Biochemical and Genetic Analysis of Methylenetetrahydrofolate Reductase in Leishmania Metabolism and Virulence*

Methylenetetrahydrofolate reductase (MTHFR; EC 1.5.1.20) is the sole enzyme responsible for generation of 5-methyltetrahydrofolate, which is required for methionine synthesis and provision of methyl groups via S-adenosylmethionine. Genome analysis showed that Leishmania species, unlike Trypanosoma brucei and Trypanosoma cruzi, contain genes encoding MTHFR and two distinct methionine synthases. Leishmania MTHFR differed from those in other eukaryotes by the absence of a C-terminal regulatory domain. L. major MTHFR was expressed in yeast and recombinant enzyme was produced in Escherichia coli. MTHFR was not inhibited by S-adenosylmethionine and, uniquely among folate-metabolizing enzymes, showed dual-cofactor specificity with NADH and NADPH under physiological conditions. MTHFR null mutants (mthfr–) lacked 5-methyltetrahydrofolate, the most abundant intracellular folate, and could not utilize exogenous homocysteine for growth. Under conditions of methionine limitation mthfr– mutant cells grew poorly, whereas their growth was normal in standard culture media. Neither in vitro MTHFR activity nor the growth of mthfr– mutants or MTHFR overexpressors were differentially affected by antifolates known to inhibit parasite growth via targets beyond dihydrofolate reductase and pteridine reductase 1. In a mouse model of infection mthfr– mutants showed good infectivity and virulence, indicating that sufficient methionine is available within the parasitophorous vacuole to meet the needs of the parasite.

Protozoan parasites of the genus Leishmania are the causative agents of human diseases that range from lethal visceral infection to severely disfiguring mucocutaneous leishmaniasis. Leishmania are transmitted by phlebotomine sand flies, where they reside extracellularly within the digestive tract. Following infection of mammalian hosts, parasites multiply intracellularly, primarily within macrophage phagolysosomes. No approved vaccines are available against these pathogens and current chemotherapy relies on pentavalent antimony, which cannot be given orally, has significant toxicity, and is facing increasing resistance in the field (1,2). New drugs and new drug targets are therefore urgently needed.
Leishmania that infect mammals are methionine auxotrophs (3,4). Methionine occupies a central place in metabolism because it is required for protein and S-adenosylmethionine (AdoMet) 2 biosynthesis. AdoMet is particularly important in anabolism as it is the methyl donor in a wide variety of biosynthetic reactions, and decarboxylated AdoMet is used in polyamine biosynthesis. In Trypanosoma brucei, enzymes that consume and recycle AdoMet are well characterized chemotherapeutic targets (5)(6)(7), however, these pathways are less well studied in Leishmania. Homocysteine arising from AdoMet-dependent methylations is released by hydrolysis of S-adenosylhomocysteine, the product of methylation reactions. Homocysteine is then remethylated by methionine synthases using 5-methyltetrahydrofolate (5-CH 3 -THF) as the methyl donor. The sole route of 5-CH 3 -THF biosynthesis in eukaryotes is reduction of 5,10-methylenetetrahydrofolate (5,10-CH 2 -THF) by methylenetetrahydrofolate reductase (MTHFR, EC 1.5.1.20) (8,9). Thus here we focused on MTHFR as a key step leading to subsequent methionine metabolism.
MTHFR is either a NADH or NADPH-dependent flavoenzyme in eukaryotes and most prokaryotes (10 -13). In mammals and yeast, NADPH-dependent reduction of 5,10-CH 2 -THF by MTHFR is physiologically irreversible and is inhibited by AdoMet (9, 14 -17). Two major domains have been identified in the mammalian MTHFR: an N-terminal catalytic domain and a C-terminal allosteric regulatory domain that binds AdoMet (Fig. 1B) (9). Inhibition by AdoMet may prevent an NADPH-dependent and thus physiologically irreversible MTHFR from depleting 5,10-CH 2 -THF through the "methyltrap" effect (9,18,19). Under physiological conditions NADPH is strongly preferred by the mammalian enzymes (20), although at high phosphate concentrations porcine MTHFR is able to use NADH or NADPH with equal effectiveness. Dual cofactor specificity is unusual as almost all pyridine nucleotide-dependent oxidoreductases are highly specific for one cofactor, with mammalian glutamate dehydrogenase being a well characterized exception (21)(22)(23). In contrast, reduction of 5,10-CH 2 -THF by NADH-dependent MTHFR enzymes appears to be physiologically reversible and these proteins either contain a C-terminal domain that does not mediate allosteric regulation, as in plants (12,17), or lack this domain entirely, as in Escherichia coli (24). A reversible MTHFR activity may also participate in salvage of host 5-CH 3 -THF in Plasmodium falciparum (25).
