Identification of potential active-site residues in the hydroxynitrile lyase from Manihot esculenta by site-directed mutagenesis.

The hydroxynitrile lyase from cassava (Manihot esculenta Crantz) (EC 4.1.2.37) catalyzes the decomposition of the achiral alpha-hydroxynitrile acetone cyanohydrin into HCN and acetone during cyanogenesis of damaged plants. This enzyme can also be used for stereoselective synthesis of a wide array of (S)-cyanohydrins by addition of HCN to aldehydes or ketones. Optically active cyanohydrins are interesting intermediates for the synthesis of alpha-hydroxy acids, alpha-hydroxy ketones, or beta-ethanolamines, all of which are important building blocks in organic synthesis. Inhibition of hydroxynitrile lyase from M. esculenta (MeHNL) by serine- and histidine-modifying reagents suggests involvement of active site seryl and histidyl residues. Furthermore, serine 80 of MeHNL is part of the active site motif Gly-X-Ser-X-Gly/Ala, often considered as the hallmark of catalytic triads having independently evolved in four groups of enzymes: the alpha/beta hydrolase fold enzymes, subtilisins, the cysteine proteases, and the eukaryotic serine proteases. By site-directed mutagenesis, three residues critical for enzyme activity have been identified: serine 80, aspartic acid 208, and histidine 236. These residues may be directly involved in MeHNL-catalyzed decomposition of cyanohydrins, providing evidence for a catalytical triad in HNLs, too. The order of the catalytic triad residues in the primary sequence of MeHNL is nucleophile-histidine-acid, suggesting that MeHNL belongs to the alpha/beta hydrolase fold group of enzymes. In contrast to all other enzymes having a catalytical triad, HNLs catalyze no net hydrolytic reactions.

To date, HNLs from various dicotyledons (7,8) as well as from Sorghum bicolor (SbHNL) 1 (9,10), a monocotyledone, and Phlebodium aureum (PhaHNL) (11), a fern, have been bio-chemically characterized. HNLs comprise a heterogenous group of enzymes with homo-and heteromers as the active enzyme, glycoproteins and nonglycoproteins, as well as flavoproteins. The molecular mass of HNL subunits ranges from 20 to 60 kDa (7)(8)(9)(10)(11). The heterogeneity of their properties suggests that most of these enzymes have independently evolved (convergent evolution). In fact, the lack of homologies among the amino acid sequences of the recently cloned HNLs from Prunus serotina (PsHNL) (12), Manihot esculenta (Me-HNL) (13), and S. bicolor (14) emphasize this idea. While PsHNL and MeHNL show none or only limited homology to other known proteins (12,13), SbHNL has astonishingly high sequence homology with serine carboxypeptidases, especially with those from wheat. It appears of particular relevance that SbHNL shares the catalytical triad Ser, Asp, and His with these enzymes (14). The catalytical triad motif evolved in the carboxypeptidase, chymotrypsin, and subtilisin group of proteases by convergent evolution but is also found in a series of other hydrolytic enzymes (15)(16)(17). The above-mentioned similarities between SbHNL and serine carboxypeptidases provoked the question of whether other HNLs also use the catalytical triad motif for catalysis. In support of this idea, almost all HNLs are inhibited by serine/cysteine and/or histidinemodifying agents (11,14,18).
In the present study, we have used site-directed mutagenesis to demonstrate that serine 80, which is part of the typical serine protease Gly-X-Ser-X-Gly/Ala consensus motif, is essential for enzyme activity of MeHNL. In addition, site-directed mutagenesis further shows that histidine 236 and aspartic acid 208 are important for enzyme activity. These observations suggest that the catalytical triad motif is also utilized in the active site of MeHNL. This work provides the first functional evidence for use of a catalytic triad by a member of the hydroxynitrile group of enzymes.

