Cloning and Characterization of Methenyltetrahydrofolate Synthetase from Saccharomyces cerevisiae *

The folate derivative 5-formyltetrahydrofolate (folinic acid; 5-CHO-THF) was discovered over 40 years ago, but its role in metabolism remains poorly understood. Only one enzyme is known that utilizes 5-CHO-THF as a substrate: 5,10-methenyltetrahydrofolate synthetase (MTHFS). A BLAST search of the yeast genome using the human MTHFS sequence revealed a 211-amino acid open reading frame (YER183c) with significant homology. The yeast enzyme was expressed inEscherichia coli, and the purified recombinant enzyme exhibited kinetics similar to previously purified MTHFS. No new phenotype was observed in strains disrupted at MTHFS or in strains additionally disrupted at the genes encoding one or both serine hydroxymethyltransferases (SHMT) or at the genes encoding one or both methylenetetrahydrofolate reductases. However, when the MTHFS gene was disrupted in a strain lacking the de novo folate biosynthesis pathway, folinic acid (5-CHO-THF) could no longer support the folate requirement. We have thus named the yeast gene encoding methenyltetrahydrofolate synthetase FAU1(folinic acid utilization). Disruption of the FAU1 gene in a strain lacking both 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) transformylase isozymes (ADE16 and ADE17) resulted in a growth deficiency that was alleviated by methionine. Genetic analysis suggested that intracellular accumulation of the purine intermediate AICAR interferes with a step in methionine biosynthesis. Intracellular levels of 5-CHO-THF were determined in yeast disrupted atFAU1 and other genes encoding folate-dependent enzymes. In fau1 disruptants, 5-CHO-THF was elevated 4-fold over wild-type yeast. In yeast lacking MTHFS along with both AICAR transformylases, 5-CHO-THF was elevated 12-fold over wild type. 5-CHO-THF was undetectable in strains lacking SHMT activity, confirming SHMT as the in vivo source of 5-CHO-THF. Taken together, these results indicate that S. cerevisiae harbors a single, nonessential, MTHFS activity. Growth phenotypes of multiply disrupted strains are consistent with a regulatory role for 5-CHO-THF in one-carbon metabolism and additionally suggest a metabolic interaction between the purine and methionine pathways.

The folate derivative 5-formyltetrahydrofolate (folinic acid; 5-CHO-THF) was discovered over 40 years ago, but its role in metabolism remains poorly understood. Only one enzyme is known that utilizes 5-CHO-THF as a substrate: 5,10-methenyltetrahydrofolate synthetase (MTHFS). A BLAST search of the yeast genome using the human MTHFS sequence revealed a 211-amino acid open reading frame (YER183c) with significant homology. The yeast enzyme was expressed in Escherichia coli, and the purified recombinant enzyme exhibited kinetics similar to previously purified MTHFS. No new phenotype was observed in strains disrupted at MTHFS or in strains additionally disrupted at the genes encoding one or both serine hydroxymethyltransferases (SHMT) or at the genes encoding one or both methylenetetrahydrofolate reductases. However, when the MTHFS gene was disrupted in a strain lacking the de novo folate biosynthesis pathway, folinic acid (5-CHO-THF) could no longer support the folate requirement. We have thus named the yeast gene encoding methenyltetrahydrofolate synthetase FAU1 (folinic acid utilization). Disruption of the FAU1 gene in a strain lacking both 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) transformylase isozymes (ADE16 and ADE17) resulted in a growth deficiency that was alleviated by methionine. Genetic analysis suggested that intracellular accumulation of the purine intermediate AICAR interferes with a step in methionine biosynthesis. Intracellular levels of 5-CHO-THF were determined in yeast disrupted at FAU1 and other genes encoding folate-dependent enzymes. In fau1 disruptants, 5-CHO-THF was elevated 4-fold over wild-type yeast. In yeast lacking MTHFS along with both AICAR transformylases, 5-CHO-THF was elevated 12-fold over wild type. 5-CHO-THF was undetectable in strains lacking SHMT activity, confirming SHMT as the in vivo source of 5-CHO-THF. Taken together, these results indicate that S. cerevisiae harbors a single, nonessential, MTHFS activity. Growth phenotypes of multiply disrupted strains are consistent with a regulatory role for 5-CHO-THF in one-carbon metabolism and additionally suggest a metabolic interaction between the purine and methionine pathways.
Although 5-formyl-tetrahydrofolate (5-CHO-THF 1 ; folinic acid or leucovorin) is the most stable derivative of the reduced folates (1), it does not have a known direct role as a one-carbon donor. Leucovorin has been employed extensively as a rescue agent in chemotherapeutic protocols, but the physiological role of this folate derivative remains poorly understood. 5-CHO-THF can arise nonenzymatically from the hydrolysis of 5,10methenyl-THF (5,10-CH ϩ -THF) under mild conditions (2). For this reason, it was considered likely that the 5-CHO-THF detected in cell extracts was an artifact of preparation rather than a normal component of folate metabolism. More recent studies designed to prevent the possibility of artifactual conversion concluded that 5-CHO-THF is indeed a naturally occurring metabolite (3). The biological source of 5-CHO-THF, however, remains in question. Stover and Schirch (4) showed that in the presence of glycine, serine hydroxymethyltransferase (SHMT) catalyzes the hydrolysis of 5,10-CH ϩ -THF to 5-CHO-THF and proposed that this reaction is the source of cellular 5-CHO-THF. Supporting evidence was provided by the observation that cells deficient in SHMT activity do not accumulate 5-CHO-THF (3,5). SHMT is an abundant enzyme that normally catalyzes the reversible conversion of serine to glycine coupled to the production of 5,10-methylene-THF (5,10-CH 2 -THF); this reaction is the major source of one-carbon units in most cells (6). The physiological significance of the SHMTcatalyzed hydrolysis of 5,10-CH ϩ -THF to 5-CHO-THF has recently been called into question. Baggott (7) studied the chemical hydrolysis of 5,10-CH ϩ -THF under mildly acidic conditions and concluded that the intermediate proposed by Stover and Schirch does not accumulate and that the likely source of 5-CHO-THF is chemical hydrolysis of 5,10-CH ϩ -THF within subcellular organelles rather than the SHMT-catalyzed reaction.
