Identification of Amino Acid Residues Responsible for the ...

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Sept. 2009, p. 5592–5599 0099-2240/09/$08.00⫹0 doi:10.1128/AEM.00301-09 Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Vol. 75, No. 17

Identification of Amino Acid Residues Responsible for the Enantioselectivity and Amide Formation Capacity of the Arylacetonitrilase from Pseudomonas fluorescens EBC191䌤† Christoph Kiziak‡ and Andreas Stolz* Institut fu ¨r Mikrobiologie, Universita ¨t Stuttgart, Allmandring 31, 70569 Stuttgart, Germany Received 6 February 2009/Accepted 27 June 2009

The nitrilase from Pseudomonas fluorescens EBC191 converted (R,S)-mandelonitrile with a low enantioselectivity to (R)-mandelic acid and (S)-mandeloamide in a ratio of about 4:1. In contrast, the same substrate was hydrolyzed by the homologous nitrilase from Alcaligenes faecalis ATCC 8750 almost exclusively to (R)-mandelic acid. A chimeric enzyme between both nitrilases was constructed, which represented in total 16 amino acid exchanges in the central part of the nitrilase from P. fluorescens EBC191. The chimeric enzyme clearly resembled the nitrilase from A. faecalis ATCC 8750 in its turnover characteristics for (R,S)-mandelonitrile and (R,S)-2-phenylpropionitrile (2-PPN) and demonstrated an even higher enantioselectivity for the formation of (R)-mandelic acid than the nitrilase from A. faecalis. An alanine residue (Ala165) in direct proximity to the catalytically active cysteine residue was replaced in the nitrilase from P. fluorescens by a tryptophan residue (as found in the nitrilase from A. faecalis ATCC 8750 and most other bacterial nitrilases) and several other amino acid residues. Those enzyme variants that possessed a larger substituent in position 165 (tryptophan, phenylalanine, tyrosine, or histidine) converted racemic mandelonitrile and 2-PPN to increased amounts of the R enantiomers of the corresponding acids. The enzyme variant Ala165His showed a significantly increased relative activity for mandelonitrile (compared to 2-PPN), and the opposite was found for the enzyme variants carrying aromatic residues in the relevant position. The mutant forms carrying an aromatic substituent in position 165 generally formed significantly reduced amounts of mandeloamide from mandelonitrile. The important effect of the corresponding amino acid residue on the reaction specificity and enantiospecificity of arylacetonitrilases was confirmed by the construction of a Trp164Ala variant of the nitrilase from A. faecalis ATCC 8750. This point mutation converted the highly R-specific nitrilase into an enzyme that converted (R,S)-mandelonitrile preferentially to (S)-mandeloamide. boxylic acids and thus in principle allow the production of the enantiomers of ␣-amino-, ␣-hydroxy-, and ␣-methylcarboxylic acids (1, 3, 10, 34). This trait has been used for the industrial production of (substituted) (R)-mandelic acid(s) from racemic (substituted) mandelonitrile(s) by dynamic kinetic resolution processes using different microorganisms (often strains of Alcaligenes faecalis) (19, 34; M. Ress-Lo ¨schke, T. Friedrich, B. Hauer, and R. Mattes, 1998, DE19848129A1, German Patent Office). An enantioselective nitrilase from A. faecalis ATCC 8750 has been purified and characterized, and the encoding gene has been cloned (4, 11, 26, 33). In previous work by our group, a different arylacetonitrilase was obtained from Pseudomonas fluorescens EBC191 (18). This enzyme converted various phenylacetonitriles (e.g., 2-PPN, Oacetoxymandelonitrile, or mandelonitrile), and also aliphatic 2-acetoxynitriles, with moderate enantioselectivities into the corresponding ␣-substituted carboxylic acids. Furthermore, with some substrates, significant amounts of the corresponding amides were also formed (5, 8, 12, 21, 27). The gene encoding the nitrilase from P. fluorescens EBC191 was recently cloned, and it was found that the nitrilases from P. fluorescens EBC191 and A. faecalis ATCC 8750 are clearly homologous to each other (12). Nevertheless, the two enzymes differ significantly in their catalytic abilities. Thus, the enzyme from A. faecalis ATCC 8750 converts racemic mandelonitrile to (R)-mandelic acid with a high enantioselectivity and forms almost no mandeloamide as a side product. In contrast, the

Nitrilases hydrolyze organic nitriles (R-C'N) to the corresponding carboxylic acids and ammonia. These enzymes have been isolated from various sources, such as bacteria, fungi, and plants. Commercially, they are a very interesting group of enzymes, because nitriles are important intermediates in the chemical industry and several biotransformations have been described that utilize the chemo-, regio-, or enantioselectivity of nitrilases (2, 6, 16, 20, 22, 29). There is an informal classification that groups nitrilases according to their substrate specificities into “benzonitrilases,” “aliphatic nitrilases,” and “arylacetonitrilases” (17, 23). The arylacetonitrilases convert substrates, such as phenylacetonitrile and ␣-substituted arylacetonitriles (e.g., 2-phenylpropionitrile [2-PPN], mandelonitrile [2-hydroxyphenylacetonitrile], or phenylglycinonitrile [2-aminophenylacetonitrile]). This group of nitrilases is especially interesting for applications in biotechnology because these enzymes can enantioselectively hydrolyze ␣-substituted racemic nitriles to optically active car-

* Corresponding author. Mailing address: Institut fu ¨r Mikrobiologie, Universita¨t Stuttgart, Allmandring 31, 70569 Stuttgart, Germany. Phone: 49 (0)711 68565489. Fax: 49 (0)711 68565725. E-mail: andreas [email protected]. ‡ Present address: Lonza AG, LBPSi Microbial Research and Development, CH-3930 Visp, Switzerland. † Supplemental material for this article may be found at http://aem .asm.org/. 䌤 Published ahead of print on 6 July 2009. 5592