EXPERIMENTAL PROCEDURES
Expression Cloning of MeHNL-The MeHNL gene was cloned in the expression vector pQE4 (Quiagen). In brief, total RNA was isolated from fresh leaves of M. esculenta using a guanidine chloride method (19). Subsequently, the coding region of the MeHNL gene was amplified from oligo(dT)-primed first strand cDNA using primer with 5Ј overhangs encompassing BamHI restriction sites (forward primer, GCA GGG CCG GAT CCC ATT TCC AAA ATG GTA ACT GCA CA; reverse primer, GCA GGG CCG GAT CCA CC AAC GTG GAA CTC TCC CAT ATT; underlined sequences correspond to positions 16 -39 and 933-910 of the MeHNL cDNA (13)). The in frame insertion of the BamHIdigested polymerase chain reaction fragment into the BamHI site of pQE4 results in the plasmid pQE4-MeHNLwt, which was transformed in Escherichia coli M15[pREP4] cells (Quiagen) for overexpression of MeHNL.
Site-directed Mutagenesis-Mutations were introduced into pQE4-MeHNLwt using the Chameleon double-stranded, site-directed mutagenesis kit (Stratagene) according to the manufacturer's recommendations. In brief, the double-stranded target plasmid DNA was heat denatured, and a selection primer, which removes a nonessential unique restriction site, as well as a primer defining the mutation of interest (Table were annealed to one strand. These primers were extended around the plasmid using T7 DNA polymerase and T4 DNA Ligase. Subsequently, the plasmid DNA was digested with the restriction enzyme that corresponds to the nonessential restriction site, which is still present in unmutated plasmids. Undigested, mutated plasmid DNA was favored, compared to digested parental plasmid DNA, in a subsequent transformation into a repair-deficient E. coli strain due to the greater transformation efficiency of circular plasmid DNA compared to linear plasmid DNA. The transformed bacteria were grown in a liquid culture overnight. To select further for mutated plasmids, DNA was isolated from the overnight culture, digested with the restriction enzyme corresponding to the nonessential unique restriction site, removed, and transformed into competent XL1-Blue cells. Mutants were identified by introduction of a new restriction site and/or by sequence analysis using the dideoxy chain termination method of Sanger et al. (20). All primers were 5Ј-phosphorylated before use. The single codon mutant plasmids are listed in Table I, in which the introduced substitution, its location, and the resultant amino acid are also summarized. A 27mer oligonucleotide having the sequence 5Ј-CAT CAT TGG AAA ACG CTC TTC GGG GCG-3Ј was used for deletion of the nonessential unique restriction site XmnI from the Amp r gene of pQE4.
Expression and Purification of Wild Type and Mutated MeHNL-Wild type and mutant MeHNL genes were expressed in E. coli strain M15[pREP4]. For induction of recombinant proteins, an overnight culture was performed at 37°C in LB medium supplemented with ampicillin (100 g/ml) and kanamycin (25 g/ml), diluted 1:10 with LB medium supplemented only with ampicillin (100 g/ml), and grown at 30°C. 90 min after dilution isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 1 mM for induction of gene expression. The cells were harvested by centrifugation after 4 h, resuspended in 50 mM sodium acetate, pH 5.4, and disintegrated by 20 pulses at 60% with a Sonoplus HD 200 sonifyer (Bandelin, Berlin, Germany). After centrifugation (16.000 rpm, 20 min, 4°C, Ja 20 rotor), wild type and mutant MeHNLs were purified from the supernatant as described previously for MeHNL from the natural source (18). Briefly, the supernatant was applied to Q-Sepharose FF at pH 5.4, and the enzyme eluted with NaCl. Final purification of MeHNL was obtained by gel exclusion chromatography on Superdex 200 (Pharmacia) and anion exchange chromatography on Mono Q (Pharmacia) at pH 7.5.
Enzyme Assays-The assay for MeHNL based on the measurement of the HCN formed during dissociation of acetone cyanohydrin. The release of HCN was determined as described by Selmar et al. (21). One unit of HNL activity is defined as the amount of enzyme that can catalyze the decomposition of 1 mol of substrate/min under the conditions described by Selmar et al. (21). The calculation of the specific activity of total protein was determined with the BCA Protein Assay Reagent from Pierce according to the manufacturer's recommendations. For inhibition studies with DFP (Sigma), phenylmethylsulfonyl fluoride (Boehringer Mannheim), and 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride (Boehringer Mannheim), 5 g of protein were incubated in the indicated concentration of the reagent for 15-30 min at room temperature. The remaining activity was then determined as described above.

