Degenerate oligonucleotide gene shuffling (DOGS): a method for enhancing the frequency of recombination with family shuffling

Gene 271 (2001) 13±20

www.elsevier.com/locate/gene

Degenerate oligonucleotide gene shuf¯ing (DOGS): a method for enhancing the frequency of recombination with family shuf¯ing Moreland D. Gibbs a, K.M. Helena Nevalainen a, Peter L. Bergquist a,b,* a

Department of Biological Sciences, Macquarie University, Sydney, New South Wales 2109, Australia Department of Molecular Medicine, University of Auckland Medical School, Auckland, New Zealand

b

Received 22 January 2001; received in revised form 13 April 2001; accepted 27 April 2001 Received by A. Dugaiczyk

Abstract Improvement of the biochemical characteristics of enzymes has been aided by misincorporation mutagenesis and DNA shuf¯ing. Shuf¯ing techniques can be used on a collection of mutants of the same gene, or related families of genes can be shuf¯ed to produce mutants encoding chimeric gene products. One dif®culty with current shuf¯ing procedures is the predominance of unshuf¯ed (`parental') molecules in the pool of mutants. We describe a procedure for gene shuf¯ing using degenerate primers that allows control of the relative levels of recombination between the genes that are shuf¯ed and reduces the regeneration of unshuf¯ed parental genes. This procedure has the advantage of avoiding the use of endonucleases for gene fragmentation prior to shuf¯ing and allows the use of random mutagenesis of selected segments of the gene as part of the procedure. We illustrate the use of the technique with a diverse family of b-xylanase genes that possess widely different G 1 C contents. q 2001 Published by Elsevier Science B.V. All rights reserved Keywords: Gene shuf¯ing; Xylanase genes; Overlap extension PCR

1. Introduction Stemmer (1995) has discussed the most effective methods to search sequence space in vitro to yield the greatest diversity of protein variants. Until recently, the most popular methods of creating combinatorial libraries are recursive strategies that seek to evolve sequences by the addition of point mutations. For in vitro evolution, inclusion of recombinant PCR (gene shuf¯ing, Crameri et al., 1996; Stemmer, 1994a,b) offers practical and theoretical advantages over simple recursive mutagenesis methods. It will rapidly ®ne tune the mutational load in several parts of the protein by recombining point mutations and wild-type sequences. The technique (and its variations) have been used to enhance enzyme activity (Crameri et al., 1998; Joo et al., 1999), substrate speci®city (Zhang et al., 1997), enzyme stability Abbreviations: DOGS, degenerate oligonucleotide gene shuf¯ing. IPTG; isopropyl b-d-thiogalactoside; X-gal, 5-bromo-4-chloro-3-indolyl b-dgalactopyranoside * Corresponding author. Research Of®ce, Macquarie University, Sydney. New South Wales 2109, Australia. Tel.: 161-2-9850-8614; fax: 161-29850-8799. E-mail address: [email protected] (P.L. Bergquist).

(Giver et al., 1998; Zhao and Arnold, 1999), and the creation of new metabolic pathways (Schmidt-Dannert et al., 2000). Family shuf¯ing is usually achieved by fragmentation of the genes to be shuf¯ed followed by PCR. It relies on homologous recombination during the PCR reassembly step. Most methods require relatively high levels of sequence similarity between the genes to be shuf¯ed as `cross-over points' appear to occur in these regions. If sequence similarity is low between the input genes, the majority products tend to be the reassembled parental genes and extensive searches need to be carried out to ®nd chimeric recombinants (Schmidt-Dannert et al., 2000; Kikuchi et al., 1999). Kikuchi et al. (1999) have described a method for gene shuf¯ing that make use of unique restriction enzyme sites in the sequences of the parental molecules. The complete restriction enzyme digestion of parental genes ensured that subsequent overlap extension gave rise to hybrid genes at high frequencies. An entirely different procedure was proposed by Ostermeier et al. (1999) that allowed the preparation of combinatorial fusion libraries by progressive truncation of coding sequences of two parental sequences followed by ligation of the fragments and selection for enzyme activity. Either parent can

0378-1119/01/$ - see front matter q 2001 Published by Elsevier Science B.V. All rights reserved PII: S 0378-111 9(01)00506-6

