Available online at www.sciencedirect.com Advances in laboratory evolution of enzymes Shimon Bershtein and Dan S Tawfik We address recent developments in the area of laboratory, or directed evolution, with a focus on enzymes and on new methodologies of generic potential. We survey three main areas: (i) library making techniques, including the application of computational and rational methods for library design; (ii) screening and selection techniques, including recent applications of enzyme screening by FACS (fluorescence activated cell sorter); (iii) new approaches for performing directed evolution, and in particular, the application of ‘neutral drifts’ (libraries generated by rounds of mutation and selection for the enzyme’s original function) and of consensus mutations to generate highly evolvable starting points for directed evolution. Address Department of Biological Chemistry, Weizmann Institute of Science, Rehovot 76100, Israel Corresponding author: Tawfik, Dan S (tawfik@weizmann.ac.il) Current Opinion in Chemical Biology 2008, 12:151–158 This review comes from a themed issue on Biocatalysis and Biotransformation Edited by Stephen G. Withers and Lindsay Eltis Available online 7th March 2008 1367-5931/$ – see front matter # 2008 Elsevier Ltd. All rights reserved. DOI 10.1016/j.cbpa.2008.01.027 Introduction Directed evolution techniques and applications have widely expanded in the past few years. The literature relevant to this area includes several hundred articles for the period of 2005–2007 that cannot be covered by this review. We therefore focus on a few major topics. Firstly, although directed evolution has now been applied to a wide range of protein functions, including ligand and protein binding, and regulatory functions, this review’s main focus is enzymes. Secondly, we primarily address new methodologies and strategies for directed evolution, rather than examples, as interesting as they may be, for the evolution of individual enzymes. The review comprises three sections that discuss recent developments in library making methodologies, screening and selection methodologies, and new approaches for performing directed evolution. The reader is referred to other recent reviews, including an extensive review on directed evolution of enzymes [1] www.sciencedirect.com and a review on selection strategies [2]. The application of directed evolution to enzyme pathways has also been reviewed [3,4]. Finally, research at the interface of fundamental and applied enzyme evolution has also been discussed [5,6]. Library making techniques Refining existing protocols Several works describe the refinement of existing library making techniques. These include variations on Stemmer’s DNaseI protocol for DNA shuffling [7,8,9] (for a comprehensive description of various in vitro recombination strategies consult the recent review [10]) and a method that aims at mixing wild-type, and improved mutant genes, in certain combinations [11]. New applications for the useful incremental truncation techniques, such as ITCHY [12], were also described [13,14]. Other works address the fact that random libraries generated by various methods, such as error-prone PCR, are clearly non-random. Each method gives a different bias of base substitutions that result in a different range of amino acid replacements, the latter also depending on the mutated gene itself. Statistical approaches that facilitate an educated choice of technique(s) have become available. These assess the sequence diversity available by each mutagenesis technique [15] and also analyze sequence and structure indicators of the resulting libraries at the protein level [16,17]. Experimental approaches aimed at increasing library sizes by ‘in situ’ mutagenesis have also been described. These are based on generating genetic diversity within the Escherichia coli, or yeast, cells in which protein expression and screening take place. Such in situ mutagenesis approaches (e.g. the use of E. coli carrying a mutator polymerase [18]) afford very large library sizes independent of transformation efficiency. Following earlier developments of in situ homologous recombination in yeast [19] and in situ mutagenesis in RNA bacteriophage [20], new methods for in situ recombination has now been described for both E. coli [21] and yeast [22]. Indels Whereas techniques for creating genetic diversity by recombination and point mutations are well developed and widely applied, methods for randomly incorporating insertions and deletions (indels) are still in their infancy. The divergence of new functions in natural enzymes often involves such changes, and their laboratory reproduction is therefore of importance. Two new methods for random incorporation of indels [23] and in-frame deletions [24] were recently described, as well as a Current Opinion in Chemical Biology 2008, 12:151–158 152 Biocatalysis and Biotransformation method for incorporation of indels in pre-defined positions in a combinatorial manner [25]. However, the wide applicability of these techniques, and in particular, a convincing demonstration that indels can yield enzymatic functions that are not accessible by point mutations and recombination, is still pending. Targeted and designed libraries The scrutiny of various library making strategies continues. One school surmises that incorporating mutations at random is the most effective approach, primarily because many function altering mutations occur at unexpected positions away from the active site. The other school focuses on the randomization of a given set of residues that