Available online at www.sciencedirect.com

(2007) 225–230www.elsevier.com/locate/jmicmeth

Journal of Microbiological Methods 71

PCR-based strategy for construction of multi-site-saturationmutagenic expression library

Jinxia Wang a,b, Sufang Zhang a, Haidong Tan a, Zongbao (Kent) Zhao a,⁎

a Dalian Institute of Chemical Physics, CAS, Dalian 116023, PR Chinab Graduate School of the Chinese Academy of Sciences, Beijing 100049, PR China

Received 31 May 2007; received in revised form 31 August 2007; accepted 4 September 2007Available online 11 September 2007

Abstract

There is an increasing demand for efficient and effective methods to engineer protein variants for industrial applications, structural biology anddrug development. We describe a PCR-based strategy that produces multi-site-saturation mutagenic expression library using a circular plasmidcarrying the wild-type gene. This restriction digestion- and ligation-independent method involves three steps: 1) synthesis of the degenerateoligonucleotide primers, 2) incorporation of the mutations through PCR, 3) transformation into the expression host. Our strategy is demonstratedthrough successful construction of an E. coli K12 malic enzyme expression library that contains members with simultaneous mutations on aminoacid residues G311, D345 and G397. This method is in principle compatible with any circular vector that can be propagated with a dam+ E. colihost to generate protein variant library with multiple changes, including mutation, short sequence deletion and insertion, or any mix of them.© 2007 Elsevier B.V. All rights reserved.

Keywords: Expression library; PCR cloning; Protein engineering; Site-saturation mutagenesis

1. Introduction

The increasing demand for proteins with superior perfor-mance has greatly accelerated the development of engineeringtechniques to attain protein variants (Hibbert and Dalby, 2005).The approach to protein variants can be distinguished, forexample, based on whether enough structural information isapplied. Irrational protein evolution approaches, such asrandom mutagenesis and sexual recombination, produce proteinvariants via a way that resembles the natural evolution withoutany structural information (Tobin et al., 2000). Indeed, there aremany successful examples using such strategies (Yuan et al.,2005). However, the challenges are that the sequence space tobe explored is astronomically large, whereas practical screeningcapacity is limited. With the progress in structural biology andbiochemistry in this post-genome era, substantial insights intothe structure–function relationships at the molecular level arenow available for many proteins of interest. Based on this

⁎ Corresponding author. Dalian Institute of Chemical Physics, 457 ZhongshanRoad, Dalian 116023, PR China. Tel./fax: +86 411 84379211.

E-mail address: zhaozb@dicp.ac.cn (Z.(K.) Zhao).

0167-7012/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.mimet.2007.09.001

information, it is now possible to envisage residues thatcontribute notably to proper function of the individualpolypeptide chain (Chica et al., 2005). In such cases, traditionalsite-directed mutagenesis can be applied for function-activityanalysis.

It is now well documented that drastic function improvementroutinely requires modifications beyond a single-site change(Bergquist et al., 2005; González-Blasco et al., 2000; Yoshikuniet al., 2006). In other words, one would encounter suchchallenge to construct multiple site-saturation mutagenic libraryof a given DNA sequence. Unfortunately, there are limitedmethods in the literature that can introduce mutations atmultiple pre-determined sites in the target gene simultaneously.

In principle, commercially available QuikChange® multisite-directed mutagenesis system (Stratagene, La Jolla, CA) cangenerate similar library, yet, it requires synthetic 5′-phospho-rated oligonucleotide primers and produces a mutated singlestranded circular DNA prior to transformation. Therefore,transformation efficiency might not be optimal to meet with themaximal diversity of mutagenic library (Patrick et al., 2003;Patrick and Firth, 2005). The method described by Parikh et al.generates a linear, blunt-ended full-length PCR product, which

Table 1Primers used in this study

Primer name Sequence a Use

G311X 5′-AATCGTCTTCCTTNNKGCAGGTTCAGCG-3′ PCR I, IIID345X 5′-AAGTCTTTATGGTCNNKCGCTTTGGCTTGC-3′ PCR IID345Xrev 5′-CAAGCCAAAGCGMNNGACCATAAAGACTTTC-3′ PCR IG397Xrev 5′-GTCTGTCCTGAGACMNNAATCAGAATATCTGG-3′ PCR II, IIISeq. R 5′-ATAATGTCCTGCGGTGTGGC-3′ SequencingpET-P1 5′-TGCTAGTTATTGCTCAGC-3′ Colony PCRpET-P2 5′-GCGAAATTAATACGACTC-3′ Colony PCRa K=G or T; M=C or A; N=A, C, G or T.