The metabolism of reduced folate cofactors is of particular interest in drug discovery, because dihydrofolate reductase (DHFR) is commonly targeted by antimalarials and antibacterials. However, current therapeutic antifolates are ineffective against Leishmania because DHFR inhibition is by-passed by pteridine reductase 1 (PTR1), which is relatively insensitive to classical antifolates (26 -28). Notably, a group of compounds derived from diaminopyrimidine, quinazoline, or pteridine scaffolds have been described that inhibit both DHFR and PTR1 and show nanomolar EC 50 values in culture. However, these inhibitors appear to additionally possess independent mode(s) of action rendering their toxicity insensitive to deletion or overexpression of PTR1 and/or DHFR-TS (29 -31). Increased methionine adenosyltransferase activity and AdoMet facilitate acquisition of methotrexate resistance in Leishmania major (32), and MTHFR has been hypothesized to be involved in this process. This suggests that 5-CH 3 -THF metabolizing enzymes such as MTHFR or methionine synthase may be additional targets of antifolates.
Here we show that L. major MTHFR (LmjMTHFR) is unique among characterized MTHFRs because it is AdoMetinsensitive and does not discriminate between NADH and NADPH. However, deletion of MTHFR and elimination of CH 3 -THF biosynthesis in L. major had no effect on viability (under conditions where methionine was not limiting), on sensitivity to antifolates, on the ability to salvage 5-CH 3 -THF, or on virulence.
Parasite Culture and Infectivity-All studies used derivatives of L. major Friedlin (MHOM/IL/80/Friedlin) clone V1. Promastigotes were routinely grown at 27°C in M199 medium (U. S. Biological Corp.) supplemented with 40 mM HEPES, pH 7.4, 50 M adenosine, 1 g ml Ϫ1 biotin, 5 g ml Ϫ1 hemin, 2 g ml Ϫ1 biopterin, and 10% (v/v) heat-inactivated fetal calf serum. Null mutants and mthfr Ϫ /ϩMTHFR clonal lines were generated as described in supplemental methods and Ref. 70. These lines were maintained in M199 containing 20 g ml Ϫ1 hygromycin and 30 g ml Ϫ1 puromycin for mthfr Ϫ and 12.5 g ml Ϫ1 G418 for mthfr Ϫ /ϩMTHFR. Growth under semi-defined conditions used RPMI medium (Invitrogen) supplemented with 30 mM HEPES, pH 7.4, 62.5 M adenosine, 2 g ml Ϫ1 biopterin, 5 g ml Ϫ1 hemin, and 1% (v/v) heat-inactivated fetal calf serum. RPMI medium was also prepared lacking methionine (RPMI-Met) or lacking folic acid and biopterin (RPMI-FA-Biop). Growth inhibition was determined by seeding parasites at 5 ϫ 10 5 cells ml Ϫ1 at various concentrations of drug and counting cells using a model Z1 Coulter counter when control cultures reached late-log phase. The virulence of wild-type (WT) and mthfr Ϫ cell lines were determined by infection of metacyclic parasites (purified by negative selection with peanut agglutinin as described (33)) into the footpad of susceptible BALB/c mice, followed by measurement of lesion growth, as described (34).
Cloning of L. major MTHFR-The MTHFR coding sequence was amplified with recombinant Pfu DNA polymerase (Stratagene) from L. major genomic DNA, which was prepared by the LiCl mini-prep method (35). PCR used the primers 5Ј-ATCAT-GTCCAAGCTTATCAG and 5Ј-TCACTCCACTAGCGCGT-TGA. The amplified gene was cloned into pGEM-T (Promega), to create pGEM-T.MTHFR (B4775) and then subcloned into the SmaI site of pXGneo (36) to create pXG-LmjMTHFR (B5128). All constructs were confirmed by DNA sequencing.
