Isolation and Characterization of the Saccharomyces cerevisiae MISl Gene Encoding Mitochondrial C1-Tetrahydrofolate Synthase*

C1-Tetrahydrofolate synthase is a trifunctional poly- peptide found in eukaryotic organisms that catalyzes 10-formyltetrahydrofolate synthetase (EC 6.3.4.3), 5,lO-methenyltetrahydrofolate cyclohydrolase (EC 3.5.4.9), and 5,lO-methylenetetrahydrofolate dehydrogenase (EC 1 .5. 1.5) activities. In Saccharomyces cerevisiae, C1-tetrahydrofolate synthase is found in both the cytoplasm and the mitochondria. The gene encoding yeast mitochondrial Cl-tetrahydrofolate syn- thase was isolated using synthetic oligonucleotide probes based on the amino-terminal sequence of the purified protein. Hybridization analysis shows that the gene (designated MISl) has a single copy in the yeast genome. The predicted amino acid sequence of mito- chondrial C1-tetrahydrofolate synthase shares 71% identity with yeast C1-tetrahydrofolate synthase and shares 39% identity with clostridial 10-formyltetra-hydrofolate synthetase. Chromosomal deletions of the mitochondrial C1-tetrahydrofolate synthase gene were generated using the cloned MISl gene. Mutant strains which lack a functional MISl gene are viable and can grow in medium containing a nonfermentable carbon source. In fact, deletion of the MISl locus has no detectable effect on cell growth. digestion of of the were plated the probed the nucleotide Plasmid YEpKS6) a hybridized to both oligonucleotide probes three rounds of isolation and rescreening. Subclones YEpKS7, YEpKS14, and YEpKS17 by inserting fragments produced by partial Sau3A digestion of plasmid YEpKS6 the BamHI of vector YEp24.

In Saccharomyces cereuisiue, Cl-THF synthase is found in the mitochondria as well as in the cytoplasm (10). Other folate-dependent enzymes reported to be present in mitochondria include serine hydroxymethyltransferase (11-13), the glycine cleavage system (14), sarcosine dehydrogenase (15, 16), dihydrofolate reductase (17), and methionyl-tRNA transformylase (18,19). Although the existence of many of these folate-dependent enzymes in the mitochondria is not yet welldocumented, this has not discouraged workers from proposing potential functions for these enzymes in mitochondrial metabolism. However, the physiological role of folate-dependent enzymes in the mitochondria has yet to be established.
We have taken advantage of the genetic manipulations possible in the yeast S. cerevisiae to investigate the function of mitochondrial Cl-THF synthase. We report here the isolation and the disruption of the gene encoding mitochondrial C,-THF synthase, which we have designated MISl. We have found, however, that a functional MISl gene is completely dispensable in yeast.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes, the large fragment of DNA polymerase I (Klenow fragment), and T4 DNA ligase were purchased from Boehringer Mannheim. T4 polynucleotide kinase, DNA polymerase I, and synthetic linkers were obtained from Bethesda Research Laboratories. All enzymes were used as recommended by the supplier. Radiochemicals were purchased from Amersham Corp. Common reagents were commercial products of the highest grade available.
Yeast Strains, Genetic Techniques, and Cell Growth-The S. cerevisiae strains used in this study are listed in Table I. The preparation of growth media and the techniques used for diploid construction, sporulation, and tetrad dissection have been described (20). Yeast were grown aerobically at 30 "C. Growth was monitored by measuring turbidity at 600 nm with a Zeiss PMQ I1 spectrophotometer. Unless otherwise indicated, cells were grown in YPD medium (1% yeast extract, 2% Bacto-peptone, 2% glucose) or, to maintain selective pressure for plasmids, in minimal medium (0.67% yeast nitrogen base, 2% glucose, plus appropriate auxotrophic supplements). Cells were harvested in late log phase (ODsw = 10). Extracts were prepared by disrupting washed cells with glass beads (0.45-mm diameter) in buffer containing 50 mM Tris-CI, pH 7.5, 10 mM KCl, 10 mM 2mercaptoethanol, 1 mM phenylmethanesulfonyl fluoride. Homogenates were centrifuged 30 min at 16,000 X g. The supernatant fractions were used for enzyme and protein assays.
Recombinant DNA Techniques-The techniques used for isolation of plasmid DNA, attachment of synthetic linkers, DNA blot analysis, nick translation, and transformation of Escherichia coli were as described by Maniatis et al. (21). Published procedures were used for the isolation of yeast genomic DNA (20) and for the transformation of yeast spheroplasts (22).
