[8] SCOPE 75

the amounts of dimer templates, making it possible to design different

types of combinatorial libraries, such as random shuffling libraries and
artificial alternative splicing libraries. The block shuffling and random
point mutations can be introduced simultaneously by performing RM-
PCR under error-prone conditions. Most of the block sequences in the
combinatorial libraries created by RM-PCR encode a long open reading
frame and are suitable for protein selection experiments. Thus, RM-PCR
promises to be a powerful approach in creating novel proteins, especially
simulating early stages of protein evolution on a rapid timescale using an
efficient and easily adaptable experimental protocol.

[8] Gene Library Synthesis by Structure-Based

Combinatorial Protein Engineering
By Paul E. O’Maille, Ming-Daw Tsai, Bryan T. Greenhagen,
Joseph Chappell, and Joseph P. Noel

Introduction
Structure-based combinatorial protein engineering (SCOPE) is a pro-
cess for the synthesis of gene libraries that lay the genetic foundation for
the exploration of the relationship between structure and function in the
encoded proteins.1 The comparative analysis both structurally and func-
tionally of protein primary, secondary, and tertiary structure generates
numerous hypotheses with which to probe the relationship between molec-
ular structure and the ensuing functional readout. SCOPE provides a tool
for constructing the gene libraries that encode rationally engineered pro-
tein variants that provide the raw material for addressing these hypothesis
using both structural and functional analyses. Mechanistic hypotheses gen-
erated from structures derived from both experiment (crystallographic and
nuclear magnetic resonance) and homology modeling is used to design
oligonucleotides that code for crossovers between genes encoding structur-
ally related proteins. A series of polymerase chain reactions (PCR), culmi-
nating in the selective amplification of crossover products, incorporates
spatial information encoded in the oligonucleotide into a full-length gene
and the resultant hybrid protein. Iteration of the process enables the
synthesis of all possible combinations of desired crossovers, producing a
hierarchical collection of chimeras in analogy to a Mendelian population.

1
P. E. O’Maille, M. Bakhtina, and M. Tsai, J. Mol. Biol. 321, 677 (2002).

Copyright 2004, Elsevier Inc.

All rights reserved.
METHODS IN ENZYMOLOGY, VOL. 388 0076-6879/04 $35.00
76 methodology [8]

The principles of the process are generally applicable and the method-
ology is easily adapted to a range of experimental objectives. At its incep-
tion, SCOPE was developed to provide a means of generating multiple
crossover gene libraries from distantly related proteins, constituting a
homology-independent in vitro recombination approach.1 This article pre-
sents the adaptation of SCOPE to the facile combinatorial synthesis of
mutant gene libraries. The newly incorporated refinements to the originally
designed SCOPE approach illustrate the underlying principles of the ex-
perimental process that make it a robust technique for the parallel explo-
ration of protein sequence space in three dimensions. This tertiary
information is embodied within the mechanistic and evolutionary under-
pinnings of protein structure and function, both of which are fundamental
aspects of biochemical adaptive change in organisms. In addition, this
information can be exploited by SCOPE for a myriad of applications in
biotechnology.

Principles
The construction of gene libraries by SCOPE involves a series of PCRs.
Other recombination techniques use multiple primers or random fragments
in a single step, thus carrying out multiple reactions in parallel.2,3 Separation
of gene synthesis into discrete steps is an essential feature of SCOPE. This
simple but critical property of SCOPE enables one to control recombination
through pairing gene fragments and genes that give rise to designed and
anticipated combinations of crossovers. As a consequence, libraries are
constructed as a series of less complex mixtures, which reduce numerical
complexity and the cost and extent of sampling required during screening,
including gene sequencing and functional assays. Crossover locations and
the frequency of genetically encoded crossovers are established by experi-
mental design and are not dictated or constrained by homology between
genes or the linear distance between multiple mutations.
An overview of the process illustrates the basic steps encompassing
SCOPE-based recombination (Fig. 1). In step I, standard PCR amplifica-
tion, using an internal and external primer pair and the appropriate tem-
plate DNA, produces chimeric gene fragments. Internal primers are
designed on the basis of one or more encoded three-dimensional structures
viewed with reference to the variable sequence space of protein homo-
logues and code for crossovers in the protein-coding region of genes.
External primers correspond to the 50 and 30 termini of a given gene, as