Expression of L. major MTHFR in Saccharomyces cerevisiae-The MTHFR coding sequence was amplified from the pGEM-T.MTHFR template using primers 5Ј-TGCAGGATCCAT-GTCCAAGCTTATCAGT (forward) and 5Ј-TGCACTG-CAGTCACTCCACTAGCGCGTT (reverse). This amplicon was digested with BamHI and PstI and cloned into the corresponding sites in pVT103-U (37). This plasmid contains URA3 for selection and an ADH1 promoter driving transgene expression. The construct and the empty pVT103-U vector were introduced into yeast strain RRY3, a met12 met13 double deletant (38). Control strains (RRY3 containing a cDNA encoding human MTHFR, and WT strain DAY4) were as previously described (12,38).
Yeast Growth and Preparation of Extracts-Cultures of strain RRY3 expressing LmjMTHFR, human MTHFR, or the corresponding empty vector control were grown as described (12). Desalted protein extracts were obtained as previously described (12). These crude extracts were stored in aliquots at Ϫ80°C after freezing in liquid N 2 . Protein concentrations were determined by the method of Bradford (39), using bovine serum albumin as standard.
Expression and Purification of Recombinant LmjMTHFR-An expression construct of LmjMTHFR was kindly provided by the Structural Genomics of Parasitic Protozoa consortium (clone Lmaj006157AAA, B5503). This used the expression vector pAVA0421 (40) that adds an N-terminal hexahistidine tag to the recombinant protein. Protein was expressed from this vector and purified by metal affinity chromatography, as described under supplemental methods.
Spectrophotometric MTHFR Assays-Spectrophotometric assays were carried out using a Beckman DU-640 spectrophotometer with 1-ml assays maintained at 37°C. The NAD(P)Hmenadione oxidoreductase activity of MTHFR was measured essentially as described (41). Assays contained 50 mM sodium phosphate, pH 7.2, 300 M EDTA, 10 M flavin adenine dinucleotide, or the same solution buffered by 50 mM Tris-Cl, pH 7.2. MTHFR was preincubated in the cuvette with NAD(P)H for 5 min before initiating reactions by adding menadione to a final concentration of 130 M. Activity was monitored by following NAD(P)H oxidation at 343 nm (the isosbestic point of menadione), using the extinction coefficient of 6220 M Ϫ1 cm Ϫ1 (11). The concentration of menadione stocks was determined as described (14).
The NADH-5,10-CH 2 -THF oxidoreductase activity of MTHFR was determined essentially as described (41) in deoxygenated phosphate assay buffer containing 10 mM formaldehyde. (6R,6S)-5,10-CH 2 -THF was produced by anaerobic addition of varying concentrations of THF 10 min before initiation of the reaction by enzyme addition. Formaldehyde stocks were made daily from paraformaldehyde and THF stocks were made directly before use by dissolving THF to 5 mM in 250 mM triethanolamine-Cl, pH 7, containing 40 mM 2-mercaptoethanol. THF solutions were kept on ice under anaerobic conditions in the dark. Activity was monitored by following NADH oxidation at 340 nm, using the extinction coefficient of 6230 M Ϫ1 cm Ϫ1 . Calculations were based on the assumptions that L. major MTHFR resembles other MTHFRs in using only (6R)-5,10-CH 2 -THF and that this isomer comprised 50% of the racemic mixture.

RESULTS
Identification of Leishmania MTHFR-L. major MTHFR (LmjF36.6390) was identified in searches of the Leishmania genome databases; orthologs were also detected in Leishmania infantum (LinJ36.0220) and Leishmania braziliensis (LbrM35.6090). The L. major gene encoded a predicted protein of 309 residues with a mass of 34.8 kDa, which shows no obvious intracellular targeting sequences. LmjMTHFR was 51 and 50% identical to Zea mays MTHFR and Arabidopsis thaliana MTHFR-2, respectively (12), but only 36% identical to E. coli (10) and 32% identical to human (46) MTHFRs. LmjMTHFR is, however, much closer in length to the 296-residue E. coli enzyme than mammalian and plant MTHFRs, due to the absence of the C-terminal extension of ϳ300 residues present in these proteins (Fig. 1). LmjMTHFR also lacked an N-terminal extension that is a site of regulatory phosphorylation in the human enzyme (47). Reanalysis of the L. major genome sequence data confirmed that these differences did not arise from assembly errors (not shown).