VOL. 75, 2009

P. FLUORESCENS EBC191 ARYLACETONITRILASE

enzyme from P. fluorescens demonstrates only a low degree of enantioselectivity for the formation of (R)-mandelic acid and forms a large amount of mandeloamide (about 16% of the totally converted mandelonitrile). We are therefore currently trying to investigate the molecular basis for these differences in order to improve the substrate specificity and enantiospecificity of nitrilases. In a previous study, we analyzed the effects of various carboxy-terminal mutations on the nitrilase of P. fluorescens EBC191. These experiments showed that deletions of 47 to 67 amino acids from the carboxy terminus of the nitrilase resulted in variant forms that demonstrated, with mandelonitrile and 2-PPN as substrates, increased amide formation and increased formation of the R acids associated with lower specific activities. Although these carboxy-terminal mutants showed increased enantioselectivity for the formation of (R)mandelic acid, the observed enantioselectivities were still much lower than those observed with the nitrilase from A. faecalis ATCC 8750 and were also associated with increased amide formation (13). Therefore, in the present study, additional mutants were generated in order to analyze the effects of amino acid exchanges close to the catalytic center of the nitrilase.

5593

harvested by centrifugation in the late exponential or early stationary growth phase. The cells were then suspended in Tris-HCl buffer (pH 7.5) to optical densities at 600 nm of 0.2 to 4.0, and the substrates (10 mM) were added. The nitrilase activity of resting cells was routinely determined in reaction mixtures (1 ml each) containing 50 ␮mol Tris-HCl buffer (pH 7.5), 10 ␮mol nitrile, and an appropriate number of cells. All nitrile stock solutions (200 mM each) were prepared in methanol. The reaction mixtures were incubated at 30°C in a thermomixer at 1,100 rpm. After different time intervals, samples (200 ␮l each) were taken, and the reactions were stopped by the addition of 1 M HCl (20 ␮l). The samples were centrifuged at 15,000 ⫻ g for 10 min, and the supernatants were analyzed using high-pressure liquid chromatography. Protein determination and sodium dodecyl sulfate-polyacrylamide gel electrophoresis were performed as described previously (13). Analytical methods. 2-PPN, mandelonitrile, and their corresponding amides and acids were analyzed by high-pressure liquid chromatography. The achiral analysis was performed as previously described (13) using a Lichrospher RP18 column with methanol (50% [vol/vol]) and H3PO4 (0.1% [vol/vol]) in H2O as the mobile phase. Separation of the enantiomers of mandelic acid, 2-phenylpropionic acid (2PPA), and the corresponding amides was achieved on a Chiral-HSA column (ChromTech AB, Ha¨gersten, Sweden). The mobile phases consisted of sodium phosphate buffer (100 mM, pH 7.0) containing 4.5% (vol/vol) acetonitrile or sodium phosphate buffer (10 mM, pH 6.0) plus 0.5% (vol/vol) 2-propanol. Chemicals. The sources of all chemicals have been described previously (13).

RESULTS MATERIALS AND METHODS Bacterial strains and culture conditions. Plasmid pIK9, which encodes a His-tagged variant of the nitrilase from P. fluorescens EBC191 (GenBank no. AAW79573), was used for the construction of all enzyme variants (12). The nitrilase gene from A. faecalis ATCC 8750 was obtained as described previously (13). All cloning experiments and plasmid preparations were carried out in Escherichia coli JM109 (32). E. coli strains were grown at 37°C in 2⫻ YT liquid medium or on 2⫻ YT agar plates (28) supplemented with 100 ␮g/ml ampicillin. Construction of mutants and chimeric enzymes. The chimeric nitrilase variant Nit(pap) was constructed via gene splicing by overlap extension (SOE) (9). In the first step, the chimeric enzyme Nit(ap1) was constructed. First, the nitrilase genes were amplified from plasmids pIK9 (encoding NitP) and pIK7 (encoding NitA) using primers S1658 plus S1319 and S1242 plus S1623, respectively. (The sequences of all relevant primers are given in the supplemental material.) This resulted in the amplification of a 474-bp fragment from the 3⬘ region of nitP and a fragment of 573 bp from the 5⬘ end of nitA. The PCR products were purified by agarose gel electrophoresis, mixed, and subjected to another round of PCR using the primers S1242 and S1319, which specifically bound to the 3⬘ and 5⬘ termini of the nitrilase genes. The amplified DNA fragment with the correct size was purified by gel electrophoresis, cleaved with NdeI and BamHI, and ligated into the vector pJOE2775, which was previously also cut with the same restriction enzymes. This construct (pCK1) was then used to transform E. coli JM109. The construct encoded a chimeric nitrilase [Nit(ap1)] in which the first 191 amino acids were derived from A. faecalis and the carboxy-terminal part was from P. fluorescens. In the next step, a part of nit(ap1) was amplified by using the primers Chimpa7-fwd and s2949_2-rev (see the supplemental material). This resulted in the amplification of a DNA fragment from nit(ap1) that encoded the “middle part” of Nit(ap1) (which was derived from NitA) and the carboxy-terminal part (which was derived from NitP) and corresponded to base pairs 406 to 1047 of nit(ap1). Finally, the PCR product was cut with BamHI and inserted in the EcoRV/ BamHI-cut plasmid pIK9 (encoding the N-terminal part of NitP). The amino acid exchanges in positions 48 and 164 of NitP were also introduced by SOE. The “general flanking 5⬘ and 3⬘ primers” S3182_2 and S2949_2-rev were used in combination with the “specific primers” E48D-fwd and E48D-rev, E48Qfwd and E48Q-rev, E48S-fwd and E48S-rev, and C164A-fwd and C164A-rev. The same strategy was also used for the introduction of site-specific mutations in positions 165, 166, and 167 of NitP and position 164 of NitA. (For the sequences of the oligonucleotides used, see the supplemental material.) DNA preparation, DNA manipulation, transformation, PCR, DNA sequencing, database searches, and sequence alignments were performed as described previously (13). Enzyme assays. The induction of the nitrilase was achieved by the addition of rhamnose (0.2% [wt/vol]) to the dYT growth medium (12). The cells were

