Bacterial Conversion of Folinic Acid Is Required for Antifolate Resistance*

Antifolates, which are among the first antimicrobial agents invented, inhibit cell growth by creating an intracellular state of folate deficiency. Clinical resistance to antifolates has been mainly attributed to mutations that alter structure or expression of enzymes involved in de novo folate synthesis. We identified a Mycobacterium smegmatis mutant, named FUEL (which stands for folate utilization enzyme for leucovorin), that is hypersusceptible to antifolates. Chemical complementation indicated that FUEL is unable to metabolize folinic acid (also known as leucovorin or 5-formyltetrahydrofolate), whose metabolic function remains unknown. Targeted mutagenesis, genetic complementation, and biochemical studies showed that FUEL lacks 5,10-methenyltetrahydrofolate synthase (MTHFS; also called 5-formyltetrahydrofolate cyclo-ligase; EC 6.3.3.2) activity responsible for the only ATP-dependent, irreversible conversion of folinic acid to 5,10-methenyltetrahydrofolate. In trans expression of active MTHFS proteins from bacteria or human restored both antifolate resistance and folinic acid utilization to FUEL. Absence of MTHFS resulted in marked cellular accumulation of polyglutamylated species of folinic acid. Importantly, MTHFS also affected M. smegmatis utilization of monoglutamylated 5-methyltetrahydrofolate exogenously added to the medium. Likewise, Escherichia coli mutants lacking MTHFS became susceptible to antifolates. These results indicate that folinic acid conversion by MTHFS is required for bacterial intrinsic antifolate resistance and folate homeostatic control. This novel mechanism of antimicrobial antifolate resistance might be targeted to sensitize bacterial pathogens to classical antifolates.

The challenge of bacterial antibiotic resistance has traditionally been approached in two ways: discovery of brand new anti-biotics and use of semisynthetic derivatives of known antibiotics to circumvent existing resistance mechanisms (1). An alternative strategy to these two pathways is to develop inhibitors of resistance mechanisms (1). In this approach, an antibiotic is co-administered with a "potentiator" that neutralizes the targeted resistance mechanism, thereby resensitizing the resistant organisms to the accompanied antibiotic. An example of this strategy is coadministration of a ␤-lactam with a ␤-lactamase inhibitor. This approach has the advantage of extending the utility of antibiotics of known pharmacology, toxicology, and treatment schedule (1,2). Insight into molecular mechanisms of antibiotic resistance could reveal novel strategies for such potentiation approaches.
Because of the vital role of folates in multiple metabolic processes of the cell, folate antagonism has been used successfully in chemotherapeutic treatment of multiple diseases including cancers, malaria, psoriasis, rheumatoid arthritis, graft-versushost disease, and bacterial infections (3,4). Folate antagonists (or antifolates) have been used extensively for the treatment of infectious diseases from the late 1930s to the 1960s, but their use has declined because of the emergence of resistant strains, cytotoxicity, and most importantly the introduction of more effective drugs (4,5). Nevertheless, combination therapies using trimethoprim and sulfonamides to create synergistic effects are still used effectively today to treat some infectious diseases such as urinary tract infection, shigellosis, Pneumocystis jiroveci pneumonia, and prophylaxis against recurrent and drug-resistant infections (5)(6)(7). The absence of enzymes required for complete de novo folate biosynthesis in humans and other mammals makes this pathway an attractive and potential target for antibiotic development (8). Whereas enzymatic activities involved in folate metabolism are rather well known, the current antimicrobial antifolates exclusively target two steps in the folate biosynthetic pathway (8 -10). Trimethoprim inhibits the reduction step through the inhibition of dihydrofolate reductases, whereas sulfonamides and sulfone drugs are paraaminobenzoic acid (pABA) 4 analogs that outcompete pABA in the condensation with a pteridine group catalyzed by dihydropteroate synthase (8 -10). Recent studies suggesting that antifolate combinations such as co-trimoxazole (trimethoprim plus sulfamethoxazole) might be effective against tuberculosis (TB) have renewed interests in the exploitation of antifolates to treat multidrug-resistant and extensively drug-resistant TB (11)(12)(13). Importantly, the World Health Organization has officially called for the widespread use of this drug combination for the prophylactic treatment of HIV patients (14), which raises the possibility of increased resistance backgrounds of bacterial pathogens to classical antifolates (14,15). Inhibition of mechanisms governing intrinsic antifolate resistance could help to sensitize pathogenic bacteria such as Mycobacterium tuberculosis to classical antifolates and thus allow expanding the use of these already available antibiotics.
Here, we report the identification of a novel determinant of intrinsic antifolate resistance that exists in two bacterial species of distantly related phyla, Mycobacterium smegmatis of Grampositive Actinobacteria and Escherichia coli of Gram-negative Proteobacteria. This mechanism thus provides a potential target for antifolate potentiation.