That LmjMTHFR indeed has MTHFR activity in vivo was shown by complementation in a yeast methionine auxotroph strain. The met12 met13 double deletant RRY3, which lacks MTHFR activity (38), grew in the absence of methionine after transformation with a LmjMTHFR expression plasmid, but not a control plasmid (Fig. 2). The growth rate of these transformants was similar to that of the WT strain DAY4.
Characterization of Leishmania MTHFR Activity-LmjMTHFR activity was studied both in crude yeast extracts and using purified enzyme obtained from engineered E. coli, with radiometric (yeast) or spectrophotometric assays (purified enzyme). Yeast expressing LmjMTHFR showed high levels of activity with 5,10-CH 2 -THF (about 80 nmol min Ϫ1 mg Ϫ1 protein), whereas none was detected in the mthfr Ϫ RRY3 mutant containing vector alone (data not shown). A purification procedure was developed that yielded recombinant enzyme of Ͼ95% purity in a single chromatographic step (supplemental data Fig. S1). Purified LmjMTHFR showed an absorbance spectrum typical of a flavoprotein, with prominent peaks at 381 and 454 nm. The 454-nm peak disappeared upon addition of excess NADH or NADPH, indicating reduction of a flavin cofactor (supplemental data Fig. S2).
When assayed in crude yeast extracts, LmjMTHFR was found to use either NADH or NADPH as reductant (Fig. 3A), which was unexpected because previously characterized MTHFRs strongly prefer one or the other. In addition, LmjMTHFR was insensitive to AdoMet, consistent with the absence of the C-terminal domain containing the AdoMetbinding site in mammalian and yeast MTHFRs (9,17). As a control, extracts of yeast expressing human MTHFR were tested similarly (Fig. 3B). Here, the human enzyme strongly preferred NADPH and was AdoMet-sensitive, as previously reported (12), with about 50% inhibition by 1 mM AdoMet. That the activity of human MTHFR was only ϳ25% of that of LmjMTHFR may reflect differences in stability or expression levels, because the specific activity of purified LmjMTHFR was comparable with that of mammalian enzymes (see below).
Purified LmjMTHFR tended to lose activity upon dilution, which was prevented by addition of flavin-adenine dinucleotide (data not shown), as with the E. coli enzyme (24). The pyridine nucleotide cofactor specificity of the purified LmjMTHFR was identical to that shown when assayed in yeast extracts. Using menadione as the electron acceptor, kinetic constants were determined for NADH and NADPH in phosphate and Tris buffers ( Table 1). The specificity constants for NADH and NADPH were the same in both buffers, showing that LmjMTHFR was efficiently reduced by either nucleotide. 5,10-CH 2 -THF was also an effective substrate, with a k cat /K m 1.3-fold lower than the NADH/menadione reaction and K m(app) and k cat for 5,10-CH 2 -THF within the range of previously determined values (10,11). In contrast to the E. coli and pig liver enzymes (10,11), substrate inhibition was not seen, up to a maximal concentration tested of 250 M (6R)-5,10-CH 2 -THF. In addition, no folic acid, dihydrofolic acid, biopterin, or dihydrobiopterin reductase activities could be detected, in the pH range of 8 to 5 (sensitivity was Ͻ0.1 mol min Ϫ1 mg Ϫ1 ).
LmjMTHFR activity expressed in crude yeast extracts was tested for inhibition by a variety of pteridine analogs shown previously to inhibit the pteridine reductases PTR1 and DHFR (29), using NADH as reductant. None of these compounds significantly inhibited MTHFR activity (supplemental data Fig.  S3). Using engineered L. major mutants described below, neither deletion nor overexpression of MTHFR had a significant effect on parasite sensitivity toward methotrexate or three of    Table S1).
