Regulation of the Balance of One-carbon Metabolism inSaccharomyces cerevisiae *

One-carbon metabolism in yeast is an essential process that relies on at least one of three one-carbon donor molecules: serine, glycine, or formate. By a combination of genetics and biochemistry we have shown how cells regulate the balance of one-carbon flow between the donors by regulating cytoplasmic serine hydroxymethyltransferase activity in a side reaction occurring in the presence of excess glycine. This control governs the level of 5,10-methylene tetrahydrofolate (5,10-CH2-H4folate) in the cytoplasm, which has a direct role in signaling transcriptional control of the expression of key genes, particularly those encoding the unique components of the glycine decarboxylase complex (GCV1,GCV2, and GCV3). Based on these and other observations, we propose a model for how cells balance the need to supplement their one-carbon pools when charged folates are limiting or when glycine is in excess. We also propose that under normal conditions, cytoplasmic 5,10-CH2-H4folate is mainly directed to generating methyl groups via methionine, whereas one-carbon units generated from glycine in mitochondria are more directed to purine biosynthesis. When glycine is in excess, 5,10-CH2-H4folate is decreased, and the regulation loop shifts the balance of generation of one-carbon units into the mitochondrion.

tienzyme complex (GDC) (see Fig. 1 and Table I, reaction 3) also contributes to the one-carbon pool by generation of 5,10-CH 2 -H 4 folate (2), whereas formate activation to 10-HCO-H 4 folate occurs via the synthetase activity of the C1-tetrahydrofolate synthase trifunctional enzymes (see Fig. 1 and Table I, reactions 4c and 6c) (3). Each of the one-carbon H 4 folate pools are interconverted by the cyclohydrolase (see Fig. 1 and Table I, reactions 4b and 6b) and dehydrogenase (see Fig. 1 and Table I, reactions 4a and 6a) activities of the C1-tetrahydrofolate synthase enzymes, whereas flow between the mitochondrion and the cytoplasm is mediated by the three one-carbon donor molecules. The result is a dynamic metabolic system in which one-carbon units are interconverted between the mitochondrial and cytoplasmic compartments as well as between oxidation states. This flow has been the subject of extensive biochemical analyses (4 -9) that have shown that glycine, serine, and formate can each act to supplement all of the different one-carbon pools. Much less, however, is known of how cells regulate these flows.
In contrast to the other enzyme activities for the generation and interconversion of one-carbon units, glycine catabolism via the GDC is strictly mitochondrial. This enzyme is composed of four subunits each encoded separately. These are the P-protein (encoded by GCV2), the H-protein (encoded by GCV3), the T-protein (encoded by GCV1), and the L-protein (encoded by LPD1). Each of the GCV genes is unique to the GDC and on addition of exogenous glycine has been shown to be induced transcriptionally (10 -12). For GCV1 and GCV2 expression this was due to a 17-base pair sequence with the hexanucleotide 5Ј-CTTCTT-3Ј at its core in their promoters (13), and similar sequences have been identified in the GCV3 promoter. This work also indicated that H 4 folate or one of its derivatives may play a role in signaling the need for increased GCV transcription when glycine is added to the medium. This possibility led us to the hypothesis that the glycine effect is only one component of a broader one-carbon metabolism regulatory system. This has been confirmed from genome-wide transcript analysis in which the addition of glycine to yeast cells growing in minimal medium led to the induction of other genes central to one-carbon metabolism (14).
Here, we have used a combination of genetics and biochemistry to identify which derivative of H 4 folate acts to signal this response, and we propose a model for 5,10-CH 2 -H 4 folate-mediated regulation of metabolism through transcription. Because of the large body of work on the phenotypes of many one-carbon metabolic mutants (6,(15)(16)(17), strains were available that were specifically altered with respect to the different reduced H 4 folate pools. Gene expression studies in these strains were combined with 13 C NMR experiments to monitor one-carbon flow. These data indicated that control over cytoplasmic serine hydroxymethyltransferase activity is responsible for determining how GCV gene transcription is regulated. By monitoring the levels of the 5,10-CH 2 -H 4 folate pool (which is normally derived from serine), cells can up-regulate glycine catabolism for one-carbon generation when 5,10-CH 2 -H 4 folate is limiting or spare the breakdown of serine when glycine is in surplus.