14

M.D. Gibbs et al. / Gene 271 (2001) 13±20

be used to provide 5 0 sequence for the hybrid gene. This procedure, termed iterative truncation for the creation of hybrid enzymes (ITCHY), can accommodate recombination between genes with as little as 50% sequence similarity and was found to give a wider range of crossovers compared with standard gene shuf¯ing techniques. We recently isolated a gene coding for a thermophilic bxylanase that had superior performance in the bleaching of paper pulp (Morris et al., 1998), and the three-dimensional structure of the enzyme has been solved (McCarthy et al., 2000). We wished to investigate the possibility of obtaining mutant derivatives that had enhanced stability and an altered pH optimum. Experiments using error-prone PCR and misincorporation mutagenesis followed by gene shuf¯ing allowed the identi®cation of mutant genes that coded for a limited sample of the variations in sequence space but required extensive screening for their identi®cation. Gene shuf¯ing following DNAseI fragmentation of related genes (family shuf¯ing) overwhelmingly gave wild type parental sequences as the major products. After several trials of methods designed to reduce the background, we devised a technique that allows shuf¯ing of genes that differ widely in sequence similarity and G 1 C content and greatly reduces the regeneration of wild type genes. Furthermore, the primer extension conditions may be modi®ed to bias the resulting progeny genes towards any one (or more) of the parental input genes. We term this procedure Degenerate Oligonucleotide Gene Shuf¯ing (DOGS) and note its compatibility with other recursive point mutation techniques. 2. Materials and methods 2.1. Source of genes Family 11 xylanase genes were obtained from the following bacterial strains: Dictyoglomus thermophilum strain Rt46B.1 xynB (Morris et al., 1998); Clostridium stercorarium xynB (Sakka et al., 1993); Bacillus sp. strain V14 (Yang et al., 1995); Caldicellulosiruptor sp. strain Rt69B.1 xynD (Morris et al., 1999), Clostridium thermocellum xynV (Fernandes et al., 1999) and Streptomyces roseiscleroticus xyl3 (Elegir et al., 1994). Each gene was PCR ampli®ed using the respective gene-speci®c primers listed in Table 1. Four of these xylanase genes were from thermophiles and coded for enzymes that had high temperature optima and the Bacillus and Streptomyces xylanase genes coded for mesophilic proteins that performed well in pulp bleaching tests but had signi®cantly lower temperature optima than the enzymes from the thermophiles. 2.2. Degenerate-end complementary primer pairs for ef®cient PCR ampli®cation of gene segments and overlapextension of adjacent segments The nucleotide sequences of the genes were aligned and

degenerate primers were designed based on the conserved amino acid motifs found in all of the genes. The genes coding for the xylanases were divided into eight fragments on the basis of alignment of the conserved regions. Degenerate forward and reverse primers were designed which allowed ampli®cation of the DNA in the eight segments when combined, as appropriate, with the nested 5 0 and 3 0 common primers (Table 1). In some cases, the degenerate oligonucleotides were designed using an extension of the CODEHOP method (Rose et al., 1998; see Section 3 for a full description). These primers consisted of a non-degenerate central core ¯anked by degenerate ends of 6±7 nucleotides. All of the segments for each gene were ampli®ed with the degenerate oligonucleotides and individually gel-puri®ed from the PCR mixtures. PCR conditions for this step were: one cycle of 958C for 1 min; then 35 cycles 958C (denaturation), 30 s; annealing at 358C, 20 s; and extension at 728C, 40 s with a ®nal incubation at 728C for 5 min. Platinum Pfx polymerase (Life Technologies Pty Ltd, Victoria, Australia) was used for all PCRs.

Table 1 Oligonucleotide primers used for gene isolation and gene segment ampli®cation Gene speci®c primers DTF DTR CSF CSR CTF CTR BSF BSR RTF RTR SRF SRR