Cloning of the MIS1 Gene-Oligonucleotides 1 (5"AAYGAYGAR-ATHCARGC-3') and 2 (5'-AARCAYCCNAAYTTYAARCC-3'), which correspond to the amino-terminal sequence of mitochondrial C,-THF synthase (lo), were prepared with an Applied Biosystems Model 280A DNA synthesizer and purified by silica gel thin-layer chromatography (23 32P]ATP to a specific activity of lo9 cpm/pg using T4 polynucleotide kinase (21). The labeled oligonucleotides were used to screen a yeast genomic library which was constructed by M. Carlson (Columbia University, New York). The library was constructed by inserting fragments produced by partial Sau3A digestion of total yeast DNA into the BamHI site of vector YEp24. Bacterial colonies containing the yeast genomic library were plated and transferred onto nitrocellulose (24), and the filters were probed with the oligonucleotides and washed with 3.0 M tetramethylammonium chloride at 51 "C (oligonucleotide 2) (25). Plasmid DNA (designated YEpKS6) was recovered from a single clone that hybridized to both oligonucleotide probes through three rounds of isolation and rescreening. Subclones YEpKS7, YEpKS14, and YEpKS17 were constructed by inserting fragments produced by partial Sau3A digestion of plasmid YEpKS6 into the BamHI site of vector YEp24. Enzyme Assays-Assay of 10-formyl-THF synthetase and 5,lOmethylene-THF dehydrogenase depended on the spectrophotometric measurement of 5,lO-methenyl-THF concentrations in acidified (Ama = ASm) reaction mixtures (26). 10-Formyl-THF synthetase activity was assayed as previously described (27) except the triethanolamine buffer was replaced with 50 mM potassium/HEPES, pH 7.5, and KC1 and ammonium formate were added to 0.1 M each. 5,lO-Methylene-THF dehydrogenase activity was assayed as described (26) except that assay mixture contained 50 mM potassium/HEPES, pH 7.5,O.l M KCl, 0.6 mM NADP', 100 pM 2-mercaptoethanol, and 1.5 mM methylene-THF in a final volume of 1.0 ml. Activities are expressed in milli-international units (nanomoles of 5,lO-methenyl-THF formed or hydrolyzed per min) based on an extinction coefficient of 24,900 M" cm" for 5,lO-methenyl-THF. Protein concentrations were determined by the dye-binding assay of Bradford (28) with bovine serum albumin as a standard.
DNA Sequencing-A 5.7-kb PstI fragment from YEpKS6 was cloned into vector pEMBL19 (29), and subclones were generated by digestion with exonuclease I11 (30). Single-stranded templates were prepared from the subclones by superinfection with phage fl (29) and were sequenced by the dideoxy method (31) as modified by Biggin et al. (32). The region outside the PstI fragment was sequenced from single-stranded templates prepared from DNA restriction fragments cloned into pEMBL18 or pEMBL19. In some cases, synthetic oligonucleotides were used to prime sequencing reactions (33). Greater than 95% of the nucleotide sequence was determined from both strands.
Disruption of the MISl Locus-Plasmid pKS31 was constructed by inserting, via BamHI linkers, the 4.9-kb PuuII fragment from YEpKS6 into the BamHI sites of vector pUC4-K (34). Plasmid pKS33, which contains the MISl gene disruption, was constructed by converting the XbaI site of pKS31 to a SalI site using synthetic linkers, cleaving with SalI to remove linkers plus 1.7 kb of the MIS1 coding sequence, and inserting, via SalI linkers, a 1.14-kb HindIII fragment containing the URA3 gene. Diploid strain KSY4 was transformed with plasmid pKS33 linearized with BamHI (35), and Ura+ transformants were selected. One transformant, designated strain KSY5, was chosen for further study.

RESULTS
Isolation of the Gene Encoding Mitochondrial C1-THF Synthase-We designed two degenerate oligonucleotide probes based on the amino-terminal sequence of mitochondrial Cl-THF synthase (10). The probes were used to screen a yeast genomic library in an episomal vector (YEp24) by the colony hybridization technique (24). A single plasmid (YEpKS6) carrying a 10.1-kb genomic insert was isolated which hybridized to both oligonucleotides (Fig. 2). An ade3 deletion strain which lacks cytoplasmic C1-THF synthase (36) was transformed with YEpKS6 DNA. The YEpKS6-transformed strain had levels of synthetase and dehydrogenase activities approximately 10-fold greater than the same strain transformed with vector DNA alone (Table 11). These levels of synthetase and dehydrogenase activities are comparable to those found in ADE3 wild-type strains; however, transformation with YEpKS6 DNA did not alleviate the adenine and histidine requirement of d e 3 parent strain. We purified synthetase   activity from the YEpKS6-transformed strain and found that the purified protein is indistinguishable from mitochondrial CI-THF synthase with respect to its purification properties and its subunit molecular weight (data not shown).