Strains and Media
The yeast strains relevant to this study are listed in Table II. Yeast were grown in minimal medium as described previously (10) with auxotrophic requirements added at 40 mg liter Ϫ1 . To elicit the glycine response, glycine was added to a final concentration of 10 mM; a 2 M sodium formate stock (pH 7.0) was used to supplement media where indicated.

Gene Expression Analyses
Constructs and Yeast Transformation-Plasmids carrying the fulllength GCV2 promoter::lacZ fusion (pRH2 and pRH1), full-length GCV1 promoter::lacZ fusion, and full-length GCV3 promoter::lacZ fusion have been described previously (13). These were integrated individually in appropriate strains as a single copy at the URA3 locus using the lithium acetate transformation method (22).
␤-Galactosidase Assays-Yeast were grown to early log phase in the media indicated and harvested for assays as described previously (10).

NMR
Cell Growth, 13 C-Labeling, and Extract Preparation-NMR studies followed the method described in Ref. 6. A 1-liter culture of cells was grown to late log phase in yeast minimal medium with or without 100 mg liter Ϫ1 [2-13 C]-labeled glycine. Cells were harvested and resuspended in 20 ml of 0.3 N HCl and warmed over a boiling water bath for 1 h. This solution was centrifuged, the supernatant was replaced with fresh HCl, and the process was repeated. The supernatants from the two extractions were pooled, reduced using a Rotavapor, and dried in a Speedivac. Samples were stored at this stage at Ϫ80°C until analysis.
Immediately prior to NMR, samples were resuspended in 1 ml of deuterated dimethyl-d6-sulfoxide.
NMR Analysis-NMR spectra were recorded on a Bruker DMX 600 spectrometer equipped with a TXI-XYZ probe at a temperature of 298 K. Chemical shifts are relative to internal 3-trimethylsilyl propionate-2,2,3,3-d 4 , sodium salt set to 0 ppm. (This results in our measured shifts being ϳ1.4 ppm lower in frequency than those reported relative to tetramethylsilicone.) Metabolites were identified by comparison with previously published data on chemical shifts (Table III) (6). Relative intensities of lines within "triplets" produced by mixtures of single 13 C and double 13 C-13 C isotopomers were extracted with line shape deconvolution using the Bruker WINNMR program. 13 C NMR spectra were acquired using power-gated decoupling. A total of 2400 scans with an acquisition time of 1.14 s and recycle delay of 5.0 s were acquired for each sample, resulting in an overall time of 4 h 7 min per sample. The spectral width was 190 ppm, and 64,000 points were collected. The pulse lengths were 9.0 s, with a 78-degree flip angle. Spectra were processed with exponential multiplication and line broadening of 3 Hz.  (23). Equimolar proportions of H 4 folate and formaldehyde (5% [ 14 C]formaldehyde) were combined in a stoppered oxygen-and light-free container. The solution was brought to pH 5 by the dropwise addition of 1 N NaOH, and the mixture was incubated at room temperature for 15 min. The reaction mixture was left at 4°C for 60 min after which the pH was adjusted to 9.5 using 1 N NaOH. The 5,10-CH 2 -H 4 folate was purified at 4°C on a 10-ml DEAE-Sephadex column pre-equilibrated with 50 mM (NH 4 ) 2 CO 3 and washed immediately before loading with 100 ml of 50 mM 2-mercaptoethanol. After loading, the column was washed with 20 ml of 50 mM 2-mercaptoethanol and then eluted with a 100-ml gradient of 10 -500 mM (NH 4 ) 2 CO 3 containing 50 mM 2-mercaptoethanol (progress was monitored by absorbance at 295 nm). Fractions (2 ml) were collected, and those containing 5,10-CH 2 -H 4 folate were identified by absorbance and scintillation counting. Concentration and purity were estimated by correlating the pteridine ring concentration (⑀ 295 ϭ 30,000 M Ϫ1 cm Ϫ1 ) with the amount of radioactivity (1 mmol ϭ 1.3 ϫ 10 11 dpm) in a given fraction. Maximum purity was estimated at greater than 90%.
