The conjugal transfer of genes has been much used in the
development of Lactococci strains for dairy industry
applications [35]. Conjugation was used to construct a
nisin-producing strain of Lactococcus lactis cremoris that
inhibits growth of pathogenic and spoilage organisms in
cheeses. The transfer of determinants for both nisin
production and resistance from the non-commercial donor
strain – L. lactis lactis – to commercial recipient strains of
L. lactis cremoris was achieved by mating, followed by
curing of marker plasmids from L. lactis cremoris
transconjugants [36].
The combined use of conjugative and integrative
elements is an efficient method for disseminating meta-
bolic functions. Intra- or inter-species gene transfer by
conjugation and transposition allows for both construction
of defined genetic backgrounds and investigation of host
mutations that could augment a metabolic pathway.
Manipulation of global regulatory systems
As controllers of physically and functionally separate
sequence elements, the components of global regulatory
systems rank as leading targets for metabolic modification
[37]. Although the recognition and study of global regulatory
systems have existed for some time, efforts towards their
systematic modification were attempted only recently.
Global regulators can exert opposing effects on discrete
metabolic pathways; thus, it is possible to enhance one
pathway and decrease the effects of competing pathways by
modification of a single regulator. Gene expression analyses,
conducted on rationally engineered or mutated organisms,
have led to the identification of global regulators that
determine a specific phenotype by modulating the activities
of multiple peripheral pathways. An ethanol-tolerant E. coli
mutant, LY01, was evolved by adaptation of the ethanolo-
genic strain, KO11. The use of gene-array technology to
investigate the expression changes between the mutant and
wild-type strains identified a non-functional fumarate
www.sciencedirect.com
Vol.24 No.1 January 2006 33
nitrate reductase regulator (fnr) gene in LY01 [38]. Recent
examination of transcript levels in classically improved
overproducing strains of Saccharopolyspora erythraea and
Saccharopolyspora fradiae indicated altered regulation of
antibiotic biosynthetic enzymes and enzymes involved in
precursor metabolism, which therefore provide additional
targets for rational engineering and directed evolution [39].
An early example of global regulatory network modifi-
cation was the disruption of carbon storage regulator (Csr)
as a way to increase the supply of precursors for aromatic
amino acid biosynthesis. The CsrA protein regulates the
expression of about 25 different enzymes in carbon
metabolism by binding to specific mRNAs; the central
metabolite, phosphoenolpyruvate (PEP), is a limiting
precursor in aromatic amino acid biosynthesis. CsrA
regulates three enzymes that metabolize PEP directly and
reduce its availability for aromatic amino acid biosynthesis.
Elevated PEP concentrations resulting from CsrA disrup-
tion translated into a twofold increase in phenylalanine
production [40].
Coordination of the transcriptional control of biosyn-
thetic genes has emerged as a mechanism for the production
of secondary metabolites. Such regulation of these biosyn-
thetic pathways can be achieved using specific transcription
factors (TFs). TFs are typically multi-domain proteins that
have evolved in nature through protein-domain shuffling
[41]. The modular nature of TFs has led to the idea that,
through protein engineering, specific modules of TFs can be
designed to regulate the desired gene(s). Furthermore,
pathway-wide regulation can potentially be achieved by a
single ‘master’ TF (MTF) (Figure 2), which is capable of
regulating all promoters in a particular pathway and
functioning in all cell types. The plausibility of developing
TF TF G
(a) A B C D E F
TF TF
MTF TF
G
(b) A B C D E F
TF
TF
MTF TF G
(c) A B C D E F
TF TF
TRENDS in Biotechnology
Figure 2. Schematic representation of pathway regulation by a master transcription
factor (MTF). (a) Natural pathway regulated by several TFs; (b) MTF designed to up-
regulate multiple steps can direct the pathway toward product F or (c) product G
(enlarged boxes for products F and G represent increased accumulations).
34 Review TRENDS in Biotechnology
engineered MTFs is supported by the fact that multigene
regulatory TFs exist in nature, and the mutations in these
TFs are implicated in broadening phenotypic diversity [42].
In recent years, impressive advances have been made in the
engineering of TF-like proteins – zinc-finger proteins, for
example. These validate the feasibility of engineering DNA-
binding modules for targeting specific sequences[43,44].