are within the active site and in direct contact with the substrate [26,80]. Indeed, in certain cases, targeted active-site libraries yield preferable results [27–29,69], but in others, the results of random and targeted libraries were comparable [30]. It therefore appears that mutations in both direct contact, and distal, mutations can improve activity, and the former can be particularly successful in cases where a small number of active-site residues that trigger the required change in substrate specificity can be identified [27]. The snag is that in most cases, the choice of active-site residues for diversification is far from obvious. Active-site walls comprise numerous residues, and the libraries in which these are simultaneously randomized are far too large and comprise almost only inactive variants. This difficulty can be addressed by iteratively exploring the active-site residues, namely by libraries that randomize one or two residues at a time and then combine the advantageous mutations [31,32]. This strategy might be limited, however, by the non-additivity (epistasis) of evolutionary pathways, namely, in many cases where the order by which mutations accumulate is crucial [33]. Alternatively, combinatorial methods that generate random subsets of the entire set of randomized active-site residues have been explored [25,30]. Finally, recent works indicate that a successful choice of which positions should be diversified, and how (e.g. into which amino acids) could also be made using a variety of statistical and computational methods [34,35–37]. While space limitations do not enable a detailed discussion of these methodologies, it appears that computationally designed libraries will play a growingly important role in protein engineering. The facile making of such ‘smart’ libraries can be greatly facilitated by sophisticated, automated DNA synthesis methods [38]. Screening/selection techniques FACS based enzyme screens Modern FACS machines can easily sort >104 events/s, accurately and reproducibly, using multiple parameters. FACS therefore holds much potential in the area of enzyme evolution. What remains a main challenge, Current Opinion in Chemical Biology 2008, 12:151–158 however, is maintaining the linkage between the enzyme, a diffusable product, and the enzyme coding gene. In some cases, where the target reaction involves the modification of a hydrophobic fluorescent substrate with a charged group, the unmodified substrate can be washed out of the cells, while the product remains within (Figure 1a). For example, modification of the fluorophore 7-amino-4-chloromethyl coumarin with glutathione resulted in its entrapment, allowing large libraries of E. coli cells carrying active glutathione-Stransferases (GSTs) variants to be selected [39,40]. In another example, fluoroscein di-b-galactopyranoside with a lipohilic tail that retains the fluorescent product upon cleavage by b-galactosidase was applied [7]. An alternative scheme for the entrapment of a fluorescent product by coupled enzymatic reactions that occur within the sorted E. coli cells was devised for the selection of glycosyltransferases [41]. Another type of application involved the display of enzyme on cell surfaces and the entrapment of the product onto the surface (Figure 1b). The protease OmpA was displayed on the surface of E. coli, and a fluorescence resonance energy transfer (FRET) substrate was added that adheres to the surface of the bacteria. Enzymatic cleavage of the substrate releases a quencher group, and the resulting fluorescent cells were isolated by FACS. This system was used for a dual selection (a selection for a cleavage of a given peptide substrate against the cleavage of others), and a variant of OmpT proteinase with a change in selectivity of >106 relative to wild-type was obtained [[42], see also note added in proof]. A complementary approach was developed whereby the substrates of surface-displayed enzymes (esterases or lipases) release a tyramide–biotin moiety that is then covalently linked to the cell surface using peroxidase-activated tyramide conjugation. This labeling enables the isolation of E. coli cells presenting active enzyme variants by either FACS or magnetic-bead selections [43,69]. The entrapment of fluorescent products can also be achieved by compartmentalization in emulsion droplets (in vitro compartmentalization, or IVC) (Figure 1c). The technology makes use of the standard water-in-oil (w/o) droplets to compartmentalize the gene, the encoded enzyme, and the resulting product. Conversion of the primary emulsion to a double water-in-oil-in-water (w/o/ w) emulsion provides an external aqueous phase that enables sorting of the droplets by FACS. Double emulsions were recently applied for the directed evolution of different enzymatic systems. A library of b-galactosidase variants was translated in vitro in the primary w/o emulsion and then FACS sorted in a double emulsion [44]. In the case of PON1, intact E. coli cells in which the library variants are expressed were emulsified and FACS sorted [30]. The same strategy has been recently applied to the www.sciencedirect.com Advances in laboratory evolution of enzymes Bershtein and Tawfik 153 Figure 1 Screening enzyme libraries by FACS (fluorescence activated cell sorter). A wealth of enzyme reactions can be screened using fluorogenic substrates, provided that the enzyme, the product, and the gene encoding the enzyme can be linked during the sort. Various strategies were developed that meet this prerequisite: (a) Entrapment of the product within the enzyme