2
W. P. Stemmer, Nature 370, 389 (1994).
3
F. J. Perlak, Nucleic Acids Res. 18, 7457 (1990).
[8] SCOPE 77

Fig. 1. Overview of the SCOPE library synthesis process.

with any typical primer pair used in a standard PCR amplification. The
template consists of a plasmid or PCR product that contains the gene of
interest. In step II, in vitro recombination occurs between a gene frag-
ment(s) and a new template; in other words, gene fragments serve as a new
set of primers, which anneal and are extended to produce single-stranded
full-length chimeras. In step III, a new external primer set directs the
selective amplification of recombination products by virtue of the unique
genetic identity encoded at their termini. Repetition of steps II and III
using various pairs of gene fragments from step I and crossover products
from step III allows the production of genetically diverse, multiple cross-
over libraries in high yield.
Careful oligonucleotide design is central to the SCOPE recombination
process. A discussion of the properties of the synthetic oligonucleotides, in
relationship to specific applications, illustrates how SCOPE can be adapted
to construct multiple crossover libraries from distantly related proteins or
combinatorial mutant libraries from mechanistically related proteins.

Internal Primers
Shuffling exons or ‘‘equivalent’’ structural elements between homo-
logues require chimeric oligonucleotides. These are composed of approxi-
mately equal halves of two distinct genes and code for a crossover region.
An example of their use is illustrated in the process overview (Fig. 1).
78 methodology [8]

Without a priori knowledge of the optimal point of fusion in regions of

low identity or the compatibility of ‘‘equivalent’’ structural elements of low
sequence identity, linkage variability can be introduced. This entails de-
signing a set of chimeric oligonucleotides (for a given crossover region),
which code for a series of insertions and/or deletions around a fixed point
to explore alternative crossovers. The corresponding collection of gene
fragments can be used together in recombination reactions (step II). Vari-
able connections between ‘‘equivalent’’ structural elements proved to be
an essential aspect of design in producing functional hybrids from distantly
related DNA polymerases.1
Combinatorial mutagenesis by SCOPE requires mutagenic oligonucleo-
tides that code for either a specific or a random set of mutations, insertions,
or deletions directed at a given site in the gene of interest. Alternatively,
bridging oligonucleotides, which code for stretches of native sequence
between mutations, can be used to mediate recombination between wild-
type and/or mutant genes. Mutagenic and bridging oligonucleotides are
employed in PCR in a similar manner as chimeric oligonucleotides, although
refinements to the process were required for their efficient incorporation
into the desired chimeric sequence as described later.

External Primers
Polymerase chain reaction amplification of mutant or chimeric genes
(step III, Fig. 1) is the final step of the SCOPE cycle. Like any conventional
amplification, a primer set that flanks the target gene of interest is required
during this final amplification step. Additionally, the inclusion of restriction
or recombination sites into the final primer set for the efficient cloning of
the resultant collection of genes is often desirable. However, a fundamental
aspect of SCOPE is that the ‘‘proper’’ primer set be used for the selective
amplification of a particular crossover product from a recombination reac-
tion, which may contain a mixture of products. In the chimeragenesis of
distantly related proteins, the termini of each gene are unique and can be
exploited in this way for selective amplification.
SCOPE, as applied to the combinatorial synthesis of mutant libraries,
where the termini of wild-type genes and crossover products are indistin-
guishable, required the design of alternative external primers and a book-
keeping system for their successful implementation and hierarchical
organization and storage. Primary amplification primers (PAPs) code for
DNA sequences flanking the gene (like any generic external primer), but
contain an additional and unique 50 sequence tag. Their use in gene frag-
ment synthesis (step I) links a unique sequence to a particular mutation.
Following recombination, secondary amplification primers (SAPs), which
[8] SCOPE 79

