The Primitive Protozoon Trichomonas vaginalisContains Two Methionine γ-Lyase Genes That Encode Members of the γ-Family of Pyridoxal 5′-Phosphate-dependent Enzymes*

Methionine γ-lyase, the enzyme that catalyzes the breakdown of methionine by an α,γ-elimination reaction and is a member of the γ-family of pyridoxal 5′-phosphate-dependent enzymes, is present in high activity in the primitive protozoan parasite Trichomonas vaginalisbut is absent from mammals. Two genes, mgl1 andmgl2, encoding methionine γ-lyase, have now been isolated from T. vaginalis. They are both single copy, encode predicted proteins (MGL1 and MGL2) of 43 kDa, have 69% sequence identity with each other, and show a high degree of sequence identity to methionine γ-lyase from Pseudomonas putida (44%) and other related pyridoxal 5′-phosphate-dependent enzymes such as human cystathionine γ-lyase (42%) and Escherichia coli cystathionine β-lyase (30%). mgl1 andmgl2 have been expressed in E. coli as a fusion with a six-histidine tag and the recombinant proteins (rMGL1 and rMGL2) purified by metal-chelate affinity chromatography. rMGL1 and rMGL2 were found to have high activity toward methionine (10.4 and 0.67 μmol/min/mg of protein, respectively), homocysteine (370 and 128 μmol/min/mg of protein), cysteine (6.02 and 1.06 μmol/min/mg of protein), and O-acetylserine (3.74 and 1.51 μmol/min/mg of protein), but to be inactive toward cystathionine. Site-directed mutagenesis of an active site cysteine (C113G for MGL1 and C116G for MGL2) reduced the activity of the recombinant enzymes toward both methionine and homocysteine by approximately 80% (rMGL1) and 90% (rMGL2). In contrast, the activity of mutated rMGL2 toward cysteine andO-acetylserine was increased (to 214 and 142%, respectively), whereas that of mutated rMGL1 was reduced to 39 and 49%, respectively. These findings demonstrate the importance of this cysteine residue in the α,β-elimination and α,γ-elimination reactions catalyzed by trichomonad methionine γ-lyase.

Trichomonas vaginalis is a parasitic protozoon that is believed to represent one of the earliest branches of the eukaryotic lineage, diverging close to the time when mitochondria were acquired (1)(2)(3). The phylogenetic position of trichomonads at the base of the eukaryotic tree suggested that at least some of their enzymes may have features intermediate between those characteristic of prokaryotic or eukaryotic isoenzymes. Trichomonads are aerotolerant, anaerobic organisms that rely upon fermentative pathways of catabolism, especially glycolysis (4). Phylogenetic analysis of glycolytic enzymes, however, gave equivocal results concerning their evolutionary origin (5). The terminal steps of carbohydrate breakdown take place in an unusual organelle, the hydrogenosome, which produces energy by the catabolism of pyruvate to acetate, CO 2 and H 2 (4). Hydrogenosomes only occur in anaerobic eukaryotes, including rumen ciliates and fungi as well as some free-living flagellates and ciliates (6), and contain enzymes such as pyruvate:ferredoxin oxidoreductase and hydrogenase, which are typical of anaerobic bacteria but absent from mitochondria (4,7). This led to the suggestion that trichomonad hydrogenosomes may have evolved from an anaerobic prokaryotic ancestor. The recent discovery of mitochondrial-like proteins in trichomonads (2, 8 -10) and that the mechanism whereby proteins are targeted to the organelles is similar to that of mitochondria (3,11) suggests, however, that hydrogenosomes and mitochondria are related and that the former may have evolved from mitochondria. Nevertheless, the anaerobic nature of hydrogenosomes and the trichomonads themselves suggests that the primitive protozoon may have acquired some features from anaerobic prokaryotes.
Another metabolic feature of T. vaginalis which resembles the situation in some anaerobic prokaryotes is sulfur amino acid metabolism. Trichomonads contain high activity of methionine ␥-lyase (EC 4.4.1.11), which catalyzes the ␣,␥-elimination reaction of methionine to ␣-ketobutyrate, methanethiol, and ammonia (12,13). Methionine ␥-lyase has been detected only in anaerobic microorganisms such as Pseudomonas, Aeromonas, and Clostridium (14) and is not believed to be present in yeast, plants, or mammals (15)(16)(17)(18). 1 Structurally the enzyme is related to other enzymes involved in the synthesis and interconversion, via the trans-sulfuration pathway, of cysteine and homocysteine, proteins that together comprise the ␥-family of pyridoxal 5Ј-phosphate-dependent enzymes (20). 1 These catalyze ␥-elimination or ␥-replacement, and also ␤-elimination, reactions with sulfur-containing amino acids. The structural * This work was supported in part by the Wellcome Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AJ000486  information available for members of the ␥-family has been limited until recently such that their evolutionary relationships were unclear. The sequences of several genes are now available which has allowed a phylogenetic analysis, 1 but only one crystal structure (cystathionine ␤-lyase from Escherichia coli) has been solved (21) and the reaction mechanism elucidated (22).