Fig. 1. Schematic illustration of PCR-based multi-site-saturation mutagenicexpression library construction. Three sites, namely G311, D345 and G397 of E.coli K12 NAD-ME, whose gene was cloned on the expression plasmid pET24b,were mutated using 4 synthetic degenerate primers, G311X, D345X, D345Xrevand G397Xrev. The PCR I product in pink, PCR II product in red, and PCR IIIproduct in blue, contained saturation mutagenesis at G311/D345, D345/G397,and G311/D345/G397, respectively. PCR IV was performed using the PCR IIIproduct as the primer and the circular methylated pET24b-ME isolated fromE. coli DH5α as template. The PCR III product was thus extended into a circularDNA molecular with double spatially-separated nicks. The parental vector wasdigested with DpnI and the double nicked, circular, double-stranded DNA (PCRIV product), was transformed into E. coli BL21 (DE3) cells to afford thecorresponding expression library. ⊗ denoted mutagenic amino acid residue. Inthe site ⊗ of the four primers, multiple changes, including mutation, shortsequence deletion and insertion, or any mix of them can be made, so it is inprinciple capable of giving rise to the large diverse library. (For interpretation ofthe references to colour in this figure legend, the reader is referred to the webversion of this article.)

226 J. Wang et al. / Journal of Microbiological Methods 71 (2007) 225–230

is subsequently self-ligated to regenerate the mutagenic plasmidin the presence of phage T4 DNA ligase and T4 DNApolymerase (Parikh and Matsumura, 2005). This strategy,however, involves five-round PCR and one self-ligation stepto obtain the mutagenic plasmid DNA harboring threesaturation mutagenic residues of E. coli β-galactosidase.Other method such as combinatorial active site saturationtesting (CASTing) is restricted to randomize spatially closepositions around the active site (Reetz et al., 2005).

In our on-going research, we are building a library ofmutated E. coli K12 NAD+-dependent malic enzyme (NAD-ME) with random mutagenesis of multiple sites to engineermolecular interactions between the pyridine nucleotide cofactorand the enzyme. In this study, we would like to report aneffective PCR-based method that affords a multiple site-saturation mutagenic expression library using the expressionplasmid harboring the wild-type E. coli K12 NAD-ME gene.

2. Materials and methods

2.1. Bacterial strains and plasmid

E. coli BL21 (DE3) was grown in LB liquid medium or SOCmedium (5 g/L yeast extract, 20 g/L tryptone, 0.5 g/L NaCl,2.5 mM KCl, 10 mM MgCl2, 10 mM MgSO4, and 20 mMglucose) or on SOB agar plate (5 g/L yeast extract, 20 g/Ltryptone, 0.5 g/L NaCl, 2.5 mM KCl, 10 mM MgCl2, 20 mMMgSO4 and 1.5% agar) at 37 °C. Electrocompetent E. coliBL21 (DE3) cells were prepared according to known methodexcept that the cell pellet was suspended with 10% (v/v)glycerol in the last step (Sambrook and Russel, 2001). E. coliK12 NAD-ME gene expression plasmid pET24b-ME wasconstructed as previously described (Wang et al., 2007). Theplasmid was propagated and isolated from E. coli DH5α, aDNA adenine methylase+ (dam+) strain.