Confirmation of the proposed catalytic triad. It was previously proposed that nitrilases contain a catalytic triad that is composed of a cysteine, a glutamate, and a lysine residue and that during catalysis the cysteine residue forms an interim covalent adduct with the carbon atom originating from the nitrile group (14, 15, 31). Previously, detailed multiple-sequence alignments were performed with the nitrilase from P. fluorescens EBC191 (NitP) (13). These alignments suggested that the relevant amino acid residues were also conserved in this nitrilase and occupied the positions Glu48, Lys130, and Cys164 (13). The position of the catalytic center was verified by the construction of the mutant Cys164Ala and the subsequent demonstration that this mutant was devoid of nitrilase activity with the substrate 2-PPN. Subsequently, the ability of the glutamate residue in position 48 to be replaced by an aspartate or a glutamine residue was tested, but these mutants were also completely inactive. Construction of a chimeric enzyme that contained the central part of a highly enantioselective arylacetonitrilase embedded in the nitrilase of P. fluorescens EBC191. It was previously demonstrated that NitP showed only weak enantioselectivity for the formation of (R)-mandelic acid from racemic mandelonitrile (12). In contrast, an arylacetonitrilase from A. faecalis ATCC 8750 (NitA) that demonstrated 64% sequence homology converted the same substrate to almost pure (R)-mandelic acid (33, 34). Chimeric enzymes that combined the aminoterminal part of NitP with up to 79 amino acid residues from the C-terminal part of NitA had already been constructed. These chimeric enzymes demonstrated only a low degree of enantioselectivity during the conversion of mandelonitrile and thus clearly resembled NitP (13). Therefore, using SOE, a chimeric enzyme [Nit(pap)] was constructed which contained the amino-terminal region of NitP (amino acid residues 1 to 138), a central part of NitA (amino acid residues 139 to 183), and a carboxy-terminal region also originating from NitP (amino acid residues 184 to 350) (9) (see Materials and Methods). Nit(pap)

5594

KIZIAK AND STOLZ

APPL. ENVIRON. MICROBIOL.

FIG. 1. Conversion of 2-PPN by resting cells of E. coli JM109 synthesizing the nitrilases from P. fluorescens EBC191 (NitP) and A. faecalis ATCC 8750 (NitA) or different nitrilase variants. The respective strains of E. coli were grown and induced as described in Materials and Methods. The cells were harvested by centrifugation and resuspended to different optical densities in 1 ml Tris-HCl (50 mM; pH 7.5) in order to obtain similar substrate turnover rates. The cell suspensions were incubated for 10 min at 30°C in a thermoshaker. The reactions were started by the addition of 53 ␮l of 200 mM 2-PPN (in methanol). Aliquots (100 ␮l each) were taken at different time intervals, and the reactions were terminated by the addition of 100 ␮l of 0.05 M HCl. The concentrations and enantiomeric compositions of 2-PPA (A, D, and G), the relative proportions of amide formed (B, E, and H), and the enantiomeric compositions of 2-PPAA (C, E, and I) are presented. The following symbols are used for the different nitrilases and their variants: left column NitP (ⴱ), NitA ( ), and Nit(pap) ( ); middle column, NitP(Ala165Phe) (Œ), NitP(Ala165Tyr) (}), NitP(Ala165Trp) (F), NitP(Ala165His) (f), NitP(Ala165Gly) (‚), NitP(Ala165Gln) (〫), NitP(Ala165Arg) (E), and NitP (ⴱ); right column, NitP(Ile168Ala⫹Gln169Ala) ( ), NitP(Ile168Leu⫹Gln169Ser) ( ), and NitP (ⴱ).

differed from NitP in 16 amino acid residues, which were all located close to the catalytically active cysteine residue. Conversion of mandelonitrile and 2-PPN by the chimeric enzyme variant and the wild-type enzymes. The three recombinant E. coli clones that synthesized NitA, NitP, or Nit(pap) were induced by the addition of rhamnose, and resting cells were incubated with 2-PPN or mandelonitrile. The chimeric enzyme was enzymatically active and converted 2-PPN and mandelonitrile, although with lower specific activities. The observed activities (normalized to a defined optical density of the recombinant resting cells) for the cells expressing NitP, NitA,

or Nit(pap) with mandelonitrile were 100:21:3, and with 2-PPN they were 100:13:3, respectively. The turnover experiments with 2-PPN confirmed that NitP converted 2-PPN to 2-PPA with a slight preference for the formation of the S enantiomer (12, 13). In contrast, NitA demonstrated the opposite enantiopreference, although it showed a rather similar degree of enantioselectivity (Fig. 1A). Surprisingly, it was found that, among the investigated nitrilases, the chimeric enzyme demonstrated the highest degree of enantioselectivity, and an enantiomeric excess (ee) for the formation of (R)-2-PPA of 94% at about 30% conversion was calculated (Fig. 1A).

VOL. 75, 2009


5595

FIG. 2. Conversion of mandelonitrile by the nitrilases from P. fluorescens EBC191 (NitP) and A. faecalis ATCC 8750 (NitA) or different nitrilase variants. Cells were cultivated, and the resting-cell experiments were performed as described in the legend to Fig. 1. Shown are the results for the formation of mandelic acid (MA; A, D, and G), the relative proportions of mandeloamide (MAA) formed (B, E, and H), and the enantiomeric excess of mandeloamide (C, F, and I), as described in Fig. 1. For the symbols used, see the legend to Fig. 1.