5 0 -GAAAACTGCAGTAGATGCAAACGTCTATAACACT 5 0 -GTTCTACTGGATCCTTAAGAAAAAGTATTTTGTG 5 0 -GAAAACTGCAGTAGATGCTCGCCGGGCGAATAAT 5 0 -GTTCTACTGGATCCTTATCTGATTTCATTCTTGT 5 0 -GAAAACTGCAGTAGATGCGCGCTGATGTGGTAAT 5 0 -GTTCTACTGGATCCTTAGTTGCCAACAGTAATTG 5 0 -GAAAACTGCAGTAGATGGCCCATGCGAGAACCAT 5 0 -GTTCTACTGGATCCTTAGTTGCCAATAAACAGCT 5 0 -GAAAACTGCAGTAGATGCAGGCAGCCATGACATT 5 0 -GTTCTACTGGATCCTTAAGTAAATGTATTCTGTG 5 0 -GAAAACTGCAGTAGATGCACGCCGCCACTACCAT 5 0 -GTTCTACTGGATCCTTAACCGCTGACCGTGATGT

Nested primers SHF SHR

5 0 -GAAAACTGCAGTAGATG 5 0 -GTTCTACTGGATCCTTA

Degenerate primers XINTF1C 5 0 -GGBTACDACTATGAACTATGGAARGA XINTR1C 5 0 -TCYTTCCATAGTTCATAGTHGTAVCC XINTF2C 5 0 -AAYATHRACAATGCATTATTCAGWAMAGG XINTR2C 5 0 -CCTKTWCTGAATAATGCATTGTYDATRTT XINTF3C 5 0 -GGNAAYTCCTATCTATGTATYTAYGG XINTR3C 5 0 -CCRTARATACATAGATAGGARTTNCC XINTF4B 5 0 -TGGGGHACCTGGCGTCCVMCNGG XINTR4B 5 0 -CCNGKBGGACGCCAGGTDCCCCA XINTF5 5 0 -ACCCGWGTWAATCAGCC XINTR5 5 0 -GGCTGATTWACWCGGGT XINTF6 5 0 -AARMGWACAAGYGGWAC XINTR6 5 0 -GTWCCRCTTGTWCKYTT XINTF7 5 0 -GAAGGWTAYCARAGCAG XINTR7 5 0 -GTWCCRCTTGTWCKYTT

M.D. Gibbs et al. / Gene 271 (2001) 13±20

2.3. Overlap extension PCR The segments of each gene were mixed in the appropriate ratio to give the desired level of chimerization. For example, using the six candidate genes G1±G6, where G1 is the D. thermophilum Rt46B.1 xynB gene, and deciding that this sequence should predominate in the shuf¯ed progeny, the pooled PCR segments for each gene were mixed in the ratio of 8.75 G1 to 1:1:1:1:1 other genes, to give chimeras with a predicted average of ®ve out of eight segments from Rt46B.1 xynB. Fifty to 100 ng of mixed segments were then used as templates for overlap extension (Ho et al., 1989) using the following conditions: one cycle of 958C for 1 min.; then 35 cycles 958C (denaturation), 30 s; annealing at 358C, 20 s; and extension at 728C, 40 s and a ®nal incubation at 728C for 5 min. The high ®delity Platinum Pfx polymerase was used for all overlap extension and PCR to decrease the frequency of PCR misincorporation mutations. A second reason for using a high ®delity polymerase is its lack of terminal transferase activity as addition of 5 0 nucleotides would interfere with template priming during the overlap extension procedure. 2.4. Chimera ampli®cation Full length chimeric genes were generated by using the overlap-extended products (50±100 ng) as a template for PCR using the common ¯anking nested 5 0 - and 3 0 - primers under the following conditions: one cycle of 958C for 1 min to activate the enzyme; then 20 cycles 958C (denaturation), 30 s; annealing at 508C, 20 s; extension at 728C, 40 s. and a ®nal incubation at 728C for 5 min using Platinum Pfx DNA polymerase. 2.5. Cloning of shuf¯ed products DOGS-generated PCR products were digested with the restriction enzymes BamHI and HindIII and ligated to pBSII KS- (Stratagene, San Diego, CA) that had been digested with the same restriction enzymes, and treated with Shrimp alkaline phosphatase (Roche Diagnostics Australia, NSW, Australia). The ligated DNA was used to transform E. coli strain DH5a to ampicillin resistance and spread onto plates containing the antibiotic, X-gal and IPTG according to the instructions provided by the vector manufacturer. Individual colonies were picked and patched in duplicate onto new plates and screened for the expression of xylanase activity by the Congo Red overlay method (Teather and Wood, 1982). 2.6. Assays for xylanase activity Conditions for determining temperature optima, thermostability, pH optima and the reducing sugar assay have been described previously (Morris et al., 1998; 1999). Whole cell extracts containing recombinant xylanase were prepared for enzyme assays using the protein extraction reagent BPER II