Isolation of Antifolate-sensitive Mutants of M. smegmatis-The pMycoMar vector that carries a Himar1 transposon was used to construct a mutant library (16,17). Wild-type M. smegmatis mc 2 155 was transformed with pMycoMar vector. Transformed bacteria were cultivated at 28°C overnight to recover and allow multiplication before plating on LB agar plates containing 50 g ml Ϫ1 kanamycin. After incubation for 5 days at 39°C, single colonies were picked and cultured separately in 96-well plates in 7H9 medium and 50 g ml Ϫ1 kanamycin for 2 days. These plates were used as "master plates" to replicate to plates of solid NE medium (16) containing serial concentrations of sulfachloropyridazine (10, 15, 20, 25, and 50 g ml Ϫ1 ) or trimethoprim (1.25, 1.5, 2, 2.5, and 3 g ml Ϫ1 ). Five wells at different positions of 96-well plates inoculated with wild-type M. smegmatis strain mc 2 155 were used as growth controls. Colonies that grew on NE-kanamycin plates but failed to grow on plates supplemented with antifolates were subjected to two rounds of additional replication to confirm drug susceptibility patterns. Minimum inhibitory concentrations (MICs) of selected mutants to antifolates were determined by serial dilution assays (see below). Mapping of transposon insertion sites in the mutants by using an arbitrary PCR method was carried out as described previously (16,18).
The entire open reading frame of M. smegmatis msmeg_5472 was deleted using the recombineering method as described previously (18). The 616-bp DNA region upstream of msmeg_5472 was PCR-amplified using primers fuel-Del1 and fuel-Del2 (supplemental Table S2). Similarly, the 492-bp downstream region was amplified using primers fuel-Del3 and fuel-Del4. These DNA arms were cloned into pYUB854 (22) flanking the built-in hygromycin cassette to create pVN842. The ⌬msmeg5472::⍀hyg linear allelic exchange substrate was removed from pVN842 by SpeI/KpnI digestion and used to transform M. smegmatis mc 2 155 cells induced to express the recombineering system from pVN701B (18). Plasmid pVN701B was later removed from ⌬msmeg5472 mutant as described previously (18).
Genetic Complementation-The 1223-bp DNA fragment including msmeg_5472 ORF and its 605-bp upstream region (P FUEL ) was PCR-amplified from M. smegmatis genomic DNA using primers 5472p1.Xb and 5472p2.BH (supplemental Table  S2). PCR products were ligated to pGEM-T Easy vector (Promega, Madison, WI), and the nucleotide sequence was verified by sequencing. DNA fragments were then excised with SpeI and HindIII and cloned into pMS2 (supplemental Table S1 Table S1) by NdeI/HindIII ligations. Transformants were selected for resistance to kanamycin plus hygromycin.
Site-directed Mutagenesis of H. sapiens MTHFS-The mutated allele mthfs(D154A) was made using the primers mthfsD154Af and mthfsD154Ar in combination with mthfs1.NE and mthfs2.BH in a two-round PCR protocol (16) and plasmid pVN808 as a template. This procedure changed the GAC (aspartic acid) codon to GCC (alanine), corresponding to amino acid 154. The mutated gene was cloned to pVN805 and pVN871 (NdeI/HindIII) to create pVN811 and pVN887, respectively (supplemental Table S1). These plasmids were transformed, respectively, to FUEL and ⌬msmeg_5472 as described above.
Production and Purification of Recombinant MTHFS Proteins from E. coli-For expression in E. coli host, the genes encoding MTHFS homologs were excised from pVN778, pVN786, and pVN793 (supplemental Table S1) by NdeI/ BamHI digestions and cloned to vector pET15b (Novagen) cut with the same restriction enzymes. The constructed plasmids (pVN795, pVN814, pVN815, and pVN816; supplemental Table  S1) were used to transform the E. coli strain ArcticExpress (DE3)RP (Stratagene) expressing the cold-adapted chaperonins Cpn10 and Cpn60 from the psychrophilic bacterium Oleispira antarctica to facilitate protein folding at reduced temperatures. For strain and plasmid maintenance, ArcticExpress (DE3)RP cultures expressing MTHFS proteins were grown in LB medium in the presence of ampicillin (100 g ml Ϫ1 ) plus tetracycline (34 g ml Ϫ1 ) and gentamycin (20 g ml Ϫ1 ). Overnight cultures (37°C at 200 rpm) were diluted 50ϫ in LB without antibiotic and grown at 30°C with shaking (250 rpm) for about 3 h. Exponentially growing cultures (A 600 nm ϳ0.5) were shifted to 10°C, and protein production was induced by isopropyl 1-thio-␤-D-galactopyranoside (0.1-1 mM). To stabilize the overproduced enzyme, 0.1 mM (6R,6S)-5-CHO-H 4 PteGlu 1 was added to the medium at the same time. The induced cultures were incubated at 10°C for 24 -48 h on an orbital table. Bacteria were harvested by centrifugation and stored at Ϫ20°C. Next, the frozen cells were washed twice with TBS buffer supplemented with a protease inhibitor mixture (Roche Applied Science), resuspended in 1 ⁄ 10 volume of TBS buffer, and disrupted by sonication (four times for 20 s with 1-min intervals). The protein concentration in the total cell lysate was determined by the Bradford method (Bio-Rad). Unbroken cells and cell debris were removed by centrifugation at 12,000 rpm in an SS-34 rotor (Sorvall) for 15 min to yield soluble fraction. Expression of MTHFS proteins and their solubility were assessed by SDS-PAGE through a 10% acrylamide gel followed by Coomassie Blue staining.