Whereas the biological source of 5-CHO-THF remains controversial, an enzyme that catalyzes its conversion back to 5,10-CH ϩ -THF has been identified and characterized from several prokaryotic and eukaryotic sources (8 -11). Methenyltetrahydrofolate synthetase (MTHFS; EC 6.3.3.2) catalyzes the irreversible (12) ATP-dependent reaction, substrate. The existence of this enzyme from bacteria to humans strongly suggests that 5-CHO-THF plays a universally important role in folate-mediated one-carbon metabolism. 5-CHO-THF has been proposed to function as a stable storage form of one-carbon units (3) and as a regulator of the flow of one-carbon units (13), based on its ability to inhibit several folate-utilizing enzymes, including SHMT. Indeed, Girgis et al. (14) have presented evidence from cultured cells that 5-CHO-THF regulates homocysteine remethylation via inhibition of SHMT.
In an effort to better understand the metabolic role of 5-CHO-THF and MTHFS, we have cloned and disrupted the gene encoding MTHFS (YER183c) from the yeast Saccharomyces cerevisiae. We have purified and characterized the yeast enzyme. Genetic experiments revealed that this enzyme is required for folinic acid utilization in yeast, and we have named the gene FAU1. We have also characterized yeast strains lacking MTHFS in combination with other enzymes of folic acid metabolism and measured cellular 5-CHO-THF levels in these strains. These studies confirm the role of MTHFS in the metabolism of 5-CHO-THF in yeast and reveal several interesting metabolic interactions between methionine and purine biosynthesis.

EXPERIMENTAL PROCEDURES
Materials-All chemicals were of the highest available commercial quality. Unless otherwise noted, chemicals were obtained from Sigma. Difco media components were obtained from VWR (West Chester, PA). The pentaglutamate derivative of (6R,6S)-5-formyl-THF was purchased from Schircks Laboratories (Jona, Switzerland). Restriction enzymes, shrimp alkaline phosphatase, calf intestinal alkaline phosphatase, and T4 DNA ligase were purchased from Invitrogen. Purified rabbit SHMT was obtained from V. Schirch (Virginia Commonwealth University); purified human cytoplasmic SHMT (15) was obtained from P. Stover (Cornell University).
Escherichia coli XL1-Blue (Stratagene) was used as the host strain for plasmid manipulations. E. coli stain BL21(DE3) (Stratagene) was used for protein expression experiments. E. coli were transformed with plasmid DNA using the rubidium chloride method (17). Plasmid DNA was isolated from bacterial hosts by the modified alkaline lysis method of Feliciello and Chinali (18) or by using a QiaPrep kit (Qiagen, Valencia CA). DNA was extracted from agarose gels using the QIAquick gel extraction kit (Qiagen, Valencia, CA). Genomic DNA was isolated from yeast using the method of Sherman et al. (19) or by using the DNeasy tissue kit (Qiagen). PCR products were purified for sequencing and cloning using the QIAquick PCR purification kit (Qiagen). DNA for cloning was amplified from genomic DNA using the proofreading Pfu polymerase (Stratagene). Noncloning PCR experiments relied on Taq polymerase (PerkinElmer Life Sciences). Primers for PCR and sequencing were made by IDT (Coralville, IA). All plasmid constructs were completely sequenced at the University of Texas at Austin DNA Analysis Facility.
Isolation of the Yeast Gene Encoding MTHFS-The ORF encoding the putative yeast MTHFS (YER183c; designated FAU1) was PCRamplified from yeast genomic DNA using the primers 5Ј-GTGGAC-CATATGGCCACTAAGCAA-3Ј and 5Ј-CCTGCCTCGAGTTGAAAC-CAATGTATGGA-3Ј, which include the restriction sites for NdeI and XhoI, respectively. The restriction sites are underlined, and the first and last codons of the yeast ORF are in boldface type. Following restriction digestion and gel purification, the PCR product was cloned into the NdeI and XhoI sites of pET16b (Novagen) and transformed into XL1-Blue. The resulting plasmid, pET-MTHFS, carries the 211-residue yeast MTHFS with a 10-histidine tag plus a 9-amino acid linker fused to its N terminus and 22 additional vector-encoded amino acids fused to its C terminus. The MTHFS ORF was also cloned into the yeast expression vector pVT-101U (20) from PCR-amplified yeast genomic DNA. Primers 5Ј-GAATTCTCGAGAAGCTTATGGCCACTAAGCAA-3Ј and 5Ј-CCTAGGCTCGAGCTGAGCAATTATTGAAACCAATGTAT-3Ј were designed to include the restriction sites for HindIII and XhoI, respectively (underlined). Following restriction digestion and gel purification, the PCR product was cloned into the HindIII and XhoI sites of pVT-101U and transformed into XL1-Blue. The resulting plasmid, pVT-MTHFS, carries the 211-residue yeast MTHFS (no N-or C-terminal fusions) under transcriptional control of the ADH1 promoter.
Enzyme Assay of MTHFS-MTHFS activity was determined by following production of 5,10-CH ϩ -THF at 360 nm (⑀ 360 ϭ 25.1 ϫ 10 3 M Ϫ1 cm Ϫ1 ) using a modification of a previously published method (8). The standard reaction mixture contained 1 mM MgATP, 500 M (6R,6S)-5-CHO-THF (calcium salt of monoglutamate derivative), 0.5% Triton-X-100, 14 mM 2-mercaptoethanol, and 50 mM K⅐HEPES, pH 7.0, in a final volume of 300 l. Reactions were performed at 30°C and monitored with a SpectraMax Plus spectrophotometer. One unit of activity represents 1 mol of 5,10-CH ϩ -THF formed/min. Assay of MTHFS activity in crude cell extracts required a gentle disruption method. Yeast spheroplasts, prepared from midlog cultures as described (21), were lysed by mild sonication. Cellular debris was removed by centrifugation, and aliquots of the supernatant were assayed using the standard reaction mixture. Steady state kinetic parameters for purified MTHFS were determined using 0.5 g of purified His-tagged enzyme in each reaction. The concentration of (6R,6S)-5-CHO-THF monoglutamate was varied from 1 to 500 M at saturating MgATP concentration (1.0 mM). Similarly, the concentration of MgATP was varied from 0.025 to 5 mM at saturating (6R,6S)-5-CHO-THF concentration (500 M). Both substrate series were performed in triplicate. Substrate-velocity data were plotted and fit to the Michaelis-Menten equation using nonlinear regression with Deltagraph Pro3 software on a Macintosh computer.