All three nitrilases converted only less than 1% of 2-PPN to 2-phenylpropionamide (2-PPAA) (Fig. 1B). Although a chiral analysis of the rather small amounts of 2-PPAA formed was difficult, it could be demonstrated that the nitrilase from A. faecalis, and also the chimeric enzyme, preferentially formed the R enantiomer of 2-PPAA and thus had a stereopreference opposite to that of the enzyme from P. fluorescens (Fig. 1C). The results obtained with mandelonitrile confirmed the results determined with 2-PPN. Also, with mandelonitrile, the chimeric enzyme demonstrated the highest degree of enantioselectivity, and an ee value of about 99% was calculated, which in a direct comparison was higher than the ee value of 96% calculated for NitA (Fig. 2A). In contrast, for NitP, an ee value of only 31% was determined, which confirmed earlier reports (5, 12, 13). The greater similarity between NitA and Nit(pap) was also confirmed by the amounts of mandeloamide formed. Thus, these two nitrilases formed only 0.7% and 0.5% amide from mandelonitrile, respectively (relative to the

amount of mandelic acid formed). In contrast, NitP formed about 16% amide during the conversion of mandelonitrile (Fig, 2B). In contrast to the two wild-type nitrilases, the chimeric enzyme preferentially formed the R enantiomer of mandeloamide (Fig. 2C). This, together with the higher enantioselectivity for the formation of (R)-mandelic acid, suggested that the chimeric enzyme in general demonstrated higher R selectivity. Construction of various mutants in close proximity to the catalytic center. The chimeric nitrilase Nit(pap) differed in 16 amino acid residues from NitP. In the subsequent experiments, we attempted to confine the number of amino acid residues that were potentially involved in the enantioselectivity of arylacetonitrilases. First, amino acid residues in close proximity to the catalytic center were analyzed by sequence alignments, and it was found that NitP differed from almost all other nitrilases (and also cyanide dihydratases) by the presence of an alanine residue (Ala165) directly adjacent to the catalytically active

5596

KIZIAK AND STOLZ


TABLE 1. Comparison of different nitrilase variants carrying different amino acid exchanges in position 165 with the wild-type nitrilases from P. fluorescens EBC191 and A. faecalis ATCC 8750 and the chimeric nitrilase variant Nit(pap) Value Parametere NitP

Formation of 2-PPA (%)a 100 65 (S) ee 2-PPAb (preferentially formed enantiomer) 2-PPAA (%)b,f 0.2 7 (R) ee 2-PPAAb (preferentially formed enantiomer) Relative activity MN/2-PPN 7.3 Formation of MA (%)a 100 31 (R) ee MAb (preferentially formed enantiomer) b,g MAA (%) 19 84 (S) ee MAAb (preferentially formed enantiomer)

NitP (A165G)

NitP (A165F)

NitP (A165Y)

NitP (A165W)

NitP (A165H)

NitP (A165E)

NitP (A165R)

Nit(pap)

NitA

NitA (W164A)

102 69 (S)

26 91 (R)

16 89 (R)

19 69 (R)

21 82 (R)

0.3 27c (R)

3 37 (R)

3 94 (R)

13 64 (R)

7 91 (S)

4.3 88 (S)

0.1 100 (R)

0.3 100 (R)

0.2 100 (R)

0.5 85 (R)

0.9c 97c (R)

16.6 87 (R)

0.2 100 (R)

0.1 95 (R)

3.9 0

6.2 87 65 (R)

2.1 8 79 (R)

1.7 4 89 (R)

1.7 5 71 (R)

21.4 62 91 (R)

11.1 0.5 84 (R)

8.0 3 84 (R)

1.6 3 99 (R)

12 21 96 (R)

2.5 2 61 (R)

3 90 (S)

1.7 72 (R)

6 0 (S)

2 9 (S)

41 48 (S)

0.5 48 (R)

0.7 38 (S)

70 89 (S)

23 83 (S)

⬍0.5 NDd

a In comparison to a resting cell suspension of the recombinant E. coli strain induced under the same conditions and diluted to the same optical density (activity of NitP ⫽ 100%). b Compared at about 30% substrate conversion. c Extrapolated from Fig. 1. d ND not detectable. e MN, mandelonitrile; MA, mandelic acid; MAA, mandeloamide. f 2-PPAA/(2-PPA ⫹ 2-PPAA). g MAA/(MA ⫹ MAA).

cysteine residue. In contrast, in almost all other nitrilases (and also cyanide dihydratases) from various bacteria and plants, a highly conserved sequence motif, Cys-Trp-Glu, was observed. The only exceptions to this rule among biochemically validated enzymes were NitP and a nitrilase from Klebsiella ozaenae. The importance of this amino acid residue for the substrate and/or reaction specificity of nitrilases was also confirmed by previously performed homology modeling, which suggested that the side chain of this residue would point into the substrate binding pocket of the enzyme (13). Furthermore, other members of the nitrilase superfamily could be easily differentiated by the conserved sequences found adjacent to the catalytically active Cys residue. Thus, D-carbamoylases or “aliphatic” amidases possessed at the relevant position the conserved sequence motif Cys-Tyr-Glu or Cys-(Glu)-Glu-Gly, respectively (25). Therefore, Ala165 was replaced in plasmid pIK9 by a tryptophan, phenylalanine, tyrosine, histidine, glutamate, arginine, or glycine residue. The recombinant E. coli JM109 strains obtained were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and it was found that all mutant variants produced approximately the same amount of recombinant enzyme. All mutant strains were able to hydrolyze mandelonitrile and 2-PPN, although in some cases only very low activities were detected. This was especially evident for the variants carrying the Ala165Arg or Ala165Glu mutation (Table 1). All variants in which the alanine residue was replaced by larger amino acid residues demonstrated with 2-PPN an inversion of enantioselectivity for the synthesis of 2-PPA similar to that observed with Nit(pap) (Fig. 1D). In addition, it was found that substitution of the alanine residue for aromatic residues resulted in enzyme variants with the highest enantioselectivities for the formation of (R)-2-PPA. In contrast, a decrease in the size of the substituent (in Ala165Gly) resulted in a mutant form that still preferentially formed (S)-PPA (Fig. 1D).