Fig. 2. Refinements to SCOPE. (A) Source of wild-type background. (B) Alternative

fragment amplification strategy for the suppression of the wild-type sequence.

correspond to the 50 unique sequence, are employed in the final amplifica-

tion (step III) to select for the desired recombination products.
Refinements to the original SCOPE process were necessary for the
efficient incorporation of unique sequence tags, their linkage to mutations,
and the suppression of wild-type background genes. During step I amplifi-
cation, single-stranded DNA or ‘‘long’’ product is produced from the
extension of each primer on the plasmid template. When derived from
PAPs, these single-stranded products code for the wild-type gene (Fig. 2A)
and, if carried over into other steps of the process, give rise to a small but
significant population of wild-type genes in the background. Separating
step I into two reactions alleviates this contamination problem (Fig. 2B).
In step IA, the internal primer and template are mixed and single-stranded
DNA containing the mutation(s) is synthesized. The product of step IA is
treated with DpnI to digest the wild-type plasmid template, leaving only
the nascent, single-stranded, mutagenic DNA. This restriction step elim-
inates the formation of long products that contribute to a wild-type back-
ground. A portion of step IA product is then used in step IB, where it
serves as a template for conventional PCR amplification with an internal
primer and a PAP. In the original development of SCOPE, gel purification
of gene fragments (step I) was an essential step of the process. The
aforementioned refinements enable the entire sequence of PCR reactions
80 methodology [8]

Scheme 1. Product specificity of two closely related terpene cyclases: henbane

premnaspirodiene synthase (HPS) and tobacco 5-epi-aristolochene synthase (TEAS).

(steps I through III) to be conducted without purifying intermediates until

the final amplification step.
An illustration of the SCOPE methodology as applied to the combina-
torial synthesis of mutant gene libraries of the terpene cyclase enzyme
known as tobacco 5-epi-aristolochene synthase (TEAS) is presented as a
proof of principle example of the power and ease of this newly revised
version of SCOPE. Previously, the product specificity of TEAS was con-
verted from 5-epi-aristolochene to premnaspirodiene, the product of the
closely related terpene cyclase premnaspirodiene synthase (HPS) from
henbane, by nine sequential mutations.4 Site-directed mutants were de-
signed using the three-dimensional structure of TEAS5 and homology
modeling of HPS. The products of these cyclases are shown in Scheme 1.
Because these mutations were made sequentially, the question remains, are
all nine mutations required for product selectivity? What combinations, if
any, of these nine mutations are sufficient for this property change? More-
over, in a broader evolutionary sense, what is the mechanistic and energetic
landscape that links such specificity switches that are encoded by multiple
changes at sites throughout a given protein? Constructing all combinations
of nine mutations (2n combinations, where n is the number of mutations)
and analyzing their product specificity and kinetic properties will provide
answers to some of these questions.
4
B. T. Greenhagen, Dissertation thesis. Department of Agriculture at the University of
Kentucky, 2003.
5
C. M. Starks et al., Science 277, 1815 (1997).
[8] SCOPE 81

Terpene cyclases are an ideal proof of the principal system for explor-
ing the utility of SCOPE given their (1) unusual mechanism employing the
conformationally directed production of reactive carbocation intermedi-
ates, (2) well-defined three dimensional structures, (3) ease of product
identification and quantification using high-throughput GC-MS analysis,
(4) evolutionarily diverse distribution of protein sequences and small mol-
ecule products across multiple kingdoms, and (5) biotechnological poten-
tial for the biosynthesis of unique small molecules representing a currently
untapped region of natural product space.