SDS-Polyacrylamide Gel Electrophoresis and Western Blot
Analysis-Electrophoresis was performed on 13.5% (w/v) polyacrylamide gels according to Laemmli (22). After electrophoresis proteins were silver stained according to Blum et al. (23) or transferred to nitrocellulose (Schleicher & Schuell) according to Towbin et al. (24) for Western blot analysis. Western blot analysis was carried out using anti-MeHNL antisera and peroxidase-conjugated anti-mouse (IgG ϩ IgM) antibodies as described previously (18).

RESULTS AND DISCUSSION
Expression of the Wild Type MeHNL Gene-For cloning of the MeHNL gene, BamHI site-containing primers encompassing the whole coding region of the gene were designed with the PRIMER program of Husar (Heidelberg UNIX Sequence Analysis Resources). Subsequently the MeHNL gene was amplified by reverse polymerase chain reaction from RNA of young leaves of M. esculenta and cloned into the expression vector pQE4 (Quiagen). The MeHNL gene was sequenced to confirm the identity of the MeHNL sequence after polymerase chain reaction amplification (data not shown) and transfected in E. coli M15[pREP4] cells. Recombinant MeHNL was purified from supernatants of induced cell lysates as described previously for the natural enzyme isolated from cassava leaves (18). Recombinant and natural MeHNL were indistinguishable in this purification procedure. Recombinant MeHNL possess the same specific activity compared to MeHNL isolated directly from the plant (Table II). MeHNL has been reported to exist as a tetramer. Using gel filtration analysis, we compared the oligomeric structure of recombinant and natural MeHNL. The analysis showed that both enzyme preparations elute in the same fractions corresponding to an estimated molecular mass of 110 kDa (Table II), which is in good accordance with a tetrameric structure of the enzyme (13). Kinetic analysis of acetone cyanohydrin cleavage by recombinant MeHNL showed a K m of 101 mM (Table II), which is similar to the value determined for the natural enzyme (13).
MeHNL Is Strongly Inhibited by Serine-modifying Agents-The effects on MeHNL activity of well known serine-modifying reagents like DFP, phenylmethylsulfonyl fluoride, and 4-(2aminoethyl)-benzenesulfonyl fluoride hydrochloride were examined using purified enzyme. Both (natural MeHNL as well as recombinant MeHNL) were almost completely inhibited by DFP at 1 mM, and a partial inhibition was observed with phenylmethylsulfonyl fluoride and 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride in the mM range (data not shown). All the tested active site reagents are well known inhibitors of serine proteases. Therefore, we investigated whether MeHNL also exhibits a protease activity. However, using resorufin-labeled casein as a substrate, proteolytic activity of MeHNL was not detectable, neither under reducing or non-reducing conditions nor at different pH values (data not shown). Low molecular weight protease inhibitors such as leupeptin and pepstatin A did not affect the MeHNL-catalyzed acetone cyanohydrin cleavage at pH 5.4.
Site-directed Mutagenesis Identifies the Catalytical Triad Ser 80 -Asp 208 -His 236 in MeHNL-The above data obtained with serine-modifying reagents indicated a critical functional role of a serine residue in MeHNL activity. For a molecular identification of catalytically important residues, site-directed mutagenesis was used. The rationale for performing site-directed mutagenesis of serine 80 was based on sequence analysis showing serine 80 as the best candidate for a serine of catalytical importance because it is the only serine of MeHNL within a Gly-X-Ser-X-Gly/Ala consensus motif. This motif defines the serine of the catalytic triad in serine proteases. The putative functional importance of serine 80 is further underscored by the fact that the consensus motif is conserved (maximum: 33.2% identity in a 258-amino acid overlap) between MeHNL and two proteins of unknown function from rice (25) (Fig. 2), whose sequence is deposited in the EMBL data bank (accession numbers Z34270 and Z34271). We selected alanine as a substituent for serine 80. Expression of the S80A mutant enzyme in E. coli revealed no differences to wild type MeHNL according to oligomeric structure, molecular weight, and behavior in the standard purification procedure (Table II, Fig. 3). However, the FIG. 1. Hydroxynitrile lyase-catalyzed cleavage of cyanohydrins. Cyanohydrins can be decomposed by hydroxynitrile lyases to HCN and a carbonyl compound. Although this decomposition can occur spontaneously in neutral or basic pH, the rate is increased by the presence of HNLs in the acidic milieu of damaged plant tissues. If the residues R and R1 differ, the ␣-C atom is optically active. All known HNLs exert an exclusive stereoselectivity for one of the two possible enantiomeres of such cyanohydrins. mutated enzyme was completely inactive in the standard acetone cyanohydrin cleaving assay (Table II).