expressing cells [7,39,40,41]. The enzyme-transformed substrate is modified in a way that prevents its diffusion out of the cells (usually by the attachment of a polar or charged group). The excess of unreacted substrate is washed out, and cells containing the entrapped fluorescent product are isolated. (b) Entrapment of the product onto the surface of cells displaying the target enzyme [42,43,69]. The surface-displayed (or periplasmic) enzyme converts a non-fluorescent substrate into a fluorescent product that is then attached to the cell surface (either covalently, or non-covalently). (c) Entrapment of the fluorescent product in an emulsion droplet carrying the enzyme and its encoding gene. In this case, the enzyme can be expressed in E. coli cells (either in the cytoplasm, or on the surface), or by cell-free translation. Conversion of the emulsion to a double, w/o/w emulsion enables the FACS sorting of the droplets [30,44–46]. selection of lactonase activity [45], using an oxo-lactone substrate whose enzymatic hydrolysis generates a thiol that was subsequently detected with a fluorogenic probe. Detailed protocols for the preparation and sorting of double emulsions are now available [46]. Other screening/selection methodologies The development and application of other emulsionbased selections are described in a recent review [47], with the most recent developments being the selection of thermostable polymerase variants for the amplification of damaged DNA samples, such as ancient DNA [48]; selection by product driven PCR amplification of enzyme coding genes [81]; and the selection of restriction endonucleases [82]. www.sciencedirect.com In vitro techniques other than IVC have also been applied for directed evolution. Most recently, messenger RNA display was used to select a library of >1012 in vitro translated protein variants derived from a novel zincfinger scaffold, and thus identify novel RNA ligases [49]. In vitro (cell-free) transcription/translation has also been used in conjunction with standard screens in 96-well plates [29]. Besides the development of novel approaches, the application of traditional methodologies, such as genetic screens and selections and screens in 96-well plates, has produced interesting results. A book describing a wide range of generally applicable approaches has been recently published [50]. Screening assays for highly useful Current Opinion in Chemical Biology 2008, 12:151–158 154 Biocatalysis and Biotransformation enzyme types such as hydroxynitrile lyases [51] and epoxygenases [52] have been developed and applied for library selections. Despite their low throughput, assays in microtiter plates (96- or 384-plates) are advantageous because they can be applied to complex enzyme systems such as the adenylation domains involved in combinatorial biosynthesis [53] or new glycosyl transferases [54]. The throughput of plate assays can also be increased by pooling up to 10 colonies in a single well [55]. Notable extensions for the application of traditional genetic in vivo screens to enzyme library selections have been also described [56], including the construction of a chemosensor E. coli host strain that recognizes the product of the targeted enzymatic conversion [57] and the application of tunable expression vectors for extending the dynamic range of in vivo selections [58]. New approaches for laboratory enzyme evolution Multi-parameter selections Whereas traditional screening and selection strategies aim at isolating variants that exhibit an increase in a single target activity, more sophisticated schemes arise that screen for more than one parameter. Multi-parameter screens can include many substrates [45,59,60]. The aim can be, for example, to maintain the enzymatic properties by screening for multiple substrates while increasing expression levels and stability (for example see reference [61]), or enhancing only one enzymatic activity without altering others [62]. Coevolving stability and enzymatic activity is another target for multiparameter selections [63,64]. Finally, it is becoming apparent that while improving the activity toward a weak, promiscuous substrate is easily achieved, under selection for a ‘new’ activity, the original, native activity of the enzyme often persists [65,66]. Hence, a ‘dual selection’ – a selection for the new function and against the old one – needs often to be applied to accelerate the evolution of a new function while erasing the original one [[42,67,69], see also note added in proof]. Neutral drift—a novel approach for directed evolution An emerging development in the area of laboratory evolution regards the reproductions of ‘neutral drifts’, that is, the gradual accumulation of mutations under selection to maintain the protein’s original function and structure [7,45,60,68,70]. Theoreticians have predicted fascinating properties of ‘neutral networks’ or quasi-species such as higher mutational robustness and evolvability that develop when proteins drift and create a ‘cloud’ of sequences around an existing sequence [71–73]. Although a neutral enzyme drift was described a while ago, but its potential in the evolution of new function has not been demonstrated [7]. More recently, neutral drifts of two different enzymes PON1 and P450 were perCurrent Opinion in Chemical Biology 2008, 12:151–158 formed by applying iterative rounds of mutation and selection for the enzymes’ native or primary activity. These showed how the potential for new functions develops when the neutral network of a protein expands [45,60]. The PON1 drift led to the isolation