Experimental Procedures

Materials
PCR components: 10 cloned pfu reaction buffer and pfu turbo DNA
polymerase (Stratagene, La Jolla, CA), dNTPs (Invitrogen, Carlsbad, CA),
and bovine serum albumin (BSA; New England Biolabs, Beverly, MA).
PCR reactions are carried out using a PTC 200 Peltier thermal cycler (MJ
Research, Waltham, MA). All PCR products are purified by gel extraction
(Qiagen, Valencia, CA), cloned into pDONR 207 using Gateway cloning
technology (Invitrogen) according to the manufacturer’s recommended
conditions. Plasmid DNA from gentamicin-resistant transformants is mini-
preped by the Salk Institute Microarray facility for sequencing at the Salk
Institute DNA sequencing/quantitative PCR facility. The cDNA of TEAS
is cloned into pH8GW (an in-house gateway destination vector) and this
plasmid DNA is used as a template for PCR.
All PCR are carried out using a master mix of a standard set of PCR
components for a 50-
l scale reaction:
5
l of 10 cloned pfu reaction buffer to give 1
1
l of pfu turbo DNA polymerase (Stratagene) (2.5 U/
l) to give
0.05 U/
l
0.5
l of BSA (10 mg/ml) to give 0.1 mg/ml
8
l of dNTP mix (1.25 mM) to give 200
M each dNTP

Primers
Oligonucleotides are from Integrated DNA Technologies (IDT) and
are listed in Table I. For both mutagenic and chimeric primers, the muta-
tion(s) or crossover point(s) is located in the center of the oligonucleotide,
such that the flanking sequence is complementary to a given gene; ideally,
this should be 18 to 24 nucleotides (or have a Tm greater than or equal

to 50 ) for effective PCR. SAPs are designed to consist of 21 nucleotides
82 methodology [8]

TABLE I
Oligonucleotides Used for SCOPE Combinatorial Mutagenesis

a
Bold and underlined characters indicate the sites of designed mutations. Shading is used
to indicate part of the attB1 and attB2 recombination sequences; complete attB sites are
generated by amplification of the target gene(s) from the destination vector pH8GW.
b
Mutageneic primers are named according to the amino acid substitutions they code for.
[8] SCOPE 83

and have a Tm greater than or equal to 55 . PAPs contain 24 bases (in
addition to their unique sequence), which correspond to partial Gateway
attB sites; the remaining attB sequence becomes incorporated into PCR
products by amplification from pH8GW. Tm values are calculated based on
nearest-neighbor thermodynamic parameters.6

Gel Electrophoresis
Analysis of PCR fragments and separation of products for gel purifica-
tion are performed using 2% (w/v) agarose gels in 1 TAE buffer contain-
ing 0.1
g/ml ethidium bromide. Concentrations of PCR products (steps IB
and III) are estimated by comparison to a standard of known concentra-
tion, such as the low DNA mass ladder (Invitrogen) using densitometry
software such as ImageJ (http://rsb.info.nih.gov/ij/).

Method
Prior to library construction, all primers are tested to ensure that they
result in unique amplification products of the expected size. Like any
standard PCR amplification, optimization of cycling parameters for specific
template and primer sets may be necessary.

Step IA: Mutagenic/Chimeric Single-Stranded DNA Synthesis

Procedure. Reactions are mixed on ice using the following:

14.5
l of PCR master mix (as defined earlier)
1
l internal primer (5
M stock) to give 0.1
M
1
l plasmid DNA template (10 nM stock) to give 200 pM
33.5
l filter-sterilized H2O added to give a 50-
l reaction volume
The master mix is added last and the resultant reaction is mixed by
 
pipetting. Cycling program: 96 for 5 min, followed by 50 cycles of 96 for
 
30 s, 55 for 30 s, and 72 for 1 min/kb of product followed by incubation at

4 at the completion of cycling.
Comments. The amount of single-stranded product formed is limited
by the amount of template DNA and the number of cycles performed.