The methionine ␥-lyases of Pseudomonas putida and T. vaginalis have been characterized in some detail at the biochemical level (13,14,(23)(24)(25) and, in the case of P. putida, the molecular level (26,27). They share some biochemical properties, including subunit composition and the variety of substrates that can be catabolized. The methionine ␥-lyase of T. vaginalis has been purified and shown to be a tetramer that has activity not only toward methionine but also related compounds including homocysteine, cysteine, and O-acetylserine (13). The functional significance of methionine ␥-lyase activity to the parasite is uncertain, but it has been proposed that the conversion of ␣-ketobutyrate to propionate with the concomitant production of ATP could make an important contribution to energy metabolism and also that the methanethiol may play a role in enabling the parasite to avoid damage from the immune response of the host (28). Whether trichomonads contain other enzymes of the ␥-family of pyridoxal 5Ј-phosphate-dependent enzymes and are able to interconvert cysteine and homocysteine via the trans-sulfuration pathway are unknown. Mammals apparently lack methionine ␥-lyase (17), and there is evidence that this difference between the parasite and host can be exploited by pro-drugs that are activated by the enzyme (29). In addition, inhibitors are likely to be detrimental to the survival of the parasite in its host and so have potential as chemotherapeutic agents against pathogens that have methionine ␥-lyase activity.
This study of trichomonad methionine ␥-lyase was undertaken with two main aims: to analyze the similarity of ␥-family pyridoxal 5Ј-phosphate-dependent enzymes of this primitive eukaryote with their bacterial homologs; and to obtain data that will, in the longer term, enable the design of drugs that exploit the presence of methionine ␥-lyase in microbial pathogens. To this end we have analyzed the structure and organization of genes that encode methionine ␥-lyase in T. vaginalis, expressed the genes in E. coli, examined the biochemical properties of the recombinant enzymes, and analyzed the functional importance of an active site cysteine residue.

MATERIALS AND METHODS
Growth of Parasites and Preparation of DNA, RNA, and First Strand cDNA-A clonal cell line (G3) of T. vaginalis was grown axenically in modified Diamond's medium as described previously (13). Cells were harvested in late log phase of growth (1-2 ϫ 10 6 /ml) and washed twice in 0.25 M sucrose. DNA was isolated using a Nucleon II kit (Scotlab, Coatbridge, Scotland). Total RNA was isolated using Trizol (Life Technologies, Paisley, Scotland), and poly(A) ϩ RNA was isolated using poly(A) ϩ Quik columns (Stratagene). First strand cDNA for use in PCR 2 was generated in a 50-l reaction using 1 g of poly(A) ϩ RNA, 100 ng of oligo(dT) primer, and 10 units of Moloney murine leukemia virus reverse transcriptase (Promega). After synthesis the reaction was diluted to 500 l with 10 mM Tris⅐HCl, pH 7.5, 0.1 mM EDTA and heated to 65°C for 15 min.
cDNA Cloning of mgl1 and mgl2-The degenerate oligonucleotide primers used for the PCR of T. vaginalis methionine ␥-lyase genes were based on consensus sequences from cystathionine ␥-lyase from human (30), rat (31), and yeast (32). The sequence of the 5Ј-and 3Ј-oligonucleotides (using IUB codes) were: 5Ј-GCAAGCTTGTITGGATHGAGAC-ICCIACNAA-3Ј and 5Ј-GCCTCGAGCCGTTIATRTAYTTIGTNGC-3Ј. Inosine (I) residues were introduced at some positions of 4-fold degen-eracy, and HindIII and XhoI restriction sites (underlined) were added to the 5Ј-ends of the oligonucleotides to facilitate cloning. The degenerate primers were used in PCRs with T. vaginalis first strand cDNA as the template. The 50-l reaction mixture contained 5 ng of T. vaginalis cDNA, 5 ng of each of the primers, nucleotides at a final concentration of 250 M, and 5 units of Taq DNA polymerase (Promega). The following amplification protocol was employed. An initial denaturation step of 4 min at 95°C was followed by 30 cycles at 94°C for 1 min, 42°C for 1 min, and 72°C for 1 min. A final step of 5 min at 72°C was used to complete the extension. The amplified product was cloned into HindIII/ XhoI-digested pBluescript (SKϪ) (Stratagene) and sequenced using Sequenase 2.0 (Amersham). Sequence data were analyzed using the Wisconsin GCG sequence software. Two distinct groups of PCR products were found to have homology to yeast cystathionine ␥-lyase. These were used to screen a [lamda]ZapII T. vaginalis cDNA library as described previously (33). Positive plaques were isolated, and the inserts were rescued with R408 helper phage into pBluescript. Two positive clones, pMGL4100 and pMGL5100, each of which hybridized to one of the 200-bp PCR products, were selected for further analysis. These cDNA clones were sequenced on both strands as described above.