Some zinc finger TFs are capable of regulating endogenous
gene expression, and synthetic zinc finger TFs can be
selected to target specific promoters, without requiring in-
depth knowledge of regulator-binding elements [45]. Fur-
thermore, these artificial zinc-finger TFs can function as
regulators of targeted genes in plant cells [46]; however, the
development of a MTF for plant pathway-regulation
requires additional manipulation of the transcriptional
activity of TFs, and the ability to interact with transcrip-
tional co-factors.
Construction of artificial regulatory circuits
One ingenious strategy to interface a heterologous path-
way with cellular metabolite pools is to subject the
pathway to an artificial regulatory circuit. This approach
is designed to minimize the deleterious effects on cellular
processes caused by the imposition of a non-native
pathway in a cell. Synthetic regulatory circuits have
been studied as a means to understand the fundamental
processes guiding intracellular networks, and for control-
ling cell phenotypes [47]. The periodic production of
increased levels of green florescent protein (GFP),
achieved by transcription from an artificial regulator
composed of the three transcriptional repressors, lacI,
tetR, and cI, shows the possibility of generating new
functions from the regulatory components of unrelated
systems [48]. Control of pathway function through a
designed, non-native regulatory mechanism can effec-
tively couple heterologous gene expression to metabolite
availability by sensing the physiological state of the cell.
In its simplest configuration, components of an artificial
circuit comprise: a signal; a sensor; a regulator of
transcription and its target promoter; gene(s) for rate
Sensor and/or regulator
glnAp2
NRI
Promoter
Figure 3. Design of an artificial regulatory circuit for increased lycopene production from Escherichia coli [49]. The artificial circuit comprises a signal; a sensor; a regulator of
transcription and its target promoter; and gene(s) for rate-limiting steps. The presence of excess glucose, the feedstock for lycopene production, triggers the formation of
acetyl phosphate (AcP). AcP is a signal for the NRI protein, used as both a sensor and regulator. Phosphorylation of NRI results in activation of the glnAp2 promoter to
transcribe the idi and pps genes, which encode the rate-limiting enzymes isopentyl diphosphate and phosphoenolpyruvate synthase, respectively.
www.sciencedirect.com
Vol.24 No.1 January 2006
limiting steps. The design of the circuit requires a
mathematical model of transcriptional regulation in
addition to knowledge of the target metabolic pathway,
and of two-component regulatory systems. An artificial
regulatory circuit was designed, and successfully applied,
for the purpose of improving production of lycopene in
engineered E. coli [49] (Figure 3). The presence of surplus
glucose, the feedstock for lycopene, was signaled by the
formation of acetyl phosphate (AcP), a known signal for
various two-component regulators. A modified Ntr regulon
was used to construct the remainder of the circuit where
the NRI protein of the Ntr system was used as both a
sensor and a regulator. NRI is phosphorylated by AcP, in a
strain lacking the NRII protein, which induces the glnAp2
promoter to transcribe selected genes in the lycopene
pathway. In the engineered pathway shown in Figure 3,
isopentenyl diphosphate isomerase (idi-encoded) and
phosphoenolpyruvate synthase (pps-encoded) control the
flux of substrates to produce lycopene. A lycopene-
producing strain containing artificially-regulated idi and
pps genes exhibited a 50% increase in lycopene titer and a
threefold increase in productivity, compared with a strain
in which the respective genes were induced from a tac
promoter. More importantly, the coupling of heterologous
gene expression with substrate sufficiency avoids growth
retardation due to metabolite imbalance. In addition to
synchronizing product formation with cellular metabolite
levels, the continuous monitoring of product formation
from an artificially regulated pathway could reveal
previously unidentified bottlenecks in the pathway.
The response of a population of cells was coordinated
through the use of a designed circuit integrated with the
quorum-sensing molecule, acyl-homoserine lactone (AHL)
[50]. The circuit programmed a bacterial population to
maintain a lower cell density than that which would be
predicted by nutrient availability. The circuit comprised
two plasmids, each containing the luxI, luxR, and ccdB3
genes. AHL, synthesized by LuxI, activates the LuxR
transcriptional regulator, which in turn, induces
expression of the CcdB3 cell-death activity. A non-induced
Glucose
Excess
Limiting
pathway Signal
enzymes (AcP)
Target genes (idi, pps)
Product
(Lycopene)
TRENDS in Biotechnology
 Review TRENDS in Biotechnology Vol.24 No.1 January 2006 35

culture, not expressing luxI and luxR, reached stationary
phase upon nutrient depletion, whereas the induced
culture maintained a steady state, at a culture density
ten times lower than the induced culture.