Homozygous Replacement of MTHFR in L. major-A homozygous null mthfr Ϫ mutant was produced by serial replacement of the MTHFR open reading frame with puromycin and then hygromycin resistance gene open reading frames (supplemental methods). Gene replacement was confirmed by Southern analysis with a probe flanking the MTHFR locus (supplemental data Fig. S4). As discussed below, the null mutant grew well in standard M199 media. As a positive control, an episomal copy of MTHFR was reintroduced using the pXG expression vector to produce an "add-back" clonal line termed mthfr Ϫ /ϩMTHFR. The absence of MTHFR in mthfr Ϫ and its presence in mthfr Ϫ /ϩMTHFR was confirmed by PCR with primers to the MTHFR open reading frame (data not shown).
Lack of 5-CH 3 THF in mthfr Ϫ Cells-Given the lack of an obvious phenotype in the null mutant, folates were analyzed in the WT, mthfr Ϫ , and mthfr Ϫ /ϩMTHFR lines to exclude the possibility that an alternative pathway could generate 5-CH 3 -THF. In contrast to WT, where 5-CH 3 -THF represents 60% of the intracellular folate pool, no 5-CH 3 -THF was detectable in mthfr Ϫ cells (140 Ϯ 50 versus less than 0.6 pmol mg Ϫ1 , respectively: Table 2). There were no other marked differences between the folate pools of WT, mthfr Ϫ , and mthfr Ϫ /ϩMTHFR lines. As expected, reintroduction of MTHFR in the mthfr Ϫ / ϩMTHFR line restored 5-CH 3 -THF production ( Table 2), to a level of 230 pmol mg Ϫ1 , which exceeds that seen in WT. This may result from overexpression of MTHFR from the multicopy episomal vector, as seen with other proteins (36,48). Total folate was estimated to be 230 pmol mg Ϫ1 protein in WT and 140 pmol mg Ϫ1 in mthfr Ϫ cells. The WT value corresponds to an intracellular concentration of ϳ79 M, similar to a previous estimate of 24 M (49).
Nutritional Phenotypes of mthfr Ϫ Null Mutants-The mthfr Ϫ mutant grew normally in standard M199 medium, which contains 100 M methionine. Amino acid analysis of supernatants of stationary phase cultures under these conditions showed little difference in methionine consumption between WT and mthfr Ϫ cell lines (31 versus 28 nmol of methionine consumed per 10 7 WT and mthfr Ϫ cells, respectively). However, WT and mthfr Ϫ L. major differed in their ability to grow in a minimal semidefined media (RPMI) at methionine concentrations below about 25 M (Fig. 4), whereas at higher concentrations growth was similar (Fig. 4). In the absence of exogenous methionine, neither WT nor mthfr Ϫ grew (Figs. 4 and 5), establishing L. major as a methionine auxotroph.
Homocystine was able to support growth of WT, although at a rate lower than that seen with methionine. As expected, homocystine did not support growth of mthfr Ϫ cells (Fig.  5B). Interestingly, the add-back lines (mthfr Ϫ /ϩMTHFR) grew well in the presence of homocystine, comparable with methionine or WT controls (Fig. 5C). This presumably reflects overexpression of MTHFR from the episomal expression vector and suggests that under some conditions, MTHFR activity could be limiting for growth. The growth of WT in RPMI-Met containing 200 M homocystine and 200 M glycine betaine was not significantly different from that in medium containing homocystine alone and addition of betaine did not allow growth of the mthfr Ϫ mutant on homocystine (data not shown). Assuming betaine is taken up, this suggests that L. major lacks significant homocysteine/betaine methyltransferase activity.