2.2. Mutagenic primer design

Four degenerate oligonucleotide primers, G311X, D345X,D345Xrev and G397Xrev, were designed to introduce satura-tion mutagenesis at G311, D345 and G397 of E. coli K12 NAD-ME (Table 1). The targeted wild-type codons were set close tothe middle of the degenerate primers and replaced with an NNKsequence (K=G or T) in primers G311X/D345X or MNNsequence (M=A or C) in primers D345Xrev/G397Xrev. The

nucleotide sequences upstream and downstream the NNK orMNN were identical to those of the wild-type gene. All thesesynthetic primers were about 30 bases in length, with a meltingtemperature (Tm) higher than 65 °C and PAGE purified(TaKaRa, Dalian, China).

Fig. 2. Agarose gel analysis of PCR products. A) Products of PCR I, II and III.Lane 1, PCR I; lane 2, PCR II; lane 3, PCR III; M, DNA ladder, 2 kb, 1.6 kb,1.0 kb, 750 bp, 500 bp, 250 bp, 100 bp. B) Product of PCR IV. Lane 1, thenegative control (without Pfu DNA polymerase); lane 2, PCR IV product (thearrow indicated the band with a size of about 7 kb); M, 1 kb DNA ladder (the topband corresponded to a linear DNA fragment with a size of 8 kb).

Fig. 4. Agarose gel analysis of colony PCR products. Lane 1, the negative control(cells harboring an authentic pET24b vector); Lane 2, the positive control (cellsharboring the wild-type pET24b-ME vector); Lanes 3–8, randomly pickedmutants (cells harboring the mutagenic pET24b-ME vector); M, DNA ladder,from top to bottom, 2.0 kb, 1.6 kb, 1.0 kb, 750 bp, 500 bp, 250 bp.

227J. Wang et al. / Journal of Microbiological Methods 71 (2007) 225–230

2.3. Multiple site-saturation mutagenesis of NAD-ME gene

Two reactions were carried out with primer pair G311X/D345Xrev for PCR I and primer pair D345X/G397Xrev forPCR II. The reaction mix in a total volume of 50 μl containing50 ng of the ancestral pET24b-ME plasmid, 500 nM eachprimer, 200 μM dNTP, 1×Pfu reaction buffer and 0.5 μl of PfuDNA polymerase (2 U/μl) (Dingguo Biotech, Beijing, China),was heated at 95 °C for 2 min, followed by 24 cycles of 95 °Cfor 30 s, 70 °C for 45 s (lowering 0.5°C every cycle from thesecond cycle), 72 °C for 1 min and 15 cycles of 95 °C for 30 s,59 °C for 45 s, 72 °C for 1min. The reactionwas further incubatedat 72 °C for 10 min, and stored at 4 °C. The PCR productswere purified using the DNA fragment extraction kit (DingguoBiotech, Beijing, China) as directed by the manufacturer.

The PCR I and II products were mixed each with a finalconcentration of 0.75 nM, and a short overlap extensionreaction (PCR III) was carried out using the primers G311X(500 nM) and G397Xrev (500 nM) and 0.5 μl of Pfu DNApolymerase (2 U/μl) in a total volume of 50 μl (Parikh andMatsumura, 2005). The reaction was heated at 95 °C for 3 min,followed by 39 cycles of 95 °C for 30 s, 65 °C for 45 s, 72 °C for1 min. It was further incubated at 72 °C for 10 min, and stored at4 °C. The PCR products were gel-purified according to theabove procedure.

2.4. PCR cloning ofmulti-site-saturationmutagenicNAD-MEgene

Amixture of PCR III product (400 ng), pET24b-ME (50 ng),1×Pfu reaction buffer, and water in a total volume of 50 μl wasplaced in an Eppendorf Mastercycler thermal cycler at 95 °C for2 min. The thermal cycler was turned off to allow the samples tocool slowly to room temperature. The reaction mixture wasplaced on ice while 1.5 μl of Pfu DNA polymerase (2 U/μl) and

Fig. 3. Sequence alignment of the partial antisense strand of

1.5 μl dNTP mix (10 mM) were added. The reaction mix wasthen slowly heated to 68 °C and incubated at the sametemperature for 5 min. The segments were amplified in 36 cyclesof 95 °C for 1 min, 55 °C for 1 min and 68 °C for 14 min,followed by incubation at 68 °C for 20 min. Once completed,10 μl of the reaction mixture (PCR IV product) were treated with0.5 μl DpnI at 37 °C for 2 h to digest the methylated parentalplasmid.