6
H. T. Allawi and J. Santa Lucia, Jr., Biochemistry 36, 10581 (1997).

c
Primary amplifications primers are named according to their unique sequence tag
(A through F for forward and 1 through 6 for reverse as listed in Fig. 4) and gateway
recombination sequence (b1 for attB1 and b2 for attB2).
d
Unique sequence tags are labeled according to their corresponding primary amplification
primer (A through F for forward and 1 through 6 for reverse as listed in Fig. 4).
84 methodology [8]

Estimated yields for the aforementioned reaction (using 50 cycles and

200 pM plasmid) are about 10 fmol of the final single-stranded product;
this amount is well in excess of what is required for subsequent amplification
reactions. A 0.1
M concentration of internal primer (>103 molar excess
of plasmid template) is sufficient; higher primer concentrations may pro-
mote the formation of alternative products in subsequent amplification
steps (step IB).

DpnI Digestion of Plasmid DNA

Procedure. The addition of 1
l of DpnI (20 U/
l, New England Bio-

labs) with mixing is followed by incubation at 37 for 1 h for digestion of

the original DNA template and 20 min at 80 for heat inactivation.

Step IB: Mutagenic/Chimeric Double-Stranded DNA

Fragment Amplification
Procedure. Reactions are mixed on ice using the following:
14.5
l of PCR master mix (as defined earlier)
2
l internal primer (5
M stock) to give 0.2
M
1
l primary amplification primer (5
M stock) to give 0.1
M
1
l of step IA reaction as template to give 1–10 pM single-stranded
DNA
31.5
l filter-sterilized H2O added to give a 50-
l reaction volume
The master mix is added last with pipetting to mix reactions. Cycling
  
program: 96 for 5 min, followed by 40 cycles of 96 for 30 s, 55 for 30 s, and
 
72 for 1 min/kb of product followed by incubation at 4 at the completion of
cycling. Amplification products are verified by agarose gel electrophoresis.
Comments. Internal primers should be in excess of external primers;
keeping the concentration of external primers below saturation and in-
creasing the number of cycles ensure their depletion. This is sufficient to
suppress wild-type background arising from ‘‘long’’ products generated
during subsequent amplification steps from carryover of the external prim-
er. The step IA product can be diluted up to 10,000-fold and still provide
enough template for robust amplification.

Step II: Recombination

Single Mutants/Crossovers
Procedure. Reactions are mixed on ice using the following:
5.8
l of PCR master mix (as defined earlier)
1
l of step IB reaction to give 10 nM (or 1–5 ng/
l) gene fragment
[8] SCOPE 85

1
l plasmid DNA template (10 nM stock) to give 200 pM final

(1 ng/
l for a 7-kb plasmid)
12.2
l filter-sterilized H2O added to give a 20-
l reaction volume
The master mix is added last with pipetting to mix reactions. Cycling
program: 96 for 5 min, followed by 15 cycles of 96 for 30 s (þ200 /cycle),
  
55 for 30 s, and 72 for 1 min/kb of product followed by incubation at 4 at
the completion of cycling.

Multiple Mutants/Crossovers
Procedure. Same as just described, except the gel-purified full-length
mutant/chimeric gene (step III product) at 1.0 ng/
l (1 nM final
concentration) is substituted for plasmid DNA.

Multiplex Recombination
Procedure. A mixture of gene fragments (step IB products) corres-
ponding to a collection of mutations or alternative crossovers is pooled, and
1
l (to give 10 nM) is used with either a plasmid or a full-length mutant/
chimeric gene (step III product) as the template in a recombination reaction.
Comments. The amount of full-length, single-stranded recombination
product produced in step II is limited by the amount of gene fragment from
step IB added to the reaction mixture. Gene fragments should ideally be
1- to 10-fold molar excess of the plasmid or mutant gene that it is recom-
bining with. This is particularly important in the case of single mutants/
crossovers, where only one terminus can be exploited in the following step
for selective amplification. The plasmid concentration should be kept to a
minimum; about 10 pM is the lowest concentration that can be used to give
the amplifiable recombination product in step III.