We performed deletion analysis to localize the boundaries of the mitochondrial Cl-THF synthase gene on YEpKS6. Various fragments generated by partial Sau3A digestion of YEpKS6 were cloned into YEp24; and the resulting plasmids, YEpKS7, YEpKS14, and YEpKS17 (Fig. 2), were used to transform an ade3 deletion strain. Only the strain containing plasmid YEpKS17 had levels of synthetase and dehydrogenase activities comparable to those found in the YEpKS6transformed strain (Fig. 3). These data suggest that sufficient information to overexpress mitochondrial C1-THF synthase is carried on the PuuII fragment of YEpKS6. DNA blots probed with a plasmid carrying this PuuII fragment confirmed the presence of the cloned insert in yeast genomic DNA. The fragments observed in these blots were identical to those predicted from the restriction map of the cloned insert (data not shown). These data indicate that mitochondrial Cl-THF synthase is encoded by a single copy gene which we have designated MISl for mitochondrial Cl-THF synthase.
Nucleotide Sequence of the MISl Gene-The nucleotide were measured in extracts prepared from the transformed strains. Data are expressed as the specific activity in cells transformed with a recombinant plasmid to that in cells transformed with vector alone or "overexpression." sequence of the MISl gene is shown in Fig. 4. Analysis of the DNA sequence revealed one open reading frame which encodes a protein of 975 amino acids with a calculated molecular weight of 106,235. The predicted amino acid sequence of residues 35-74 is identical to the amino-terminal sequence of mitochondrial Cl-THF synthase (10). The sequence predicted for residues 1-34 is highly enriched for arginine, leucine, and serine, which is typical of mitochondrial targeting sequences (37). The removal of these initial 34 residues yields a "mature" protein with a calculated molecular weight of 102,251, which

VATVRALK] s [HGG] a p d v k p G q p l p s s y t e E N i c f v e K G a~N m c k q I a N i k q F G~P v [VVAIN] VATVRALK] s [HGG] a p n v k p G q s l p k e y t c E N i d f v a K G v s N l v k q l c N i k t F G i P v [VVAIN]
VATVRALK] m [HGG] v a k c d 1 G . . . . . . .

k F c T D t e g E i a = i r k a a l c a G a f c A v t s n h w a c G G k C a i d L A k a V i~.~s n q p " d F h f [LYDV] n s S v r F c T D s q a E i c v i k k~e I n a G s s h A v t s n h w m c G G k G a v e L A h a V v d .~t k c p k n F n f
[LYDV] nsSi

~F p T D t c s E k q l v f d k c k e m G v d v A i s d v f . a k G G d G g v c L A q k V i d v c c n k k s d F k v [LYDV] eeSi e d K 1 T s I v q k m Y g g~k i c v s p c A q k k I d t y k l ; q G f g n [LPIC] i [AKTQYS] IShDpsLkGvPrGF e d K l T t I v q k m Y g g a a i d i l p e A q~k I d m y k e q G f g n [LPIC] i [AKTQYS] 1ShDatLkGvPtGF p c K i T k I a k e i Y r s d k v n f s k a A k k q I a c l e k l G l d k [LPIC] m [AKTQYS] fSdDpnLlGaPeGF
tical (Fig. 5). The similarity extends for the entire length of the coding sequences except for the putative "targeting sequence" at the amino terminus of the mitochondrial isoenzyme. A comparison of the three protein sequences showed 39% identity when gaps were introduced to obtain maximal alignment (Fig. 5). The bacterial protein aligned with the carboxyl-terminal region of the yeast proteins mitochondrial C1-THF synthase (residues 344-975) and C1-THF synthase (residues 320-946). Disruption of the MISl Gene-Mitochondrial Cl-THF synthase is present in ade3 mutants that lack Cl-THF synthase, yet the mitochondrial isoenzyme cannot satisfy the requirement for adenine and histidine that is characteristic of most ade3 mutant strains. This observation suggests that C1-THF synthase and mitochondrial C,-THF synthase have different metabolic roles. To determine how the lack of mitochondrial C1-THF synthase affects cell growth, we disrupted the MISl gene in vitro and introduced the nonfunctional gene into the yeast chromosome. A 1.7-kb fragment of the MISl coding sequence was replaced with a DNA fragment carrying the URA3 gene (Fig. 6A). The disrupted gene was excised from the plasmid and was used to transform a diploid urd-52/ urd-52 ade3-130/ade3-130 serllserl strain (KSY4) to uracil prototrophy. Hybridization analysis of genomic DNA isolated from a Ura+ transformant confirmed that the disrupted gene had integrated into the MISl locus (Fig. 6B). The pattern of fragments seen in these blots suggests that the diploid transformant carries one wild-type copy of the MISl gene and one disrupted copy.