The digest solution (5 μl), which contained the doublenicked, circular, double-stranded DNA, was used to transforminto 150 μl of electrocompetent E. coli BL21 (DE3) cells. Thetransformation mixture was taken by 1 ml of SOC medium,incubated with shaking at 37 °C for 1 h. The culture was ten-fold diluted, plated on 90-mm SOB agar plates supplementedwith 30 μg/ml kanamycin each with a culture volume of 100 μl,and grown at 37 °C for 16 h.

2.5. Multi-site-saturation mutagenic library validation

Two colonies were randomly picked, propagated in 5 ml ofLB liquid medium. Cells were collected and submitted forsequence analysis by TaKaRa (Dalian, China) using Seq.R asthe primer (Table 1).

Colony PCR was also carried out using a primer pair pET-P1/pET-P2 that was flanking immediately outside the NAD-ME gene on the pET24b vector. Overnight cultures wereboiled for 8 min to form a cell lysate. Colony PCR reaction mixin a total volume of 25 μl contained 5 μl of the above celllysate, 500 nM each primer (pET-P1/pET-P2), 200 μM dNTP,1×Taq plus reaction buffer and 0.25 μl of Taq plus DNApolymerase (2 U/μl) (Dingguo Biotech, Beijing, China). Thereaction was heated at 95 °C for 3 min, followed by 30 cyclesof 95 °C for 30 s, 52 °C for 30 s, 72 °C for 2 min 10 s, andincubated at 72 °C for 10 min. The PCR products were analyzedby agarose gel electrophoresis.

To further assay the library, individual colonies were pickedand inoculated in 200 μl of LB liquid medium containing

NAD-ME gene between clones M1, M2 and wild-type.

Fig. 5. Preliminary evaluation of the multi-site-saturation mutagenesisexpression library. Experiments were performed with 90 colonies in the 96-well microtiter plate in the presence of 1 mM NAD+ and 5 mM L-malate. Theabsorbance at 340 nm was monitored within 40 min. Samples in wells A1 andA2 were the blanks; Samples in wells A3 and A4 were the negative control (cellsharboring an authentic pET24b vector); Wells A5 and A6 were the positivecontrol (cells harboring the wild-type pET24b-ME vector); Samples in otherwells were the mutants (cell harboring the mutagenic pET24b-ME vector).

228 J. Wang et al. / Journal of Microbiological Methods 71 (2007) 225–230

30 μg/ml kanamycin in 96-well plates (Knaust and Nordlund,2001). The cultures were grown overnight at 37 °C, 200 rpm.After adding 4 μl of the above culture into 0.2 ml of LB liquidmedium containing 30 μg/ml kanamycin, the cells were grownat 37 °C for 3 h and then induced with 10 μM IPTG followedby an additional 12 h incubation at 30 °C, 200 rpm. Cells werepelleted at 3000 rpm for 15 min and lysed with 20 μl lysisreagent (50 mM HEPES pH 7.5, 1% Triton X-100, and 1 mg/ml lysozyme) at 37 °C for 30 min. The supernatant containingsoluble protein was collected by centrifugation at 3000 rpm for15 min, and used for activity evaluation.

Activity assay was performed according to a publishedprocedure (Wang et al., 2007). Briefly, the reaction mixcontained 50 mM HEPES pH 7.2, 1 mM NAD+, 5 mM L-malate, and 5 mM MnCl2 in a total volume of 0.1 ml. Afteradding 5 μl of crude cell extract, the absorbance at 340 nm wasmonitored at 25 °Cwithin 40min with PowerWaveXS universalmicroplate spectrophotometer (Bio-Tek instruments Inc., USA).