Serine 80 is part of the consensus motif Gly-X-nucleophile-X-Gly/Ala, which is considered as the hallmark of proteins having the catalytic triad nucleophile-histidine-acid. To identify the histidine residue involved in a putative catalytic triad of MeHNL, we have systematically replaced histidine residues by alanine. We have restricted this alanine scanning to histidines conserved between MeHNL and the above-mentioned proteins from rice. Exchange of histidine 236 results in a mutant enzyme, which is unable to catalyze the decomposition of acetone cyanohydrin, indicating critical involvement in catalysis. All other histidine mutants show no or only a limited (Ͻ15%) decrease in enzyme activity. In particular K m and V max values remain unaffected (Table II). Identification of the two enzymatically inactive MeHNL mutants S80A and H236A gives compelling evidence for involvement of a catalytic triad in this enzyme. All known non-proteolytic enzymes having a catalytic triad belong to the ␣/␤ hydrolase fold group of enzymes. In all these enzymes, the linear order of the catalytic residues in the primary sequence is nucleophile-histidine-acid. To identify the catalytic triad acid of MeHNL, we have constructed mutants with alanine substituting for aspartic acid residues located between serine 80 and histidine 236 in the primary sequence and which are conserved with respect to the before-mentioned rice proteins (Asp 95 , Asp 130 , and Asp 208 ). A greater 80% inhibition of enzyme activity was found when aspartic acid 208 was replaced by alanine    CAT TGT TGG TGA GGC CTG TGC AGG   250-276  S80A   GC  GGT AAC TGC AGC TAA AGT TCT  Potential Active-site Residues in the Hydroxynitrile Lyase ( Fig. 3), whereas the substitutions of aspartic acid 95 and 130, respectively, have no significant influence on enzyme activity and kinetic parameters (Table II). The MeHNL D208A mutant was still inhibited by mM concentrations of the serine-modifying reagent DFP (Table II). Kinetic analysis suggests that the K m value of this mutant increases from 101 mM for wild type MeHNL to over 200 mM. However, a reliable calculation of K m and V max was not possible due to the pronounced base catalyzed cleavage of the substrate at high substrate concentrations.
Substitution of a single residue may cause a decrease in enzymatic activity not directly related to the catalytic site. Such critical residues often influence protein folding and stability and may lead to decreased levels of expression. However, most amino acid substitutions do not have obvious effects on protein stability. All active MeHNL mutants employed here can be stored for several days at room temperature without loss of activity and are similarly affected by increasing concentrations of urea and SDS, indicating that the introduced substitutions cause no changes in protein stability (data not shown). Moreover, we found no alterations in expression levels in comparison to the wild type protein (Fig. 3). Further, during the purification procedure, no evidence for grossly altered biochemical properties of the mutant proteins became apparent, because in the gel filtration step all established mutants of Me-HNL elute in the expected tetrameric form, and in the anion exchange chromatography step all mutants elute with 110 -120 mM NaCl (Table II). Furthermore, the K m values of all mutants not affected in the putative catalytic triad residues are comparable to the K m value of the wild type protein (Table II). Therefore, we assume that the introduced substitutions do not change general protein integrity. Therefore, as mutations S80A, D208A, and H236A all dramatically reduce enzyme activity, serine 80, aspartic acid, and histidine 236 are likely to be directly involved in catalysis or represent important sites of substrate recognition.