5Ј-RACE and Inverse PCR-The 5Ј-ends of the mRNA transcribed from mgl1 and mgl2 were determined using a 5Ј-RACE System Kit (Life Technologies, Inc.). For RACE-PCR, total cellular RNA from T. vaginalis (1 g) was transcribed into single-stranded cDNA using oligo(dT) as primer. Excess dNTPs and primer were removed from the DNA, and a homopolymeric tail of dCs was added to the 3Ј-end. 5 l of tailed cDNA was used for the PCRs containing the components of the 5Ј-RACE System Kit and a gene-specific primer.
The following gene-specific primers were used: mgl1, 5Ј-ATCG-GAGATTAAGTGATCTCCGGCC-3Ј; and mgl2, 5Ј-AACAACGTGG-GAGGTGTCAAGGAC-3Ј. 30 cycles at 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min, and then a final step of 10 min at 72°C were applied. Products from the first round of amplification were then used as template for PCRs with a nested primer: mgl1, 5Ј-GCAATGGCAC-CCATGCCAGAAGATG-3Ј; and mgl2, 5Ј-GCAGCAATAGCACCCAT-GCCAGAAG-3Ј. The amplification protocol was the same as outlined above except that 35 cycles were carried out. The 5Ј-RACE products were cloned directly into pTAg vector (Ingenius) and sequenced.
For inverse PCR, 1 g of T. vaginalis genomic DNA was cleaved to completion with ClaI in a volume of 10 l. After heat inactivation of the ClaI (68°C for 30 min), the DNA was circularized in a 50 l of T4 DNA ligase reaction and then purified (Wizard DNA Clean-Up, Promega) with elution in 40 l of water. PCR was performed on 4-l aliquots with Expand polymerase (Boehringer Mannheim) according to the manufacturer's instructions; 33 cycles and an annealing temperature of 57°C were employed. The outwardly directed primer pairs, corresponding to sequences near the 5Ј-and 3Ј-ends of the mgl1 and mgl2 coding regions, respectively, were: 5Ј-CATGGATGCATGCTGTTGCTG-3Ј with 5Ј-TAT-TGAAGATGCCGACGAACTC-3Ј; and 5Ј-GGAAAGTGTGTCTGTATGT-GTTG-3Ј with 5Ј-GGCTGTGAGAACGTTCAGGAT-3Ј. Gel electrophoresis of the inverse PCR products revealed bands of approximately 2.4 and 1.6 kb (data not shown). With the addition of approximately 1.1 kb corresponding to the distance between each of the primer pairs, these product sizes agree with the ClaI fragment sizes for mgl1 and mgl2 (3.5 and 2.6 kb, respectively) observed by Southern blotting (see Fig. 2). The inverse PCR products were cloned and sequenced on both strands by primer walking.
Southern and Northern Blot Analyses-T. vaginalis genomic DNA (2 g) was restricted, size fractionated on a 0.7% agarose gel, and blotted to Hybond N membrane as described (34). A 1.3-kb KpnI/SpeI fragment from pMGL4100, which contained the open reading frame for mgl1, and a 1.3-kb KpnI/SpeI fragment from pMGL5100, which was nine nucleotides short at the 5Ј-end of being full-length for mgl2, were labeled by random priming (Stratagene). Hybridizations to T. vaginalis genomic DNA and subsequent washing steps were performed as described above for the library screen. T. vaginalis total (10 g) and poly (A) ϩ (1 g) RNA were denatured and run on a 1.5% agarose/formaldehyde gel as described previously (34). RNA was transferred to Hybond N membrane. A 1-kb EcoRI/XhoI fragment from pMGL4100 and an 800-bp EcoRI fragment from pMGL5100 were labeled by random priming. Hybridizations and washing steps were carried out under high stringency as for the Southern hybridizations.