3. Results

3.1. Multi-site-saturation mutagenesis of NAD-ME gene

We employed four degenerate oligonucleotide primers tointroduce saturation mutagenesis at G311, D345 and G397 of E.coli K12 NAD-ME (Fig. 1). Firstly, two separate reactions wererun using pET24b-ME as template and with primer pairsG311X/D345Xrev andD345X/G397Xrev to give PCR I productand PCR II product, respectively. Both products were gel-purified and mixed in equimolar ratio. Next, an overlapextension reaction (PCR III) was run using the above mix astemplate and G311X/G397Xrev as the primer pair. Agarose gelelectrophoresis analysis of these PCR products was shown inFig. 2A. Three clear bands appeared in well accordance withtheir corresponding sizes, 130 bp, 187 bp and 288 bp forproducts of PCR I, II and III, respectively.

3.2. PCR cloning ofmulti-site-saturationmutagenicNAD-MEgene

Linear amplification reaction (PCR IV) was performed byPfu DNA polymerase in the presence of the PCR III productand plasmid pET24b-ME. As indicated in Fig. 2B, one clearband of about 7 kb was obtained, suggesting a successfulamplification. As illustrated in Fig. 1, the PCR IV productshould anneal to form double-stranded circular DNA moleculescontaining a) one nick on each strand, and b) site-saturationmutations at the target sites.

The PCR mix was treated with DpnI to digest the methylatedparental plasmid, and transformed into electrocompetent E. coliBL21 (DE3) cells. The transformants were grown on SOB agarplates to afford a library containing about 105 independentclones (∼1000 clones/90 mm plate, 115 plates).

3.3. Library validation

Two clones (M1 and M2) were randomly picked andsequenced. Alignment with the wild-type gene revealed that themutations occurred at all of the target sites, and that one-, two-and three-nucleotide replacements were found (Fig. 3). CloneM1 encoded a protein variant G311C/D345N/G397S, withthree amino acid changes. However, Clone M2 would encode atruncated protein as a stop codon was introduced at G397 site.Colony PCR with 6 randomly picked clones afforded clearproducts with apparently identical size to that of the positivecontrol (Fig. 4). These results indicated that full-length DNAsequence corresponding to the wild-type NAD-ME gene waspresent in the library.

To further characterize the library, 90 colonies were grown inLB liquid medium, and induced with IPTG. Cells were collectedand lysed. The crude cell extracts were tested for enzyme activityin the presence of NAD+ and L-malate. These datawere presentedin Fig. 5. It was clear that most of clones lost over 90% activitycomparing with the wild-type controls (A5 and A6). Only 4clones retained above 19% activity, and the best one, E3, showed75% activity of that of the wild-type clones. This was likely due tomutations at the targeted sites, leading to a diminished capabilityto accept NAD+ as the cofactor. Therefore, these datademonstrated that our strategy was effective to attain multiplesite-saturation mutagenic expression library.

4. Discussion

We have engineered novel pyridine nucleotide-dependentdehydrogenases using multi-site-saturation PCR-based muta-genesis and E. coli K12 NAD-ME. Homology modeling, andstructure analysis based on those NAD-ME structures (Changand Tong, 2003; Rao et al., 2003) revealed that G311, D345 andG397 are three vital residues that interact with the adenosinemonophosphate part of NAD+ cofactor (data not shown).Therefore, we decided to construct an expression library thatcovers all amino acid possibilities at these sites.

Conventional methods, such as QuikChange® multi site-directed mutagenesis system, and those used by Parikh et al.(Parikh and Matsumura, 2005) and Reetz et al. (Reetz et al.,

229J. Wang et al. / Journal of Microbiological Methods 71 (2007) 225–230

2005) were found unsatisfactory to construct such a library. Wetherefore developed a novel strategy to construct multi-site-saturation mutagenic expression library. As shown in Fig. 1,site-saturation mutagenesis was introduced via degenerateprimers solely based on PCR. DNA fragments containingthree site-saturation mutagenic modifications were obtained byPCR III, which was essentially an overlap extension PCR basedon equimolar mix of the PCR I and PCR II products. Themutagenic gene was incorporated into the vector using highfidelity Pfu DNA polymerase that extended the PCR III productover the entire plasmid. Subsequently, the parental vector wasdigested by DpnI and the mixture was transformed intoelectrocompetent E. coli BL21 (DE3) cells for proteinexpression. Such PCR cloning greatly improved library qualityand diversity, yet it utilized no restriction digestion and ligationenzymes.