The directed evolution of designed circuits is an
additional tier at which pathway properties can be
altered. A designed circuit might be evolved for various
properties, including recognition of multiple or altered
signals, altered pathway specificities, and altered regu-
latory modes, through simultaneous changes in protein–
small molecule, DNA–protein and protein–protein inter-
actions. A functional circuit was derived from an
improperly functioning one using error-prone PCR
(EP-PCR) amplification of specific circuit components
and screening for the desired output [51]. The directed
evolution of designed circuits is analogous to the directed
evolution of synthetic pathways, whereby physiologically
meaningful solutions, which complement rational design,
can be obtained in the absence of fully elucidated
regulatory control mechanisms. Furthermore, the diver-
sity and modular architecture of microbial signaling
systems, as revealed by comparative genomics, presents
an increasing array of sensory and transduction com-
ponents for the experimental manipulation of cellular
behavior through combinatorial techniques [52].
Homologous recombination-based pathway evolution
Single-gene shuffling and whole genome shuffling (WGS)
formats have been applied successfully in the improve-
ment of metabolic pathways. Arsenate resistance, encoded
by the arsRBC operon, was evolved by three iterations of
single-gene shuffling formats applied to the entire operon
[53]. An E. coli strain containing a shuffled operon grew in
arsenate of up to 0.5 M concentration, exhibiting a 40-fold
increase in resistance compared with the wild-type
operon, which was subjected to selection without prior
shuffling. Shuffled variants contained mutations in the
genes encoding the arsenite membrane pump and
arsenate reductase. Whereas the native plasmid remained
episomal, the evolved operon reproducibly integrated into
the bacterial chromosome. It is possible that the coordi-
nated effects of unexpected mutations, which improved
the overall function of the pathway, might not have been
achieved through rational approaches, or through
improvement of individual enzyme properties. Family
shuffling will be useful for rapidly recombining mutations
that are distributed over multiple pathway components
from different variants. Single gene and family shuffling
formats require a contiguous set of structural genes and
regulatory components, and previous synthesis of operons
is needed for cases in which phenotypes are determined
by widely distributed genes and unlinked regulatory
elements.
WGS formats can target phenotypes at the genomic
level in organisms for which well-established genetic tools
are unavailable and limited information exists on genome
sequence and metabolic pathway configuration. The
potential advantage of WGS over classical strain improve-
ment is the ability to access multiple genotypes simul-
taneously, by including a population of strains in the WGS
experiment, and to create complex phenotypes that derive
from distributed sequences across a genome. Three
applications of WGS are described in Table 4.
A key advantage of WGS over other recombination-
based formats for metabolic pathway evolution is the
ability to access dispersed genes determining a specific
phenotype, simultaneously. Additionally, functional infor-
mation on uncharacterized metabolic networks can be
derived from WGS, when performed in conjunction with
transcription and metabolic profiling, high-throughput
DNA sequencing, and phenotype microarray analyses.
Directed WGS is not limited to prokaryotes. Eukaryotic
genomes, including regenerable plant and animal cell
types, can be recombined recursively for accelerated
phenotype improvement; although, it should be noted
that the library sizes generated by the shuffling of entire
genomes are potentially larger than most high-through-
put screening programs can accommodate. A large
fraction of the library will probably contain variants
with non-relevant genotypes with respect to the pheno-
type of interest. To identify functional variants with the
relevant gene-trait correlations, rapidly, it will be necess-
ary to devise very-high-throughput screening methods
(capable of screening libraries of w109–1010 variants),
such as genetic selections, complementation of auxotro-
phies, fluorescence-activated cell sorting (FACS) or

Table 4. Applications of whole genome shuffling (WGS)

Organism
Streptomyces fradiae
Lactobacillus
Sphingobium chlorophenolicum
WGS application Refs
The titer of the polyketide antibiotic tylosin, produced by S. fradiae, was improved by two [71]
iterations of protoplast fusions of mixed populations. Genome-shuffled variants that attained
the titer of current industrial production strains were identified in one year following screening
of w24 000 variants. In comparison, the industrial production strains were derived through
twenty years of strain improvement techniques, and w1 000 000 assays.