Production of Recombinant MGL1 and MGL2-rMGL1 and rMGL2 were produced in E. coli using the pQE vectors (Qiagen). The mgl1 coding region was amplified by PCR from the cDNA clone pMGL4100 using the oligonucleotides 5Ј-CGCCATGGCTCACGAGAGAATGAC-3Ј and 5Ј-GCAGATCTTAAAAGAGCGTCAAGGCCC-3Ј. The amplified PCR product was cloned into NcoI/BglII pQE60 to give plasmid clone pGL14. A 1.3-kb BamHI/XhoI fragment from pMGL5100 was cloned into BamHI/SalI pQE30 to give plasmid clone pGL15. This resulted in a recombinant protein containing the sequence MRGSHHHHHHG-SPGLQEFGTSA at the NH 2 terminus. The alanine residue is the first mgl2-encoded amino acid (see Fig. 1) and corresponds to the fourth residue of the mgl2 translation product after the start methionine. rMGL1 and rMGL2 were purified from M15pREP4 E. coli after induction with isopropyl-␤-D-thiogalactopyranoside. Briefly, 50 ml of an overnight culture of E. coli containing pGL14 or pGL15 were inoculated into 400 ml of Luria-Bertani broth containing 100 g/ml ampicillin and 25 g/ml kanamycin and grown until they reached an A 600 nm of 0.6 -0.8. The cultures were then induced with 2 mM and 0.2 mM isopropyl-␤-Dthiogalactopyranoside, respectively, and grown for a further 2.25 h before harvesting. The E. coli were resuspended in 5 ml of sonication buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl) and pyridoxal 5Ј-phosphate added to a final concentration of 20 M. The cells were lysed by sonication and the soluble fraction recovered by centrifugation at 10,000 ϫ g for 30 min at 4°C. Soluble fraction was loaded on to a 4-ml Ni 2ϩ -NTA resin column (Qiagen) that had been pre-equilibrated with sonication buffer. After sample application, the column was washed with 30 ml of sonication buffer and with 30 ml of wash buffer (50 mM sodium phosphate, pH 6.0, 300 mM NaCl, 10% v/v glycerol). Recombinant protein was eluted from the column by an FPLC-generated linear gradient of 0 -500 mM imidazole in wash buffer over 100 min at 0.5 ml/min. Fractions containing rMGL1 or rMGL2 were identified by A 280 and by SDS-PAGE. Peak fractions were pooled and dialyzed against 100 mM sodium phosphate buffer, pH 6.5, 300 mM NaCl, 20% v/v glycerol, 20 M pyridoxal 5Ј-phosphate, 15 M dithiothreitol. Dialyzed rMGL1 or rMGL2 was combined 1:1 with enzyme stabilization solution (80% v/v glycerol, 40 M pyridoxal 5Ј-phosphate, 30 M dithiothreitol, 100 mM sodium phosphate buffer, pH 6.5) and stored at Ϫ20°C. Protein concentrations were determined using the BCA protein assay microtiter plate method (Pierce), with bovine serum albumin as standard.
Site-directed Mutagenesis-The production of mutated rMGL1 (C113G) and rMGL2 (C116G) was achieved using the PCR-based QuikChange TM site-directed mutagenesis kit (Stratagene). For mgl1, pGL14 was used as a template directly to give mutated plasmid pGL16. For mgl2, mutagenesis was performed on the pMGL5100 plasmid (pBS backbone) and then subsequently subcloned into pQE30 as described above for pGL15 to give pGL17. Mutagenesis was performed on both plasmids using a pair of oligonucleotide primers complementary to opposite strands of the cDNA clones, each containing a point mutation to convert the respective cysteine codon (TGC) to a glycine codon (GGC), as follows: 5Ј-TGCCTTTATGGCGGCACACATGCTCTCT-3Ј (sense) and 5Ј-AAGAGAGCATGTGTGCCGCCATAAAGG-3Ј (antisense). After PCR-mediated amplification, plasmids containing the desired mutations were identified by sequencing on both strands using gene-specific internal oligonucleotide primers. Production and purification of rMGL1 (C113G) (from plasmid pGL16) and rMGL2 (C116G) (from plasmid pGL17) were performed as described above. Data obtained for two independently derived mutants for each gene were comparable, and only one set of data for each is presented.
Western Blot Analysis-The generation of polyclonal antisera in rabbits against purified rMGL1 and rMGL2 was performed using standard immunization protocols (35). A T. vaginalis homogenate was prepared by the lysis of cells in 0.25 M sucrose, 0.25% Triton X-100 containing proteinase inhibitors (10 M E-64, 2 mM 1,10-phenanthroline, 4 M pepstatin A, and 1 mM phenylmethylsulfonyl fluoride), the addition of an equal volume of 2ϫ SDS loading buffer, and boiling for 5 min. To prepare supernatant fraction, the lysate was centrifuged at 15,000 ϫ g in a microcentrifuge for 10 min. The pellet was resuspended in an equal volume of lysis buffer. rMGL1 and rMGL2 were prepared by boiling in SDS loading buffer. Western blot analysis was carried out as described (34). Filters were blocked with 5% (w/v) low fat dried milk and 10% (v/v) horse serum in Tris-buffered saline (20 mM Tris, pH 7.6, 137 mM NaCl) and incubated with polyclonal preimmune or immune rabbit serum diluted (1:500) in blocking reagent. Bound antibody was detected by using horseradish peroxidase-coupled secondary antibodies and ECL Western blotting detection reagents (Amersham).