The soluble fraction was further diluted with buffer A (50 mM sodium phosphate, 300 mM NaCl, and 10 mM imidazole) and loaded to a cobalt metal affinity spin column (Pierce) pre-equilibrated with the same buffer. The column was washed three times with 2 column volumes of buffer A followed by elution by 3 volumes of the same buffer A supplemented with a higher concentration of imidazole (150 mM). Protein purity was demonstrated by a single band on SDS-PAGE. Fractions containing purified protein were pooled, 0.2% (v/v) Tween 80 was added, and fractions were stored at 4 or Ϫ20°C for further studies.
Chemical Complementation-Cultures of M. smegmatis strains were grown until midlog phase and normalized to A 600 nm of 1. Aliquots (50 l) of the cultures were cast in soft agar (0.5%) and placed on top of NE medium plates supplemented with or without 0.3 mM pABA or a folate derivative. Paper discs embedded with 0.25 or 1 mg of sulfachloropyridazine (BD Diagnostic Systems) were placed at the center of the plates. Growth inhibition was visualized as the inhibition zones surrounding the antibiotic disc after 5 days of incubation at 37°C. A successful complementation was regarded as the disappearance of the inhibition zones.
Extraction and Determination of Cellular Levels of Folate Derivatives-Sample preparation was handled under subdued light to minimize folate degradation. M. smegmatis cultures growing in 7H9-glucose medium were harvested by centrifugation and resuspended in extraction buffer (Ϫ75°C) containing 80% methanol, 0.1% ascorbic acid, and 20 mM ammonium acetate (pH 6.2). 100 l of 0.01 mM 5-[ 13 C]CHO-H 4 PteGlu 1 was added to the suspension to serve as an internal standard. Aliquots of identical cultures were also harvested for determination of total protein and wet weight. Cells were disrupted by sonication (9 ϫ 20 s) with 1-min intervals. Unbroken cells and debris were removed by centrifugation and filtered through 0.45-m filters. The extracts were dried under N 2 gas and resuspended in liquid chromatography starting mobile phase solvent that consisted of 95% eluent A (0.1% of formic acid in water) and 5% eluent B (0.1% of formic acid in acetonitrile). Protein was precipitated by centrifugation, and the supernatant was analyzed by high performance liquid chromatography and tandem mass spectrometry (HPLC-MS/MS) as described previously (24,25). Conditions for HPLC and mass spectrometric analyses using a 4000Q-trap mass spectrometer (Applied Biosystems, Foster City, CA), preparation of calibrators, and data analyses were carried out as before (24,25). 5,10-Methenyltetrahydrofolate Synthase Assay-The MTHFS enzymatic activities were quantified as described previously (26) through the formation of 5,10-methenyltetrahydrofolate (5,10-CH ϩ -H 4 PteGlu n ) that results in an increase in absorbance at 360 nm (⑀ ϭ 25,000 M Ϫ1 cm Ϫ1 ). The assay mixture contained, in 1 ml, 50 mM MES buffer (pH 6.0), 10 mM magnesium acetate, 1 mM DTT, 0.1 mM EDTA, 0.1 mM ATP, and 100 M (6S)-5-CHO-H 4 PteGlu n . Reactions were conducted in quartz cuvettes of 0.5-cm optical path and were initiated by the addition of MTHFS sources. The absorbance was recorded using an Evolution-300 spectrophotometer (Thermo Scientific) equipped with VisionLife software. The temperature was maintained at 30°C by the air-cooled Peltier SPG-1A (Thermo Scientific). One unit of activity corresponds to 1 mol of 5,10-CH ϩ -H 4 PteGlu n formed in 1 min. Specific activity is expressed as units mg Ϫ1 of protein. Protein concentrations were determined by the Bradford method.
