Cytoplasmic Serine Hydroxymethyltransferase Mediates Competition between Folate-dependent Deoxyribonucleotide and S -Adenosylmethionine Biosyntheses*

Folate-dependent one-carbon metabolism is required for the synthesis of purines and thymidylate and for the remethylation of homocysteine to methionine. Methionine is subsequently adenylated to S -adenosylmethi-onine (SAM), a cofactor that methylates DNA, RNA, proteins, and many metabolites. Previous experimental and theoretical modeling studies have indicated that folate cofactors are limiting for cytoplasmic folate-dependent reactions and that the synthesis of DNA precursors competes with SAM synthesis. Each of these studies con-cluded that SAM synthesis has a higher metabolic priority than dTMP synthesis. The influence of cytoplasmic serine hydroxymethyltransferase (cSHMT) on this competition was examined in MCF-7 cells. Increases in cSHMT expression inhibit SAM concentrations by two proposed mechanisms: (1) cSHMT-catalyzed serine synthesis competes with the enzyme methylenetetrahydrofolate reductase for methylenetetrahydrofolate in a gly-cine-dependent manner, and (2) cSHMT, a high affinity 5-methyltetrahydrofolate-binding protein, sequesters this cofactor and inhibits methionine synthesis in a gly-cine-independent manner. Stable isotope tracer studies indicate that cSHMT plays an important role in is a metabolic switch synthesis trimethylchlorosilane (Pierce) and acetonitrile, and heated at 140 °C for 30 min. The trimethylsilane-base derivatives were separated on a HP-5MS column. Isotopic enrichment was determined in positive ionization mode by gas chromatography-mass spectrometry using a model 6890 gas chromatograph and model 5973 mass spectrometer (Hewlett-Pack- ard Corp., Palo Alto, CA). Selected ion monitoring was conducted at a mass-to-charge ratio m / z 255–257 for thymine, m / z 280–283 for ade- nine, and m / z 368–371 for guanine.

Folate is present in cells as a family of coenzymes that carry one-carbon units and function in both the mitochondrial and cytoplasmic compartments (1)(2)(3). Mitochondrial folate metabolism is necessary for the conversion of serine to glycine and formate (a one-carbon unit) (1,3), whereas cytoplasmic folate metabolism utilizes mitochondria-derived formate for the biosynthesis of purines (supplies the #2 and #8 carbons of the purine ring), thymidine (conversion of deoxyuridine monophos-phate (dUMP) to deoxythymidine monophosphate (dTMP) and for the generation of methionine from homocysteine (1, 3) (see Fig. 1 below). Methionine, in turn, can be converted to Sadenosylmethionine (SAM), 1 a cofactor for many methylation reactions, including the methylation of proteins, phospholipids, neurotransmitters, RNA, and DNA (4,5). Serine hydroxymethyltransferase (SHMT) catalyzes the reversible transfer of the hydroxymethyl group of serine to tetrahydrofolate (THF) to form methyleneTHF and glycine (6). The enzyme is present in both the mitochondria and cytoplasm (7,8). This reaction is a major source of THF-activated one-carbon units in mammalian cells (6). Loss of mitochondrial SHMT (mSHMT) function cannot be rescued by the activity of the cytoplasmic SHMT (cSHMT) isozyme in cultured cells; some evidence suggests that cSHMT may be a serine synthase in the cytoplasm (7,9).
Cellular folate derivatives are sequestered by a variety of proteins collectively called folate-binding proteins (2,10). The cellular concentration of folate-binding proteins exceeds that of folate derivatives, and therefore the concentration of free folate in the cell is negligible (6,11,12). This implies that folate-dependent biosynthetic pathways must compete for folate cofactors (12,13). This competition for folate cofactors is most pronounced for reactions that utilize methyleneTHF, a derivative that serves as a cofactor in three known enzymatic reactions in the cytoplasm (see Fig. 1). It is required for the conversion of dUMP to dTMP, catalyzed by thymidylate synthase (TS); for the conversion of glycine to serine, catalyzed by cSHMT; and for the synthesis of 5-methylTHF, catalyzed by methylenetetrahydrofolate reductase (MTHFR), a reaction that commits one-carbon units to the methionine cycle.
Because the MTHFR reaction is virtually irreversible in vivo (1,2), methionine synthase (MS) activity is essential for recycling 5-methylTHF to other folate cofactor forms. Otherwise, 5-methylTHF accumulates at the expense of all other folate derivatives, impairing folate-dependent deoxyribonucleotide synthesis. This phenomenon is known as the "methyl trap," a state of functional folate deficiency and impaired DNA synthesis (3, 13) (see Fig. 1). SAM inhibits MTHFR and thereby provides feedback regulation that protects against a folate methyl trap (see Fig. 1) and ensures that, during methionine repletion, folate-activated one-carbon units are spared for DNA precursor synthesis. During B 12 deficiency, apo-MS cannot convert 5-methylTHF to THF, resulting in the accumulation of cytoplasmic folate as 5-methylTHF, homocysteine accumulation and in decreased SAM synthesis (4,14). Depleted SAM leads to an increase in MTHFR activity and accelerated 5-methylTHF synthesis, thereby exacerbating the metabolic dysfunction.