Purification of Recombinant Yeast MTHFS-The pET-MTHFS construct and the pREP4-GroESL plasmid (22) were co-transformed into E. coli BL21(DE3) cells. Transformants containing both plasmids were used to inoculate a 3-ml 37°C starter culture in 2YT medium (1.6% tryptone, 1% yeast extract, 0.5% NaCl) containing 50 g/ml ampicillin and 50 g/ml kanamycin. When the A 550 of the culture reached 0.6, an aliquot was used to inoculate a 25-ml 2YT/ampicillin/kanamycin culture at A 550 ϭ 0.1. The 25-ml culture was grown at 37°C to A 550 ϭ 0.6, at which point the entire culture was used to inoculate a 1-liter culture in 2YT/ampicillin/kanamycin. The 1-liter culture was incubated in a 37°C orbiting incubator until A 550 ϭ 0.6 -0.8 was reached. The culture was then cooled in a water bath to 21°C, and protein production was induced by the addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration of 2 mM. The induced culture was incubated at room temperature for 8 -10 h on an orbital table. The cells were harvested by centrifugation at 8,000 rpm for 15 min in a GSA rotor (Sorvall). The medium was discarded, and the cell pellet was frozen at Ϫ20°C until use.
The frozen cells were resuspended in an equal volume of column buffer (0.1 M KPO 4 , 10% glycerol (v/v), 1.0 mM NaATP, 0.05% Triton X-100, 2.0 mM imidazole, pH 7.0) and disrupted by two passes through a French pressure cell at 12,000 p.s.i. Debris was removed by centrifugation at 12,000 rpm in an SS-34 rotor (Sorvall) for 15 min. The supernatant was passed through a 0.22-m cellulose acetate syringe filter (Whatman) to yield the cell-free extract. Expression of MTHFS was confirmed by SDS-PAGE through a 12% acrylamide gel. The cellfree extract was loaded onto an 8-ml cobalt immobilized metal affinity column (Talon; CLONTECH) equilibrated in column buffer. The column was washed with 10 volumes of column buffer followed by 10 volumes of column buffer containing 20 mM imidazole. Protein was eluted with a 50 -500 mM imidazole linear gradient in column buffer and collected in 5-ml fractions. Protein purity was demonstrated by a single band after SDS-PAGE, and fractions containing pure protein were pooled for study. Concentration of the pooled protein was determined by densi-tometry of Coomassie Blue-stained SDS-PAGE gels, using bovine serum albumin as a standard.
Construction of Mutant Yeast Strains-Disruptions in yeast genes were effected using a PCR-based gene targeting method (23). The URA3 cassette of pJR-URA3 (24) was amplified with primers that contain URA3 sequences in their 3Ј-ends and FAU1 sequences in their 5Ј ends. The upstream primer (5Ј-GTAAGTCACTTCCTGAAAACCATTCT-GCTTGCGAGCCCGGTAAAACGACGGCCAGT-3Ј) includes 39 nucleotides of sequence 300 bp upstream of the FAU1 start codon. The downstream primer (5Ј-CCAATGTATGGATCCATCTCCGCATACTATACAA TCCATGGCAGCTATGACCATGATTACGCC-3Ј) includes sequences complementary to codons 196 -209 of the 211-codon ORF. The 5Ј underlined nucleotides of each primer direct the fragment to recombine at the FAU1 locus. The amplified fragment was gel-purified and used to transform diploid strain DAY4a/␣ to uracil prototrophy. PCR using Taq polymerase was performed directly on positive colonies (25) with FAU1-specific primers 5Ј-CCATCAAAGGCCTGTTTGCT-3Ј and 5Ј-CTAGCGAA-CAAGGGAATCCA-3Ј to confirm the FAU1 disruption. FAU1-disrupted diploids were sporulated on solid medium (26) for 3-4 days. Tetrads were prepared for dissection using lyticase (Sigma) and dissected using a Zeiss microscope. The resulting spores were replica-plated onto selective and rich media. Several spores were tested for mating type by mating with an a and ␣ test strain. The URA3 cassette was evicted as described by Roca et al. (24). Strains that had excised the URA3 cartridge were identified by plating onto medium containing uracil (2 mg/ml) and 5-fluoroorotic acid (1 mg/ml). These strains were screened by colony PCR to demonstrate excision of the URA3 cartridge. The resulting ⌬fau1 disruption strain was designated WHY1 (Table I). Haploid strains disrupted at the FAU1 locus in combination with disruptions at other loci involved in folate metabolism were obtained through crosses of WHY1 with strains carrying the disruption(s) of interest. Following mating, the strains were sporulated, and tetrads were dissected onto YPD plates. Progeny were screened for disrupted genes by colony PCR or in some cases by growth phenotype. The new haploid strains (Table I) were characterized by their ability to grow on selective medium and in some cases by growth rates in selective media. WHY1.1 and WHY1.2 were derived from a cross of WHY1␣ with ATY3.1. WHY1.4, WHY1.5, and WHY1.6 were derived from a cross of WHY1␣ with EKY3. WHY1.7, WHY1.8, and WHY1.9 were derived from a cross of WHY1␣ with RRY3. WHY3 was derived from a cross of ATY3.1 with TR3. WHY1.3 was derived from a cross of WHY3 with WHY1␣. WHY1.3.1 was derived from a cross of WHY1.3 with YCY38B. YCY38B was derived from a cross of DAY4 with TR3. WHY3.1 was obtained from WHY1.3.1 by disruption of the HIS4 gene with a URA3 cassette using the PCR-based gene targeting method described above. The primers used were HIS4KO5Ј (5Ј-ATGGTTTTGCCGATTCTACCGTTAATTGATGATGTA AAACGACGGCCAGT-3Ј) and HIS4KO3Ј (5Ј-AAACTTTAAGGCATCCG AATCACAGTCAGTCAGCTATGACCATGATTACGCC-3Ј) (URA3 se-quences underlined). Subsequent excision of the URA3 cassette was confirmed by PCR.