MIC Assay-Antifolates were prepared from stocks (50 mg ml Ϫ1 ). Serial 2ϫ dilutions of sulfonamides or trimethoprim (1-1024 g ml Ϫ1 ) were prepared in 100-l volumes of a liquid medium (Mueller Hinton broth for E. coli and Middlebrook 7H9 plus glucose and Tween 80 for M. smegmatis) in 96-well, U-bottom microplates. For trimethoprim/sulfonamides combinations (0.039 -20 g ml Ϫ1 ), the two drugs were mixed in a 1:19 ratio. An E. coli strain with known MIC values (ATCC 25922; Ref. 27) was used as an internal control. Single colonies from strains were grown to exponential phase, and optical density was adjusted to 0.12 on the MicroScan turbidity meter (28), which is equivalent to the McFarland standard of 0.5 (29). Next, the culture suspensions were diluted 25 times with phosphatebuffered saline, and 5-l aliquots were used to inoculate each well of the prepared 96-well plates using a multiwell inoculator. This procedure produced ϳ5 ϫ 10 5 cfu/well (30) as assayed by colony counts of the inoculum immediately after inoculation. The inoculated plates were sealed and incubated at 37°C for 18 (E. coli) or 72 h (M. smegmatis) under static conditions. Growth was assessed with indirect light using a reading mirror (Dynatec Laboratories). MIC was defined as the lowest drug concentration that inhibited at least 80% of growth as recommended by the Clinical and Laboratory Standards Institute (31) for rapidly growing mycobacteria.

Identification of FUEL, a Novel Determinant of Intrinsic
Antifolate Resistance-We used a chemogenomic profiling approach to screen for whole-genome antifolate resistance determinants in mycobacterial species. A transposon mutant library, currently composed of ϳ15,000 individual mutants of M. smegmatis deposited in separate wells of 96-well plates, was con- structed using the Himar1 minitransposon system originally isolated from the horn fly Hematobia irritans (17,(32)(33)(34)(35). Screening this library, we identified several mycobacterial determinants of intrinsic antifolate resistance. The mutants were further screened for restoration of antifolate resistance by exogenous reagents of the folate pathways to elucidate whether and how their antifolate susceptibility relates to folate starvation. 5 One of the mutants identified in this screen was of special interest as it displayed a dramatically increased susceptibility to multiple antifolate drugs. The sensitivity spectrum of the mutant includes trimethoprim and several sulfonamides tested (sulfamethoxypyridazine, sulfadimethoxine, sulfamethiazole, sulfadiazine, sulfamethoxazole, sulfathiazole, and sulfachloropyridazine (SCP)). Antifolate susceptibility tests showed that the mutant became hypersusceptible to multiple combinations of trimethoprim and sulfonamides with MIC values 3-6 log 2 dilutions lower than wild type (Fig. 1A). By contrast, growth of the mutant in the absence of antifolates was identical to that of wild-type M. smegmatis (Fig. 1B), suggesting that the function of the disrupted gene is related to folate metabolism. Chemical complementation data showed that whereas folic acid (Pte-Glu 1 ) and pABA completely abolished the antifolate susceptibility of the mutant addition of monoglutamylated 5-formyltetrahydrofolate (5-CHO-H 4 PteGlu 1 ; also known as folinic acid or leucovorin) to the growth medium did not affect its sensitivity. This result further suggested that the inserted gene might involve metabolism of this folate species (Fig. 1C). Thus, the mutant was named FUEL.
Genetic Mapping of FUEL-Genetic mapping using nested PCR followed by sequencing identified Himar1 insertion into msmeg_5472 ( Fig. 2A). The Himar1 transposon insertion into msmeg_5472 gene in FUEL was further confirmed by PCR amplification of the mutant locus using primers flanking the putative open reading frame (5472p1.NE and 5472p2.BH; supplemental Table S2). The mutant gene generated a larger fragment corresponding to the inserted transposon (2199 bp), resulting in a decreased mobility of the PCR product on an agarose gel (Fig. 2B). Sequencing of the msmeg_5472-Himar junction region from this PCR product identified the insertion site at the dinucleotide 145 TA 146 that introduced a stop codon after the triplet encoding the Ala 48 residue (Fig. 2A, bottom). To confirm interruption of msmeg_5472 expression, Western blots were done using a polyclonal antibody raised against the purified recombinant protein (Josman LLC). The antibody recognized a protein band of ϳ23 kDa corresponding to the predicted molecular mass of Msmeg_5472 in the cell lysate of wildtype M. smegmatis that is absent in the lysate of FUEL or a targeted msmeg_5472 deletion mutant (Fig. 2C).