15

(Pierce Chemical Company, Rockford, IL). Xylanase activity was determined using the method of Lever (1973) with Birchwood xylan (Sigma-Aldrich, Sydney, Australia) as substrate. The standard assay reaction mixture consisted of 0.5% (w/v) xylan in 120 mM universal buffer (Britton and Robinson, 1931) pH 6.5, and enzyme to give a ®nal volume of 0.03 ml. The reaction mixture was incubated at 608C for 20 min. The reaction was carried out in 200 ml PCR-tubes using a hot-top PCR machine to prevent evaporation of the mixture. The effects of temperature and pH on the activity of the recombinant enzymes were determined as described by Morris et al. (1999). 3. Results 3.1. Rationale An overview of the DOGS technique is given in Fig. 1. We developed this procedure after ®nding a high frequency of parental non-shuf¯ed products were generated in our early experiments using the established technique with DNAseI fragmentation introduced by Stemmer (1994a). This observation made extensive screening necessary for the identi®cation of candidate chimeric genes, particularly when distantly related genes were used in family shuf¯ing. The restriction enzyme approach by Kikuchi et al. (1999) has the merits of simplicity as the complete restriction enzyme digestion of each parental gene lowers the frequency of regeneration of wild-type genes. However, this technique relies on there being regions of conserved homology adjacent to the restriction sites used in order for overlap extension to occur. We considered that it may be important to be able to control the input of parental genes and levels of chimerism generated by the shuf¯ing procedure, especially if mutants of one particular gene were of interest. We had observed the rarity with which chimeras arose when genes with limited sequence similarity were used as parents, and devised the DOGS technique to overcome this problem. For any library generated using the DOGS technique, the total number of possible chimeric combinations (C) is described by the formula C ˆ gS 2 g where g is the number of genes shuf¯ed, and (S) is the number of segments per gene. When one parental gene is used as the major input DNA (major parent), such that on average there are (p) major parent DNA segments per chimera, then the total number of possible chimeric (C) combinations is described by the formula Cˆ

S! …g 2 1†…S2p† p!…S 2 p†!

We provide below a description of the products from experiments entailing shuf¯ing of six xylanase genes

16

M.D. Gibbs et al. / Gene 271 (2001) 13±20

Fig. 1. An overview of DOGS technique.

M.D. Gibbs et al. / Gene 271 (2001) 13±20

using Dictyoglomus thermophilum Rt46B.1 xynB6 (Morris et al., 1998) as the major parent DNA. 3.2. Degenerate primer design The most commonly used strategy to isolate distantlyrelated sequences by PCR has been to design degenerate primers which bind to highly conserved regions of DNA sequence. The dif®culty with this method is that as the primer degeneracy increases to accommodate more divergent genes, the total number of primer molecules in a PCR that can correctly prime extension decreases. Non-speci®c ampli®cation may then occur because of the abundance of primers that do not participate in ampli®cation of the target gene, and so are available to prime non-speci®c synthesis, especially as low stringency annealing conditions are usually needed to detect mismatched templates. Rose et al. (1998) have described a strategy that overcomes problems of degenerate methods for primer design called Consensus-Degenerate Hybrid Oligonucleotide Primers (CODEHOP). CODEHOP primers consist of a relatively short 3 0 degenerate end and a 5 0 non-degenerate consensus clamp. Reducing the length of the 3 0 end to a minimum decreases the total number of individual primers in the degenerate primer pool. Hybridization of the 3 0 degenerate end with the target template is stabilized by the 5 0 non-degenerate consensus clamp, which allows higher annealing temperatures without increasing the degeneracy of the pool. Although potential mismatches may occur between the 5 0 consensus clamp of the primer and the target sequence during the initial PCR cycles, they are situated away from the 3 0 hydroxyl extension site, and so mismatches between the primer and the target are less disruptive to the priming of the polymerase extension reaction. Further ampli®cation of primed PCR products during subsequent rounds of primer hybridization and extension is enhanced by the sequence similarity of all primers in the