Determination of Cellular 5-Formyltetrahydrofolate Levels-Intracellular 5-CHO-THF levels were determined spectrophotometrically by measuring the formation of the ternary complex SHMT-glycine-5-CHO-THF (27). Midlog phase yeast cultures growing in YPD were harvested by centrifugation and washed once in extraction buffer (50 mM K⅐BES, pH 7.0). The cell pellet was evenly divided, and the cells were resuspended in extraction buffer (1 ml/g, wet weight) either with or without 50 mM 2-mercaptoethanol and 50 mM ascorbate. Cells were disrupted by vortexing with glass beads at high speed for 10 min in a cold room (4°C). The extract was clarified by centrifugation at 18,000 ϫ g for 5 min and transferred to fresh tubes. Aliquots of the extract were reserved for protein determination, and the remainder was boiled for 5 min (if 2-mercaptoethanol and ascorbate were included) or 20 min (if no reductant added) and then centrifuged at 18,000 ϫ g for 5 min. For samples without 2-mercaptoethanol and ascorbate, 100-l aliquots of the supernatant were brought to pH 12 with NaOH and then boiled for 10 min to destroy folates other than 5-CHO-THF. Only the 5-CHO-THF derivative is stable under these conditions (27,28). After cooling on ice, the samples were adjusted to pH 7 with HCl and then centrifuged at 18,000 ϫ g.
The 5-CHO-THF content of the supernatant was determined in a reaction containing 50 mM glycine and 8.2 M SHMT in 50 mM K⅐BES, pH 7.0, in a total volume of 300 l. Spectrophotometric readings at 502 and 550 nm were taken (initial absorbance, I), and then SHMT was added to initiate the binding reaction. The reaction was incubated at room temperature, and readings at 502 and 550 nm were taken after 20 min (final absorbance, F). Absorption of the ternary complex was given by (A 502 . Subtraction of the difference in absorption at 550 nm allows correction for the addition of SHMT to the sample. The extinction coefficient for the ternary complex at 502 nm is 40,000 M Ϫ1 cm Ϫ1 (29).

RESULTS
Identification and Cloning of Yeast MTHFS-An ORF (YER183c) with homology to human MTHFS was identified in the Saccharomyces Genome Databank using a BLAST search against the human MTHFS sequence (30). When aligned with human and rabbit MTHFS using the ClustalW algorithm (31) (Fig. 1), the three proteins were found to share 23% identities and 25% conservative substitutions. The yeast ORF encodes a 211-amino acid protein with a predicted molecular mass of 24.1 kDa. The amino acid sequence was analyzed using PSORTII (humangen.med.ub.es/tools/PSort2_form.html) and is pre- Purification and Kinetic Analysis of Recombinant Yeast MTHFS-The His-tagged recombinant protein was expressed from the pET-16b vector in E. coli BL21(DE3) following induction by isopropyl-␤-D-thiogalactopyranoside. Expression of soluble MTHFS required coexpression of the E. coli GroEL and GroES proteins from the pREP4-GroESL plasmid (22) along with a shift in induction temperature from 37 to 21°C. The pREP4-GroESL plasmid, a gift from Dr. Martin Stieger (Hoffman-La Roche), contains the E. coli genes encoding GroEL and GroES behind the Lac promoter/operator element. This construct allowed for production of these chaperonins in conjunction with expression of MTHFS. Without coexpression of the chaperone proteins, all MTHFS protein was found to be in the particulate fraction at both temperatures. However, when the co-transformed cells were induced at 21°C, an adequate portion of the expressed protein was found in the soluble fraction.
Enzyme activity was found to be unstable when the bacterial cells were disrupted by sonication; therefore, cells were disrupted with a French pressure cell. Chromatography on a cobalt metal affinity column yielded a pure protein that migrated at ϳ28 kDa on SDS-PAGE (Fig. 2), consistent with the predicted molecular weight of the recombinant protein with its Nand C-terminal extensions. The purified protein precipitated rapidly during concentration by ultrafiltration or during dialysis. The addition of 10% glycerol and 1 mM ATP and storage at 4°C stabilized the protein for up to 1 week.
Purified His-tagged recombinant protein was used to demonstrate that the cloned ORF encodes an active MTHFS. Preliminary experiments revealed that enzymatic activity was inhibited by salt; thus, MgATP was used instead of MgCl 2 and NaATP. Activity was followed spectrophotometrically by monitoring the formation of 5,10-CH ϩ -THF at 360 nm. The purified protein exhibited a specific activity of 7 units/mg of protein.
The steady state kinetic parameters for ATP and (6R,6S)-5-CHO-THF monoglutamate were determined by nonlinear re-gression analysis of the kinetic data (Fig. 3). The K m values obtained for ATP and (6R,6S)-5-CHO-THF were 43 M and 33 M, respectively. These values are similar to those obtained for the human enzyme (Table II), confirming the identity of this protein as a functional MTHFS.