and characterization of 311 neutral variants that maintain wild-typelike activity and expression level. About half of these neutral variants exhibited significant changes in promiscuous activities, specificities, or inhibition (the latter simulating the emergence of drug resistance) [45]. Similar fluctuations in promiscuous activities were observed in the P450 neutral variants [60]. A number of the PON1 neutral variants isolated were one, or even two, mutations closer to a new enzyme specificity (aryl esterase, thiolactonase, phosphotriesterase, or drug resistance). Indeed, a subsequent screen of these 311 variants with a new substrate (an analog of the nerve agent cyclosarin) yielded two variants with up to 72-fold higher activity relative to wild-type PON1 [45]. Additional variants exhibiting large improvements with other promiscuous substrates were recently identified (Gupta D et al., unpublished results). How can a small library of only 311 members yield a range of improved variants? The standard libraries used for directed evolution comprise a majority of ‘dead’ variants owing to deleterious mutations, and hence demand the screening of much larger diversities. The neutral drift libraries comprise enzyme variants that are all stable and active. Nevertheless, these variants carry numerous mutations in, and around, the active site. Some of these mutations have the ability to mediate new specificities by increasing or decreasing existing promiscuous activities while exerting little effect on native activity (Figure 2). Boosting enzyme evolvability An oft-forgotten point is that the mutation rates practiced in directed laboratory evolution are orders of magnitude higher than those of natural organism (>1 mutation/gene/ generation, in contrast to 10 6 in most of the natural organisms). The prevailing view is that proteins tolerate most mutations with no loss of function. But in fact, although proteins generally withstand the first one or few mutations, with no apparent loss of activity, their stability is severely compromised, and so is their ability to stomach more mutations [68]. Most proteins used as starting points for laboratory evolution never evolved to withstand high mutational loads, but because thermodynamic stability correlates with mutational tolerance, the application of thermodynamically stable proteins, either natural thermophylic proteins (for example see reference [74]), or laboratory evolved stabilized proteins, as starting points for library making is one way of solving this problem [75,79]. Neutral drifts provide another way of tackling the problem of sensitivity to mutations. As previously predicted www.sciencedirect.com Advances in laboratory evolution of enzymes Bershtein and Tawfik 155 Figure 2 [72,73], neutral drifts can prompt the emergence of mutational robustness. The neutral drift performed with a P450 demonstrated that increased stability and mutational tolerance underlies large polymorphic populations, and not in a drift taken through a bottleneck, that is, when a single variant was passed from one generation to another [70]. A drift we performed with TEM-1 b-lactamase by selecting for its native penicillinase activity showed that several mutations that act as ‘global suppressors’ were enriched that increase TEM-1’s stability and suppress the effect of a broad range of deleterious destabilizing mutations. These global suppressors enabled a higher number of mutations to accumulate without deleterious consequences and increased the frequency of positive variants observed in a selection for a new function (degradation of a synthetic, ‘third-generation antibiotic) (Bershtein S, et al., unpublished results). Interestingly, all the global suppressors we identified emerged in positions where the sequence of TEM-1 deviates from its family consensus, and thereby bring TEM-1 closer to its consensus. Such changes, which can be readily predicted from sequence alignments, have been shown to stabilize numerous proteins [76–78]. Neutral drifts and the evolution of new enzyme variants. (a) The neutral drift of serum paraoxonase (PON1) introduced a wide range of apparently neutral mutations in the periphery and within PON1’s active site [45]. The active site pocket of PON1 comprises 32 residues and is presented as a blue mesh. The highly conserved calcium ion (magenta sphere), and histidine dyad H115 and H134 (green spheres) that mediate PON1’s native activity as a lactonase are shown. Active site residues found to be mutated in variants isolated during three rounds of neutral drift are colored in either red, or yellow. The red colored residues were identified in variants with native lactonase activity and expression levels that are essentially identical to wild-type PON1, whereas yellow residues refer to variants in which a significant lactonase activity is maintained although not at wild-type levels [45]. Note that the active site wall in the vicinity of the 115–134 dyad remained intact in contrast to extensive changes seen in other parts. (b) A schematic representation of PON1’s putative neutral network based on the experimentally characterized neutral variants and variants identified in previous directed evolution experiments PON1 (adopted from reference [45]). The neutral variants comprise a ‘cloud’ of sequences scattered around the wild-type sequence that is primarily a lipo-lactonase. The expansion of PON1’s neutral network, through the mutations highlighted in panel a enables gradual (by single mutation steps) and smooth (without loss of protein stability or function) transitions to a