The variants carrying the Ala165Gly and Ala165Arg mutations formed 4.3% and 16.6% amide, respectively, and thus produced significantly more amide than the wild-type enzyme (and the other mutant forms) (Fig. 1E). The Ala165Gly mutant preferentially formed (S)-2-PPAA, and the Ala165Arg mutant mainly formed (R)-2-PPAA (Fig. 1F). The turnover experiments with mandelonitrile demonstrated that all variant enzymes formed the R enantiomer of the acid with higher enantioselectivities than NitP. The highest ee values (91%) were observed with the mutant NitP (Ala165His) (Fig. 2D). In addition to its increased enantioselectivity, the NitP (Ala165His) mutant also demonstrated an increased relative activity for the formation of mandelic acid. The mutant formed mandelic acid from mandelonitrile about 21 times faster than 2-PPA from 2-PPN. In contrast, this value was about 7:1 with the wild-type enzyme (Table 1), and it was observed for the mutants NitP(Ala165Phe), NitP(Ala165Tyr), and NitP (Ala165Trp) that the relative activities for mandelic acid formation were only about twice as high as those found with 2-PPN. A similar shift had already been observed for the chimera Nit(pap), which carries a Trp residue in the relevant position (see above). The variants NitP(Ala165Gly) and NitP(Ala165Arg) formed increased amounts of mandeloamide (Fig. 2E), and also 2-PPAA (see Fig. 1E). In contrast, all other mutants showed a significant decrease in the amounts of amide formed compared to the wild-type enzyme. Thus, it was found that the mutant NitP(Ala165Phe) formed less than 0.5% mandeloamide from the mandelonitrile converted (Fig. 2E). There were also significant differences observed in the enantiomeric compositions of the mandeloamides formed by the different mutants. Thus, most of the mutants preferentially formed (as the wild-type enzyme) (S)-mandeloamide. In contrast, almost no enantiopreference was observed with NitP

VOL. 75, 2009

(Ala165Glu) and NitP(Ala165His). The variant NitP(Ala165Trp) demonstrated stereoinversion and preferentially formed (R)mandeloamide (Fig. 2F). Modifications of the amino acids in positions 166 and 167. The previously performed sequence comparisons demonstrated a signature sequence for nitrilases in direct proximity to the catalytic center consisting of Cys-(Trp)-Glu-His, and all enzymatically active members of the nitrilase superfamily harbor the general motif Cys-X-Glu-X at this position. The previously performed sequence analysis suggested that these residues form the bottom of the substrate-binding pocket (13). In order to analyze the importance of these substituents, the mutants Glu166Ala and His167Ala were generated from NitP. The mutation Glu166Ala resulted in an enzyme that showed only marginal activities (about 0.1 to 0.3% of the activities of the wild-type enzyme with 2-PPN and mandelonitrile). In contrast, the mutant His167Ala still possessed about 30% of the activity of the wild-type enzyme. With 2-PPN, both mutant forms showed no significant changes in the ee values of the products and the degree of amide formation. In contrast, with mandelonitrile as a substrate, decreases in the relative amounts of amide formed and also in the enantiomeric excess of the (R)-mandelic acid formed were observed (ee values were about 13 at 30% conversion). Exchange of the amino acids Ile168 and Gln169. NitP differed from Nit(pap) in the amino acids Ile168 and Gln169, which are located in the C-terminal region near the catalytically active cysteine residue. These amino acid residues were replaced in Nit(pap) (and thus in NitA) by a leucine and a serine residue. Therefore, the amino acids were replaced in NitP by the corresponding amino acids from NitA. Furthermore, a second double mutant was constructed in which both amino acids were replaced by two alanine residues. The corresponding mutants, NitP(Ile168Leu⫹Gln169Ser) and NitP (Ile168Ala⫹Gln169Ala), were initially tested with 2-PPN, and it was found that the two double mutants clearly resembled NitP in respect to the enantioselectivity of acid formation and the amounts of amide formed (Fig. 1G and H). Nevertheless, the two mutants significantly differed from the wild type in their increased tendency to form (R)-2-PPAA (Fig. 1I). Surprisingly, it was found that the two double mutants showed rather different behavior during the conversion of mandelonitrile. NitP(Ile168Leu⫹Gln169Ser) demonstrated a slight increase in the formation of (R)-mandelic acid and a significant decrease in the amount of amide formation (Fig. 2G and H). In contrast, the “double-alanine mutant” showed a much more pronounced increase in the R selectivity of mandelic acid formation and also a pronounced increase in the degree of amide formation. The mandeloamide formed possessed the S configuration (Fig. 2I). Thus, NitP(Ile168Ala⫹ Gln169Ala) clearly resembled the mutant Ala165Gly, in which an amino acid “downstream” of the catalytically active cysteine residue had also been replaced by a smaller amino acid. Site-directed exchange of W164 in the enantioselective arylacetonitrilase from A. faecalis ATCC 8750. The results obtained with the nitrilase from P. fluorescens EBC191 clearly demonstrated the importance of the amino acid in position 165. In order to generalize this observation, the corresponding amino acid residue (W164) was also mutagenized in the enantioselective nitrilase from A. faecalis ATCC 8750. Thus, the