Disruption of Yeast Gene Encoding MTHFS-The gene encoding the yeast MTHFS was disrupted in the diploid strain DAY4a/␣ using a PCR-based gene targeting method (23). DAY4a/␣ was transformed to uracil prototrophy with a PCR fragment containing the URA3 cartridge flanked by sequences from the MTHFS ORF at both ends. Recombination at the correct locus results in the replacement of about 900 bp of the MTHFS gene, including the entire ORF, with the 1.2-kbp URA3 cartridge. Uracil prototrophs were screened for proper insertion of the URA3 cartridge by colony PCR using a primer pair designed to amplify from 150 bp upstream to 150 bp downstream of the expected insertion site. A heterozygous diploid gave the expected 1043-bp band for the wild type locus and a 1481-bp band for the disrupted locus (Fig. 4, lane 2). A haploid strain carrying the disrupted gene (DAY4 mthfs::URA3) was obtained by sporulation and tetrad dissection onto YPD plates. The spore progeny were screened for uracil prototrophy and by PCR (Fig. 4, lanes 3-6) to identify those carrying the disrupted gene. Finally, the URA3 cassette was evicted, and the resulting haploid disruptant strain was designated WHY1 (Table I).
MTHFS Activity in Yeast-Extracts from the disrupted and wild type strains were assayed for MTHFS enzyme activity. No activity was detected in extracts prepared by glass bead disruption. Instead, cells were treated with lyticase, and the resulting spheroplasts were lysed by brief sonication. Extract was cleared of debris by centrifugation, and the supernatant was used for the enzyme assay. MTHFS activity in the wild type strain, DAY4, was readily detectable (1.46 milliunits/mg of protein), whereas no MTHFS activity could be detected in the disrupted strain WHY1 (data not shown).
Phenotype of Yeast Lacking MTHFS-5-CHO-THF has been proposed as a storage form of readily available one-carbon units and of reduced folate in fungal spores (3) and as a regulator of SHMT (13). Cells lacking MTHFS activity might be expected to accumulate 5-CHO-THF. If S. cerevisiae spores normally accumulate 5-CHO-THF, a possible phenotype of mthfs mutants might be poor germination of spores incapable of converting the stored cofactor to its active form. To determine the effect on spore germination, a homozygous disrupted diploid strain, WHY1a/␣ (Table I) was created through mating. Sporulation was induced by nitrogen starvation, and tetrads were dissected onto rich medium. No difference in germination rate was observed between the mutant and the wild type diploid strain DAY4a/␣ (data not shown).
Accumulation of 5-CHO-THF in growing cells may also lead to inhibition of SHMT activity, producing a phenotype similar to that of strains lacking one or both SHMTs. In yeast cells carrying a ser1 mutation, which blocks the synthesis of serine from glycolytic intermediates (32), SHMT activity is required to satisfy serine requirements. High levels of glycine (100 mg/ liter) can provide both the two-carbon unit and the 5,10-methylene-THF (via the glycine cleavage system) required for serine synthesis via SHMT (33,34). ser1 mutants lacking mitochondrial SHMT have longer doubling times than the parental strain when grown on high glycine (35). The addition of adenine to the high glycine medium completely rescues the growth of SHMT ϩ cells, but mitochondrial SHMT mutants grow very slowly (9-h doubling time) (35). We used this medium to test whether deletion of MTHFS affected SHMT activity in vivo. The mthfs deletion strain WHY1 was grown in high glycine minimal medium supplemented with adenine, and the growth rate was directly compared with the parental strain DAY4. Doubling rates for the strains were identical (2.9 h), suggesting no impairment in the conversion of glycine to serine by SHMT in the MTHFS mutant.
We next took a genetic approach to detecting a phenotype resulting from deletion of the MTHFS gene. S. cerevisiae possesses a de novo pathway for the synthesis of folic acid that begins with GTP and ends with dihydrofolate (Fig. 5). Dihydrofolate is then reduced to the active coenzyme, tetrahydrofolate, by dihydrofolate reductase (DHFR). Dihydrofolate synthase (DHFS), encoded by the FOL3 gene, catalyzes the addition of a glutamate residue to dihydropteroate to produce dihydrofolate. fol3 mutants are thus unable to synthesize tetrahydrofolate and require the addition of either folic acid or 5-CHO-THF (folinic acid) to the medum for growth (36). Utilization of folinic acid by fol3 mutants should be dependent on a functional MTHFS, whereas utilization of folic acid requires only DHFR and should thus support the growth of fol3/mthfs double mutants (Fig. 5). To test this prediction, the mthfs knockout WHY1 was mated with a fol3 mutant strain (CD208 -2B) lacking DHFS. If the two loci segregate independently, tetratype tetrads will predominate, where one of four spores in each tetrad harbors both mutations. Following sporulation and

TABLE II Kinetic parameters of purified MTHFS enzymes
The kinetic parameters for all three species were determined at 30°C. The human and rabbit enzymes were assayed at pH 6.0 (40,41). The yeast enzyme (this study) was assayed at pH 7.0. The folate K m values were determined using the R,S mixture in each case.  4. Disruption of ORF YER183c in yeast. PCR primers amplify a 1043-bp product from the wild-type gene and a 1481-bp product from the disrupted gene. Lane 1, wild-type haploid (DAY4); lane 2, heterozygous diploid; lanes 3-6, haploid progeny of a dissected tetrad. MW, 1-kbp ladder molecular weight markers.
FIG. 5. Two metabolic routes to tetrahydrofolate. The de novo pathway begins with GTP and requires the activities of dihydrofolate synthase (DHFS; FOL3 gene) and dihydrofolate reductase (DHFR; DFR1 gene). DHFR catalyzes the reduction of folic acid to dihydrofolate and dihydrofolate to tetrahydrofolate. Folinic acid (5-CHO-THF) can also satisfy the tetrahydrofolate requirement in a pathway that requires the activities of methenyl-THF synthetase (MTHFS; FAU1 gene) and methenyl-THF cyclohydrolase (ADE3 gene) to produce 10-formyl-THF, which yields tetrahydrofolate after donation of the formyl group in the transformylase reactions of purine synthesis. tetrad dissection, spore germination was examined on medium supplemented with either 5-CHO-THF or folic acid versus medium with no supplementation (Table III). Tetrads dissected onto unsupplemented rich medium gave two viable and two nonviable progeny, consistent with the folate requirement of fol3 mutants. On medium with 5-CHO-THF added, three of four spores were observed to germinate and produce colonies. Folic acid supplementation was found to support complete germination of tetrads, demonstrating that the failure to germinate on other media is due to folate starvation. If the diploid was first transformed with a multicopy plasmid containing the MTHFS gene, germination was nearly 100% when tetrads were dissected onto plates containing 5-CHO-THF (data not shown). These results confirm the function of MTHFS in vivo, and we have named the gene FAU1 (for folinic acid utilization).