Step III: Selective Amplification of Recombination Products

Amplification of Single Mutants/Crossovers

Procedure. Reactions are mixed on ice using the following:
14.5
l of PCR master mix (as defined earlier)
2
l secondary amplification primer (5
M stock) to give 0.2
M
2
l primary amplification primer (5
M stock) to give 0.2
M
1
l of step II reaction as template to give 100–200 pM single-
stranded DNA
30.5
l filter-sterilized H2O added to give a 50-
l reaction volume
The master mix is added last with pipetting to mix reactions. Cycling
  
program: 96 for 5 min, followed by 30 cycles of 96 for 30 s, 55 for 30 s,
86 methodology [8]

 
and 72 for 1 min/kb of product followed by an additional 10 min at 72 and

incubation at 4 at the completion of cycling. Amplification products are
verified by agarose gel electrophoresis.

Amplification of Multiple Mutants/Crossovers

Procedure. Same as amplification of single mutants/crossovers, but only

secondary amplification primers are used.
Comments. The final step in a cycle of SCOPE is a standard PCR
amplification of full-length mutant/chimeric genes with unique sequence
tags at both 50 and 30 ends. In synthesis of the first generation of mutants,
only one SAP can be used for selective amplification. This corresponds to
the unique sequence of the PAP used in step IB. A PAP is directed at the
opposite terminus, where it incorporates unique sequence at this terminus.
Because this primer is directed to the flanking sequence of the gene, it can
efficiently prime any carryover long product (single-stranded wild-type
DNA) from step IB or any plasmid from step II. Therefore, it is important
to eliminate the long product from step IB and minimize the amount of
plasmid in step II, as the single-stranded product generated at this step has
the potential to carry over into subsequent rounds of synthesis. In the step
III amplification of multiple mutants/crossovers, SAP combinations are
chosen to allow selective amplification of desired recombination products.

Product Isolation and Cloning

Full-length mutant genes from step III are gel purified using the Qiagen
gel extraction kit according to the manufacturer’s recommended proce-
dures. Gel-purified attB PCR products are cloned into pDONR207 via the
gateway BP reaction according to the manufacturer’s recommendations.

Controls

DpnI Treatment. Digestion is omitted and the step IB reaction is

performed using the undigested step IA product as the template. Because
plasmid DNA is carried over into the step IB reaction, PAPs can be
extended to produce wild-type single-stranded DNA as described previ-
ously (Fig. 2A). As a result, wild-type genes can be amplified efficiently
using a 1-
l portion of step IB as the template and a PAP and SAP primer
pair. If the step IB reaction is performed using a 10-fold molar excess of
mutagenic primer, then the amount of amplifiable wild-type gene decreases
markedly. In fact, the combination of increasing the number of cycles in
step IA to 100 (resulting in 2-fold more template) and using a 10-fold
excess of internal mutagenic primer in step IB enables the suppression of
[8] SCOPE 87

wild-type background and a mutagenesis efficiency of 80%, as apparent

from terpene cyclase libraries produced in this manner.
Step IB Products. The selectivity of amplification (or the suppression of
wild-type sequences) is evaluated using the step IB reaction as the tem-
plate. If DpnI digestion is complete, no amplifiable wild-type product is
observed. In the case where restriction digestion is omitted, wild-type
product is observed.