Improved lactic acid production and low pH-tolerance by Lactobacilli was achieved by WGS. A [72]
set of parental Lactobacillus strains with subtle improvements in pH-tolerance was initially
obtained by point mutagenesis and selection. This population contains the genetic sequence
diversity for subsequent genome shuffling protocols. Five iterations of protoplast fusions of the
parental population yielded improved variants with threefold higher lactic acid production
compared with the wild-type strains. Additionally, other shuffled variants grew at pH 3.8 – a pH
at which wild-type strains ceased to grow.
WGS was applied to improve tolerance to, and degradation of, the pesticide pentachlorophenol [73]
(PCP) by Sphingobium chlorophenolicum. Three iterations of genome shuffling of S.
chlorophenolicum resulted in variants able to grow in the presence of 6 to 8 mM PCP, whereas
the wild-type strain was unable to grow at PCP concentrations greater than 0.6 mM. Shuffled
variants were capable of complete degradation of 3 mM PCP, whereas significant degradation
was not achieved by the wild-type strain.

www.sciencedirect.com
36 Review TRENDS in Biotechnology Vol.24 No.1 January 2006

Table 5. Metabolic pathway modification by directed evolution of key pathway enzymes

Metabolic pathway
Polyhydroxyalkanoate (PHA)
biosynthesis
Doramectin biosynthesis
Carotenoid biosynthesis
Terpene biosynthesis
Key enzyme(s) and directed evolution formats Refs
Enzymes of the polyhydroxyalkanoate (PHA) biosynthetic pathway have been targeted for [74,75]
novel polymer compositions and improved yield. The enzyme PHA synthase catalyzes the
polymerization of hydroxyacyl-CoA thioesters into PHA polymers. Single gene shuffling of
PHA synthase yielded variants with increased activity towards 3-hydroxybutyryl-CoA, and
increased accumulation of 3-hydroxybutyrate (3HB) and a copolymer of 3HB and 3-
hydroxyhexanoate (3-HB-co-3-HHX). In a separate study, PHA synthase was evolved in vivo
using the E. coli mutator strain XL1-Red, producing variants capable of synthesizing PHAs
of increased molecular mass, and increased yields.
The product of the aveC gene, of unknown catalytic function in the pathway to doramectin, [76]
is known to determine the ratio of avermectin B1 to avermectin B2, the latter being an
unwanted by-product in the pathway. The aveC gene was evolved by semi-synthetic
shuffling, which resulted in a 23-fold improvement in the ratio of B1:B2.
Directed evolution of a carotene desaturase by error prone-PCR (EP-PCR) led to the [77]
formation of an increased number of conjugated double bonds and therefore, to novel
carotenoid structures.
Genes encoding three different monoterpene synthase isozymes with different product [78]
specificities were introduced from lemon into tobacco, individually, and subsequently
combined by crossing, which generated recombination events in vivo between the 3 genes.
The gene-stacking hybrid tobacco produced a pool of monoterpenes with greater structural
complexity than the total monoterpenes produced by the individual transformants.

gene-trait mapping, before developing assays for bio-
chemical activity or the presence of the fermented product.
Directed evolution of individual pathway components
Key enzymatic steps in a metabolic pathway can be
evolved to alter properties of the pathway; the cases
summarized in Table 5 illustrate the successful use of this
strategy. The directed evolution of selected enzymes in a
pathway is a readily executed approach to pathway
evolution for cases in which limiting steps are known,
enzymes are characterized, or potentially promiscuous
functions exist: as in the case of the desaturase and
monoterpene synthase enzymes.
The astonishing capacity of an organism to alter its
DNA in response to environmental signals forms the basis
of the experimental evolution of metabolism. Recognition
that genome alteration is a regulated process, and that
specific molecular systems are activated in response to
environmental stimuli, has led to the identification of
some of the biochemical mechanisms underlying genetic
change. The experimental evolution of metabolism will be
enabled by access to the cellular components that signal
and affect concerted changes in the genome. The challenge
will be to integrate cellular signal transduction systems
with genome restructuring, mutation and repair appar-
atus to evolve specific metabolic functions.