whole range of new enzyme variants, each possessing a new and different substrate specificity. www.sciencedirect.com These experiments point out an alternative approach that can yield interesting and useful results despite the use of relatively small libraries. The latter is particularly important in a wide range of emerging applications for which screening can only be performed with limited throughput, for example, under industrially relevant conditions, or for substrate/reaction that is not amenable to high throughput screens. The neutral drift (i.e. repetitive rounds of mutation and selection for the original function of the starting gene) requires relatively small libraries because, unlike a selection for a new function where active variants are rare, most library-mutated variants are active. The resulting library contains large, yet non-deleterious, variation. As such, neutral drift libraries – although relatively small in size – can possess a vastly increased potential to evolve new functions. Another useful approach that emerges concerns libraries enriched with consensus mutations. These mutations could be spiked into libraries using synthetic oligonucleotides and DNA shuffling [25,34]). The presence of these mutations can boost the protein’s tolerance to mutations, and compensate for the deleterious effects of function-altering mutation, thereby increasing the frequency of variants conferring new functions (Bershtein S, et al., unpublished results). Note added in proof Varadarajan et al. report the application of enzyme display and FACS (see Ref. [42]; Figure 1b) and the engineering of a remarkable series of site-specific endopeptidases capable of cleaving a wide range of peptide sequences with high selectivity and catalytic efficiency (Varadarajan N, Rodriguez S, Hwang B-Y, Georgiou G, Iverson BL: A Current Opinion in Chemical Biology 2008, 12:151–158 156 Biocatalysis and Biotransformation family of engineered endopeptidases. Nature Chem Biol; in press). These enzyme variants were obtained by iterative rounds using different diversification techniques, and by applying both positive selection for variants cleaving the desired peptide substrate, and negative selection to purge variants cleaving other peptides. Acknowledgements Research grants from the Minerva Foundation, the EU MiFEM consortium, and the Estate of Fannie Sherr are gratefully acknowledged. References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: of special interest 1. Kaur J, Sharma R: Directed evolution: an approach to engineer enzymes. Crit Rev Biotechnol 2006, 26:165-199. 2. Boersma YL, Droge MJ, Quax WJ: Selection strategies for improved biocatalysts. FEBS J 2007, 274:2181-2195. 3. Johannes TW, Zhao H: Directed evolution of enzymes and biosynthetic pathways. Curr Opin Microbiol 2006, 9:261-267. 4. Chatterjee R, Yuan L: Directed evolution of metabolic pathways. Trends Biotechnol 2006, 24:28-38. 5. O’Loughlin TL, Patrick WM, Matsumura I: Natural history as a predictor of protein evolvability. Protein Eng Des Sel 2006, 19:439-442. 6. Peisajovich S, Tawfik DS: Protein engineers turned evolutionists. Nat Methods 2007, 4:991-994. 7. Bergquist PL, Reeves RA, Gibbs MD: Degenerate oligonucleotide gene shuffling (DOGS) and random drift mutagenesis (RNDM): two complementary techniques for enzyme evolution. Biomol Eng 2005, 22:63-72. A useful modification of the DNase shuffling technique that decreases contamination by parental genes and enables targeting of random mutations to specified fragments of the shuffled genes. The article also described a FACS based enzyme screening strategy and raises the potential utility of neutral drift as a general strategy for generating libraries. 8. Zhao H, Zha W: In vitro ‘sexual’ evolution through the PCRbased staggered extension process (StEP). Nat Protoc 2006, 1:1865-1871. 9. Rosic NN, Huang W, Johnston WA, DeVoss JJ, Gillam EM: Extending the diversity of cytochrome P450 enzymes by DNA family shuffling. Gene 2007, 395:40-48. 16. Wong TS, Roccatano D, Schwaneberg U: Steering directed protein evolution: strategies to manage combinatorial complexity of mutant libraries. Environ Microbiol 2007, 9:2645-2659. 17. Wong TS, Roccatano D, Zacharias M, Schwaneberg U: A statistical analysis of random mutagenesis methods used for directed protein evolution. J Mol Biol 2006, 355:858-871. 18. Camps M, Naukkarinen J, Johnson BP, Loeb LA: Targeted gene evolution in Escherichia coli using a highly error-prone DNA polymerase I. Proc Natl Acad Sci U S A 2003, 100:9727-9732. 19. Swers JS, Kellogg BA, Wittrup KD: Shuffled antibody libraries created by in vivo homologous recombination and yeast surface display. Nucleic Acids Res 2004, 32:e36. 20. Makeyev EV, Bamford DH: Evolutionary potential of an RNA virus. J Virol 2004, 78:2114-2120. 21. Wang PL, Lo BK, Winter G: Generating molecular diversity by homologous recombination in Escherichia coli. Protein Eng Des Sel 2005, 18:397-404. 22. Zumarraga M, Camarero S, Shleev S, Martinez-Arias A, Ballesteros A, Plou FJ, Alcalde M: Altering the laccase functionality by in vivo assembly of mutant libraries with different mutational spectra. Proteins 2007, 71:250-260. 23. Fujii R, Kitaoka M, Hayashi K: RAISE: a simple and novel method of generating random insertion and deletion mutations. Nucleic Acids Res 2006, 34:e30. 24. Simm AM, Baldwin AJ, Busse K, Jones DD: Investigating protein structural plasticity by surveying the consequence of an amino acid deletion from TEM-1 beta-lactamase. FEBS Lett 2007, 581:3904-3908. 