Results and Discussion

SCOPE was applied to create a library representing all possible combi-
nations of nine point mutations in the terpene cyclase, TEAS. The location
of mutations in the amino acid and nucleotide sequences of TEAS are
indicated in Fig. 3. The nine positions were recombined as six units (shown
in boxes). Some mutations were clustered, requiring a plurality of internal
primers. For example, amino acid positions 436, 438, and 439 required a
collection of seven internal primers to code for all permutations: three
single, three double, and a triple mutant. A system was developed to
introduce unique sequence tags; PAPs link mutation (or a collection there-
of) to a unique sequence during gene fragment amplification and SAPs
enable selective amplification of the desired combinations of mutations.
An illustration of their use and the nomenclature system is described in
Fig. 4.
An attribute of SCOPE synthesis is the fractionation of complex mix-
tures into many simpler ones. This has the benefit of reducing the numerical
complexity and hence the screening requirements necessary to verify and
identify the collection of desired changes. This effect arises from sampling
probability as described by the following mathematical expression:
X  
n 1
n! ðn iÞ k
pðnÞ ¼ 1 ð 1Þiþ1 (1)
i¼1
i!ðn iÞ! n

Fig. 3. Location of mutations in TEAS recombined by SCOPE. The nucleotide positions

and corresponding amino acid changes are shown above and below, respectively
88 methodology [8]

Fig. 4. Recombination units and tagging system. The recombined positions and their
associated unique sequence tags give rise to a naming system to describe the recombination
product created by SCOPE.

where k is the sample size, n is the number of unique members, and p is the
probability that a sample of size k contains at least one representative of
each unique member. As complexity increases, the amount of oversampling
required to achieve the same probability of screening the library increases,
where oversampling refers to sample size (k) in multiples of library
complexity (n). This can be shown graphically in Fig. 5.
Each iteration of the process ends with a conventional PCR amplifica-
tion step, and after multiple iterations, additional mutations accumulate.
The overall frequency of nondesired additional mutations in the popula-
tion analyzed is 5.5%. No strong bias for the type of error or its location
within the gene was observed. The nondesired mutation rate after the first
round was 2.67%, which matches previous measures of pfu error fre-
quency.7 However, the random mutation rate increases as a function of

7
J. Cline, J. C. Braman, and H. H. Hogrefe, Nucleic Acids Res. 24, 3546 (1996).
[8] SCOPE 89

Fig. 5. Sampling probability as a function of oversampling. The probability that a sample

contains one copy of each unique clone for a given complexity (n) is calculated using Eq. (1).
Probability is calculated for a range of sample sizes (k) that are in multiples of a fixed library
complexity (n) and the results are fit to a sigmoidal curve.

iterations of SCOPE and, after four iterations, reached 8.9%. Using a

higher fidelity polymerase can minimize random mutation rates. Alterna-
tively, products from step III amplification reactions can be cloned and the
SCOPE cycle started anew (from step IA). Bridging oligonucleotides may
be useful in this case to recombine various mutations. Also, gene fragments
(from step IB) can be made to include multiple mutations from previous
cycles.
In the example used here, 512 mutants were made from a series of
simpler mixtures, the most complex of which contained 21 unique mem-
bers. To achieve a 50% probability ( p ¼ 0.5) of identifying by screening
every unique member in a mixture of that complexity requires 3.38-fold
oversampling. To achieve the same probability of identifying all unique
members of a library by screening a mixture of 512 unique possibilities
requires 6.6-fold oversampling. Given the exponential relationship be-
tween sample size and library complexity, this difference equates to a
reduction in numerical complexity of a factor of 25 for the entire library.

Library Analysis
Over 600 colonies from discrete mixtures, representing about half of the
complexity of the TEAS library (241 unique members), were picked and
their sequences determined. A summary of the results is listed in Table II.
Of the clones sequenced, only 24 wild-type genes (3.5%) were found. This
library was synthesized prior to addition of the DpnI restriction step (as
described earlier), and while the efficiency of the first round of mutagenesis
was 80%, the overall efficiency of the entire process reached 96.5%
90 methodology [8]

TABLE II
Sequence Analysis Results

Library statistics

Clones sequenced 692

Wild-type genes 24
Percentage of mutants 96.5%
Complexity screened 241
Unique clones identified 193
Fold oversampled 2.8
Complexity covered 80.1%
Total library complexity 512
Percentage of verified mutants 37.70%
Additional mutations
Silent 9
Frameshift 16
Point mutants 13
Total 38
Mutation rate 5.49%

(Table II). Mutations become incorporated into the wild-type sequence

during recombination reactions in subsequent iterations of the process. As
a result, wild-type sequences vanish in multiple crossover populations.
Aside from the low-level appearance of wild-type sequence and ran-
dom mutations likely arising from PCR errors, the actual distribution of
mutations in a given mixture was as designed experimentally. Some recom-
bination reactions produce a single product having several designed muta-
tions, such as A1236. In reactions containing multiple mutations, the
reaction distribution appears random.