The fau1 deletion strain was mated with strains lacking other enzymes involved in folate utilization to generate double or triple disruptants (Table I). The resulting strains were examined for growth requirements on plates. Strains that carried fau1 disruptions in combination with disruptions of the genes encoding mitochondrial and/or cytosolic SHMTs (WHY1.4, WHY1.5, WHY1.6) and either one or both MTHFRs (WHY1.7, WHY1.8, WHY1.9) demonstrated no growth differences from the parental strains. Strains that carried the fau1 disruption in combination with disruptions of the genes encoding either of the bifunctional AICAR transformylase/IMP cyclohydrolases (WHY1.1, WHY1.2) also showed no new phenotype. However, when FAU1 was disrupted in combination with both ade16 and ade17, a new methionine deficiency was detected in the triple mutant strain (WHY1.3). The methionine-deficient strain demonstrated a normal growth rate in minimal medium supplemented with methionine, but growth was reduced in medium lacking methionine (Table IV). Neither parental strain (ATY3.1, WHY1) exhibited a methionine deficiency. S-adenosylmethionine, but not homocysteine, could rescue the methionine deficiency (data not shown), suggesting a homocysteine remethylation defect.
ade16 ade17 double mutants require adenine (37) and might be expected to accumulate the substrate of AICAR transformylase, AICAR (Fig. 6). AICAR has been reported to inhibit Sadenosylhomocysteine hydrolase and adenosine deaminase (38), two enzymes in the homocysteine remethylation pathway, and thus might interfere with methionine synthesis. AICAR is actually produced in two pathways: purine and histidine biosynthesis (Fig. 6). To test whether AICAR might be involved in the methionine deficiency of the triple mutant, a new strain was generated (WHY3.1; ⌬fau1 ⌬ade16 ⌬ade17 ⌬his4 ade2) ( Table I). ADE2 and HIS4 encode enzymes that catalyze steps preceding AICAR in the purine and histidine biosynthetic pathways, respectively (Fig. 6). WHY3.1 is thus unable to make AICAR from either pathway. When grown in minimal media, WHY3.1 grew at the same rate with or without methionine (Table IV). This suggests that the combination of the fau1 mutation with the ade16/ade17 mutations leads to excessive accumulation of AICAR, which affects some aspect of homocysteine remethylation, leading to the methionine deficiency.

Intracellular 5-CHO-THF Levels-
The new methionine deficiency observed in the ⌬fau1 ⌬ade16 ⌬ade17 strain (WHY1.3) suggested that 5-CHO-THF might also be elevated. Intracellular 5-CHO-THF levels were determined by spectrophotometrically measuring the formation of the ternary complex SHMTglycine-5-CHO-THF. The ternary complex has an extinction coefficient of 40,000 M Ϫ1 cm Ϫ1 at 502 nm (29); thus, 5-CHO-THF concentrations as low as 0.25 M (A 502 ϭ 0.01) may be reliably determined. Because SHMT forms a light-absorbing ternary complex with all reduced folates except 10-CHO-THF (39), samples were prepared in the absence of reducing agents and boiled; under these conditions, only the 5-CHO-THF derivative is stable (27,28). The accuracy of the ternary complex formation method of determining 5-CHO-THF was demonstrated by testing a known concentration of the pentaglutamate derivative of 5-CHO-THF. The standard was found to be stable under the conditions of preparation, and the concentration determined by the binding assay was the same as the concentration given by spectrophotometric determination (data not shown).
To obtain an estimate of the total cellular folate pool, one set of extracts was prepared using 2-mercaptoethanol and ascorbate to protect reduced folates other than 5-CHO-THF. Using the values from these samples, a ratio of 5-CHO-THF to total TABLE III Tetrad dissection of fau1 fol3 diploids Parental strains used in the cross were WHY1a (a ⌬fau1) and CD208 -2B (␣ ⌬fol3). Twenty-two or more tetrads were dissected onto each of three rich plates supplemented as indicated.
folate can be estimated (Table V, percentage of total folate).
The results indicate that a higher proportion of folates were in the form of 5-CHO-THF in the ⌬fau1 strain WHY1 (30%) when compared with the wild type strain DAY4 (10%). In the strains with no SHMT activity, (EKY3 and WHY1.6), 5-CHO-THF levels were below detection, and this ratio could not be calculated. In the methionine-deficient strain (WHY1.3), 5-CHO-THF represented 80% of the total cellular folate.

DISCUSSION
The experiments described here confirm that the yeast gene, YER183c, encodes a functional MTHFS, with properties very similar to the mammalian homologs. The ORF predicts a protein of 211 residues with 23% identity to the human and rabbit enzymes. An N-terminally His-tagged version of the yeast protein was expressed in E. coli and purified to homogeneity by metal affinity chromatography on a cobalt column. Like the rabbit (8) and Lactobacillus casei (9) enzymes, the yeast enzyme was inhibited by high salt. The removal of salt and inclusion of glycerol and the nonionic detergent, Triton-X-100, were found to increase the stability and activity of the enzyme. The K m values for (6R,6S)-5-CHO-THF monoglutamate (Table  II) were found to be similar to those obtained for the human (40) and rabbit enzymes (41), confirming the identity of this protein as a functional MTHFS. The rabbit enzyme has a K m of 0.2 M for the pentaglutamate form of 5-CHO-THF (10), indicative of a polyglutamate binding site on the enzyme. Although we have not examined the yeast enzyme with polyglutamate substrates, the high sequence homology suggests the yeast enzyme will also exhibit a polyglutamate binding site.