5597

mutant enzyme NitA(Trp164Ala) was constructed. This mutant enzyme preferentially formed (S)-2-PPA from (R,S)-2PPN and thus demonstrated a pronounced inversion of enantioselectivity compared to the wild-type enzyme NitA (Table 1). The same trend (increased amounts of the S enantiomer formed) was also observed during the conversion of (R,S)mandelonitrile. The mutant also produced significantly increased amounts of amides. This was especially evident during the conversion of (R,S)-mandelonitrile, because the mutant enzyme produced almost 70% amide (relative to the total amount of products formed). In contrast, the wild-type form of NitA under the same conditions produced less than 1% mandeloamide. This demonstrated that a single amino acid exchange could change a nitrilase that preferentially formed (R)-mandelic acid into an enzyme that preferentially formed (S)-mandeloamide. DISCUSSION The present study clearly demonstrated the importance of the amino acid residue directly adjacent to the catalytically active cysteine residue for the enantioselectivity of arylacetonitrilases. Thus, all enzyme variants that carried rather large substituents at this position formed increased proportions of the R enantiomers of mandelic acid and 2-PPA. The importance of the sizes of the relevant residues was visualized by plotting the molecular weights of the amino acid side chains present in position 165 against the ee values of the acids formed (Fig. 3). This indicated the importance of steric hindrance, because the highest enantiomeric excesses were observed with the mutants carrying histidine, phenylalanine, or tyrosine residues, which show different polarities but have similar sizes. Interestingly, it was found that the enzyme variant Ala165Phe, in which the aromatic ring of the introduced phenylalanine residue is identical to the benzene ring of 2-PPN, showed the highest degree of enantioselectivity for the formation of (R)-PPA. This might indicate that in this enzyme variant the benzene ring of phenylalanine occupies the position in the active center of the enzyme that in the wild-type enzyme allows binding of (S)-2-PPN. Surprisingly, as in most bacterial nitrilases, a slight decrease in enantioselectivity was observed for both substrates analyzed when Ala165 was replaced by a tryptophan residue. This could suggest that in known enantioselective arylacetonitrilases (such as the one from A. faecalis ATCC 8750) it might be worthwhile to replace the corresponding tryptophan residue with a smaller aromatic amino acid. The crucial importance of the amino acid residue in the position corresponding to Ala165 in NitP for the enantioselectivity of arylacetonitrilases was also confirmed by the mutant of NitA generated in the course of the present study. This mutant preferentially converted 2-PPN to (S)-2-PPA. This was in sharp contrast to the wild-type form of NitA, which demonstrated the opposite enantiopreference. This effect was surprisingly strong, and (S)-2-PPA was formed by this mutant with a rather high enantiomeric excess (ee ⫽ 90% at 30% conversion), which was even higher than the ee value found for NitP under the same conditions (ee ⫽ 60%) (Fig. 2). The amino acid exchange Trp164Ala in NitA also resulted in a significant decrease in enantioselectivity during the conversion of mandelonitrile. Thus, the ee value decreased from 96% for (R)-man-

5598

KIZIAK AND STOLZ


FIG. 3. Relationship between enantioselectivity of acid formation and amide-forming activity during the conversion of mandelonitrile and 2-PPN correlated with the size of the substituent in position 165 of the wild-type nitrilase of P. fluorescens EBC191 (NitP) and different enzyme variants. The ee values of mandelic acid () and 2-PPA (F) and the relative amounts of mandeloamide (‚) and 2-PPAA (E) formed were taken from Table 1 and plotted against the molecular weights of the relevant amino acid side chains (the molecular weight of amino acid 56 for the peptide backbone).

delic acid for the wild-type enzyme to 61% for the mutant. Furthermore, this enzyme variant formed significantly more (S)-mandeloamide than (R)-mandelic acid from racemic mandelonitrile and thus seems to have a higher affinity for (S)mandelonitrile than for (R)-mandelonitrile (Table 1). These results clearly demonstrated that the amino acid residue that is positioned carboxy terminally in direct proximity to the catalytically active cysteine residue has a pronounced influence on the enantioselectivity of acid formation. Nevertheless, it also became evident that additional amino acids are involved in the determination of enantioselectivity. This was indicated by the observation that none of the point mutants of NitP that were mutated in position 165 demonstrated a high degree of enantioselectivity similar to that of Nit(pap). Furthermore, it was also observed that mutations in amino acid positions 167 to 169 of NitP caused significant alterations in enantioselectivity during the conversion of mandelonitrile. It is very probable that not only steric interactions, but also some charge-related interactions, control the enantioselectivity of the nitrilase reaction, because NitP, NitA, and the various mutant enzymes converted (R,S)-mandelonitrile almost generally to much higher proportions of the R enantiomers than was observed during the conversion of (R,S)-2-PPN. Furthermore, some mutants were also identified (e.g., the “double mutants” in positions 167 and 168) that demonstrated significant changes in enantioselectivity during the hydrolysis of mandelonitrile, but not in the conversion of 2-PPN. There are several examples in the literature showing that, in addition to the acids, with certain nitriles nitrilases also form the corresponding amides as (by)products (7, 24, 30), and it was observed that with many substrates NitP produces extraordinarily large amounts of amides (5, 12). The ability of nitrilases to form amides as (by)products can be explained by the