Concluding Remarks
Adaptation of SCOPE to combinatorial mutant library design and
construction demonstrates the broader utility of these library construction
principles. While various techniques have been developed for either ho-
mology-independent recombination or combinatorial mutagenesis, none
can efficiently do both. SCOPE provides an effective means for the crea-
tion of both global or local sequence space as demonstrated by the synthe-
sis of DNA libraries representing the genetically encoded information
spanned by distant homologues1 or closely related members of a gene
family.
[9] oligonucleotide library modules 91

Acknowledgments
We are grateful to the National Institutes of Health for the grants that supported this
work (GM43268 to M.D.T. and GM54029 to J.C. and J.P.N.). P.E.O. is an NIH Postdoctoral
Research Fellow (GM069056-01). Additionally, we thank Marina Bakhtina and Brandon
Lamarch for valuble consultations during the early phases of this work.

[9] New Enzymes from Combinatorial Library Modules

By Werner Besenmatter, Peter Kast, and Donald Hilvert

Introduction
Current strategies for the construction of combinatorial gene libraries
for directed evolution experiments generally make use of cassette muta-
genesis1,2 to insert library modules3–6 into plasmids. We have applied this
technique in a variety of formats to investigate chorismate mutase, a key
enzyme in the biosynthesis of aromatic amino acids.7 Active variants are
directly selected from gene libraries transformed into a chorismate mutase-
deficient Escherichia coli strain (Fig. 1).8 Because catalytic activity is an
extremely sensitive probe for protein integrity, a wealth of information on
structural and functional aspects of this enzyme can be derived from
sequence patterns in selected variants.9
The extent of randomization of the gene library cassettes depends
on the questions asked. For instance, to investigate the roles of indi-
vidual active site residues, one or two codons were randomized at
a time.8,10 When loops connecting secondary structural elements were
(re-)designed, we opted for formats mutagenizing three to seven codons

1
J. A. Wells, M. Vasser, and D. B. Powers, Gene 34, 315 (1985).
2
J. F. Reidhaar-Olson and R. T. Sauer, Science 241, 53 (1988).
3
S. Kamtekar, J. M. Schiffer, H. Xiong, J. M. Babik, and M. H. Hecht, Science 262, 1680
(1993).
4
G. Cho, A. D. Keefe, R. Liu, D. S. Wilson, and J. W. Szostak, J. Mol. Biol. 297, 309 (2000).
5
S. V. Taylor, K. U. Walter, P. Kast, and D. Hilvert, Proc. Natl. Acad. Sci. USA 98, 10596
(2001).
6
T. Matsuura, A. Ernst, and A. Plückthun, Protein Sci. 11, 2631 (2002).
7
E. Haslam, ‘‘Shikimic Acid: Metabolism and Metabolites.’’ Wiley, Chichester, UK, 1993.
8
P. Kast, M. Asif-Ullah, N. Jiang, and D. Hilvert, Proc. Natl. Acad. Sci. USA 93, 5043 (1996).
9
S. V. Taylor, P. Kast, and D. Hilvert, Angew. Chem. Int. Ed. 40, 3310 (2001).
10
P. Kast, J. D. Hartgerink, M. Asif-Ullah, and D. Hilvert, J. Am. Chem. Soc. 118, 3069
(1996).

Copyright 2004, Elsevier Inc.

All rights reserved.
METHODS IN ENZYMOLOGY, VOL. 388 0076-6879/04 $35.00