Samples were freeze-dried and stored under argon at Ϫ80°C away from light. Prior to use, the powder was resuspended in 50 mM 2-mercaptoethanol.

Gel Mobility Shift Assay
Assays were performed as described previously (24) using purified protein extracts obtained from heparin-Sepharose chromatography (25). Proteins were extracted from yeast strain BWG1-7A, purified, and incubated with 40-base pair DNA fragments harboring the glycine regulatory region of GCV2. These were prepared, separated by electrophoresis, and analyzed as described previously (13). The relative amount of DNA present in the DNA/protein complexes was estimated by PhosphorImager analysis. To reduce the effects of loading inconsistencies, these values were normalized to the total DNA present in each lane.

RESULTS
One-carbon metabolism comprises a complex set of reactions that contribute to many central biosynthetic pathways. Several approaches can be taken to identify components of the signaling system regulating the glycine response, although direct biochemical analyses are difficult because of the ability of the reactants to interchangeably supplement the pools of one-carbon-derived compounds and the relative instability of the H 4 folate derivatives. By combining extensive genetic analysis with biochemical approaches, several groups have provided clear indications of the flow of metabolites and the consequences of specific metabolic blocks (6,(15)(16)(17). In particular, Appling and co-workers have effectively combined genetics with the use of 13 C NMR to provide a wealth of information on the biochemistry of one-carbon metabolic flow. Here, we have combined genetic and biochemical approaches to extend our previous finding that a folate species plays a role in signaling the glycine response of the GCV genes (13). Because glycine catabolism via the glycine decarboxylase reaction directly contributes to the one-carbon pool, we first tested whether a mutant lacking GDC activity could still elicit a glycine response.
The Glycine Response Is Independent of the Ability to Metabolize Glycine-We previously proposed that changes in an intermediate or product of one-carbon metabolism (e.g. a derivative of H 4 folate) provide the signal for the glycine response. To test whether these changes require glycine catabolism, we transformed a strain mutant for glycine cleavage activity with a normally regulated GCV2::lacZ fusion (pRH2) and compared reporter gene expression in minimal medium with that in minimal medium containing 10 mM glycine (Fig. 2). A 3-to 4-fold increase in gene expression on the addition of glycine was observed, which is typical of the wild-type glycine response previously reported (10). Because this strain is mutant for the GCV1 gene product encoding the T-protein of the glycine decarboxylase complex and has no detectable GDC activity, it is unable to use glycine as a sole one-carbon and nitrogen source (26). Thus glycine catabolism has no direct effect on H 4 folatemediated one-carbon metabolism. To reconcile this result with the data indicating that GCV transcriptional signaling involves a H 4 folate intermediate, we sought a one-carbon metabolic mutant unable to elicit a normal glycine response. We reasoned that this strain would reveal how glycine acts to control onecarbon metabolism and the balance of cellular H 4 folate molecules.
Mutation of Cytosolic Serine Hydroxymethyltransferase Disrupts the Glycine Response of the GCV Genes-The cytoplasmic SHMT reaction is normally the principle route for the synthesis of one-carbon charged H 4 folate derivatives using H 4 folate and serine as substrates ( Fig. 1 and Table I, reaction 1). Fig. 3 shows the effect of a mutation eliminating cytosolic SHMT activity (shm2) on glycine induction of expression of all three GCV genes. Expression of each of the GCV genes was constitutively high in minimal medium with and without glycine. Interestingly, the elevated level of expression in this strain in minimal medium was about the same as that of the wild type (ADE3 SHM2) grown in minimal medium containing glycine. The mutation in this strain therefore elicited the same increase   in gene expression as seen for the addition of glycine to the wild type.