To determine whether a functional MISl gene is required for cell growth, we sporulated a MISl/misl::URA3 diploid (KSY5) to generate haploid segregants. Four spores in each of 12 tetrads dissected grew on rich YPD medium. Two spores per tetrad were Ura+, and two were Ura-. To test whether or not mitochondrial C,-THF synthase is produced from the disrupted MISl locus, we measured synthetase and dehydrogenase activities in extracts prepared from the spores of a single tetrad. Extracts prepared from the Ura-spores had levels of synthetase and dehydrogenase activities found in MIS1 ade3 strains, whereas extracts from the Ura+ sister  KSY4 (lanes 1 and  3 ) , and a Ura' transformant, KSY5 (lanes 2 and 4 ) , were digested with PuuII (lanes 1 and 2 ) or PstI (lanes 3 and 41, fractionated by electrophoresis on a 0.8% agarose gel, and transferred onto nitrocellulose. The filter was probed with plasmid pKS31 radiolabeled by nick translation. The molecular weight standards used were X DNAIHindIII fragments and 6x174 replicative form DNA/ HaeIII fragments. spores had no detectable activities (Table 111). These results indicate that misl mutant strains are viable and that mitochondrial C1-THF synthase is dispensable in yeast. Initiation of protein synthesis in mitochondria occurs via a unique tRNA species, tRNAN" (18), which functions as a formylated methionylated derivative. The formyl group is transferred to methionyl-tRNA'"' from 10-formyl-THF (19), a product of mitochondrial CI-THF synthase. Thus, a plau-sible function for this enzyme may be to supply 10-formyl-THF for the synthesis of formylmethionyl-tRNAMet. Reduced folates are required for mitochondrial function in yeast (39), possibly because mitochondrial protein synthesis is dependent on the formylation of methionyl-tRNAM" (40). However, we found that strains containing the disrupted MISl gene which completely lack mitochondrial CI-THF synthase grew on medium containing a nonfermentable carbon source (YPG me- These results indicate that mitochondrial Cl-THF synthase is not required for mitochondrial protein synthesis in yeast. Mitochondria contain many folate-dependent enzymes including serine hydroxymethyltransferase and the glycine cleavage system (41) as well as Cl-THF synthase (10). Genetic studies have demonstrated that certain folate-dependent reactions in the mitochondria supply one-carbon units for specific biosynthetic processes in the cytoplasm (17, 42). Thus, although mitochondrial Cl-THF synthase is not required for growth on rich media, it may offer an advantage to cells grown under more arduous conditions. However, we found that a misl::URA3 mutant strain (KSYS) and a MZSl wild-type strain (KSY10) grew at the same rate (doubling time = 2.25 h) in minimal medium (0.67% yeast nitrogen base, 2% dextrose, plus uracil and serine) (data not shown). These results suggest that the products of mitochondrial C1-THF synthase are not required for folate-dependent reactions in the cytoplasm.

DISCUSSION
We have isolated the gene encoding mitochondrial Cl-THF synthase from S. cerevisiae. In yeast, mitochondrial C,-THF synthase is encoded by a single gene which we have designated MISl. The protein encoded by MISl shows extensive homology with yeast cytoplasmic C1-THF synthase and clostridial 10-formyl-THF synthetase. Genetic and physical evidence indicates that Cl-THF synthase has two functionally independent domains (4, 36,43). Synthetase activity is catalyzed on a 70-kDa carboxyl-terminal domain, and dehydrogenase and cyclohydrolase activities are catalyzed on a 30-kDa amino-terminal domain. The alignment of the amino acid sequence predicted for monofunctional 10-formyl-THF synthetase with mitochondrial and cytoplasmic C1-THF synthases may define the synthetase domains of the trifunctional proteins. In both mitochondrial and cytoplasmic THF synthase, the putative synthetase domain is immediately preceded by a proline-rich region. These proline-rich regions probably have disordered and extended conformations that could act to separate the two domains. Proteolysis studies in our laboratory show that yeast CI-THF synthase can be cleaved with chymotrypsin to generate 70-and 30-kDa peptides. The combined proteolytic fragments have each of the activities of the intact enzyme. However, we cannot isolate an active dehydrogenase and cyclohydrolase fragment.3 Here we show that a fragment of the MISl gene which carries N. Tse and M. Williams, unpublished observations. sufficient information to encode the dehydrogenase and cyclohydrolase domain (YEpKS14) does not express a protein with dehydrogenase activity. These data may suggest that an intact synthetase domain is necessary to stabilize the dehydrogenase and cyclohydrolase domain in yeast Cl-THF synthases.