To confirm that the disruption of msmeg_5472 expression was responsible for the increased antifolate susceptibility of FUEL, the entire open reading frame of msmeg_5472 was deleted by targeted mutagenesis. The msmeg_5472 null mutant (⌬msmeg_5472) showed an antifolate susceptibility profile identical to FUEL (not shown). Next, the DNA fragment carrying the encoding gene (msmeg_5472) and its putative promoter upstream (P FUEL ) was cloned for in trans expression in M. smegmatis. Transformation of plasmids expressing msmeg_5472 in trans [pVN794 to FUEL or pVN888 to ⌬msmeg_5472] (supplemental Table S1) restored both protein expression (Fig. 2C) and resistance to multiple antifolates (Fig. 1A).
Enzymatic Activity of 5,10-Methenyltetrahydrofolate Synthases Is Required for Antifolate Resistance-The deduced amino acid sequence of msmeg_5472 shows weak homology to MTHFS (also called 5-formyltetrahydrofolate cyclo-ligase; E.C.6.3.3.2) (Fig. 3A). To investigate whether the antifolate hypersensitivity of FUEL was due to a lack of MTHFS activity, the cDNA sequence of the gene encoding the MTHFS prototype from human (H. sapiens) (17.7% identity and 16.3% similarity compared with Msmeg_5472) (Fig. 3B), whose MTHFS enzymatic activity has been well established (23,36), was optimized for codon usage in M. smegmatis (supplemental Table  S3), chemically synthesized, and in trans expressed in FUEL. Heterologous expression of H. sapiens mthfs gene from plasmid pVN808 [P FUEL :mthfs] restored the antifolate resistance to FUEL, similar to the in trans expression of msmeg_5472 (Fig.  1A), suggesting that the increased antifolate susceptibility of FUEL might be due to a lack of MTHFS activity.
The enzymatic activity of H. sapiens MTHFS is dependent on an ATP binding domain, which is in a close proximity to the 5 S. Ogwang, H. T. Nguyen, and L. Nguyen, unpublished data. substrate-binding site (37). Aspartate residue 154 in this domain, which forms hydrogen bonds to the magnesium molecule that is hydrogen-bonded to the ␣and ␤-phosphates of the ATP molecule, is required for the MTHFS enzymatic activity (37). A mutation of aspartate 154 to alanine (D154A) abolishes more than 99% of the in vitro MTHFS activity (37). To confirm that the cross-kingdom expression of H. sapiens MTHFS provides antifolate resistance to FUEL through its enzymatic activity, we constructed a vector expressing the mthfs(D154A) allele of H. sapiens MTHFS. In contrast to the intact mthfs (pVN808), the mthfs(D154A) allele (pVN811) failed to rescue the antifolate resistance defect of FUEL (Fig.  1A), strongly supporting that the increased antifolate sensitivity phenotype of FUEL was due to a loss of MTHFS activity encoded by msmeg_5472. These results also indicated that a MTHFS enzymatic activity is required for the intrinsic antifolate resistance in M. smegmatis.
MTHFS Activity Mediates Utilization of 5-CHO-H 4 PteGlu 1 in M. smegmatis-To further evaluate if the function of Msmeg_5472 in antifolate susceptibility is related to its role in cellular utilization of 5-CHO-H 4 PteGlu 1 and MTHFS activity, chemical complementation experiments were done in the presence of sulfonamide drugs to block de novo folate biosynthesis. Antifolate susceptibility of the M. smegmatis strains could be completely reversed by exogenous addition of PteGlu 1 (Fig. 4) or pABA (not shown), visualized by the disappearance of the inhibition zone surrounding sulfonamide discs (e.g. sulfachloropyridazine in Fig. 4). These results indicated that M. smegmatis is able to uptake and metabolize extracellular folate in addition to its ability of de novo synthesis. The ability of M. smegmatis to utilize extracellular PteGlu 1 was independent of msmeg_5472 gene, as PteGlu 1 provided sulfonamide resistance to both wild type M. smegmatis and FUEL (Fig. 4). In contrast to PteGlu 1 , utilization of 5-CHO-H 4 PteGlu 1 was dependent on the expression of either msmeg_5472 (pVN794) or the synthetic gene encoding H. sapiens MTHFS (pVN808) (Fig. 4). However, the MTHFS dead allele (pVN811) was unable to restore 5-CHO-H 4 PteGlu 1 utilization to FUEL (Fig. 4, far right panel). Similar results were obtained with ⌬msmeg_5472 null mutant and its derived complemented strains (not shown). Collec- tively, these experiments suggested that msmeg_5472 encodes a mycobacterial MTHFS activity responsible for 5-CHO-H 4 PteGlu 1 utilization that might be required for the intrinsic antifolate resistance in M. smegmatis.