To determine whether the effect of this mutation was due to a disturbance of one-carbon metabolism and not simply loss of the SHMT enzyme, we supplemented the cytosolic one-carbon pool with formate. From Fig. 1 and Table I it can be seen that catabolism of formate in the cytosol should compensate for any one-carbon imbalance of an ADE3 shm2 strain via the activity of C1-tetrahydrofolate synthase (ADE3 gene product; Ref. 7). This enzyme first activates the one-carbon units derived from formate to 10-HCO-H 4 folate before conversion to 5,10-CH 2 -H 4 folate. In the wild-type strain (ADE3 SHM2) addition of excess formate caused a decrease in GCV2 gene expression (Fig. 4A). This reduction was even more evident in the ADE3 shm2 strain, occurring at a lower formate concentration. This effect was eliminated by the introduction of a mutation in the ade3 gene. The lack of response to formate by the ade3-130 shm2 strain shows that assimilation of cytoplasmic one-carbon units from formate is required for the down-regulation of gene expression. Interestingly, in these experiments the effect on control of adding glycine to the wild type was similar to that of the shm2 mutation. That is, there was an ADE3-dependent reduction in GCV gene expression induced by glycine on the addition of formate (Fig. 4B). We propose, therefore, that the presence of excess glycine acts in a similar manner to the shm2 mutation by causing a reduction in the level of one of the cytoplasmic one-carbon charged H 4 folates and that this decrease then acts as the signal to increase GCV gene expression.
The Principle Effect of Mutation of Cytosolic SHMT Is a Reduction in Cytosolic 5,10-CH 2 -H 4 folate-Two principle lines of evidence support the model that the ADE3 shm2 strain is limited in its cytosolic one-carbon pool. First, when grown in minimal medium, it has a slightly slower growth rate than the wild type that can be restored by the addition of adenine to the medium (Table IV). This is also consistent with the observation made by West et al. (9) that a strain that was unable to oxidize cytoplasmic 5,10-CH 2 -H 4 folate could not generate sufficient one-carbon units in the mitochondrion for purine synthesis to support wild-type growth. This strain recovered when adenine was added to the medium.
Secondly, we used 13 C NMR to follow the incorporation of one-carbon units from the mitochondrion into molecules derived from the cytoplasmic one-carbon H 4 folate pools. When cells are grown in the presence of exogenous glycine it is incorporated into choline and adenine both directly as an intact molecule and also as a one-carbon unit from the cytoplasmic pool. Appling and co-workers (6) have shown that monitoring the incorporation of [2-13 C]glycine into choline and adenine allows an insight into the flow of mitochondrial one-carbon metabolism to the cytoplasm. Fig. 5 shows a comparison of the relevant regions of the NMR spectra for extracts of three mutants used for the gene expression studies (ADE3 SHM2, ade3-130 SHM2, and ADE3 shm2). The GCV::lacZ gene fusions were transformed as single copies into a strain mutant for the cytosolic serine hydroxymethyltransferase (ADE3 shm2) and its isogenic wild type (ADE3 SHM2). Strains are described in Table II. Error bars represent the standard deviation of ␤-galactosidase assays performed in triplicate. Dmin, minimal medium (10).

FIG. 4. Exogenous formate alters GCV gene transcription by altering the one-carbon metabolic balance.