Due to its ability to utilize NADH as a cofactor, LmjMTHFR could in principle operate in the reverse direction, and participate in 5-CH 3 -THF salvage, as has been proposed in Plasmodium falciparum (25). The role of MTHFR in salvage of extracellular folates was therefore investigated by growing WT and mthfr Ϫ parasites in media containing minimal folate levels (RPMI-FA-Biop). The only folates in this media come from the addition of 1% fetal calf serum, which will supply folate at an   approximate concentration of less than 0.1 nM, with 5-CH 3 -THF as the major species (50). Neither line was able to grow past three passages in medium lacking L-biopterin (Fig. 6). This is consistent with previous studies, showing that L. major requires unconjugated pterins for growth (51,52). However, the WT and mthfr Ϫ lines showed similar growth in RPMI-FA supplemented with L-biopterin alone. Because Leishmania require pteridines for growth (53)(54)(55), this suggests that under these conditions MTHFR is not essential for the utilization of folate in the form of 5-CH 3 -THF. Virulence of mthfr Ϫ in a Susceptible Mouse Model-Both WT and mthfr Ϫ L. major were virulent in the BALB/c mouse footpad model infection, following inoculation with infective metacyclic promastigotes (34). In most experiments, the rate of lesion growth with mthfr Ϫ was comparable with that of WT (Fig. 7A). However, when smaller inocula were used, a lag in lesion growth was consistently observed, although the length of this delay was variable (Fig.  7B). Parasites were recovered from an mthfr Ϫ containing lesion and their genotype confirmed by testing for hygromycin and puromycin resistance (data not shown). These data suggested that MTHFR was not essential for virulence in mouse infections, but its loss might produce a mild attenuation. Were this the case, restoration of MTHFR expression in mthfr Ϫ lines should result in full restoration of virulence. However, this was not found in several add-back lines, which showed only partial restoration of WT lesion formation (mthfr Ϫ /ϩMTHFR; Fig. 7B, or data not shown). It is well established that when cultured in vitro over the periods of time required to generate null mutants (which requires two rounds of gene replacement for disomic loci), Leishmania species can lose some or all virulence. Because MTHFR expression was restored in the mthfr Ϫ /ϩMTHFR lines (see above), these data argue that most of the loss of virulence in the mthfr Ϫ lines is due to culture-associated phenomena, rather than loss of MTHFR.

DISCUSSION
The data presented here establish that L. major MTHFR is unique among characterized eukaryotic MTHFRs in that it lacks the C-terminal allosteric regulatory domain, and uses NADH and NADPH with equal efficiency. Although porcine liver MTHFR can use both cofactors in vitro, the use of NADH requires high concentrations of phosphate and the enzyme is predicted to be NADPH-dependent under physiological conditions (20). Interestingly, the mammalian bifunctional 5,10-CH 2 -THF dehydrogenase/cyclohydrolase (EC 1.5.1.15, EC 3.5.4.9) also shows phosphate effects on cofactor specificity. For this enzyme, phosphate activates NAD ϩ -dependent activity and inhibits NADP ϩ -dependent activity (56). With LmjMTHFR, no such differential effects were seen, with the k cat /K m values for NADH and NADPH being essentially unchanged when assayed in phosphate-free buffer. This surprising result indicates the L. major enzyme could be reduced by either nucleotide in vivo. A few other oxidoreductases have been shown to have dual cofactor specificity, including aldose reductase from Candida tenuis (57) and glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides (58). However, in most cases one substrate is used more efficiently than the other, mammalian glutamate dehydrogenase being one of the very few exceptions (21)(22)(23).
The only published structure of an MTHFR is of the NADHdependent E. coli enzyme, which was determined with both NADH and 5-CH 3 -THF as ligands (59). Although this protein is only 36% identical to LmjMTHFR, the majority of the E. coli active-site residues are either conserved or conservatively substituted in the L. major protein (Fig. 1). The determinants of the cofactor specificity of the E. coli active site were explored by  Pejchal et al. (59) by modeling a 2Ј phosphate group onto the NADH ligand. In their model, no unfavorable contacts were predicted and long-range interactions between the 2Ј phosphate and the basic residues Arg 33 and Lys 222 might be predicted to stabilize NADPH binding. They therefore noted that the structural basis for the NADH specificity of E. coli MTHFR is unclear. If LmjMTHFR shares the same fold as its E. coli homolog, it is possible that dual cofactor specificity could result from an active site very similar to that of E. coli MTHFR.
The lack of AdoMet inhibition of LmjMTHFR is also consistent with the ability of this enzyme to link 5,10-CH 2 -THF/5-CH 3 -THF interconversion to the NAD ϩ /NADH redox couple. The reversibility of NADH-dependent MTHFRs under physiological conditions (12), which contrasts to the irreversibility of NADPH-dependent human MTHFR (60), probably limits AdoMet accumulation and makes feedback inhibition unnecessary (17). Another possible role of a reversible MTHFR activity is folate salvage through conversion of 5-CH 3 -THF monoglutamate, the predominant form of folate in mammalian plasma (18), to other forms of folate. However, we found that mthfr Ϫ L. major grew as well as WT in semidefined media containing minimal levels of external serum-derived folate (presumably 5-CH 3 -THF; Fig. 6). This suggests that exogenous 5-CH 3 -THF can enter the Leishmania folate pool through an MTHFR-independent mechanism, most likely through the action of methionine synthases.