intermediate formation of a tetrahedral complex in which the substrate is covalently bound to the catalytically active cysteine residue of the enzyme. This intermediate can then undergo either a thiol elimination (resulting in amide formation) or a release of ammonia. The release of ammonia (resulting in the formation of the acid) chemically requires a protonation of the nitrogen atom originating from the nitrile group in order to obtain a good leaving group (5, 31). The mutants obtained in the present study demonstrated significant changes, not only in the enantioselectivity of the hydrolytic reaction, but also in the degree of amide formation. In general, it was observed that the presence of smaller substituents in direct proximity to the catalytically active cysteine residue was correlated with a decreased preference for the formation of the R acids and the formation of larger amounts of amides. Increased formation of amides was described previously for various C-terminally deleted mutants of NitP. Surprisingly, in these mutants, an increased enantioselectivity for the formation of the R acids was correlated with an increased tendency for amide formation (13). Increased formation of the R acids in combination with extraordinary amounts of amides was also observed for the mutant NitP(Ala165Arg), which demonstrated behavior slightly different from that of the other nitrilase variants carrying larger substituents in position 165 (Fig. 3). This might indicate that in this mutant the presence of the positively charged guanidine group of arginine in close proximity to the tetrahedral reaction intermediate might facilitate the protonation of the thiol group of the catalytically active cysteine residue, which could produce an increased tendency for this sulfhydryl group to function as a leaving group and thus result in increased formation of amides. There were also some significant variations in the enantiomeric compositions of the amides formed as by-products dur-

VOL. 75, 2009


ing the conversion of mandelonitrile and 2-PPN. NitP mainly formed (S)-mandeloamide from mandelonitrile, and this was also the case for NitA and most of the point mutations in position 165. In contrast, (R)-mandeloamide was preferentially produced by Nit(pap) and NitP(Ala165Trp), which both carry a tryptophan residue at position 165. These two mutants also formed lower enantiomeric excesses of (R)-mandelic acid, as originally expected from the large size of the tryptophan residue present in position 165 (see above). This indicated that (at least in a “NitP background”) the large size of the tryptophan residue oriented the enzyme-bound (R)-enantiomer of the substrate in a way that increased the probability of a protonation of the thiol group of the cysteine residue of the active center and thus facilitated the release of (R)-mandeloamide. An opposite inversion of the enantiopreference was found for the conversion of 2-PPN by NitP(Ala165Gly). This mutant converted 2-PPN to (S)-2-PPAA as a by-product, which was unique among the wild-type and mutant nitrilases analyzed in this and a previous study (13). In conclusion, it can be deduced that the mutants generated in the previous and present studies using the nitrilase from P. fluorescens EBC191 in combination with a recent study of an “aliphatic nitrilase” from Rhodococcus rhodochrous ATCC 33278 (35) will allow more purposeful modifications of nitrilases, even in the absence of true crystal structures. Furthermore, the abilities of certain nitrilase mutants to form rather large amounts of amides might also allow the conversion of nitrilases into (enantioselective) amide-forming nitrile-hydratases and thus increase the catalytic usefulness of this group of enzymes.

12.

13.

14.

15.

16. 17.

18.

19. 20. 21.

22.

23. 24.

25. REFERENCES 1. Banerjee, A., P. Kaul, and U. C. Banerjee. 2006. Purification and characterization of an enantioselective arylacetonitrilase from Pseudomonas putida. Arch. Microbiol. 184:407–418. 2. Bunch, A. W. 1998. Nitriles, p. 277–324. In H. J. Rehm and G. Reed (ed.), Biotechnology, vol. 8a. Biotransformations I. Wiley-VCH, Weinheim, Germany. 3. Choi, S. Y., and Y. M. Goo. 1986. Hydrolysis of the nitrile group of aminophenylacetonitrile by nitrilase: development of a new biotechnology for stereospecific production of S-␣-phenylglycine. Arch. Pharm. Res. 9:45–47. 4. Favre-Bulle, O., J. Pierrard, C. David, P. Morel, and D. Horbez. January 2001. Industrial scale process for the preparation of 2-hydroxy-4-methylbutyric acid using a nitrilase. U.S. patent 6,180,359. 5. Fernandes, B. C. M., C. Mateo, C. Kiziak, J. Wacker, A. Chmura, F. van Rantwijk, A. Stolz, and R. A. Sheldon. 2006. Nitrile hydratase activity of a recombinant nitrilase. Adv. Synth. Catal. 348:2597–2603. 6. Gagavan, J. E., R. DiCosimo, A. Eisenberg, S. K. Fager, P. W. Folsom, E. C. Hann, K. J. Schneider, and R. D. Fallon. 1999. A Gram-negative bacterium producing a heat-stable nitrilase highly active on aliphatic nitriles. Appl. Microbiol. Biotechnol. 52:654–659. 7. Goldlust, A., and Z. Bohak. 1989. Induction, purification, and characterization of the nitrilase of Fusarium oxysporum f. sp. melonis. Biotechnol. Appl. Biochem. 11:581–601. 8. Heinemann, U., C. Kiziak, S. Zibek, N. Layh, M. Schmidt, H. Griengl, and A. Stolz. 2003. Conversion of aliphatic 2-acetoxynitriles by nitriles hydrolysing bacteria. Appl. Microbiol. Biotechnol. 63:274–281. 9. Horton, R. M., H. D. Hunt, S. N. Ho, J. K. Pullen, and L. R. Pease. 1989. Engineering of hybrid genes without use of restriction enzymes: gene splicing by overlap extension. Gene 77:61–68. 10. Kaul, P., A. Banerjee, S. Mayilraj, and U. C. Banerjee. 2004. Screening for enantioselective nitrilases: kinetic resolution of racemic mandelonitrile to (R)-mandelic acid by new bacterial isolates. Tetrahedron Asym. 15:207–211. 11. Kaul, P., A. Stolz, and U. C. Banerjee. 2007. Cross-linked amorphous ni-