One-carbon metabolic mutant strains transformed with the full-length GCV2::lacZ gene fusion were grown in minimal medium with the addition of formate (concentration as indicated) (left) or minimal medium containing 10 mM glycine with the addition of 25 mM formate (right). ␤-Galactosidase assays were performed on exponentially growing cells as described under "Experimental Procedures." Values are relative to those without the addition of formate. Error bars represent the standard deviation of ␤-galactosidase assays performed in triplicate. Dmin, minimal medium (10). line (notably at the C-4 position) in this strain was generated by the export of labeled formate from the mitochondrion and its activation and interconversion by the ADE3 gene product ( Fig.  1 and Table I, reactions 4a-4c). In contrast, the mitochondrially derived one-carbon metabolites contributed much more extensively to adenine biosynthesis because the ade3-130 mutation virtually eliminates the incorporation into all labeled positions (C-2, C-5, and C-8) of adenine, whereas the shm2 mutation had a much lesser effect (Fig. 5B). These data indicate that the majority of one-carbon units from glycine catabolism in the mitochondrion are used for purine synthesis, which is consistent with the requirement for adenine for optimal growth of the ADE3 shm2 mutant outlined above. Because for purine synthesis the major drain on the one-carbon pool is at the level of 10-HCO-H 4 folate, very little enters the 5,10-CH 2 -H 4 folate pool, which explains the significant decrease in the amount of labeled choline in this strain. These data indicated that the principle metabolic effect on folate intermediates of the shm2 mutation was the reduction of cytoplasmic 5,10-CH 2 -H 4 folate. The data also show that 10-HCO-H 4 folate levels were not as greatly affected in the ADE3 shm2 mutant. 5,10-CH 2 -H 4 folate is therefore a prime candidate as mediator of the glycine response; however other molecules derived from it (e.g. 5-CH 3 -H 4 folate) could be candidates, and 10-HCO-H 4 folate remains a possible but unlikely one.
Adenine and 10-HCO-H 4 folate Levels Do Not Influence the Glycine Response-The shm2 strain has a requirement for ad-enine for optimal growth, and hence it is possible that reduction in the cellular purine pool is responsible for the transcriptional effect. To test this, we monitored the expression of the GCV2::lacZ fusion in an ade3-130 SHM2 strain that through lack of cytoplasmic C1-tetrahydrofolate synthase activity is unable to synthesize purine metabolites de novo (27). Fig. 5B shows the NMR data for the ade3-130 SHM2 strain illustrating that there was no synthesis of adenine from [2-13 C]glycine, as occurred in the wild-type strain. However, Fig. 6 illustrates that the effect of this mutation on the glycine response was to augment the extent of gene repression seen in the absence of glycine; the fully induced level of expression was unaffected. The interpretation of the data from this strain is complicated by the need to add exogenous adenine to the medium; however, expression of the GCV genes in the ade3-130 shm2 mutant was constitutively derepressed even with the addition of exogenous adenine at the same concentration (data not shown). This rules out purine nucleotides as the mediator(s) of the glycine response because the regulation would be expected to be similar between these mutants. A likely explanation for the observed difference is that in the ade3-130 SHM2 strain 5,10-CH 2 -H 4 folate would be elevated because of reduction in flow to purine synthesis, whereas in the ade3-130 shm2 strain it would be depleted. Hence 5,10-CH 2 -H 4 folate is a prime candidate as the mediator.
Interestingly, the purine auxotrophy of the ade3-130 SHM2 mutant is not thought to be a product of 10-HCO-H 4 folate deficiency (despite the lack of 5,10-CH ϩ -H 4 folate cyclohydrolase activity of the C1-tetrahydrofolate synthase enzyme) ( Fig.  1 and Table I, reaction 4b). West et al. (9) showed that an ade3-130 SHM2 strain does produce 10-HCO-H 4 folate by the action of an alternate (NAD-dependent) 5,10-CH 2 -H 4 folate dehydrogenase activity ( Fig. 1 and Table I, reaction 5) coupled with spontaneous oxidation of 5,10-CH ϩ -H 4 folate. The purine auxotrophy of the ade3 strain may therefore be due to a structural requirement for the C1-tetrahydrofolate synthase protein. This work also described a strain whose 10-HCO-H 4 folate (DAY4) levels were low because of its consumption for purine biosynthesis. There was no alteration of glycine control of GCV2 expression in this strain (see Fig. 8). Changes in 10-HCO-H 4 folate are therefore unlikely to be responsible for the changes in GCV gene expression. Whereas 5,10-CH 2 -H 4 folate remains the primary candidate, 5-CH 3 -H 4 folate derived from it has yet to be ruled out.