Conclusions
Technologies available for the manipulation of cellular
metabolism range from rational knowledge-intensive
approaches to directed evolution formats that require
minimal information for sequence-to-function corre-
lations. The choice of effective strategies for the
evolution of a specific pathway depends on the extent
to which the target pathway, and genetics and
physiology of the host, are understood, as well as the
ability to devise meaningful genetic selections and
practical high-throughput screens to evaluate the
variant libraries. The experimental evolution of path-
ways will become increasingly efficient through con-
tinuing advances in genome annotation, availability of
genome restructuring tools, expression-profiling tech-
nologies and parallel detection techniques. The ability
to alter metabolism but maintain cellular metabolite
balances will be a key determinant in the application of
modified pathways to a wider range of industrially
relevant functions. It is worth noting that pathway
evolution formats, such as transduction, conjugation,
adaptive mutagenesis and WGS, are regarded as
natural genome-altering mechanisms; therefore, organ-
isms developed through these processes would not
require labeling as ‘genetically modified’ for applications
in agriculture, food and nutrition industries.
www.sciencedirect.com
Acknowledgements
We would like to thank K. Shen for critical reading of the manuscript and
S. Copley for insightful suggestions on evolution of the PCP degradation
pathway. This work is supported in part by a grant (to L.Y.) from the
Kentucky Tobacco Research and Development Center, University
of Kentucky.
References
1 Mills, D.R. et al. (1967) An extracellular Darwinian experiment with a
self-duplicating nucleic acid molecule. Proc. Natl. Acad. Sci. U. S. A.
58, 217–224
2 Chen, K. and Arnold, F.H. (1993) Tuning the activity of an enzyme for
unusual environments: sequential random mutagenesis of subtilisin E
for catalysis in dimethylformamide. Proc. Natl. Acad. Sci. U. S. A. 90,
5618–5622
3 Stemmer, W.P. (1994) DNA shuffling by random fragmentation and
reassembly: in vitro recombination for molecular evolution. Proc. Natl.
Acad. Sci. U. S. A. 91, 10747–10751
4 Schmidt-Dannert, C. (2001) Directed evolution of single proteins,
metabolic pathways, and viruses. Biochemistry 40, 13125–13136
5 Williams, G.J. et al. (2004) Directed evolution of enzymes for
biocatalysis and the life sciences. Cell. Mol. Life Sci. 61, 3034–3046
6 Keseler, I.M. et al. (2005) EcoCyc: a comprehensive database resource
for Escherichia coli. Nucleic Acids Res. 33, D334–D337
7 Strohl, W.R. (2001) Biochemical engineering of natural product
biosynthesis pathways. Metab. Eng. 3, 4–14
8 Koffas, M. et al. (1999) Metabolic engineering. Annu. Rev. Biomed.
Eng. 1, 535–557
9 Capell, T. and Christou, P. (2004) Progress in plant metabolic
engineering. Curr. Opin. Biotechnol. 15, 148–154
 Review TRENDS in Biotechnology
10 Stephanopoulos, G.N. et al. (1998) Metabolic Engineering Principles
and Methodologies, Academic Press
11 Sauer, U. (2004) High-throughput phenomics: experimental methods
for mapping fluxomes. Curr. Opin. Biotechnol. 15, 58–63
12 Bailey, J.E. (1991) Toward a science of metabolic engineering. Science
252, 1668–1675
13 Goddard, J.P. and Reymond, J.L. (2004) Enzyme assays for high-
throughput screening. Curr. Opin. Biotechnol. 15, 314–322
14 Tobin, M.B. et al. (2000) Directed evolution: the ‘rational’ basis for
‘irrational’ design. Curr. Opin. Struct. Biol. 10, 421–427
15 Yuan, L. et al. Laboratory directed protein evolution. Microbiol. Mol.
Biol. Rev. 69, 373–392
16 Arber, W. (1991) Elements in microbial evolution. J. Mol. Evol. 33,
4–12
17 Shapiro, J.A. (1995) Mobile Genetic Elements - Frontiers in Molecular
Biology (Sherratt, D.J., ed.), IRL Press
18 Xie, G. et al. (2003) Ancient origin of the tryptophan operon and the
dynamics of evolutionary change. Microbiol. Mol. Biol. Rev. 67,
303–342
19 van der Meer, J.R. et al. (1992) Molecular mechanisms of genetic
adaptation to xenobiotic compounds. Microbiol. Rev. 56, 677–694
20 Lawrence, J.G. and Roth, J.R. (1996) Evolution of coenzyme B12
synthesis among enteric bacteria: evidence for loss and reacquisition
of a multigene complex. Genetics 142, 11–24
21 Boucher, Y. and Doolittle, W.F. (2000) The role of lateral gene transfer
in the evolution of isoprenoid biosynthesis pathways. Mol. Microbiol.