25. Herman A, Tawfik DS: Incorporating synthetic oligonucleotides via gene reassembly (ISOR): a versatile tool for generating targeted libraries. Protein Eng Des Sel 2007, 20:219-226. A multi-purpose protocol for shuffling with synthetic oligos, which enables rational design inputs to be incorporated in a combinatorial manner. The method enables partial diversification of a large set of residues (as many as 45 positions/library) within the same library so that each variant contains a different subset of mutated positions. It can be used to incorporate a variety of point mutations (e.g. consensus mutations for increased stability) as well as indels at pre-defined positions (for other applications see reference [34]). 26. Morley KL, Kazlauskas RJ: Improving enzyme properties: when are closer mutations better? Trends Biotechnol 2005, 23:231237. A valuable discussion that points out a correlation between the distance of mutations from the active site and the correspondent improvement of various enzyme properties. On the basis of a survey of many protein engineering studies, the authors conclude that in many cases (enantioselectivity, substrate specificity, and new catalytic activities) mutations closer to the active site seem to be more effective. 10. Sen S, Venkata Dasu V, Mandal B: Developments in directed evolution for improving enzyme functions. Appl Biochem Biotechnol 2007, 143:212-223. 27. Kelly RM, Leemhuis H, Dijkhuizen L: Conversion of a cyclodextrin glucanotransferase into an alpha-amylase: assessment of directed evolution strategies. Biochemistry 2007, 46:899-908. 11. Hamamatsu N, Aita T, Nomiya Y, Uchiyama H, Nakajima M, Husimi Y, Shibanaka Y: Biased mutation-assembling: an efficient method for rapid directed evolution through simultaneous mutation accumulation. Protein Eng Des Sel 2005, 18:265-271. 28. Park S, Morley KL, Horsman GP, Holmquist M, Hult K, Kazlauskas RJ: Focusing mutations into the P. fluorescens esterase binding site increases enantioselectivity more effectively than distant mutations. Chem Biol 2005, 12: 45-54. 12. Ostermeier M, Nixon AE, Shim JH, Benkovic SJ: Combinatorial protein engineering by incremental truncation. Proc Natl Acad Sci U S A 1999, 96:3562-3567. 29. Koga Y, Yamane T, Nakano H: Creation of novel enantioselective lipases by SIMPLEX. Methods Mol Biol 2007, 375:165-181. 13. Peisajovich SG, Rockah L, Tawfik DS: Evolution of new protein topologies through multistep gene rearrangements. Nat Genet 2006, 38:168-174. 30. Aharoni A, Amitai G, Bernath K, Magdassi S, Tawfik DS: Highthroughput screening of enzyme libraries: thiolactonases evolved by fluorescence-activated sorting of single cells in emulsion compartments. Chem Biol 2005, 12:1281-1289. 14. Yadid I, Tawfik DS: Reconstruction of functional beta-propeller lectins via homo-oligomeric assembly of shorter fragments. J Mol Biol 2007, 365:10-17. 15. Volles MJ, Lansbury PT Jr: A computer program for the estimation of protein and nucleic acid sequence diversity in random point mutagenesis libraries. Nucleic Acids Res 2005, 33:3667-3677. Current Opinion in Chemical Biology 2008, 12:151–158 31. Clouthier CM, Kayser MM, Reetz MT: Designing new BaeyerVilliger monooxygenases using restricted CASTing. J Org Chem 2006, 71:8431-8437. 32. Reetz MT, Carballeira JD: Iterative saturation mutagenesis (ISM) for rapid directed evolution of functional enzymes. Nat Protoc 2007, 2:891-903. www.sciencedirect.com Advances in laboratory evolution of enzymes Bershtein and Tawfik 157 The article describes a series of protocols for the systematic modification of enzymes by saturation mutagenesis of specific positions, to enhance enantioselectivity and thermostability, by several iterative rounds, each involving the screening of a relatively small library. 46. Miller OJ, Bernath K, Agresti JJ, Amitai G, Kelly BT, Mastrobattista E, Taly V, Magdassi S, Tawfik DS, Griffiths AD: Directed evolution by in vitro compartmentalization. Nat Methods 2006, 3:561-570. 33. Dean AM, Thornton JW: Mechanistic approaches to the study of evolution: the functional synthesis. Nat Rev Genet 2007, 8:675688. 47. Griffiths AD, Tawfik DS: Miniaturising the laboratory in emulsion droplets. Trends Biotechnol 2006, 24:395-402. 34. Fox RJ, Davis SC, Mundorff EC, Newman LM, Gavrilovic V, Ma SK, Chung LM, Ching C, Tam S, Muley S et al.: Improving catalytic function by ProSAR-driven enzyme evolution. Nat Biotechnol 2007, 25:338-344. A statistical analysis of protein sequence–activity relationships was developed for enzyme optimization. The method is based on the statistical identification of beneficial mutations in early rounds variants, including variants with reduced activity, and combining these mutations by the incorporation of synthetic oligos and DNA shuffling. The method was applied toward the evolution of a halohydrin dehalogenase for the synthesis of the cholesterol-lowering drug Lipitor, and the optimized variant carried >35 mutations. 35. Meyer MM, Hochrein L, Arnold FH: Structure-guided SCHEMA recombination of distantly related beta-lactamases. Protein Eng Des Sel 2006, 19:563-570. 36. Yoshikuni Y, Ferrin TE, Keasling JD: Designed divergent evolution of enzyme function. Nature 2006, 440:1078-1082. 37. Mena MA, Treynor TP, Mayo SL, Daugherty PS: Blue fluorescent proteins with enhanced brightness and photostability from a structurally targeted library. Nat Biotechnol 2006, 24:1569-1571. 38. Cox JC, Lape J, Sayed MA, Hellinga HW: Protein fabrication automation. Protein Sci 2007, 16:379-390. 39. Griswold KE, Aiyappan NS, Iverson BL, Georgiou G: The evolution of catalytic efficiency and substrate promiscuity in human theta class 1-1 glutathione transferase. J Mol Biol 2006, 364:400-410. 