Decreased 5-CH 3 -H 4 folate Does Not Eliminate the Glycine Response-5,10-CH 2 -H 4 folate can be converted to 5-CH 3 -H 4 folate by the action of methylene-H 4 folate reductase (MTHFR) (Fig. 1 and Table I, reaction 7). In yeast this activity  Table III.  6. The glycine response of GCV2 is normal in a strain lacking cytoplasmic C1-tetrahydrofolate synthase activity. The full-length GCV2::lacZ gene fusion was transformed as a single copy into a strain mutant in the gene encoding the cytoplasmic C1-tetrahydrofolate synthase (ade3-130 SHM2). Error bars represent the standard deviation of ␤-galactosidase assays performed in triplicate. Dmin, minimal medium (10).
is encoded by the MET13 gene product that supplies onecarbon units for the synthesis of methionine (28). This pathway transfers the methyl moiety via S-adenosylmethionine to the C-4 position of choline (resonating at ϳ55.3 ppm in 13 C NMR spectra; Table III). Fig. 5A shows that there was a significant reduction in labeling of this carbon in the ADE3 shm2 strain, indicating lowered levels of 5-CH 3 -H 4 folate in the cell. The possibility that changes in the 5-CH 3 -H 4 folate pool mediate the glycine response was therefore analyzed in a met13 strain lacking MTHFR activity.
The met13 strain retained the ability to elicit the glycine response (Fig. 7). Furthermore, in contrast to the constitutively high expression in the ADE3 shm2 mutant, the level of GCV2 expression in minimal medium was markedly lower than in the wild-type strain (less than 5%). This was not an artifact of methionine addition to the growth medium because the ADE3 shm2 strain retained constitutively high levels of expression under the same culture conditions (data not shown). This result was not surprising because we have previously noted that excess methionine caused a slight increase in GCV2 expression (13). This experiment excludes the possibility that low 5-CH 3 -H 4 folate is the cause of increased GCV2 expression in the ADE3 shm2 strain. Moreover under all four conditions described above in which the cytoplasmic 5,10-CH 2 -H 4 folate pool would be in surplus, the level of GCV gene expression was down-regulated. This effect was most dramatic in the met13 mutant when the cytoplasmic pool of 5,10-CH 2 -H 4 folate would be greatest because of the block to synthesis of 5-CH 3 -H 4 folate and because one-carbon metabolism via the SHMT reaction is directed more toward the generation of methyl groups (methionine, choline, and dTMP) than of purines (see "The Principle Effect of Mutation of Cytosolic SHMT Is a Reduction in Cytosolic 5,10-CH 2 -H 4 folate" above and "How Is 5,10-CH 2 -H 4 -folate Used in the Cell to Maintain a One-carbon Metabolic Balance").
How Could the Addition of Glycine to Cells Affect 5,10-CH 2 -H 4 folate Levels?-The evidence above indicates that 5,10-CH 2 -H 4 folate is the molecule mediating the regulation of the GCV genes. Because metabolism of glycine via the GDC is not essential to elicit a glycine response, we sought to explain how the addition of glycine to cells could modulate the cytoplasmic concentration of 5,10-CH 2 -H 4 folate. In a detailed analysis of the regulation and catalytic mechanism of SHMT from rabbit liver and Escherichia coli, Schirch et al. (29) have shown that glycine in the range of 4 -37.5 mM inhibited the conversion of serine to glycine and 5,10-CH 2 -H 4 folate. Hence end-product inhibition of the SHMT reaction is one explanation. More significantly, however, was the discovery that 5-CH 3 -H 4 folate or 5-HCO-H 4 folate binds SHMT in conjunction with glycine to form a dead-end complex that dramatically inhibits the activity of the enzyme (30). The 20-fold reduction of GCV gene expression in the met13 strain (Fig. 7) when grown in minimal medium may thus reflect the consequences of additive metabolic disturbances leading to a build up of cytoplasmic 5,10-CH 2 -H 4 folate levels: disruption of the use of 5,10-CH 2 -H 4 folate for 5-CH 3 -H 4 folate synthesis and loss of 5-CH 3 -H 4 folate inhibition of SHMT activity.