Enzyme Assays-The activity of rMGL1 and rMGL2 toward a variety of substrates was measured by monitoring 2-ketoacid production as described (36) or the production of hydrogen sulfide from homocysteine (13). The breakdown of cysteine and O-acetylserine (both at 10 mM) by the two recombinant proteins was determined by assaying for the production of pyruvate as described (37). All assays were carried out at 37°C using 0.1 M imidazole buffer, pH 6.5. Kinetic calculations were performed using the computer program Grafit (Erithacus Software). Homocysteine desulfurase activity in native polyacrylamide gels was detected by immersion at 37°C in a reaction mixture containing 3.3 mM DL-homocysteine, 0.33 mM lead acetate, 10 mM 2-mercaptoethanol, and 100 mM imidazole buffer, pH 6.5. The effect of propargylglycine (Sigma) was determined by immersion of the gel in 10 mM propargylglycine in 0.1 M imidazole buffer, pH 6.5, for 5 min at 37°C before immersion in the homocysteine desulfurase detection mixture. Native PAGE using 12% gels was the same as for SDS-PAGE except that SDS was omitted from the gel and buffers, reducing agent was omitted from the sample buffer, and the sample was not boiled.

RESULTS
Isolation and Characterization of T. vaginalis mgl1 and mgl2-In the absence of data on the protein sequence of the trichomonad methionine ␥-lyase or the gene sequence of methionine ␥-lyases from other organisms, the approach we adopted to clone the trichomonad methionine ␥-lyase gene was to use PCR with degenerate oligonucleotide primers based on regions of homology identified in the related enzyme cystathionine ␥-lyase. This is another member of the ␥-family of pyridoxal 5Ј-phosphate-dependent enzymes, one for which sequence information was available from yeast, rat, and human (30 -32). The peptide sequences chosen were VWIETPTN (5Јprimer) and ATKY(M/I)NG (3Ј-primer). The 3Ј-primer corresponds to the pyridoxal 5Ј-phosphate binding domain, and the 5Ј-primer corresponds to a region highly conserved in cystathionine ␥-lyase molecules and differing from other known pyridoxal 5Ј-phosphate-dependent enzymes of the ␣-and ␤-elimination families (20). PCR using T. vaginalis first strand cDNA resulted in the amplification of 200-bp DNA fragments that were of the expected size. Cloning and sequencing of these revealed that they belonged to two independent groups, both of which had predicted peptide similarity to yeast cystathionine ␥-lyase but only about 50% identity with each other. The 200-bp insert of a representative clone from each group was used to screen a T. vaginalis cDNA library. Of the clones isolated, pMGL4100 contained a 1.3-kb insert with the complete open reading frame of the methionine ␥-lyase 1 gene (mgl1), whereas pMGL5100 contained a truncated gene (mgl2) missing the 5Ј-end including the ATG start codon.
Genomic clones containing the mgl1 or mgl2 genes were isolated by inverse PCR. A 2.4-kb fragment containing part of the mgl1 gene and flanking sequence was amplified from a 3.5-kb ClaI (see Fig. 2A, lane 1) fragment, while a 1.6-kb fragment containing part of the mgl2 gene and flanking sequence was amplified from a 2.5-kb ClaI fragment (see Fig. 2B, lane 1). Sequence data for the these PCR fragments were combined with the cDNA sequences determined from pMGL4100 and pMGL5100 to give the complete sequence of the mgl1 and mgl2 genes. 5Ј-RACE was used to map the putative site of initiation of transcription for mgl1 and mgl2. mgl1 and mgl2 were found to have very short 5Ј-untranslated regions of 13 and 14 bp, respectively. The sequences from the 5Ј-termini to the A of the ATG start codon were ATTTTTAGACAAC (mgl1) and ACTTTATATAAAAG (mgl2). mgl1 is predicted to encode a protein (MGL1) of 396 amino acids with a molecular mass of 42.9 kDa, and mgl2 predicts a protein (MGL2) of 398 amino acids with a molecular mass of 43.1 kDa (Fig. 1). Amino acid sequence comparison of MGL1 and MGL2 revealed that the two trichomonad proteins had 69% identity. Data bank searches revealed the highest level of amino acid sequence identity for the two trichomonad proteins with the two methionine ␥-lyases from P. putida (26,27) (44 -45%) and cystathionine ␥-lyases from yeast (32) and human (30) (43-44%). Significant levels of sequence identity (30 -38%) were also observed for MGL1 and MGL2 with a number of other sulfur amino acid-metabolizing enzymes, such as O-acetylhomoserine sulfydrylase from yeast (38) and cystathionine ␥-synthase from E. coli (39), which are also members of the ␥-family of pyridoxal 5Ј-phosphate-dependent enzymes (20). 1 An alignment of methionine ␥-lyases from T. vaginalis and P. putida and cystathionine ␥-lyases from human and yeast is shown in Fig. 1. The two underlined sequences correspond to the areas of consensus to which the PCR primers were designed. 3/8 residues of the 5Ј-consensus site in MGL1 and MGL2 differed from those of yeast cystathionine ␥-lyase, whereas all 8 residues were the same in the 3Ј-consensus site which corresponds to the pyridoxal 5Ј-phosphate binding region and contains the essential lysine residue. 35 residues that

FIG. 1. Amino acid sequence alignment of T. vaginalis methionine ␥-lyases, MGL1 and MGL2 (this study), with two methionine ␥-lyases from P. putida (PpMGL1 (26) and PpMGL2 (27)) and cystathionine ␥-lyases from yeast (ScCYS3 (32)) and human (HsCGL (30)).