Purified msmeg_5472-encoded Recombinant Protein Exhibits MTHFS Activity in Vitro-To further determine whether the msmeg_5472 gene de facto encodes an MTHFS activity (Fig.  5A), it was PCR-amplified from M. smegmatis genomic DNA and cloned to pET15b vector, such that the recombinant protein would be expressed as a N-terminal 6xHis-tagged peptide from the built-in isopropyl 1-thio-␤-D-galactopyranoside-inducible T7 promoter. First attempts to express the protein in E. coli host BL21(DE3) using standard induction protocol (1 mM isopropyl 1-thio-␤-D-galactopyranoside, 37°C) resulted in majority of the recombinant protein insoluble (not shown), similar to previously reported studies with other MTHFS proteins (38). To optimize the solubility of the protein, the E. coli strain ArcticExpress (DE3)RP, that co-expresses the coldadapted chaperonin proteins from Oleispira Antarctica, was used to facilitate protein folding at reduced temperatures. Induction was carried out at 10°C with a lower concentration of isopropyl 1-thio-␤-D-galactopyranoside (0.5 mM). In addition, extended induction time and addition of 5-CHO-H 4 PteGlu 1 (0.1 mM) to the medium helped enhance the expression level and solubility of the recombinant protein (not shown), respec-tively. In this optimized condition, the recombinant protein was expressed at a high level in soluble form (Fig. 5B).
Cobalt affinity chromatography was successfully used to purify the 6xHis-tagged protein to homogeneity (Fig. 5B). The purified protein migrated as a single band of ϳ 25 kDa on SDS-PAGE, consistent with the predicted molecular weight of the recombinant protein with its N-terminal-tag extension (Fig.  5B). The identity of the protein was confirmed by Western blot using antibody raised against the 6xHis tag (Fig. 5B, lower  panel). Addition of 0.2% Tween 80 and storage at 4°C stabilized the protein for up to 3 months (not shown). The purified 6xHis-tagged protein was subjected to MTHFS assay measuring its activity to convert 5-CHO-H 4 PteGlu 1 to 5,10-CH ϩ -H 4 PteGlu 1 (Fig. 5C). A steady shift of spectrophotometric absorbance characteristic of the substrate (5-CHO-H 4 PteGlu n , 285 nm) to its product (5,10-CH ϩ -H 4 PteGlu n , 360 nm) upon addition of the purified protein demonstrated the MTHFS enzymatic activity of the recombinant protein encoded by msmeg_5472 (Fig. 5C). MTHFS activities of the purified H. sapiens MTHFS protein and its dead mutant MTHFS(D154A) were respectively used as positive and negative controls (Fig. 5D). The presence of 5,10-CH ϩ -H 4 PteGlu 1 in the chemical reactions was also confirmed by LC-MS/MS (m/z 456/412) and 5,10-CH ϩ -H 4 PteGlu 1 standard (Fig. 5E) (24, 25). These biochemical analyses confirmed   (39). To investigate whether MTHFS activity provides antifolate resistance to other bacteria besides M. smegmatis, the gene encoding the MTHFS homolog in the Gram-negative proteobacterium E. coli (ygfA) was deleted using a targeted recombineering method. In two different E. coli backgrounds, MG1655 and BW25113, deletion of ygfA led to increased susceptibility to multiple sulfonamide drugs (Table 1), although the levels were not as dramatic as those observed in M. smegmatis. Addition of trimethoprim to these sulfonamides did not lead to pronounced synergistic effects (not shown), which is in agreement with a recent study in E. coli (40). Importantly, heterologous expression of M. smegmatis MTHFS Ms (msmeg_5472) restored antifolate resistance to the ⌬ygfA mutants ( Table 1), indicating that these two homologs are functionally identical and species-interchangeable. Collectively, these results suggested that MTHFS activity is ubiquitously required for the intrinsic antifolate resistance in bacteria.

DISCUSSION
Folate metabolism, which plays a central role in the growth and proliferation of cells across all kingdoms of life, is generally divided into two stages: biosynthesis (upstream) and utilization (downstream) (41). The upstream de novo folate biosynthesis, absent in human and mammalian cells, involves multiple steps that synthesize and condense a pteridine group, pABA, and glutamate altogether to form dihydrofolate (H 2 PteGlu n ) that is reduced to form tetrahydrofolate (H 4 PteGlu n ) (Fig. 8) (41). The downstream folate utilization includes multiple one-carbon metabolic reactions in which different reduced forms of H 4 PteGlu n participate in distinct reactions donating or accepting one-carbon units for the formation of several molecules  including purines, thymidines, pantothenate, glycine, methionine, S-adenosylmethionine, and formyl-methionyl-tRNA, the initiator of protein synthesis in bacteria (41,42). Therefore, defects in the biosynthesis or utilization of folate result in suspension of principal cellular processes including the biosynthesis of nucleic acids and proteins, DNA methylation, and homocysteine homeostasis, etc., leading to consequent retardations of cell growth and division. Folate is particularly important during periods of rapid cell division when supplies of the building precursors of macromolecules are most needed (43).