The second important control molecule, 5-HCO-H 4 folate, is produced from 5,10-CH ϩ -H 4 folate by the SHMT from rabbit liver and E. coli in the presence of excess glycine (31). This reaction appears in E. coli to be the sole source of 5-HCO-H 4 folate, a metabolite that is also present in yeast but whose function until now was not known. Hence both end-product inhibition and the effect of glycine in conjunction with folate derivatives in dead-end complex formation would lead to a reduction in 5,10-CH 2 -H 4 folate synthesis on addition of glycine to the cell. This situation mimics that observed above in the ADE3 shm2 strain.
To What Extent Does the Above Dead-end Complex Formation Contribute to the in Vivo Transcription Response to Glycine?-GCV gene expression was assayed in strain WHY1, mutant for the putative 5,10-CH ϩ -H 4 folate synthetase gene (YER183c). This mutant cannot convert 5-HCO-H 4 folate to 5,10-CH ϩ -H 4 folate. 2 Blocking this reaction ( Fig. 1 and Table I control of GCV2 transcription, we synthesized the compound and used it in gel mobility shift assays. Fig. 9 shows that addition of up to 2 M 5,10-CH 2 -H 4 folate led to a decrease in the binding of the protein to the specific DNA; this decrease, however, was only to 60% of the initial level. Our previous work using H 4 folate showed an increase in DNA/protein complex formation but at a higher concentration than that measured here for 5,10-CH 2 -H 4 folate (50 M-1 mM). Other commercially available folates (folic acid, folinic acid, and 5-CH 3 -H 4 folate) had no effect on complex formation over the range up to 1 mM (data not shown). DISCUSSION It is understood how metabolites flow through the various one-carbon metabolic pathways in the cell (4 -9), but less is known on mechanisms controlling this flow. It has been shown here that there is an inverse relationship between cytoplasmic 5,10-CH 2 -H 4 folate levels and GCV gene expression. We propose that this is a control mechanism whereby the cell regulates the balance of serine, glycine, and one-carbon metabolism to meet conflicting demands. This is achieved by altering the balance of one-carbon metabolism between the cytoplasm and the mitochondrion to ensure a constant supply of one-carbon units to the important biosynthetic pathways such as those concerned with purine and pyrimidine biosynthesis. At the center of this control loop is regulation of the levels of cytoplasmic 5,10-CH 2 -H 4 folate.
How is 5,10-CH 2 -H 4 folate Used in the Cell to Maintain a One-carbon Metabolic Balance?- Fig. 1 and Table I show the many ways in which 5,10-CH 2 -H 4 folate is involved in metabolism in the cytoplasm. Our data and those of many others show that the SHMT-catalyzed reaction serves as the principle source of this molecule, which is the primary contributor to the one-carbon pool (1). Pathways that are involved in the consumption of 5,10-CH 2 -H 4 folate include methionine synthesis via the action of the MTHFR enzyme and 5,10-CH ϩ -H 4 folate synthesis via the methylene-H 4 folate dehydrogenase function of the C1-tetrahydrofolate synthase enzyme. When each of these functions was disturbed, alterations in the level of 5,10-CH 2 -H 4 folate signaled the need for the cell to synthesize onecarbon units from the alternate donor molecule glycine and thus shift the balance of one-carbon metabolism into the mitochondrion. From these data, we have been able to gain an insight into the regulation and flow of one-carbon metabolites in the cell under normal metabolic conditions.