37, 703–716
22 Alves, R. et al. (2002) Evolution of enzymes in metabolism: a network
perspective. J. Mol. Biol. 320, 751–770
23 Rison, S.C. and Thornton, J.M. (2002) Pathway evolution, structurally
speaking. Curr. Opin. Struct. Biol. 12, 374–382
24 Aharoni, A. et al. (2005) The ‘evolvability’ of promiscuous protein
functions. Nat. Genet. 37, 73–76
25 Dai, M. et al. (2003) A previously unrecognized step in pentachlor-
ophenol degradation in Sphingobium chlorophenolicum is catalyzed
by tetrachlorobenzoquinone reductase (PcpD). J. Bacteriol. 185,
302–310
26 Shapiro, J. (1997) Genome organization, natural genetic engineering
and adaptive mutation. Trends Genet. 13, 98–104
27 Wright, B.E. (2004) Stress-directed adaptive mutations and evolution.
Mol. Microbiol. 52, 643–650
28 Shapiro, J.A. (1984) Observations on the formation of clones contain-
ing araB–lacZ cistron fusions. Mol. Gen. Genet. 194, 79–90
29 Hall, B.G. (2003) The EBG system of E. coli: origin and evolution of a
novel beta-galactosidase for the metabolism of lactose. Genetica 118,
143–156
30 Hall, B.G. (1982) Evolution of a regulated operon in the laboratory.
Genetics 101, 335–344
31 Hall, B.G. (2004) Predicting the evolution of antibiotic resistance
genes. Nat. Rev. Microbiol. 2, 430–435
32 Hall, B.G. (1999) Transposable elements as activators of cryptic genes
in E. coli. Genetica 107, 181–187
33 Goryshin, I.Y. and Reznikoff, W.S. (1998) Tn5 in vitro transposition.
J. Biol. Chem. 273, 7367–7374
34 Salipante, S.J. et al. (2003) GeneHunter, a transposon tool for
identification and isolation of cryptic antibiotic resistance genes.
Antimicrob. Agents Chemother. 47, 3840–3845
35 Klaenhammer, T.R. (1991) Development of bacteriophage-resistant
strains of lactic acid bacteria. Biochem. Soc. Trans. 19, 675–681
36 Broadbent, J.R. and Kondo, J.K. (1991) Genetic construction of
nisin-producing Lactococcus lactis subsp. cremoris and analysis of
a rapid method for conjugation. Appl. Environ. Microbiol. 57,
517–524
37 Babu, M.M. et al. (2004) Structure and evolution of transcriptional
regulatory networks. Curr. Opin. Struct. Biol. 14, 283–291
38 Gonzalez, R. et al. (2003) Gene array-based identification of changes
that contribute to ethanol tolerance in ethanologenic Escherichia coli:
comparison of KO11 (parent) to LY01 (resistant mutant). Biotechnol.
Prog. 19, 612–623
39 Lum, A.M. et al. (2004) Reverse engineering of industrial pharma-
ceutical-producing actinomycete strains using DNA microarrays.
Metab. Eng. 6, 186–196
www.sciencedirect.com
Vol.24 No.1 January 2006 37
40 Tatarko, M. and Romeo, T. (2001) Disruption of a global regulatory
gene to enhance central carbon flux into phenylalanine biosynthesis in
Escherichia coli. Curr. Microbiol. 43, 26–32
41 Morgenstern, B. and Atchley, W.R. (1999) Evolution of bHLH
transcription factors: modular evolution by domain shuffling? Mol.
Biol. Evol. 16, 1654–1663
42 Grotewold, E. (2005) Plant metabolic diversity: a regulatory perspec-
tive. Trends Plant Sci. 10, 57–62
43 Jamieson, A.C. et al. (2003) Drug discovery with engineered zinc-
finger proteins. Nat. Rev. Drug Discov. 2, 361–368
44 Magnenat, L. et al. (2004) In vivo selection of combinatorial libraries
and designed affinity maturation of polydactyl zinc finger transcrip-
tion factors for ICAM-1 provides new insights into gene regulation.