40. Griswold KE, Kawarasaki Y, Ghoneim N, Benkovic SJ, Iverson BL, Georgiou G: Evolution of highly active enzymes by homologyindependent recombination. Proc Natl Acad Sci U S A 2005, 102:10082-10087. 41. Aharoni A, Thieme K, Chiu CP, Buchini S, Lairson LL, Chen H, Strynadka NC, Wakarchuk WW, Withers SG: High-throughput screening methodology for the directed evolution of glycosyltransferases. Nat Methods 2006, 3:609-614. A new high-throughput screening methodology for sialyltransferases was developed that is based on entrapment of the fluorescent product within E. coli cells (Figure 1a). The cells were genetically manipulated to enable the transport of the substrates, prevent substrate catabolism, and allow the synthesis of the donor sugar. A library of >106 sialyltransferase mutants was sorted by FACS, and a variant with 400-fold higher catalytic efficiency for a range of fluorescently labeled acceptor sugars was isolated. 42. Varadarajan N, Gam J, Olsen MJ, Georgiou G, Iverson BL: Engineering of protease variants exhibiting high catalytic activity and exquisite substrate selectivity. Proc Natl Acad Sci U S A 2005, 102:6855-6860. 43. Becker S, Michalczyk A, Wilhelm S, Jaeger KE, Kolmar H: Ultrahigh-throughput screening to identify E. coli cells expressing functionally active enzymes on their surface. Chembiochem 2007, 8:943-949. 44. Mastrobattista E, Taly V, Chanudet E, Treacy P, Kelly BT, Griffiths AD: High-throughput screening of enzyme libraries: in vitro evolution of a beta-galactosidase by fluorescenceactivated sorting of double emulsions. Chem Biol 2005, 12:1291-1300. 45. Amitai G, Devi-Gupta R, Tawfik DS: Latent evolutionary potentials under the neutral mutational drift of an enzyme. HFSP J 2007, 1:67-78. Small, yet highly effective, libraries were generated by selection for the enzyme’s native activity, and expression levels, thus demonstrating the role of neutral drifts in natural evolution, and as an emerging new tool in laboratory enzyme evolution (see also references [7,60]). Analysis of 311 neutral variants indicated the accumulation of a range of active-site mutations, and enzyme variants with new activities were readily isolated from this small repertoire (Figure 2). www.sciencedirect.com 48. d’Abbadie M, Hofreiter M, Vaisman A, Loakes D, Gasparutto D, Cadet J, Woodgate R, Paabo S, Holliger P: Molecular breeding of polymerases for amplification of ancient DNA. Nat Biotechnol 2007, 25:939-943. Selection by compartmentalization in water-in-oil emulsions and family shuffling were applied toward the directed evolution of highly useful thermostable polymerase variants capable of extending ancient DNA with a characteristic variety of mismatches and lesions. 49. Seelig B, Szostak JW: Selection and evolution of enzymes from a partially randomized non-catalytic scaffold. Nature 2007, 448:828-831. A rare, possibly the first, example of laboratory evolution of a genuinely novel enzymatic activity by selection with mRNA display from a randomized zinc-finger scaffold that possessed no enzymatic activity. Singleturnover product formation was the selection criterion applied, thus enabling the isolation of very rare library variants that exhibit even weak catalytic activity. While the scope of this selection approach (in terms of substrates and reactions) is clearly limited, this work is a long-awaited violation of the second putative rule of directed evolution: ‘you should select for what is already there’ (see reference [6]). 50. Reymond J-L (Ed): Enzyme assays: High-throughput Screening, Genetic Selection and Fingerprinting. Weinheim: Wiley-VCH; 2005. 51. Krammer B, Rumbold K, Tschemmernegg M, Pochlauer P, Schwab H: A novel screening assay for hydroxynitrile lyases suitable for high-throughput screening. J Biotechnol 2007, 129:151-161. 52. Tee KL, Schwaneberg U: A screening system for the directed evolution of epoxygenases: importance of position 184 in P450 BM3 for stereoselective styrene epoxidation. Angew Chem Int Ed Engl 2006, 45:5380-5383. 53. Otten LG, Schaffer ML, Villiers BR, Stachelhaus T, Hollfelder F: An optimized ATP/PP(i)-exchange assay in 96-well format for screening of adenylation domains for applications in combinatorial biosynthesis. Biotechnol J 2007, 2:232-240. 54. Williams GJ, Zhang C, Thorson JS: Expanding the promiscuity of a natural-product glycosyltransferase by directed evolution. Nat Chem Biol 2007, 3:657-662. 55. Polizzi KM, Parikh M, Spencer CU, Matsumura I, Lee JH, Realff MJ, Bommarius AS: Pooling for improved screening of combinatorial libraries for directed evolution. Biotechnol Prog 2006, 22:961-967. 56. Kleeb AC, Edalat MH, Gamper M, Haugstetter J, Giger L, Neuenschwander M, Kast P, Hilvert D: Metabolic engineering of a genetic selection system with tunable stringency. Proc Natl Acad Sci U S A 2007, 104:13907-13912. 57. van Sint Fiet S, van Beilen JB, Witholt B: Selection of biocatalysts for chemical synthesis. Proc Natl Acad Sci U S A 2006, 103:16931698. An E. coli host strain was engineered that recognizes the product of the biocatalytic conversion on interest through a transcriptional activator and a reporter gene. This engineered strain responded to the oxidation of benzaldehydes to benzoic acids thus enabling the selection of biocatalytically active cells. 58. Neuenschwander M, Butz M, Heintz C, Kast P, Hilvert D: A simple selection strategy for evolving highly efficient enzymes. Nat Biotechnol 2007, 25:1145-1147. 