Our results indicate that when cells are growing normally cytoplasmic 5,10-CH 2 -H 4 folate is used primarily for the production of methyl groups via methionine biosynthesis and secondly for the production of more oxidized C1-H 4 folate derivatives and metabolites such as purines. For purine synthesis there is therefore a balance between one-carbon metabolites derived from mitochondrial metabolism of glycine and the cytoplasmic conversion of serine to 5,10-CH 2 -H 4 folate. This model is supported by the fact that a ser1 allele (ade9, defective in its ability to produce serine from glycolytic intermediates) was originally identified as an adenine auxotroph when serine was limiting (32). At low serine levels, adenine limited growth of this strain, indicating that serine can be channelled preferentially into fulfilling the requirements of the cytoplasmic 5,10-CH 2 -H 4 folate pool for 5-CH 3 -H 4 folate and subsequently methionine biosynthesis at the expense of the more oxidized C1-H 4 folates required for purine biosynthesis. However, whereas flow from the cytoplasmic 5,10-CH 2 -H 4 folate pool for synthesis of purines may have lower priority, it does still play a role in normal cell growth.
Exogenous adenine was required for the shm2 strain to grow at an optimal rate, indicating that the mitochondrial one-carbon pathway cannot fully maintain the requirements for purine biosynthesis. However, Pasternack et al. (6) have shown that at least 25% of one-carbon units for purine synthesis are mitochondrially derived. They concluded that the normal pathway for glycine-derived one-carbon units is through the synthesis in the mitochondrion of formate, which is exported to the cytosol for purine synthesis. Thus the normal balance for flow of onecarbon units derived from serine is toward methyl group biogenesis, whereas glycine derived from serine in the mitochondrion supplements the supply of one-carbon units for purine biosynthesis.
Under What Circumstances Is This Flow Altered?-Under physiological conditions the cell needs to adjust the synthesis and degradation of serine and glycine to fulfil the requirements for protein synthesis as well as one-carbon metabolism and hence needs to regulate the balance of one-carbon flow. We explored two situations in which the use of glycine for onecarbon metabolism was enhanced (shown by an increase in GCV gene expression): when cells were grown with exogenous glycine and when the SHM2 gene was disrupted. These two situations resulted in a similar reduction of the levels of cytoplasmic 5,10-CH 2 -H 4 folate, which forms the basis for the signal to increase GCV gene transcription. Interestingly, in a recent study on the global transcriptional effects of the DNA-damaging reagent methyl methanesulfonate, the transcription of GCV2, GCV1, and GCV3 was increased 12.5-, 7.5-, and 2.6-fold, respectively (33). According to the regulatory mechanism proposed above, a decrease in 5,10-CH 2 -H 4 folate levels due to an increased demand on cytoplasmic one-carbon metabolism for nucleotide biosynthesis would signal a need to increase GCV gene transcription. Because the principle route for one-carbon unit synthesis is through the cytoplasmic SHMT, it was not surprising that loss of its function resulted in a decrease of 5,10-CH 2 -H 4 folate levels. It was less apparent, however, how addition of glycine also affected the levels. We have shown that the side reaction of the cytoplasmic SHMT described by Schirch and co-workers (29 -31), which results in its inhibition by ex-FIG. 9. 5,10-CH 2 -H 4 folate reduces protein binding to the glycine regulatory region of GCV2. 5,10-CH 2 -H 4 folate was synthesized as described under "Experimental Procedures" and added to gel mobility shift assays containing heparin-Sepharose-purified protein extracts from strain BWG1-7a and GCV2 promoter DNA harboring the GRR sequence (Ϫ309 to Ϫ267). Intensity values of bound DNA are expressed relative to 0 M (no 5,10-CH 2 -H 4 folate added). Error bars represent the standard deviation of quadruplicate measurements from different experiments. 5,10 mTHF, 5,10-CH 2 -H 4 folate. cess glycine, has an important physiological role. We therefore