J. Mol. Biol. 341, 635–649
45 Lund, C.V. et al. (2004) Promoter-targeted phage display selections
with preassembled synthetic zinc finger libraries for endogenous gene
regulation. J. Mol. Biol. 340, 599–613
46 Ordiz, M.I. et al. (2002) Regulation of transgene expression in plants
with polydactyl zinc finger transcription factors. Proc. Natl. Acad. Sci.
U. S. A. 99, 13290–13295
47 Gardner, T.S. et al. (2000) Construction of a genetic toggle switch in
Escherichia coli. Nature 403, 339–342
48 Elowitz, M.B. and Leibler, S. (2000) A synthetic oscillatory network of
transcriptional regulators. Nature 403, 335–338
49 Farmer, W.R. and Liao, J.C. (2000) Improving lycopene production in
Escherichia coli by engineering metabolic control. Nat. Biotechnol. 18,
533–537
50 You, L. et al. (2004) Programmed population control by cell–cell
communication and regulated killing. Nature 428, 868–871
51 Yokobayashi, Y. et al. (2002) Directed evolution of a genetic circuit.
Proc. Natl. Acad. Sci. U. S. A. 99, 16587–16591
52 Galperin, M.Y. (2004) Bacterial signal transduction network in a
genomic perspective. Environ. Microbiol. 6, 552–567
53 Crameri, A. et al. (1997) Molecular evolution of an arsenate
detoxification pathway by DNA shuffling. Nat. Biotechnol. 15,
436–438
54 Crameri, A. et al. (1998) DNA shuffling of a family of genes from
diverse species accelerates directed evolution. Nature 391, 288–291
55 Ness, J.E. et al. (2002) Synthetic shuffling expands functional protein
diversity by allowing amino acids to recombine independently. Nat.
Biotechnol. 20, 1251–1255
56 Zhao, H. et al. (1998) Molecular evolution by staggered extension
process (StEP) in vitro recombination. Nat. Biotechnol. 16, 258–261
57 DeSantis, G. et al. (2003) Creation of a productive, highly enantiose-
lective nitrilase through gene site saturation mutagenesis (GSSM).
J. Am. Chem. Soc. 125, 11476–11477
58 Coco, W.M. et al. (2001) DNA shuffling method for generating highly
recombined genes and evolved enzymes. Nat. Biotechnol. 19, 354–359
59 Abecassis, V. et al. (2000) High efficiency family shuffling based on
multi-step PCR and in vivo DNA recombination in yeast: statistical
and functional analysis of a combinatorial library between human
cytochrome P450 1A1 and 1A2. Nucleic Acids Res. 28, e88
60 Zha, D. et al. (2003) Assembly of designed oligonucleotides as an
efficient method for gene recombination: a new tool in directed
evolution. ChemBioChem 4, 34–39
61 Song, J.K. et al. (2002) Construction of DNA-shuffled and incremen-
tally truncated libraries by a mutagenic and unidirectional reassem-
bly method: changing from a substrate specificity of phospholipase to
that of lipase. Appl. Environ. Microbiol. 68, 6146–6151
62 Kitamura, K. et al. (2002) Construction of block-shuffled libraries of
DNA for evolutionary protein engineering: Y-ligation-based block
shuffling. Protein Eng. 15, 843–853
63 Petri, R. and Schmidt-Dannert, C. (2004) Dealing with complexity:
evolutionary engineering and genome shuffling. Curr. Opin. Biotech-
nol. 15, 298–304
64 Ostermeier, M. et al. (1999) A combinational approach to hybrid
enzymes independent of DNA homology. Nat Biotechnol. 17,
1205–1209
65 Sieber, V. et al. (2001) Libraries of hybrid proteins from distantly
related sequences. Nat Biotechnol. 19, 456–460
66 Kolkman, J.A. and Stemmer, W.P. (2001) Directed evolution of
proteins by exon shuffling. Nat. Biotechnol. 19, 423–428
38 Review TRENDS in Biotechnology Vol.24 No.1 January 2006

67 Bittker, J.A. et al. (2004) Directed evolution of protein enzymes using
nonhomologous random recombination. Proc. Natl. Acad. Sci. U. S. A.