Although inhibition of MTHFR by SAM certainly contributes to the prevention of methyl trapping under normal conditions, there is evidence that competition among folate-dependent enzymes also regulates the supply of one-carbon units to the methionine cycle. Fowler et al. (15) examined the formation of both methionine and serine in human fibroblasts. In fibroblasts where MS activity was deficient due to a variety of genetic mutations, serine formation was low compared with control cells, consistent with the formation of a folate methyl trap. Moreover, mutant fibroblasts with diminished MTHFR activity exhibited normal to high serine formation, indicating that MTHFR deficiency increased the availability of 5,10-methyleneTHF for cSHMT-catalyzed synthesis of serine from glycine (15).
The cSHMT enzyme is poised to regulate the metabolic competition between TS and MTHFR (see Fig. 1). The reaction catalyzed by cSHMT is reversible in vitro, and therefore its capacity to affect this competition in vivo is expected to be dependent on cellular glycine concentrations. Previously, we have demonstrated that the cSHMT enzyme is robustly regulated by heavy chain ferritin (HCF) and that HCF-induced increases in cSHMT expression enhance de novo thymidylate synthesis (16). In this study, we examined the effect of altered cSHMT expression and activity on the homocysteine remethylation pathway and the influence of glycine on cSHMT activity and folate metabolism.
Generation of Cell Lines-The MCFHCF cell line, which expresses rat heavy chain ferritin, and the MCFcSHMT21 cell line, which expresses the human cSHMT cDNA, were described previously (16). MCFGlyT cells were developed by subcloning the murine glycine transporter cDNA (GlyT1, obtained from Dr. Nathan Nelson, Roche Institute of Molecular Biology) (17) into the EcoRI/XhoI site of the vector pcDNA3. The GlyT1 construct was transfected into MCF-7 cells (2 ϫ 10 6 cells) by electroporation (0.22 kV, 950 microfarads) and cultured until stable G418 colonies formed and isolated.
Glycine Transport Assay-GlyT1 activity is dependent on the presence of Na ϩ and Cl Ϫ (17). MCFGlyT cells (1 ϫ 10 7 ) in mid-to late-log phase were incubated with 100 mM NaCl, 100 mM KCl, 10 mM HEPES (pH 7.5), 0.1 Ci of [2-3 H]glycine (PerkinElmer Life Sciences), and 1 mM unlabeled glycine. Control cells were incubated with a similar buffer lacking NaCl but containing 200 mM KCl to inhibit transport by GlyT. Transport was allowed to proceed for 30 min at room temperature; then the cells were pelleted by centrifugation and washed three times with 100 volumes of cold phosphate-buffered saline. The cell pellets were lysed with 0.5% Triton X-100 in 10 mM potassium P i (pH 7.5), and the accumulation of [2-3 H]glycine was determined using a liquid scintillation counter. Intracellular amino acid pools were determined as previously described (18).
Determination of Folate Cofactor One-carbon Distribution-Cells were cultured to 60 -80% confluence in 100-mm plates and exposed to experimental culture medium 24 h prior to labeling with [ 3 H]folinic acid (Moravek). The culture medium (␣MEM-modified (HyClone), 2.5 g/liter NaHCO 3 , 11.2% (v/v) dialyzed fetal bovine serum (HyClone), 0.1 mM methionine) lacked ribo-and deoxyribonucleosides, hypoxanthine, thymidine, folic acid, and serine unless otherwise noted. For all experiments, 20 nM [ 3 H]folinic acid was added to the medium, and glycine was added at concentrations ranging from 0 to 50 mM. After 24 h, media were removed from cells, and cells were washed twice with 2 ml of ice-cold phosphate-buffered saline. Cells were lifted from the plate by incubation with 0.75-1 ml of trypsin-EDTA at room temperature for 5-10 min. The cell suspension was pelleted, and the cells were washed with 0.75 ml of ice-cold phosphate-buffered saline. After removing the supernatant, the cell pellet was frozen at Ϫ80°C. The relative distribution of the folate one-carbon forms was quantified by high performance liquid chromatography as described previously (18).