Mitochondria use active one-carbon units for the formylation of the initiator tRNA, methionyl-tRNAmet. The formyl donor in this reaction is 10-formyl-THF, a product of C1-THF synthase and mitochondrial Cl-THF synthase. We found that mutant strains that lack both C1-THF and mitochondrial Cl-THF synthase can grow on nonfermentable carbon sources, showing that these enzymes and their products are not required for mitochondrial function. These results can best be explained if mitochondrial protein synthesis can be initiated with unformylated methionyl-tRNAmet.
Although it is generally believed that the formylation of the initiator tRNA is required for protein synthesis in bacteria (44) and in eukaryotic organelles (40, 45), it has been established that this requirement is not absolute. Streptococcus faecalis can initiate protein synthesis in the absence of formylmethionyl-tRNAmet. This organism cannot synthesize folate and normally requires this vitamin for growth. However, the requirement for folate can be entirely replaced by the addition of serine, methionine, thymine, a purine base, and pantothenate to the growth medium (46). Under these folate-deficient conditions, formylmethionyl-tRNAmet is not synthesized, and initiation of protein synthesis proceeds with non-formylated methionyl-tRNA""' (47)(48)(49). The ability of this bacteria to initiate protein synthesis without formylated methionyl-tRNAmet is due to a single modification in the initiator tRNA. In folate-sufficient cells, the uracil residue in loop IV is methylated to ribothymine; whereas in folatedeficient cells, this methylation does not occur (50). It was later shown that in S. faecalis a folate derivative, 5,lO-methylene-THF, serves as the methyl donor in this methylation (51).
Similarly, certain E. coli mutants can initiate protein synthesis with unformylated methionyl-tRNA'"'. Baumstark et al. (52) isolated mutants that grew in the absence of paminobenzoate from a strain that required p-aminobenzoate for growth. These mutants cannot synthesize folate, and the cells contain no 10-formyl-THF. Extracts of the mutant strain support protein synthesis from exogenous mRNA; however, the proteins synthesized are initiated with methionine rather than with formylmethionine. The mutant strain is deficient in tRNA methyltransferase; and as a result, tRNA, from these cells have reduced levels of ribothymidine. Thus, in both folate-deficient S. faecalis and the mutant E. coli strain, the lack of ribothymidine in the tRNA allows initiation of protein synthesis with unformylated methionyl-tRNAmet. We are now directing our efforts toward determining whether the initiator tRNAmet from our ade3 misl mutant strain is formylated in vivo to investigate whether a similar mechanism can occur in yeast mitochondria.
Because reduced folates and their derivatives are not transported across the inner mitochondrial membrane (53), it is generally believed that active one-carbon units are independently generated in the mitochondria and the cytoplasm for use in each cellular compartment. However, there is evidence to suggest that one-carbon units generated in the mitochondria are used in the cytoplasm. Yeast tmp3 mutants, which lack the mitochondrial form of serine hydroxymethyltransferase, require dTMP, methionine, histidine, and adenine (13, 54). Also, a glycine-requiring Chinese hamster cell line was found to be deficient in the mitochondrial form of serine hydroxymethyltransferase (42). More recently, Barlowe and Appling4 found that isolated rat liver mitochondria can utilize serine or sarcosine to generate one-carbon units for purine synthesis in the cytoplasm. They propose that a significant fraction of the cell's one-carbon units are generated from serine in the mitochondria via mitochondrial serine hydroxymethyltransferase. The 5,lO-methylene-THF generated in this reaction is converted to formate via mitochondrial C1-THF synthase, which can exit the mitochondria to be activatedvia cytoplasmic C1-THF synthase for use in biosynthetic reactions in the cytoplasm. This pathway for generating cytoplasmic one-carbon units depends on the integrity of mitochondrial CI-THF synthase; however, we found that the presence of mitochondrial C1-THF synthase offers no advantage to cells growing on minimal medium. Our results indicate that the mitochondrial folate pathway is not essential in yeast, although they do not rule out the possibility that this pathway can act as an alternate route for the synthesis of activated one-carbon units.