17

pool; this fact potentially allows utilization of all primers in the reaction. We describe here a modi®cation of the CODEHOP method that allows ef®cient ampli®cation of overlapping segments of related genes, and subsequent overlap-extension of adjacent segments from different genes resulting in the formation of chimeric gene fragments. The modi®cation of the CODEHOP technique entails the design of perfectly complementary pairs of primers. Each primer has a nondegenerate clamp ¯anked by both 5 0 and 3 0 prime degenerate ends, referred to here as complementary degenerate-end primers (CDE primers). As with the CODEHOP primers, the 3 0 degenerate end gives each CDE primer their templatebinding speci®city, while the non-degenerate region acts as a stabilizing clamp in subsequent rounds of the PCR. The 5 0 degenerate end is not required to contribute to the binding ef®ciency of the CDE primer during PCR. However, it plays a pivotal role in allowing ef®cient binding and subsequent overlap-extension of separate PCR products (gene segments) generated using, respectively, the forward or the reverse CDE primers. The non-degenerate clamp of CDE primers is generally based upon the corresponding coding sequence of one gene, designated the major parent gene, for shuf¯ing. This results in the formation of chimeric fragments which retain major parent gene sequence at the points of segment overlap. An example of the design strategy for making complementary oligonucleotide pairs suitable for the ampli®cation of gene segments from related genes, and for the subsequent overlap extension of segments to generate chimeric genes, is shown in Fig. 2. 3.3. Shuf¯ing with D. thermophilum Rt46B.1 xynB as the major parent Three DOGS gene shuf¯ing libraries were generated, each with differing ratios of input DNAs. Libraries R1, R2 and R3 were generated by mixing equal amounts of each of

Fig. 2. An example of CDE primers design. A conserved region of an alignment between the six xylanase genes used in this work is shown with the most conserved nucleotides shown in reverse text. The XINTF2 and XINTR2 CDE primers used in this study are aligned below with the non-degenerate clamp corresponding to the major parent gene D. thermophilum xynB also shown in reverse text.

18

M.D. Gibbs et al. / Gene 271 (2001) 13±20

the major parent gene segments with other gene segments in the ratios shown in Table 2. The ratios used were calculated to give on average 5/8, 6/8 and 7/8 major parent segments per gene, for the respective R1, R2 and R3 libraries. After shuf¯ing and PCR of input DNAs to give full length genes, chimeric fragments were digested with the appropriate restriction enzymes, directly ligated into the pBS KSvector and used to transform E. coli DH5a. A total of 100 colonies from each library were screened for the expression of xylanase activity (Morris et al., 1998). Seventy xylanasepositive transformants and 30 xylanase-negative transformants were identi®ed and the plasmid insert of each was sequenced and the resulting data compared to the parental sequences. It was possible to assign the origin of each segment in a recombinant from the sequence data, as shown in Fig. 3. Of the 100 genes sequenced, 75% were chimeric for library R1, while 54 and 52% were chimeric for libraries R2 and R3, respectively. The xylanase-positive chimeras are shown in Fig. 3. With these input ratios, we expected a dominance of parental Rt46B.1 xynB sequences from the input ratios. Some isolates that contained all parental Rt46B.1 segments were inactive in the plate test. These results can be explained by the occurrence of point mutations introduced by the degeneracy in the primers used to amplify each segment. Also, in six of the 30 xylanase-negative isolates, frameshifts were observed at a segment boundary, presumably introduced during the overlap extension of adjacent segments and resulted in truncated open reading frames in all cases observed. A further experiment was carried out in which the input ratios for all segments were 1:1:1:1:1:1. A total of twenty chimeras were sequenced and on average only one segment per chimera was derived from D. thermophilum xynB6 (data not shown). This result indicated that the xynB6-derived degenerate clamp of CDE primers did not bias recombination toward reformation of the parental gene. 3.4. Xylanase assays on shuf¯ed gene products All xylanase-positive isolates were assayed for pH opti-