Each individual sequence is numbered accordingly. Residues that are conserved in all sequences are indicated by black boxes. Residues that are conserved in the four methionine ␥-lyase sequences but are absent from the cystathionine ␥-lyase sequences are indicated by a #. The cysteine residue shown to be important for methionine ␥-lyase activity of P. putida (25) and mutated in this study is indicated by a star (ૺ). are conserved in the four methionine ␥-lyase sequences but absent from the cystathionine ␥-lyases are indicated. These residues may play a role in defining the substrate specificity of the enzymes. One of these residues, cysteine 113 (numbering for MGL1, starred in Fig. 1), is of particular interest because it has been shown to be important for methionine ␥-lyase activity in P. putida (25,40). In addition, the MGL sequences have a 7-amino acid insertion relative to the cystathionine ␥-lyases toward the NH 2 terminus (residues 49 -55 in MGL1).
Genomic Organization and Expression of mgl1 and mgl2-To assess the copy number of mgl1 and mgl2 in T. vaginalis, genomic DNA was digested with restriction enzymes and analyzed by Southern blotting using the mgl1 and mgl2 cDNAs as probes (Fig. 2). For each of the five restriction enzymes used, a single major hybridizing band was detected with each of the mgl1 and mgl2 probes, which is consistent with the presence of single copy genes. For example, the 3.5-kb ClaI fragment detected by the mgl1 probe ( Fig. 2A, lane 1), which was cloned by inverse PCR, revealed the presence of one open reading frame, and likewise one copy of mgl2 was found on the 2.6-kb ClaI fragment detected by the mgl2 probe (Fig. 2B, lane 1). At the high stringency conditions used for the Southern blot, the two genes did not cross-hybridize to a significant extent. Both mgl1 and mgl2 probes hybridized to a single 7.5-kb EcoRI band (Fig.  2, A and B, lane 2). This raises the possibility that the two genes are linked, but this has not been investigated further.
To examine the expression of mgl1 and mgl2, Northern blots were performed with total and poly(A) ϩ RNA. A single 1.3-kb transcript was detected with both the mgl1-and mgl2-specific probes (data not shown). The length of the mRNAs for the two genes corresponds well with the size of the cDNA clones isolated, the position determined for the addition of the poly[A] tails, and the sites mapped for transcription initiation.
Expression and Purification of Recombinant MGL1 and MGL2-The two T. vaginalis mgl genes were cloned into pQE vectors (Qiagen) for expression of recombinant enzyme with a six-histidine tag. The complete open reading frame of mgl1 was inserted by PCR into the pQE60 vector, which utilizes the normal ATG start codon of the mgl1 gene and adds a sixhistidine tag at the COOH terminus. The generation of an NcoI site at the NH 2 terminus, however, introduced a proline residue in place of a serine at position 2 of the recombinant protein (rMGL1). The pMGL5100 cDNA clone was used to insert the mgl2 gene into the pQE30 vector. This resulted in a fusion protein with 21 residues of vector-derived sequence, including the six-histidine tag, at the NH 2 terminus, followed by the MGL2 sequence. The pMGL5100 clone is not full-length, which resulted in the recombinant protein (rMGL2) lacking the first three amino acids predicted to be present in the NH 2 terminus of the native protein. In this construct the native stop codon was utilized. rMGL1 and rMGL2 were expressed in M15pREP4 E. coli and subsequently affinity purified by FPLC on Ni 2ϩ -NTA resin. About 10 mg of recombinant enzyme was purified from each 450 ml of culture. Denaturing SDS-PAGE analysis confirmed the predicted sizes of rMGL1 and rMGL2 (43.8 and 47.2 kDa) and showed a high degree of purity (Fig. 3A, lanes 1  and 2). The major protein detected by SDS-PAGE for each of the MGL1 and MGL2 preparations (Fig. 3B, lanes 1 and 2) was shown to be coincident with enzyme activity as detected by a native PAGE in situ assay for homocysteine desulfurase (Fig.  3B, lanes 3 and 4). Methionine ␥-lyase purified from T. vaginalis has high homocysteine desulfurase activity (13). The in situ homocysteine desulfurase activity of the two recombinant proteins was also inhibited completely by propargylglycine, an inhibitor of pyridoxal 5Ј-phosphate-linked enzymes (Fig. 3B,  lanes 5 and 6). Gel filtration chromatography showed that the molecular mass of rMGL2 was approximately 160 kDa (data not shown). This suggests rMGL2 is a homotetramer.