In the one-carbon metabolic pathway, 5-CHO-H 4 PteGlu n is formed by the hydrolytic reaction of 5,10-CH ϩ -H 4 PteGlu n catalyzed by serine hydroxymethyltransferase that also catalyzes the reaction interconverting serine and glycine (Fig. 8) (44,45). Although it is well known chemically and widely used as a medical agent, the biological function of 5-CHO-H 4 PteGlu n remains unknown (46). 5-CHO-H 4 PteGlu n is the most stable form of reduced folate species in nature, but it does not function as a cofactor in any of the one-carbon metabolic reactions thus far known (46). The increased presence of 5-CHO-H 4 PteGlu n in plant seeds and fungal spores suggests that it might function as a folate storage form required for these dormant life states (46 -48). In mammals and yeasts, 5-CHO-H 4 PteGlu n comprises 3-10% of total folate, whereas its presence may account for up to 50% of total folate in plant mitochondria during photorespiration when the glycine to serine flux is accelerated (49,50). In vitro, 5-CHO-H 4 PteGlu n is also a potential inhibitor of serine hydroxymethyltransferase and other enzymes of onecarbon metabolism; thus, it may potentially serve to regulate these metabolic reactions (49,51). MTHFS is the only enzymatic activity known to recycle 5-CHO-H 4 PteGlu n back to 5,10-CH ϩ -H 4 PteGlu n in an irreversible, ATP-dependent reaction (Fig. 5A). Deletion of MTHFS in Arabidopsis leads to a 2-8-fold increased accumulation of total 5-CHO-H 4 PteGlu n , 46-fold accumulation of glycine, reduced growth, and delayed flowering (50). In human cells, overexpression of MTHFS lowers folate levels and increases folate turnover, suggesting that MTHFS may also functions as a folate-degrading enzyme (26). Interestingly, a recent work suggested that ygfA, the gene encoding the MTHFS homolog in E. coli, is required for the formation of antibiotic persisters that become phenotypically tolerant to antibiotics (52).
In the cell, folate species are normally conjugated to a ␥-linked polyglutamyl tail of up to 8 residues, but the precise number of glutamate residues appears to be species-specific (53). The cellular function of polyglutamylation remains unclear, although work has suggested that it plays a role in the stability, cellular retention, enzyme specificity, and homeostasis of folates (53,54). It is possible that 5-CHO-H 4 PteGlu n serves as a folate reserve that contributes to bacterial survival of folate antagonism. Redundant folate inflows from the de novo biosynthesis upstream might be regularly converted to these more stable forms (5-CHO-H 4 PteGlu n ). This mechanism may be used to regulate cellular folate homeostasis as well as the rate  of nucleic acid and protein biosynthesis via controlling the production of their precursors that are synthesized by one-carbon metabolic reactions (55). When the synthesis of these macromolecules needs acceleration or de novo folate production is restricted (e.g. by sulfonamides or trimethoprim), the ATPconsuming reversion of 5-CHO-H 4 PteGlu n to 5,10-CH ϩ -H 4 PteGlu n catalyzed by MTHFS provides a folate backup to one-carbon metabolic reactions, thereby supporting temporary growth. This folate reserve may thus allow cellular activities to remain active until de novo folate synthesis resumes (e.g. due to attrition of antifolate drugs). The reserved source of 5-CHO-H 4 PteGlu n in the MTHFS mutants (M. smegmatis FUEL and ⌬msmeg_5472 and E. coli ⌬ygfA strains) is unusable due to their lack of MTHFS activity. Without this folate resource, the intracellular folate pool becomes rapidly exhausted when de novo synthesis is blocked. Therefore, these mutants become more susceptible to antifolates that inhibit folate biosynthesis (sulfonamides) or reduction (trimethoprim). An alternative model might rest on previous observations that 5-CHO-H 4 PteGlu n is able to regulate reactions of the one-carbon metabolic pathway (49,51,56). The absence of MTHFS leading to accumulation of 5-CHO-H 4 PteGlu n species could therein have led to destabilized folate homeostasis. This regulatory defect thus sensitizes bacterial cells to antifolates, which starve cells from newly synthesized folates. This regulatory malfunction might be related to the inability of MTHFS mutants to utilize exogenous 5-CH 3 -H 4 PteGlu 1 (Fig. 6) as well as the reduced concentrations of polyglutamylated 5-CH 3 -H 4 PteGlu n (Fig. 7B). However, the precise enzymatic mechanism leading to these phenotypes remains unclear. 5-CH 3 -H 4 PteGlu n is a major reduced H 4 PteGlu species in one-carbon metabolism, and impaired metabolism of 5-CH 3 -H 4 PteGlu n is associated with the so-called methylfolate trap phenomenon (57,58), which may also contribute to antifolate susceptibility. Increased cellular content of glycine in the absence of MTHFS might also attribute to the general antibiotic resistance (50,52). Glycine has been shown to disturb not only the integrity of bacterial peptidoglycan (59) in general but also the mycolate content of the Mycobacterium cell wall (60). As a consequence, this might lead to acceleration of antibiotic penetration into the cytoplasm.