mum and temperature optimum. A broad range of temperature optima were observed, ranging from 55±808C (see Fig. 3.) No detectable variation in pH optimum was observed with all isolates measured having a pH optimum of 6.5, the pH optimum of the wild type D. thermophilum xynB. 3.5. Correlations between enzyme activity and gene family segment content Such correlations are best viewed by a comparison of the segment composition of a particular shuf¯ed recombinant and its pH pro®le from 3 to 10 and its activity at temperatures from 42 and 928C at pH 6.5. The three-dimensional structure of the D. thermophilum Rt46B.1 XynB has been Ê . Examination of the structural features of the solved at 1.8 A enzyme shows that virtually every segment has some portion sitting in the catalytic groove of the enzyme (see Fig. 1; McCarthy et al., 2000). The segments which contribute least to the catalytic cleft are 1, 5, 6 and 8 (see Fig. 1; 2, this paper), and these segments are the ones most frequently found to be shuf¯ed in the active chimeric genes with 74% of the shuf¯ed segments occurring at these positions. The two glutamic acid residues directly implicated in the catalytic mechanism lie, respectively, on segment 4, and on the boundary (degenerate primer binding position) of segments 7 and 8. 4. Discussion The DOGS procedure that we have described demonstrates that it is possible to shuf¯e members of a gene family that are not particularly closely related and still generate chimeric molecules at a high enough frequency that comprehensive and time consuming screens are not necessary. The use of CDE primers has allowed the reliable PCR ampli®cation and shuf¯ing of equivalent gene segments from a diverse range of genes with low overall sequence homology. It is unclear whether the CDE primers used in this study

Table 2 Levels of chimerization in DOGS libraries Library Input ratio a

Predicted chimera Xylanase frequency b phenotype

R1

8.75:1:1:1:1:1 5/8

R2

15:1:1:1:1:1 6/8

R3

35:1:1:1:1:1 7/8

a b c d

1 2 1 2 1 2

Plasmid inserts with observed major parent segments per gene 8/8

7/8

6/8

5/8

7 1 8 9 15 nd d

4 6 7 6 13 nd

6 5 4 2 3 nd

3 0 0 1 0 nd

Ratio of amount of major parent (D. thermophilum xynB) DNA to DNA from other genes. Predicted average number of major parent segments out of eight segments per xylanase gene. Observed average number of major parent segments out of eight segments per xylanase gene. nd, not determined.

Total genes Observed chimerism c

Percentage chimeric genes

20 12 19 18 31 ±

6.72/8

75

7.24/8

54

7.39/8

52

M.D. Gibbs et al. / Gene 271 (2001) 13±20

19

Fig. 3. A diagrammatic representation of shuf¯ing results with D. thermophilum xynB as the major parent gene at a segment input ratio of 8.75:1:1:1:1:1 (library R1, pre®xed with 1.x), 15:1:1:1:1:1 (library R2, pre®xed with 2.x) and 35:1:1:1:1:1 (library R3, pre®xed with 3.x). The segments of each gene are coloured according to their parental origin. A visual approximation of halo size was used to estimate likely levels of enzyme activity in subsequent assays.

represent the maximum degree of degeneracy tolerated in order for the DOGS technique to work reliably. The complexities of optimisation of primer/template or template/template annealing make the PCR a very empirical technique. Further studies are required to determine whether CDE oligonucleotides with higher degeneracy could be used reliably. The recombination frequencies can be controlled by altering the segment input ratios so that shuf¯ing of particular fragments can be enhanced or attenuated as required. Accordingly, the procedure allows domain-swapping experiments to be conducted with relative ease, replacing older methods that rely on suitable restriction enzyme sites. It is evident that PCR-induced misincorporation or errorprone mutagenesis can be incorporated as part of the procedure to introduce even more diversity into the products. In this respect, the DOGS method lends itself to random mutagenesis of individual segments to assist in ®ne-tuning of the encoded gene product. In this case, the CDE primers would be modi®ed so that no degeneracy was available except for the segment to be mutagenized. This would give the investigator control over the extent and nature of mutagenesis of a particular segment by introducing a DNA polymerase without proof-reading activity at the appropriate stage in

the procedure. In addition, it is clear from the design of the degenerate primers and the results reported here that altered nucleotide sequences will be generated even using a high ®delity DNA polymerase because of the mismatches designed into the degenerate primers. Even greater misincorporation mutagenesis can be generated by employing a polymerase without proof-reading activity in the ampli®cation and primer extension steps. One of the objectives that led to our development of gene shuf¯ing methodologies was a desire to increase the pH optimum of XynB from D. thermophilum Rt46B.1. Extensive biochemical studies have been carried out on the Bacillus circulans xylanase (also Family 11) and this work suggested that altering the pH optimum would be a dif®cult task. The B. circulans xylanase has been shown to have two glutamic acid residues at the active site, Glu78 and Glu172. These residues are conserved in the entire family 11 xylanases and are found to point into the active-site cleft. This cleft spans the whole width of the molecule and appears to be suf®ciently long to bind ®ve or six xylose residues. Based on site-directed mutagenesis and crystallographic analysis of the active site residues of the B. circulans xylanase, it has been proposed that Glu78 and Glu172 act as a nucleophile and an acid-catalyst, respectively (Wakarchuk et al.,