59. Kurtovic S, Runarsdottir A, Emren LO, Larsson AK, Mannervik B: Multivariate-activity mining for molecular quasi-species in a glutathione transferase mutant library. Protein Eng Des Sel 2007, 20:243-256. 60. Bloom JD, Romero PA, Lu Z, Arnold FH: Neutral genetic drift can alter promiscuous protein functions, potentially aiding functional evolution. Biol Direct 2007, 2:17. 61. Aharoni A, Gaidukov L, Yagur S, Toker L, Silman I, Tawfik DS: Directed evolution of mammalian paraoxonases PON1 and Current Opinion in Chemical Biology 2008, 12:151–158 158 Biocatalysis and Biotransformation PON3 for bacterial expression and catalytic specialization. Proc Natl Acad Sci U S A 2004, 101:482-487. 62. Nakagawa Y, Hasegawa A, Hiratake J, Sakata K: Engineering of Pseudomonas aeruginosa lipase by directed evolution for enhanced amidase activity: mechanistic implication for amide hydrolysis by serine hydrolases. Protein Eng Des Sel 2007, 20:339-346. 63. Arnold FH, Wintrode PL, Miyazaki K, Gershenson A: How enzymes adapt: lessons from directed evolution. Trends Biochem Sci 2001, 26:100-106. 64. Strausberg SL, Ruan B, Fisher KE, Alexander PA, Bryan PN: Directed coevolution of stability and catalytic activity in calcium-free subtilisin. Biochemistry 2005, 44:3272-3279. 65. Aharoni A, Gaidukov L, Khersonsky O, Mc QGS, Roodveldt C, Tawfik DS: The ‘evolvability’ of promiscuous protein functions. Nat Genet 2005, 37:73-76. 66. Khersonsky O, Roodveldt C, Tawfik DS: Enzyme promiscuity: evolutionary and mechanistic aspects. Curr Opin Chem Biol 2006, 10:498-508. 67. Collins CH, Leadbetter JR, Arnold FH: Dual selection enhances the signaling specificity of a variant of the quorum-sensing transcriptional activator LuxR. Nat Biotechnol 2006, 24:708-712. 68. Bershtein S, Segal M, Bekerman R, Tokuriki N, Tawfik DS: Robustness-epistasis link shapes the fitness landscape of a randomly drifting protein. Nature 2006, 444:929-932. 69. Lipovsek D, Antipov E, Armstrong KA, Olsen MJ, Klibanov AM, Tidor B, Wittrup KD: Selection of horseradish peroxidase variants with enhanced enantioselectivity by yeast surface display. Chemistry & Biology 2007, 14:1176-1185. Yeast display (that is highly versatile and applicable to numerous proteins) has been applied for the selections of large libraries (of either random mutations, or targeted residues) of the enzyme horseradish peroxidase. The enzyme displaying cells were sorted by FACS by the covalent conjugation of fluorescent products to the displaying cells (Figure 1b), and rounds of positive selection (with the desired enantiomer) and negative selection (against variants conjugating the undesired enantiomer) were applied towards the generation of stereoselective oxidases. 70. Bloom JD, Lu Z, Chen D, Raval A, Venturelli OS, Arnold FH: Evolution favors protein mutational robustness in sufficiently large populations. BMC Biol 2007, 5:29. 71. Ancel LW, Fontana W: Plasticity, evolvability, and modularity in RNA. J Exp Zool 2000, 288:242-283. 72. Bornberg-Bauer E, Chan HS: Modeling evolutionary landscapes: mutational stability, topology, and superfunnels in sequence space. Proc Natl Acad Sci U S A 1999, 96:10689-10694. Current Opinion in Chemical Biology 2008, 12:151–158 73. van Nimwegen E, Crutchfield JP, Huynen M: Neutral evolution of mutational robustness. Proc Natl Acad Sci U S A 1999, 96:97169720. 74. Jurgens C, Strom A, Wegener D, Hettwer S, Wilmanns M, Sterner R: Directed evolution of a (beta alpha)8-barrel enzyme to catalyze related reactions in two different metabolic pathways. Proc Natl Acad Sci U S A 2000, 97:9925-9930. 75. Bloom JD, Labthavikul ST, Otey CR, Arnold FH: Protein stability promotes evolvability. Proc Natl Acad Sci U S A 2006, 103:58695874. Increased stability endows proteins with increased tolerance to mutations. This work experimentally demonstrates that a protein’s capacity to evolve is also enhanced by increased stability. Mutants derived from a stabilized P450-BM3 variant were found to exhibit new or improved functions with higher likelihood. The link between protein stability, mutational robustness, and evolvability is further established by observations of the evolution of stability under high mutational loads (reference [70]), and the induction of back-to-consensus/ancestor mutations that boost the protein’s tolerance to mutations and evolvability (Bershtein S, et al., unpublished results). 76. Lehmann M, Kostrewa D, Wyss M, Brugger R, D’Arcy A, Pasamontes L, van Loon AP: From DNA sequence to improved functionality: using protein sequence comparisons to rapidly design a thermostable consensus phytase. Protein Eng 2000, 13:49-57. 77. Amin N, Liu AD, Ramer S, Aehle W, Meijer D, Metin M, Wong S, Gualfetti P, Schellenberger V: Construction of stabilized proteins by combinatorial consensus mutagenesis. Protein Eng Des Sel 2004, 17:787-793. 78. Polizzi KM, Bommarius AS, Broering JM, Chaparro-Riggers JF: Stability of biocatalysts. Curr Opin Chem Biol 2007, 11:220-225. 79. Camp A, Herman A, Loh E, Loeb LA: Genetics constraints on protein evolution. Crit Rev Biochem Mol Biol 2007, 42:313326. An interesting discussion about the impact imposed by mutations on protein fitness and stability and the importance of compensatory mutations in adaptive protein evolution. 80. Jurkowski TP, Anspach N, Kulishova L, Nellen W, Jeltsch A: The M.EcoRV DNA-(adenine N-6)-methyltransferase uses DNA bending for recognition of an expanded EcoDam recognition site. JBC 2007, 282:36942-36952. 81. Kelly BT, Griffiths AD: Selective gene amplification. PEDS 2007, 20:577-581. 82. Zheng Yu, Roberts RJ: Selection of restriction endonucleases using artificial cells. Nucleic Acids Res 2007, 35:e83. www.sciencedirect.com