The quality of the library was evaluated. Two randomly pickedclones were found carrying mutations at all of the target sites,suggesting proper priming with those degenerate primers. ColonyPCR experiments suggested that the library contained full-lengthNAD-ME gene sequence. Activity assays of 90 clones indicatedthat none of the mutants retained over 75% activity of that of thewild-type clone. These results indicated that we had achieved ahigh-quality multiple site-saturation mutagenic expression li-brary. On the other hand, it was also indicative that the three siteswere indeed critical for binding the NAD+ cofactor.

There are a few things that one needs to be aware of this PCR-based library construction. Firstly, the DNA polymerase used inthe linear amplification reaction (PCR IV) should be devoid ofstrand displacement activity. If the polymerase displaces bothends of the mutagenic gene and continues amplifying, the risksare that the PCRproduct loses themutagenic information encodedby the degenerate primers. Fortunately, Pfu DNA polymerase issuited for such requirement. Though fewer cycles may bebeneficial to reduce the chance of strand displacement, the stranddisplacement is not a problem for the majority of clones even ifthe number of thermal cycles was up 35 to improve the yield (Vanden Ent and Lowe, 2006). PfuTurbo™ DNA polymerase fromStratagene is the preferred one, because of its good fidelity andefficiency to amplify large-sized DNA fragments up to 15 kb.Secondly, the primers must be designed individually toincorporate the desired degenerate codons. There are a fewhints to achieve more effective mutagenic primers, as alsosuggested by QuikChange®Multi Site-Directed Mutagenesis Kitmanual: 1) Primers should be between 25 and 45 bases in length,with amelting temperature (Tm) higher than 65 °C and aminimumGC content of 40%; 2) The desired degenerate codon should beclose to the middle of the primer and with ∼10–15 bases oftemplate-complementary sequence on both sides; 3) The Tm's ofboth primer sequences flanking the degenerate codon shouldmatch well. Thirdly, since long PCR fragments prefer self-annealthat decreases the effective concentration of the primers in thereaction (Wang and Malcolm, 1999), it is very important to useenough PCR fragments (Geiser et al., 2001) and hot start PCR inthe linear amplification step (PCR IV). Lastly, it is recommendedto use electrocompetent E. coli cells for transformation to securethe quality of themutagenic library. High-efficient PCR cloning is

easy to obtain the library containing about 105 independent clonesfor three sites saturation mutagenesis enough to screen thespecific substrate and retain library completeness and diversity(Patrick and Firth, 2005).

In general, site-saturation mutagenesis of multiple (N)spatially dispersed sites of any gene harbored on a reason-able-sized plasmid with this strategy would require only threesteps, 1) design (2N−2) oligonucleotide primers, 2) run (N+1)rounds of PCR, and 3) execute transformation. There are norequirements for ligation, restriction digestion process orsynthesis of 5′-phosphorylated primers. Any circular vectorwith reasonable size should be suitable as long as it is workablewith a dam+ E. coli strain, such as E. coli DH5α, HB101 andJM109. The upper size limit for this method is largelydetermined by the capacity and fidelity of the DNA polymeraseemployed in the last cloning step. A rough estimation of thosecurrent commercial enzymes would put a maximal size of 15 kbfor the plasmid harboring the target gene. It should be noted thatif the synthetic primers incorporated defined codons, in lieu ofthe degenerate NNK as employed herein, this method can bepotentially functioned as a substitute of the expensiveQuikChange® multi site-directed mutagenesis kit. Moreover,it is in principle capable of generating proteins with simulta-neous multiple changes, including mutation, short sequencedeletion and insertion, or any mix of them, once the syntheticprimers are well designed. This method should have nospacious distance restriction between those targeted positions.

In conclusion, we have developed an efficient PCR-basedstrategy to construct multi site-saturation mutagenic expressionlibrary using the expression plasmid harboring the gene ofinterest as parental template without restriction digestion orligation process. The simplicity of the primer design and theexperimental procedures as well as the high transformationefficiency makes the method attractive for library constructionand protein engineering. We are currently using this strategy toprepare similar libraries to study molecular interaction betweenproteins and their ligands.