20

M.D. Gibbs et al. / Gene 271 (2001) 13±20

1994a,b; Miao et al., 1994; Davoodi et al., 1995). McIntosh et al. (1996) describe the pKa of the acid/base catalyst (Glu172) of the Bacillis circulans family 11 xylanase and the complex situation likely to be encountered in trying to change the pH optimum of a family 11 xylanase. Their results suggest that a change in the pH optimum of XynB, may require quite elaborate structural changes to the enzyme which might not be compatible with high temperature stability. In addition, acid residues with very high pKa's like Glu172 are inherently unstable and to increase the pKa any further may destabilize the enzyme. Although no recombinants with altered pH optima were found in the mutants screened so far, the small sample of mutants examined probably precluded their identi®cation and we intend to report further after a more extensive examination. The experiments described have used only a single round of chimera formation. Clearly, more diversity can be introduced to allow exploration of the sequence space by additional rounds of DOGS. Further genes could also be easily added to the procedure to increase chimera diversity. The procedure lends itself to combination with other gene shuf¯ing and combinatorial mutagenesis techniques for the generation of novel proteins with modi®ed characteristics. Acknowledgements This research was supported by grants from the Australian Research Council and the Public Good Science Fund (New Zealand). References Britton, H.T.S., Robinson, R.A., 1931. Universal buffer solutions and the dissociation constant of veronal. J. Chem. Soc. 1, 1456±1462. Crameri, A., Raillard, S-A., Bermudez, E., Stemmer, W.P.C., 1998. DNA shuf¯ing of a family of genes from diverse species accelerates directed evolution. Nature 391, 288±291. Crameri, A., Whitehorn, E.A., Tate, E., Stemmer, W.P.C., 1996. Improved green ¯uorescent protein by molecular evolution using DNA shuf¯ing. Nat. Biotechnol. 14, 315±319. Davoodi, J., Wakarchuk, W.W., Campbell, R.L., Carey, P.R., Surewicz, W.K., 1995. Abnormally high pKa of an active-site glutamic acid residue in Bacillus circulans xylanase. Eur. J. Biochem. 232, 839±843. Elegir, G., Szakacs, G., Jeffries, T.W., 1994. Puri®cation, characterization and substrate speci®city of multiple xylanases from Streptomyces sp strain B-12-2. Appl. Environ. Microbiol. 60, 2609±2615. Fernandes, A.C., Fontes, C.M., Gilbert, H.J., Hazlewood, G.P., Fernandes, T.H., Ferreira, L.M., 1999. Homologous xylanases from Clostridium thermocellum: evidence for bi-functional activity, synergism between xylanase catalytic modules and the presence of xylan-binding domains in enzyme complexes. Biochem. J. 342, 105±110. Giver, L., Gershenson, A., Freskgard, P-O., Arnold, F.H., 1998. Directed evolution of a thermostable esterase. Proc. Natl. Acad. Sci. USA 95, 12809±12813. Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K., Pease, L.R., 1989. Sitedirected mutagenesis by overlap extension using the polymerase chain reaction. Gene 77, 51±59.