Purified rMGL1 and rMGL2 were used to immunize rabbits to generate polyclonal antibodies. Western blot analysis showed that the rabbit serum raised against rMGL1 was strongly reactive against purified rMGL1 and rMGL2 (Fig. 3C,  lanes 1 and 2). The rMGL1 antiserum also recognized a single protein of 43 kDa in T. vaginalis whole cell homogenate (lane 3) and soluble extract (lane 5), but not in a sedimented fraction (lane 4). Immunoblot analysis of rMGL2, T. vaginalis homogenate, and soluble fraction with the rabbit serum raised against rMGL2 also showed that they were strongly reactive, a 43-kDa protein being recognized (Fig. 3D, lanes 2, 3, and 5). The rMGL2 antiserum, however, did not cross-react significantly with either purified rMGL1 or a T. vaginalis sedimented fraction (Fig. 3D, lanes 1 and 4). No bands were detected when duplicate blots were probed with preimmune rMGL1 or rMGL2 antisera (not shown).
Enzymatic Activity of rMGL1 and rMGL2-Methionine ␥-lyase purified from T. vaginalis has activity toward a number of substrates, including methionine, homocysteine, cysteine, and O-acetylserine, but has no activity toward cystathionine (13). To assess the similarity between MGL1 and MGL2 and purified native methionine ␥-lyase, the enzyme activities of the recombinant enzymes were analyzed. rMGL1 and rMGL2 were found to have very high activity toward homocysteine and also to catabolize methionine, cysteine, and O-acetylserine rapidly ( Table I). The two recombinant enzymes were unable to utilize cystathionine as a substrate (activity less than 1 nmol/min/mg of protein). The kinetic parameters of the two recombinant proteins were also determined for homocysteine and cysteine (Table II). The apparent K m values of rMGL1 toward these substrates were lower than those for rMGL2, but the largest difference was for methionine for which the apparent K m values of rMGL1 and rMGL2 were 0.65 Ϯ 0.02 mM and 10.6 Ϯ 4.0 mM, respectively.
Enzymatic Activities of Mutated rMGL1 (C113G) and rMGL2 (C116G)-Chemical modification studies have revealed that the cysteine residue at position 116 of the P. putida methionine ␥-lyase is important, although not essential, for catalytic activity (25). Additionally, specific labeling of this residue with the pyridoxal 5Ј-phosphate analog N-(bromoacetyl)pyridoxamine phosphate confirmed that it is located at or near the active site (40). This cysteine residue is conserved in T. vaginalis MGL1 and MGL2 but is not present in other ␥-family enzymes. For example, a glycine residue is present at this position in the cystathionine ␥-lyase of yeast and human (Fig. 1). To test for the importance of this cysteine residue in T. vaginalis methionine ␥-lyase activity, site-directed mutagenesis was used to change this cysteine to a glycine for rMGL1 and rMGL2 (to generate rMGL1(C113G) and rMGL2(C116G)). The mutant enzymes were compared with rMGL1 and rMGL2 (Table I). Under optimal conditions, the activities of rMGL1(C113G) toward all substrates were considerably lower than those of rMGL1, and the residual levels of activity correlated well with those reported previously for chemically modified P. putida methionine ␥-lyase (25). rMGL2(C116G) also had lower activities toward homocysteine and methionine than rMGL2, but surprisingly the activity of the mutated enzyme toward cysteine and O-acetylserine was increased to 217 and 142%, respectively. Neither of the mutated enzymes exhibited activity toward cystathionine.
Comparative kinetic analyses of the mutated and wild type recombinant enzymes with respect to the catabolism of homocysteine and cysteine were performed (Table II). The slightly higher K m and markedly lower V max values of rMGL1(C113G) compared with those of wild type rMGL1 suggest reduced substrate binding efficiency of this mutated enzyme. The chemically modified P. putida methionine ␥-lyase also exhibited increased K m values for a range of cysteine-analog substrates (25). In contrast, the apparent K m of rMGL2(C116G) for cysteine was considerably lower than that of rMGL2, and this correlates with the enhanced activity (higher V max ) of the mutated enzyme toward this substrate. Curiously, the K m of rMGL2(C116G) for homocysteine was also much reduced relative to that of the wild type enzyme, despite the significantly lower V max of the mutated enzyme with this substrate. DISCUSSION A native methionine ␥-lyase activity from T. vaginalis has been purified and characterized previously (13). The two genes mgl1 and mgl2 from T. vaginalis, as reported in this study, have methionine ␥-lyase activity when expressed in E. coli. Recombinant MGL1 and MGL2 have substrate specificity profiles similar to that of the native enzyme, catalyzing the breakdown of methionine, homocysteine, cysteine, and O-acetylserine but being inactive toward cytathionine (Table I). mgl1 and mgl2 are both highly transcribed genes with similar sized mRNAs, and the protein products are both about 45 kDa (Fig.  3, C and D, lane 3), comparable with the native enzyme, which was estimated to have a molecular mass of 160 kDa composed of four apparently identical 43-45 kDa-subunits (13). Because rMGL2 was estimated to be about 160 kDa, this suggests that the recombinant enzyme is also a tetramer composed of identical subunits. It is possible that the purified enzyme charac-  terized in the earlier study was composed of a mixture of MGL1 and MGL2, although it is interesting that in the first