101, 7011–7016
68 O’Maille, P.E. et al. (2004) Gene library synthesis by structure-based
combinatorial protein engineering. Methods Enzymol. 388, 75–91
69 Kuhlman, B. et al. (2003) Design of a novel globular protein fold with
atomic-level accuracy. Science 302, 1364–1368
70 Broothaerts, W. et al. (2005) Gene transfer to plants by diverse species
of bacteria. Nature 433, 629–633
71 Zhang, Y.X. et al. (2002) Genome shuffling leads to rapid phenotypic
improvement in bacteria. Nature 415, 644–646
72 Patnaik, R. et al. (2002) Genome shuffling of Lactobacillus for
improved acid tolerance. Nat. Biotechnol. 20, 707–712
73 Dai, M. and Copley, S.D. (2004) Genome shuffling improves
degradation of the anthropogenic pesticide pentachlorophenol by
Sphingobium chlorophenolicum ATCC 39723. Appl. Environ. Micro-
biol. 70, 2391–2397
74 Kichise, T. et al. (2002) Enhanced accumulation and changed
monomer composition in polyhydroxyalkanoate (PHA) copolyester
by in vitro evolution of Aeromonas caviae PHA synthase. Appl.
Environ. Microbiol. 68, 2411–2419
75 Amara, A.A. et al. (2002) In vivo evolution of the Aeromonas punctata
polyhydroxyalkanoate (PHA) synthase: isolation and characterization
of modified PHA synthases with enhanced activity. Appl. Microbiol.
Biotechnol. 59, 477–482
76 Stutzman-Engwall, K. et al. (2005) Semi-synthetic DNA shuffling of
aveC leads to improved industrial scale production of doramectin by
Streptomyces avermitilis. Metab. Eng. 7, 27–37
77 Umeno, D. et al. (2005) Diversifying carotenoid biosynthetic
pathways by directed evolution. Microbiol. Mol. Biol. Rev. 69,
51–57
78 Lucker, J. et al. (2004) Metabolic engineering of monoterpene
biosynthesis: two-step production of (C)-trans-isopiperitenol by
tobacco. Plant J. 39, 135–14

Elsevier celebrates two anniversaries with

gift to university libraries in the developing world
In 1580, the Elzevir family began their printing and bookselling business in the Netherlands, publishing works by scholars such as John
Locke, Galileo Galilei and Hugo Grotius. On 4 March 1880, Jacobus George Robbers founded the modern Elsevier company intending,
just like the original Elzevir family, to reproduce fine editions of literary classics for the edification of others who shared his passion, other
’Elzevirians’. Robbers co-opted the Elzevir family’s old printer’s mark, visually stamping the new Elsevier products with a classic old
symbol of the symbiotic relationship between publisher and scholar. Elsevier has since become a leader in the dissemination of
scientific, technical and medical (STM) information, building a reputation for excellence in publishing, new product innovation and
commitment to its STM communities.
In celebration of the House of Elzevir’s 425th anniversary and the 125th anniversary of the modern Elsevier company, Elsevier will donate
books to 10 university libraries in the developing world. Entitled ‘A Book in Your Name’, each of the 6 700 Elsevier employees worldwide
has been invited to select one of the chosen libraries to receive a book donated by Elsevier. The core gift collection contains the
company’s most important and widely used STM publications including Gray’s Anatomy, Dorland’s Illustrated Medical Dictionary,
Essential Medical Physiology, Cecil Essentials of Medicine, Mosby’s Medical, Nursing and Allied Health Dictionary, The Vaccine Book,
Fundamentals of Neuroscience, and Myles Textbook for Midwives.
The 10 beneficiary libraries are located in Africa, South America and Asia. They include the Library of the Sciences of the University of
Sierra Leone; the library of the Muhimbili University College of Health Sciences of the University of Dar es Salaam, Tanzania; the library
of the College of Medicine of the University of Malawi; and the libraries of the University of Zambia, Universite du Mali, Universidade
Eduardo Mondlane, Mozambique; Makerere University, Uganda; Universidad San Francisco de Quito, Ecuador; Universidad Francisco
Marroquin, Guatemala; and the National Centre for Scientific and Technological Information (NACESTI), Vietnam.
Through ‘A Book in Your Name’, the 10 libraries will receive approximately 700 books at a retail value of approximately 1 million US
dollars.
For more information, visit www.elsevier.com

www.sciencedirect.com