As anticipated from studies of other Leishmania species, we showed that L. major is auxotrophic for methionine (Figs. 4 and 5). Consistent with the presence of both MTHFR and two methionine synthases in the Leishmania genome, the methionine requirement could be satisfied through provision of homocystine, albeit only partially (Fig. 5). Thus to probe the biological role of MTHFR, we generated null mutants (mthfr Ϫ ) by serial gene replacement of both MTHFR alleles. The growth of the mthfr Ϫ null mutant was unaffected in standard culture media, which typically contain high levels of methionine, also an essential amino acid for the host (Figs. 4 and 5). However, when external methionine levels were reduced sufficiently, growth of mthfr Ϫ was impaired at concentrations below about 25 M (Fig. 4). Presumably at these levels the contribution of homocysteine re-methylation via methionine synthase becomes limiting even in WT parasites, and this pathway is completely denied to mthfr Ϫ due to the absence of 5-CH 3 -THF. Correspondingly, whereas WT parasites were able to grow (albeit at a somewhat reduced rate) when homocysteine replaced methionine, as expected, the mthfr Ϫ mutant was unable to grow on homocysteine alone (Fig. 5). Analysis of reduced folate levels showed that in fact the mthfr Ϫ completely lacked 5-CH 3 -THF (Table 2), and thus its survival could not be attributed to the presence of an unexpected novel MTHFR encoded elsewhere in the parasite genome.
These data show that homocysteine methylation and de novo methionine synthesis occur in L. major promastigotes, in contrast to T. brucei and T. cruzi, which lack MTHFR and methionine synthase activity and/or genes (61). It has been proposed that Leishmania might synthesize methionine by the reverse transsulfuration pathway (62), with cystathionine ␥-lyase converting 2-oxobutyrate and cysteine to cystathionine, which could then be hydrolyzed to release homocystine, the substrate for methionine synthases. However, cystathionine ␥-lyase activity has not been detected in L. major (63) and even if this pathway were active, Leishmania methionine auxotrophy indicates that it cannot supply sufficient methionine for their metabolic needs.
The retention of MTHFR, despite the tendency for genome reduction in intracellular parasites (64,65), suggests the ability to recycle homocysteine might confer a selective advantage to L. major under some conditions. Foremost among these would be the amastigote stage, which, unlike the promastigote stage residing extracellularly in sand flies and studied in most of our work here, multiplies within the macrophage phagolysosome. This site poses a challenge to obtaining nutrients, which must enter across the phagolysosomal membrane or through endocytic trafficking pathways. Whereas no direct measurements of amino acid levels within nor flux through the Leishmania parasitophorous vacuole have been reported, indirect evidence based upon studies of the effects of manipulation of parasite proteases or glucose metabolism suggest that amino acids are likely to be abundantly available (66 -69).
The description of the phagolysosome as an amino acid-rich environment is consistent with the finding that the mthfr Ϫ mutant showed little if any attenuation specifically attributable to loss of MTHFR in a mouse infection (Fig. 7). This supports the model that parasites can salvage from the host sufficient methionine for their metabolic needs. Whether MTHFR plays a critical role in Leishmania survival in parasite stages not studied here, for example, in promastigotes forms within the sand fly midgut (as opposed to culture vessels), or persistent parasites potentially residing in cells other than phagocytic macrophages, remains to be determined.
Because we have shown MTHFR is not an essential gene for in vitro or in vivo growth in L. major, this enzyme is unlikely to be a useful target for monochemotherapy. Potentially it could be exploited as a target in combination therapy with other antifolates, as suggested by the role of 5-CH 3 -THF and AdoMet metabolism in the acquisition of methotrexate resistance (32). Moreover, our studies also suggest that MTHFR is unlikely to be involved in the action of a group of hydrophobic pteridine analogues described previously. These compounds inhibit the Leishmania pteridine reductases PTR1 and DHFR and are highly toxic toward parasites, yet are insensitive to changes in PTR1 and DHFR levels and are thus likely to have additional targets (29).