Proposed Reaction Mechanism of MeHNL-Chymotrypsin is a well studied example of an enzyme using a catalytic triad for catalysis. The enzymatic activity of chymotrypsin critically depends on histidine 57 and serine 195. Both are located near each other in the active site. Analogous to the reaction mechanism proposed for the hydrolysis of peptide bonds by chymotrypsin, one can formulate a possible general base catalyzed mechanism for cyanohydrin cleavage as follows: In free Me-HNL, the hydroxyl group of serine 80 is hydrogen-bonded to the imidazole nitrogen of histidine 236, which in turn is stabilized by a hydrogen bond with aspartic acid 208. If the enzyme encounters a substrate molecule, the proton of serine 80 is rapidly transferred to an imidazole nitrogen of histidine 236. The resulting oxyanion of serine 80 functions as a strong base in a nucleophilic attack of the substrate hydroxyl proton, leading to a negatively charged oxyanion on the substrate. This oxyanion could be stabilized by an oxygen hole formed by amide nitrogen of the backbone of serine 80 and glycine 78 (Fig. 4). Both residues belong to the consensus motif, which is typical for nucleophils located in a catalytic triad. The stabilized oxyanion could further increase the local positive charge on the ␣-C atom of the cyanohydrin, allowing the cyanide group to leave the molecule.
Cyanohydrin Cleavage by MeHNL Defines a Novel Type of Catalytic Triad-catalyzed Reaction-To date, four groups of enzymes have been found to utilize the catalytic triads as active site principle: subtilisins, the cysteine proteases, the eukaryotic serine proteases, and the ␣/␤ hydrolase fold enzymes (17). It is commonly accepted that this stable active site principle was evolved by convergent evolution (17). All members of the first three groups are proteases, whereas the recently defined group of ␣/␤ hydrolase fold enzymes also contains lipases, dehalogenases, and esterases (17). The members of this group of enzymes are of widely differing phylogenetic origin but share the structural core motif of eight ␤ sheets connected by ␣ helices. Ollis et al. (17) suggested that the ␣/␤ hydrolase fold group of enzymes represents molecules, where the catalytic subsite framework was conserved during evolution. These enzymes have therefore the same linear order of active site residues in their primary sequences.

Potential Active-site Residues in the Hydroxynitrile Lyase
We have identified the components of a putative catalytic triad in MeHNL using site-directed mutagenesis. The order of these residues is identical to those in enzymes of the ␣/␤ hydrolase fold group, suggesting that MeHNL also belongs to this group of molecules. Cleavage of cyanohydrins into HCN and carbonyle compounds by MeHNL defines a new reaction executed by ␣/␤ hydrolase fold enzymes. Interestingly, this reaction is not a net hydrolytic reaction, like all other reactions catalyzed by enzymes having a catalytic triad. The recently cloned HNL from S. bicolor (SbHNL) shows extended sequence homologies to serine carboxypeptidase from wheat, which is also a member of the ␣/␤ hydrolase fold enzymes (14,17). Thus, SbHNL is likely to be a member of this enzyme group. MeHNL and SbHNL share no significant sequence homologies. Moreover, the high sequence homologies between SbHNL and serine carboxypeptidase from wheat indicate that SbHNL evolved only recently. In fact, the SbHNL type of HNLs is restricted to the sorghum group of gramineae. 2 In contrast, the putative common ancestor of all ␣/␤ hydrolase fold enzymes should be placed early in molecular evolution due to lack of sequence homologies among these enzymes. Therefore, MeHNL and Sb-HNL would have to be independently evolved from two different members of the ␣/␤ hydrolase fold group. In other words, SbHNL and MeHNL appear to be the result of convergent evolution on the background of molecules, which were in turn evolved by divergent evolution.
The well known HNL from P. serotina, which is unusal in containing a FAD without catalyzing a net oxido/reduction reaction, was also recently cloned (12). And in fact, this enzyme also possesses exactly one nucleophilic residue, which is part of the before-mentioned consensus motif, defining the nucleophil in a catalytic triad.