Acknowledgment

This work was supported by the National Natural ScienceFoundation of China (NO. 20472084).

References

Bergquist, P.L., Reeves, R.A., Gibbs, M.D., 2005. Degenerate oligonucleotidegene shuffling (DOGS) and random drift mutagenesis (RNDM): twocomplementary techniques for enzyme evolution. Biomol. Eng. 22, 63–72.

Chang, G.G., Tong, L., 2003. Structure and function of malic enzymes: a newclass of oxidative decarboxylases. Biochemistry 42, 12721–12733.

Chica, R.A., Doucet, N., Pelletier, J.N., 2005. Semi-rational approaches toengineering enzyme activity: combining the benefits of directed evolutionand rational design. Curr. Opin. Biotechnol. 16, 378–384.

Geiser, M., Cèbe, R., Drewello, D., Schmitz, R., 2001. Integration of PCRfragments at any specific site within cloning vectors without the use ofrestriction enzymes and DNA ligase. BioTechniques 31, 88–92.

González-Blasco, G., Sanz-Aparicio, J., González, B., Hermoso, J.A., Polaina, J.,2000. Directed evolution of β-glucosidase A from Paenibacillus polymyxa tothermal resistance. J. Biol. Chem. 275, 13708–13712.

230 J. Wang et al. / Journal of Microbiological Methods 71 (2007) 225–230

Hibbert, E.G., Dalby, P.A., 2005. Directed evolution strategies for improvedenzymatic performance. Microb. Cell Fact. 4, 29.

Knaust, R.K.C., Nordlund, P., 2001. Screening for soluble expression ofrecombinant proteins in a 96-well format. Anal. Biochem. 297, 79–85.

Parikh,M.R.,Matsumura, I., 2005. Site-saturationmutagenesis ismore efficient thanDNA shuffling for the directed evolution of β-fucosidase from β-galactosidase.J. Mol. Biol. 352, 621–628.

Patrick, W.M., Firth, A.E., 2005. Strategies and computational tools forimproving randomized protein libraries. Biomol. Eng. 22, 105–112.

Patrick, W.M., Firth, A.E., Blackburn, J.M., 2003. User-friendly algorithms forestimating completeness and diversity in randomized protein-encodinglibraries. Protein Eng. 16, 451–457.

Rao, G.S.J., Coleman, D.E., Karsten, W.E., Cook, P.F., Harris, B.G., 2003.Crystallographic studies onAscaris suumNAD-malic enzyme bound to reducedcofactor and identification of an effector site. J. Biol. Chem. 278, 38051–38058.

Reetz,M.T., Bocola,M., Carballeira, J.D., Zha,D., Vogel, A., 2005. Expanding therange of substrate acceptance of enzymes: combinatorial active-site saturationtest. Angew. Chem. Angew. Chem. Int. Ed. Engl. Suppl. 44, 4192–4196.

Sambrook, J., Russel, D.W., 2001. Molecular Cloning: A Laboratory Manual,3rd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Tobin, M.B., Gustafsson, C., Huisman, G.W., 2000. Directed evolution: the‘rational’ basis for ‘irrational’ design. Curr. Opin. Struck. Biol. 10, 421–427.

Van den Ent, F., Lowe, J., 2006. RF cloning: a restriction-free method for insertingtarget genes into plasmids. J. Biochem. Biophys. Methods 67, 67–74.

Wang, W., Malcolm, B.A., 1999. Two-stage PCR protocol allowing introductionof multiple mutations, deletions and insertions using QuikChange site-directed mutagenesis. BioTechniques 26, 680–682.

Wang, J., Tan, H., Zhao, Z., 2007. Over-expression, purification, and charac-terization of recombinant NAD-malic enzyme from Escherichia coli K12.Protein Expr. Purif. 53, 97–103.

Yoshikuni, Y., Ferrin, T.E., Keasling, J.D., 2006. Designed divergent evolutionof enzyme function. Nature 440, 1078–1082.

Yuan, L., Kurek, I., English, J., Keenan, R., 2005. Laboratory-directed proteinevolution. Microbiol. Mol. Biol. Rev. 69, 373–392.