The gene encoding MTHFS is not essential in yeast. The gene was first disrupted in a diploid yeast strain. This heterozygous diploid was then sporulated, and tetrads were dissected to obtain haploid disruptants. Loss of one MTHFS locus in the diploid did not affect sporulation or subsequent spore germination. Haploid disruptants grew normally under all conditions tested, on both liquid and solid media. The only phenotype observed in the haploid disruptants was loss of MTHFS enzyme activity (measured in crude extracts), indicating that the yeast genome possesses only one gene for MTHFS. Kruschwitz et al. (3) reported that 5-CHO-THF polyglutamates are the major folate derivatives in Neurospora crassa spores. Because 5-CHO-THF is the most stable reduced folate derivative, they hypothesized that this form of the coenzyme serves as a storage form of reduced folates and one-carbon units in dormant cells. A functioning MTHFS might then be essential to convert the stored 5-CHO-THF into the active one-carbon pool as the cell resumes vegetative growth. If S. cerevisiae spores accumulate 5-CHO-THF in a similar manner, then spores from a homozygous diploid disruptant might display reduced germination rates or impeded growth. However, a homozygous diploid MTHFS disruptant (WHY1a/␣) sporulated efficiently, and the germination and growth of its haploid progeny was the same as that of the wild type diploid DAY4a/␣.
Although the folate status of yeast spores was not measured, the inability to metabolize 5-CHO-THF via MTHFS clearly did not affect the processes of sporulation or spore germination under the conditions tested.
In vivo function of the yeast MTHFS was demonstrated by a genetic approach. Yeasts are capable of synthesizing folic acid by a pathway that begins with GTP (Fig. 5). Disruption of the folic acid biosynthetic pathway results in yeast auxotrophic for folate; this requirement can be satisfied by the addition of 5-CHO-THF (folinic acid) to the media (36,42). As illustrated in Fig. 5, utilization of 5-CHO-THF should require MTHFS activity. Folic acid, on the other hand, can bypass both the DHFS and MTHFS steps, requiring only a functional DHFR for its conversion to THF. Haploid strains WHY1 (MTHFS disruptant) and CCD208B (fol3 disruptant) were crossed to produce an mthfs/MTHFS fol3/FOL3 heterozygous diploid. Upon sporulation of this diploid, one of four spores in each tetrad should harbor both mutations, since the two loci are on separate chromosomes. As expected, when tetrads were dissected onto media containing 5-CHO-THF, on average only three of four spores germinated (Table III). Folic acid supplementation was found to support complete germination of tetrads. If the diploid was first transformed with a multicopy plasmid containing the yeast MTHFS gene, germination was nearly 100% on plates containing 5-CHO-THF. These data confirm the role of MTHFS in the utilization of folinic acid in vivo. We have thus designated the YER183c gene FAU1 for folinic acid utilization.
The metabolic role of MTHFS was further investigated by studying strains carrying the fau1 disruption in combination with mutations in other folate enzyme genes. No new growth phenotypes were observed when the fau1 disruption was combined with disruption of the two SHMT genes (SHM1 and SHM2; Refs. 30 and 38) or the two methylenetetrahydrofolate reductase genes (MET12 and MET13; Ref. 39). However, when the fau1 disruption was combined with disruptions of two genes in purine biosynthesis, a striking methionine deficiency was observed, and it appeared to be related to the purine intermediate, AICAR.
AICAR is produced in one of the latter steps of the de novo purine biosynthetic pathway. It is converted to IMP by the two activities of the bifunctional AICAR transformylase/IMP cyclohydrolase (Fig. 6). Yeast have two isozymes of this bifunctional enzyme, encoded by the ADE16 and ADE17 genes (37). Either of these enzymes is sufficient to support purine biosynthesis in yeast, whereas the double disruptant (⌬ade16⌬ade17) is an adenine auxotroph. In addition to a purine requirement, an ⌬ade16⌬ade17 strain is also auxotrophic for histidine (45). The connection between purine and histidine metabolism is not well understood, but AICAR is also generated as a by-product of histidine biosynthesis (32), and this AICAR can presumably be utilized in purine biosynthesis (Fig. 6). Yeast that are ade3 mutants also exhibit a secondary histidine requirement (46). The ADE3 gene encodes the trifunctional enzyme C 1 -THF syn- thase (47), which produces the 10-CHO-THF substrate required in the AICAR transformylase reaction. We would predict that AICAR would accumulate in strains lacking AICAR transformylase activity (⌬ade16⌬ade17 mutants) or in strains unable to synthesize cytoplasmic 10-CHO-THF (ade3 mutants). These observations led Tibbetts and Appling (45) to propose that AICAR inhibits histidine biosynthesis by feedback regulation, explaining the histidine requirement of both ade16ade17 and ade3 mutants. This hypothesis is strengthened by the observation that blocks in the de novo purine pathway that precede AICAR formation do not result in histidine auxotrophy. The fau1 disruption was introduced into an ade16 ade17 background by crossing and tetrad dissection to create the haploid strain WHY1.3 (⌬fau1⌬ade16⌬ade17), which lacks all AICAR transformylase and MTHFS activities. In addition to the adenine and histidine requirements caused by the ade16ade17 disruptions, this strain exhibited a new methioninedependent growth defect (Table IV). The parental strains WHY1 (⌬fau1) and ATY3.1 (⌬ade16⌬ade17) were expected to accumulate 5-CHO-THF and AICAR, respectively. Since neither parental strain had a methionine deficiency, we speculated that the effect required the simultaneous accumulation of both metabolites. This was tested by introducing additional mutations that block the synthesis of AICAR. Mutation of the HIS4 gene blocks AICAR production from the histidine biosynthetic pathway, and mutation of ADE2 blocks its production from the de novo purine pathway (32) (Fig. 6). If just one of the pathways was blocked, the resulting strains remained methionine-deficient (data not shown). When both pathways were blocked, however, the strain WHY3.1 (⌬fau1⌬ade16⌬ade17 ⌬ade2⌬his4) regained normal growth on media lacking methionine (Table IV). Therefore, it appears that the accumulation of AICAR, combined with elevated 5-CHO-THF, leads to a defect in methionine synthesis. There is some precedent for an interaction between AICAR and methionine metabolism. AICAR was found to inhibit S-adenosylhomocysteine hydrolase in extracts of blood cells from rabbit and human, although significant inhibition required 1 mM AICAR (38). S-Adenosylhomocysteine hydrolase is involved in the remethylation cycle, and its inhibition would lead to decreased homocysteine levels, thereby slowing methionine synthesis. However, homocysteine depletion is probably not the mechanism in this case, since the methionine deficiency was not alleviated by the addition of homocysteine in the medium (data not shown).