Compared with M. smegmatis, deletion of ygfA in E. coli resulted in a lower reduction of antifolate resistance (Table 1) and only a 6-fold increase in total cellular content of 5-CHO-H 4 PteGlu n (39). The reason behind this species-specific difference is currently unclear. A recent study showed that conversion of 5-CHO-H 4 PteGlu n (to free H 4 PteGlu n ) could be catalyzed by the moonlighting activity of glutamate formiminotransferase (FT) in some bacteria, especially those lacking MTHFS activity (39). This enzyme might thus help those bacteria to compensate for the absence of MTHFS by lowering the toxic accumulation of 5-CHO-H 4 PteGlu n species. Nevertheless, both E. coli and mycobacteria (Ref. 39 and our BLAST search) do not have genes encoding formiminotransferase homologs; thus, they are unlikely to be able to recycle 5-CHO-H 4 PteGlu n by the moonlighting activity of formiminotransferase. The precise mechanism(s) underlying the role of MTHFS in bacterial antifolate resistance thus remains to be further characterized.
The function of MTHFS in antifolate resistance was observed in two distant bacteria. M. smegmatis belongs to the high GϩC Gram-positive Actinobacteria group, whereas E. coli represents Gram-negative Proteobacteria. In addition, M. tuberculosis MTHFS homolog (rv0992c) was able to complement both the antifolate resistance (Fig. 1A) and folinic acid utilization (not shown) of the M. smegmatis MTHFS mutants, suggesting that it plays the same role in M. tuberculosis. These observations suggest that the MTHFS-mediated antifolate resistance mechanism is ubiquitous among bacteria, making it an attractive target for new drug development. Antifolates have been used for the treatment of mycobacterial infections: p-aminosalicylic acid, an approved TB drug with an unknown mech-SULFONAMIDES TRIMETHOPRIM FIGURE 8. Role of MTHFS in folate metabolism and antagonism. A simplified scheme illustrates two stages of folate metabolism: upstream de novo synthesis and downstream one-carbon metabolic network. Folate antagonists include sulfonamides, that outcompete pABA in the condensation with pterin, and trimethoprim that inhibits folate reduction by dihydrofolate reductase. Chemicals used to suppress activity of antifolates are dependent on enzymatic activities that convert them to other folate forms. MTHFS is the only activity known to convert 5-CHO-H 4 PteGlu n to 5,10-CH ϩ -H 4 PteGlu n . Cellular 5-CHO-H 4 PteGlu n might function in folate storage, or regulation of one-carbon interconversion reactions. Absence of MTHFS also leads to accumulation of glycine that might affect cell wall integrity and permeability. DHFS, dihydrofolate synthase; DHFR, dihydrofolate reductase; DHPS, dihydropteroate synthase; Hcy, homocysteine; SHMT, serine hydroxymethyltransferase; MS, methionine synthase; 5,10-CH 2 -H 4 PteGlu n , 5,10-methylenetetrahydrofolate; 10-CHO-H 4 PteGlu n , 10-formyltetrahydrofolate; H 2 PteGlu n , dihydrofolate.
anism, was recently found to target folate metabolism (61), and dapsone, a sulfone drug that also acts as a pABA analog, has been used in monodrug regimens to treat leprosy and other mycobacterial infections for many decades (62). Interestingly, a recent study further suggested that the frontline TB drug isoniazid might also target folate metabolism through its inhibitory action on dihydrofolate reductase, the same enzyme inhibited by trimethoprim (63). With the current epidemic of multidrugresistant and extensively drug-resistant TB, there is a renewed interest in the use of alternative drugs such as antifolates (11)(12)(13). Inhibitors of MTHFS might not only sensitize M. tuberculosis to classical antifolates but also the current TB drugs that happen to target folate pathways such as p-aminosalicylic acid and isoniazid. One potential obstacle for the use of MTHFS inhibitors might be their nonspecific inhibition toward human MTHFS. However, the low homologies of MTHFS proteins indicate the possibility to identify species-specific inhibitors. Trimethoprim, which specifically inhibits bacterial dihydrofolate reductase but not the human counterpart, represents an encouraging example for such possibilities. Future study will focus on the identification of M. tuberculosis MTHFS-specific inhibitors and their synergy with antifolates to inhibit growth of M. tuberculosis including multidrug-resistant and extensively drug-resistant clinical isolates.