round of purification of the native enzyme by ion exchange chromatography two peaks of activity were observed. Only one of the peaks, corresponding to 90% of the activity, was purified further (13). The K m of the purified native enzyme toward methionine was 4.3 mM, roughly halfway between that determined for rMGL1 (0.65 mM) and rMGL2 (10.6 mM). The similarity between MGL1 and MGL2 at the primary sequence level means that heterotetramers could occur in vivo, and they may differ in activity from homotetramers. Because the subunit sizes of MGL1 and MGL2 are very similar, it is unlikely that heterotetramers and homotetramers would have been distinguished in the original analysis of the native enzyme. The finding that the K m values of the two recombinant enzymes toward homocysteine were both higher than that reported for the native enzyme (0.5 mM) does not assist in judging whether just one of the genes encodes the major native trichomonad activity. Thus whereas it seems very likely that mgl1/mgl2 encode methionine ␥-lyase activity in T. vaginalis, it is not currently known how much each gene contributes toward the total enzyme activity of the parasite. The level of sequence identity between the trichomonad enzymes is 69%, but they are significantly less similar (44%) to the Pseudomonas enzymes. The trichomonad and bacterial methionine ␥-lyases are more closely related phylogenetically to each other than to other members of the ␥-family of pyridoxal 5Ј-phosphate-dependent enzymes (20), 1 and they share several features. Both catabolize methionine and homocysteine; however, the Pseudomonas recombinant enzyme can also use cystathionine as a substrate (27) whereas the trichomonad enzyme cannot. Interestingly, it was reported that the native P. putida enzyme was not active toward cystathionine, whereas the Aeromonas enzyme had such activity (14). A low activity apparently of cystathionine ␥-lyase was detected in crude lysates of T. vaginalis (41), but the enzyme responsible for this is unknown, and it could simply reflect a broad substrate specificity of another enzyme. The success of the strategy adopted for the cloning of the two trichomonad methionine ␥-lyase genes and the finding that the use of the degenerate primers (based on conserved regions in cystathionine ␥-lyases) did not lead to the amplification of fragments of other genes suggest that T. vaginalis does not contain cystathionine ␥-lyase itself. There is little information on the presence, or otherwise, of other trans-sulfuration enzymes, and other members of the ␥-family of pyridoxal 5Ј-phosphate-dependent enzymes, in trichomonads. They contain high levels of an activity designated serine sulfydrase (19) that catalyzes the conversion of cysteine to serine and hydrogen sulfide. This activity is thought to be mediated by cystathionine ␤-synthase, the first enzyme in the trans-sulfuration pathway, in many cells (15). Cystathionine ␤-synthase is a member of the ␤-family of pyridoxal 5Ј-phosphate-dependent enzymes (20). To date, however, cystathionine synthesis, and indeed trans-sulfuration reactions in general, have not been shown to occur in trichomonads. Thus the current data clearly suggest that sulfur amino acid metabolism of T. vaginalis differs considerably from that in mammals and that the mgl genes of T. vaginalis are more closely related to bacterial genes than any detected so far in eukaryotes.
Chemical modification studies on the P. putida methionine ␥-lyase showed that a cysteine residue (Cys-116, P. putida numbering) was important, but not essential, for activity of the enzyme (25,40). The equivalent cysteine residue is present in the two trichomonad methionine ␥-lyases, but not in other ␥-family pyridoxal 5Ј-phosphate-dependent enzymes (Fig. 1). For example, yeast cystathionine ␥-lyase has a glycine in this position and indeed has no cysteine residues at all (32). This suggests that the cysteine residue at this position cannot be directly involved in the catalytic mechanism, although clearly it could be involved in mediating substrate binding and so determining specificity. Mutation of this cysteine residue to a glycine in rMGL1 reduced ␣,␥-elimination activity, for example the breakdown of homocysteine, by 90% and ␣, ␤-elimination activity, for example the breakdown of cysteine, by 61%, which correlates well with data obtained for P. putida (25). Mutation of this cysteine residue in rMGL2 to a glycine, however, although reducing ␣,␥-elimination activity by 80%, resulted in more than twice the ␣,␤-elimination activity. These data confirm the importance of this cysteine residue in substrate specificity. They also show that despite the high similarity between the two trichomonad enzymes there clearly are differences in the active sites which have important implications for the catalytic mechanisms of ␣,␤-elimination and ␣,␥-elimination reactions. We must await data on the three-dimensional structure of the active site of the two trichomonad enzymes for a thorough interpretation of these differences and the role of the cysteine residue. The recent success in solving the structure of the ␥-family enzyme cystathionine ␤-lyase (21) should enable some useful modeling of the active site of trichomonad methionine ␥-lyases. However, for a definitive study data on the crystal structure of the trichomonad protein are required. To this end, rMGL1 has recently been crystallized, and structural studies are under way.