Joo, H., Lin, Z., Arnold, F.H., 1999. Laboratory evolution of peroxidemediated cytochrome P450 hydroxylation. Nature 399, 670±673. Kikuchi, M., Ohnishi, K., Harayama, S., 1999. Novel family shuf¯ing methods for the in vitro evolution of enzymes. Gene 236, 159±167. Lever, M., 1973. Colorimetric and ¯uorometric carbohydrate determination with p-hydroxybenzoic acid hydrazide. Biochem. Med. 7, 274±281. McCarthy, A.A., Morris, D.D., Bergquist, P.L., Baker, E.N., 2000. Structure of XynB, a highly thermostable b-1,4-xylanase from Dictyoglomus thermophilum Rt46B.1, at 1.8 A resolution. Acta Crystallogr. D. Biol. Crystallogr. 56, 1367±1375. McIntosh, L.P., Hand, G., Johnson, P.E., Joshi, M.D., Korner, M., Plesniak, L.A., Ziser, L., Wakarchuk, W.W., Withers, S.G., 1996. The pKa of the general acid/base carboxyl group of a glycosidase cycles during catalysis: A 13C-NMR study of Bacillus circulans xylanase. Biochemistry 335, 9958±9966. Miao, S., Ziser, L., Aebersold, R., Withers, S.G., 1994. Identi®cation of glutamic acid 78 as the active site nucleophile in Bacillus subtilis xylanase using electrospray tandem mass spectroscopy. Biochemistry 33, 7027±7032. Morris, D.D., Gibbs, D.D., Chin, C.W., Koh, M.H., Wong, K.K.Y., Allison, R.W., Nelson, P.J., Bergquist, P.L., 1998. Cloning of the xynB gene from Dictyoglomus thermophilum strain Rt46B.1 and action of the gene-product on kraft pulp. Appl. Environ. Microbiol. 64, 1759±1765. Morris, D.D., Gibbs, M.D., Ford, M., Thomas, J., Bergquist, P.L., 1999. Family 10 and 11 xylanase genes from Caldicellulosiruptor isolate Rt69B.1. Extremophiles 3, 103±111. Ostermeier, M., Shim, J.H., Benkovic, S.J., 1999. A combinatorial approach to hybrid enzymes independent of DNA homology. Nat. Biotechnol. 17, 1205±1209. Rose, T.M., Schultz, E.R., Henikoff, J.G., Pietrokovski, S., McCallum, C.M., Henikoff, S., 1998. Consensus-degenerate hybrid oligonucleotide primers for ampli®cation of distantly related sequences. Nucleic Acids Res. 26, 1628±1635. Sakka, K., Kojima, Y., Kondo, T., Karita, S., Ohmiya, K., Shimada, K., 1993. Nucleotide sequence of the Clostridium stercorarium xynA gene encoding xylanase A: identi®cation of catalytic and cellulose binding domains. Biosci. Biotechnol. Biochem. 57, 273±277. Schmidt-Dannert, C., Umeno, D., Arnold, F.H., 2000. Molecular breeding of carotenoid biosynthetic pathways. Nat. Biotechnol. 18, 750±753. Stemmer, W.P.C., 1994a. DNA shuf¯ing by random fragmentation and reassembly: in vitro recombination for molecular evolution. Proc. Natl. Acad. USA 91, 10747±10751. Stemmer, W.P.C., 1994b. Rapid evolution of a protein in vitro by DNA shuf¯ing. Nature 370, 389±391. Stemmer, W.P.C., 1995. Searching sequence space. Bio/Technol. 13, 549± 553. Teather, R.M., Wood, P.J., 1982. Use of Congo Red polysaccharide interaction in enumeration and characterisation of cellulolytic bacteria from bovine rumen. Appl. Environ. Microbiol. 43, 777±780. Wakarchuk, W.W., Campbell, R.L., Sung, W.L., Davoodi, J., Yaguchi, M., 1994a. Mutational and crystallographic analyses of the active site residues of the Bacillus circulans xylanase. Protein Prot. Science Sci. 3, 467±475. Wakarchuk, W.W., Sung, W.L., Campbell, R.L., Watson, D.C., Yaguchi, M., 1994b. Thermostabilization of the Bacillus circulans xylanase by the introduction of disul®de bonds. Protein Prot. Engineering Eng. 7, 1379±1386. Yang, V.W., Zhuang, Z., Elegir, G., Jeffries, T.W., 1995. Alkaline-active xylanase produced by an alkaliphilic Bacillus sp isolated from kraft pulp. J. Industrial Ind. Microbiol. 15, 434±441. Zhang, J-H., Dawes, G., Stemmer, W.P.C., 1997. Directed evolution of a fucosidase from a galactosidase by DNA shuf¯ing and screening. Proc. Natl. Acad. Sci. USA 94, 4504±4509. Zhao, H., Arnold, F.H., 1999. Directed evolution converts subtilisin E into a functional equivalent of thermitase. Protein Eng. 12, 47±53.