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

5599

trilase aggregates for enantioselective nitrile hydrolysis. Adv. Synth. Catal. 349:2167–2176. Kiziak, C., D. Conradt, A. Stolz, R. Mattes, and J. Klein. 2005. Nitrilase from Pseudomonas fluorescens EBC191: cloning and heterologous expression of the gene and biochemical characterization of the recombinant enzyme. Microbiology 151:3639–3648. Kiziak, C., J. Klein, and A. Stolz. 2007. Influence of different carboxyterminal mutations on the substrate-, reaction-, and enantiospecifity of the arylacetonitrilase from Pseudomonas fluorescens EBC191. Prot. Eng. Design Sel. 20:385–396. Kobayashi, M., H. Izui, T. Nagasawa, and H. Yamada. 1993. Nitrilase in biosynthesis of the plant hormone indole-3-acetic acid from indole-3-acetonitrile: cloning of the Alcaligenes gene and site-directed mutagenesis of cysteine residues. Proc. Natl. Acad. Sci. USA 90:247–251. Kobayashi, M., H. Komeda, N. Yanaka, T. Nagasawa, and H. Yamada. 1992. Nitrilase from Rhodococcus rhodochrous J1. Sequencing and overexpression of the gene and identification of an essential cysteine residue. J. Biol. Chem. 267:20746–20751. Kobayashi, M., and S. Shimizu. 1994. Versatile nitrilases: nitrile-hydrolysing enzymes. FEMS Microbiol. Lett. 120:217–224. Kobayashi, M., N. Yanaka, T. Nagasawa, and H. Yamada. 1990. Purification and characterization of a novel nitrilase of Rhodococcus rhodochrous K22 that acts on aliphatic nitriles. J. Bacteriol. 172:4807–4815. Layh, N., A. Stolz, S. Fo ¨rster, F. Effenberger, and H.-J. Knackmuss. 1992. Enantioselective hydrolysis of O-acetylmandelonitrile to O-acetylmandelic acid by bacterial nitrilases. Arch. Microbiol. 158:405–411. Liese, A., K. Seelbach, and C. Wandrey. 2006. Industrial biotransformations, 2nd ed. Wiley-VCH, Weinheim, Germany. Martinkova ´, L., and V. Kren. 2002. Nitrile- and amide-converting microbial enzymes: stereo-, regio- and chemoselectivity. Biocatal. Biotrans. 20:79–93. Mateo, C., A. Chmura, S. Rustler, F. van Rantwijk, A. Stolz, and R. A. Sheldon. 2006. Synthesis of enantiomerically pure (S)-mandelic acid using an oxynitrilase-nitrilase bienzymatic cascade. A nitrilase surprisingly shows nitrile hydratase activity. Tetrahedron Asym. 17:320–323. Mathew, C. D., T. Nagasawa, M. Kobayashi, and H. Yamada. 1988. Nitrilasecatalysed production of nicotinic acid from 3-cyanopyridine in Rhodococcus rhodochrous J1. Appl. Environ. Microbiol. 54:1030–1032. Nagasawa, T., J. Mauger, and H. Yamada. 1990. A novel nitrilase, arylacetonitrilase, of Alcaligenes faecalis JM3. Eur. J. Biochem. 194:765–772. Osswald, S., H. Wajant, and F. Effenberger. 2002. Characterization and synthetic applications of recombinant AtNIT1 from Arabidopsis thaliana. Eur. J. Biochem. 269:680–687. Pace, H. C., and C. Brenner. 2001. The nitrilase superfamily: classification, structure and function. Genome Biol. 2:0001.1–0001.9. Rey, P., J.-C. Rossi, J. Taillades, G. Gros, and O. Nore. 2004. Hydrolysis of nitriles using an immobilized nitrilase: applications to the synthesis of methionine hydroxyl analogue derivatives. J. Agric. Food Chem. 52:8155–8162. Rustler, S., A. Mu ¨ller, V. Windeisen, A. Chmura, B. C. M. Fernandes, C. Kiziak, and A. Stolz. 2007. Conversion of mandelonitrile and phenylglycinenitrile by recombinant E. coli cells synthesizing a nitrilase from Pseudomonas fluorescens EBC191. Enzyme Microb. Technol. 40:598–606. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Schulze, B. 2002. Hydrolysis and formation of C-N bonds, p. 699–715, In K. Drauz and H. Waldmann (ed.), Enzyme catalysis in organic synthesis, vol. II. Wiley-VCH, Weinheim, Germany. Stevenson, D. E., R. Feng, F. Dumas, D. Groleau, A. Mihoc, and A. C. Storer. 1992. Mechanistic and structural studies on Rhodococcus ATCC39484 nitrilase. Biotechnol. Appl. Biochem. 15:283–302. Stevenson, D. E., R. Feng, and A. C. Storer. 1990. Detection of covalent enzyme-substrate complexes of nitrilase by ion-spray mass spectroscopy. FEBS Lett. 277:112–114. Vieira, J., and J. Messing. 1982. The pUC plasmids, an M13mp7-derived system for insertion mutagenesis and sequencing with synthetic universal primers. Gene 19:259–268. Yamamoto, K., I. Fujimatsu, and K.-I. Komatsu. 1992. Purification and characterization of the nitrilase from Alcaligenes faecalis ATCC 8750 responsible for enantioselective hydrolysis of mandelonitrile. J. Ferment. Bioeng. 73:425–430. Yamamoto, K., K. Oishi, I. Fujimatsu, and K.-I. Komatsu. 1991. Production of R-(⫺)-mandelic acid from mandelonitrile by Alcaligenes faecalis ATCC 8750. Appl. Environ. Microbiol. 57:3028–3032. Yeom, S.-J., H.-J. Kim, J.-K. Lee, D.-E. Kim, and D.-K. Oh. 2008. An amino acid at position 142 in nitrilase from Rhodococcus rhodochrous ATCC 33278 determines the substrate specificity for aliphatic and aromatic nitriles. Biochem. J. 415:401–407.