Whatever the mechanism, it appears to require elevated 5-CHO-THF as well as AICAR. Intracellular 5-CHO-THF was indeed elevated in fau1 mutants and was highest in WHY1.3 (⌬fau1⌬ade16⌬ade17). The intracellular 5-CHO-THF concentration was estimated to be 5 M in wild type cells (Table V). This increased to 16 M in the ⌬fau1 mutant (WHY1) and to 62 M in the ⌬fau1⌬ade16⌬ade17 mutant. The polyglutamate forms of 5-CHO-THF are known inhibitors of several mammalian folate-dependent enzymes, including AICAR transformylase (48) and SHMT (29), with K i values in the low micromolar range. If the yeast enzymes have similar kinetic properties, the 5-CHO-THF levels observed in the MTHFS mutants are clearly sufficient to inhibit AICAR transformylase and SHMT activity in vivo. In yeast grown with serine, where the cytoplasmic SHMT operates in the direction of glycine and 5,10-CH 2 -THF production (35), inhibition of SHMT would lead to a deficiency of cytoplasmic 5,10-CH 2 -THF and thus the 5-methyl-THF required for methionine synthesis. The extreme elevation of 5-CHO-THF in the ⌬fau1⌬ade16⌬ade17 strain (12-fold) may explain why the methionine deficiency is seen only when both MTHFS and AICAR transformylase activities are missing in the same strain.
Additional evidence that 5-CHO-THF can inhibit SHMT activity in vivo comes from the recent work of Piper et al. (49) on the regulation of one-carbon metabolism in yeast. Based on experiments with various mutant strains, they concluded that yeast respond to a deficiency of 5,10-CH 2 -THF by increasing expression of the mitochondrial glycine cleavage system, which produces 5,10-CH 2 -THF from glycine. They showed that strain WHY1 (⌬fau1) has altered control of the glycine cleavage system gene expression, indicative of low cytoplasmic 5,10-CH 2 -THF levels. This effect is consistent with the elevated 5-CHO-THF in that strain inhibiting SHMT, causing the deficiency of 5,10-CH 2 -THF.
We can only speculate as to why the loss of AICAR transformylase activity is associated with additional accumulation of 5-CHO-THF in the fau1 mutant. Based on studies with Salmonella typhimurium grown under folate starvation conditions, Bochner and Ames (50) proposed that 5-aminoimidazole carboxamide riboside 5Ј-triphosphate (ZTP; the 5Ј-triphosphate derivative of AICAR) is an "alarmone" that signals a cellular deficiency of 10-CHO-THF. AICAR (ZMP) accumulates in the folate-starved cells and is subsequently phosphorylated to the triphosphate level (ZTP). The authors proposed that this ZTP then triggers alterations in cellular metabolism to remedy the one-carbon unit deficiency (50). If AICAR does influence the cellular content of one-carbon units, these might be stored as 5-CHO-THF. In the absence of MTHFS activity, the effect would be the trapping of the one-carbon unit and perhaps more importantly the coenzyme as 5-CHO-THF. ZTP can be detected in eukaryotic cells (51,52), but there are no reports on its existence in yeast. AICAR itself, however, has been shown to be an activator of the mammalian AMP-activated protein kinase (53,54). Members of the AMP-activated protein kinase subfamily are found in most eukaryotes, where they are the central component of a protein kinase cascade that acts as a metabolic sensor for the AMP/ATP ratio (55). S. cerevisiae also has an AMP-activated protein kinase homologue, the SNF1 protein kinase. In yeast, the SNF1 protein kinase plays a central role in the glucose repression response but is also involved in the regulation of sporulation, glycogen storage, peroxisome biogenesis, and lipid metabolism as well (56). Although the intracellular signal in yeast is not known, changes in the AMP/ATP levels correlate reasonably well with SNF1 kinase activity under a variety of nutritional conditions (57). Attempts to demonstrate a direct effect of AMP on the activity of the purified yeast kinase in vitro have been unsuccessful, but AICAR was not tested (57). Thus, it remains a possibility that the elevated AICAR in the ade16ade17 mutants, acting through the SNF1 kinase cascade, alters folate-mediated one-carbon metabolism, and in cells lacking the ability to move 5-CHO-THF into the active one-carbon pool, this form accumulates to the extreme levels observed in the ⌬fau1⌬ade16⌬ade17 strain.
Finally, our data also shed new light on the source of 5-CHO-THF in vivo. Stover and Schirch (4) showed that SHMT catalyzes the conversion of 5,10-CH ϩ -THF to 5-CHO-THF in vitro, and proposed that this is the primary source of 5-CHO-THF in vivo. Consistent with this proposal is the observation that E. coli and N. crassa cells deficient in SHMT activity do not accumulate 5-CHO-THF (3,5). On the other hand, Baggott (7) has argued that all cellular 5-CHO-THF is due to the nonenzymatic hydrolysis of 5,10-CH ϩ -THF in mildly acidic subcellular organelles, based on the in vitro observation that the equilibrium between 5,10-CH ϩ -THF and 5-CHO-THF favors 5-CHO-THF at pH 4.5. In the present study, 5-CHO-THF was undetectable in yeast strains entirely lacking SHMT activity (EKY3, WHY1.6). Deletion of fau1 was shown to cause accu-mulation of 5-CHO-THF but only in strains with active SHMT (compare WHY1 and WHY1.6; Table V), confirming that SHMT is the only significant source of 5-CHO-THF in vivo. Thus, results in S. cerevisiae, E. coli, and N. crassa strains with reduced or absent SHMT activity all support the hypothesis that the cellular source of 5-CHO-THF is the SHMT-catalyzed hydrolysis of 5,10-CH ϩ -THF.