Determination of SAM and SAH Concentrations-SAM and SAH concentrations were determined in cells cultured as described above, except that the methionine concentration was 10 M and folinic acid was 50 nM. Cell extracts for SAM and SAH analyses were prepared by a modification of the previously described procedure (19). Cells were suspended in 500 l of 0.1 M NaAcO buffer (pH 6), and cellular proteins were precipitated by adding 312 l of 10% perchloric acid to each sample. After vortexing, samples were centrifuged at 2000 ϫ g for 10 min at 4°C. The supernatant was neutralized with 1 M sodium phosphate (pH 11.5) and diluted 2-fold with deionized water. The pellet was saved and stored at Ϫ80°C for subsequent measurement of total protein. The supernatant was applied to C18 Sep-Pak Plus cartridges (Waters Corp.) primed with 5 mM 1-heptanesulfonic acid (Acros) in methanol. The column was washed with 5 ml of deionized water before SAM and SAH were eluted in 2 ml of methanol. After addition of 50 l of 3 M NaAcO (final pH of 5.5) to the eluate, the sample was dried under vacuum. SAM and SAH were converted to their fluorescent derivatives by adding 50 l of chloroacetaldehyde and incubating at 60°C for 1 h. Purified SAM and SAH from Sigma-Aldrich were used to generate standard curves from which sample SAM and SAH content were determined. Each standard was prepared as described for the tissue extracts.
Samples were loaded onto a C8 column (5 m, 250 ϫ 4.6 mm, from Phenomenex) fitted with a C18 guard column (5 m, 30 ϫ 4.6 mm) operated by a Shimadzu SCL-10A system controller module connected to an LC-10AS solvent delivery unit and an RF-10A spectrofluorometric detector. Fluorescence of eluted compounds was monitored using ex ϭ 270 nm and em ϭ 410 nm. A two-buffer elution system was used: both buffers were 10 mM 1-heptanesulfonic acid (Acros) in 25 mM sodium phosphate (pH 3) with Buffer A and Buffer B containing 18 and 30% methanol by volume, respectively. Elution of SAM and SAH was achieved at a flow rate of 1 ml/min with the following parameters: 0 -10 min, 100% A; 10 -20 min, linear gradient to 100% B; 20 -30 min, 100% B; 30 -35 min, linear gradient to 100% A; 35-40 min, 100% A. SAM and SAH values were normalized to cellular protein content that was determined using the Lowry-Bensadoun method (20).
Deoxyuridine Suppression Assays-The efficiency of de novo thymidine biosynthesis was determined using a modified deoxyuridine suppression assay that was described previously (16).
Protein pellets were suspended in 6 N HCl (100 l) in vacuum hydrolysis tubes and heated at 100°C for 20 h. The amino acids were purified by cation exchange chromatography (21-23). Amino acids were converted to heptafluorobutyryl n-propyl ester derivatives (21) and were separated on an HP-5MS column (30 m ϫ 0.25 mm). Isotopic enrichment was determined in electron capture negative ionization mode by gas chromatography-mass spectrometry using a model 6890 gas chromatograph and model 5973 mass spectrometer (Hewlett-Packard Corp., Palo Alto, CA). Selected ion monitoring was conducted at a mass-to-charge ratio m/z 519 -523 for serine, m/z 305-308 for dehydroalanine (DHA), m/z 349 -353 for leucine, m/z 367-370 for methionine, and m/z 293-295 for glycine.
DNA samples were dried under nitrogen and suspended in formic acid (1 ml) and hydrolyzed at 150°C for 45 min in vacuum hydrolysis tubes. After drying at 55°C under nitrogen, the bases were dissolved in 0.2 ml of a 1:1 mixture of N,O-bis-[trimethylsilyl]trifluoroacetamide /1% trimethylchlorosilane (Pierce) and acetonitrile, and heated at 140°C for 30 min. The trimethylsilane-base derivatives were separated on a HP-5MS column. Isotopic enrichment was determined in positive ionization mode by gas chromatography-mass spectrometry using a model 6890 gas chromatograph and model 5973 mass spectrometer (Hewlett-Packard Corp., Palo Alto, CA). Selected ion monitoring was conducted at a mass-to-charge ratio m/z 255-257 for thymine, m/z 280 -283 for adenine, and m/z 368 -371 for guanine.

Isolation and Characterization of MCFGlyT Cells-Previous
studies indicate that cellular glycine levels influence serine synthesis and that folate-dependent serine synthesis comes at the expense of methionine synthesis. Activation of cSHMT in neuroblastoma by removal of its endogenous inhibitor 5-formylTHF depletes 5-methylTHF pools, impairs homocysteine remethylation, elevates cellular serine levels, and increases the methionine requirement for maximal cell growth (18). In other studies, Penafiel et al. (24) observed that adult mice injected with high levels of glycine display a 14-fold increase in hepatic glycine concentrations and a 6-fold rise in hepatic serine, suggesting that high levels of glycine were able to drive the direction of the SHMT reaction toward serine synthesis. Additionally, an isotope labeling study by Petzke et al. (25) showed that the rate of glycine-toserine conversion is increased in the hepatocytes of rats that are fed glycine-rich diets. Indeed, the clinical manifestations of nonketotic hyperglycinemia may be partly explained by the influence of glycine on cSHMT activity and utilization of methyleneTHF for serine synthesis. Measurements of metabolites in cerebrospinal fluid from nonketotic hyperglycinemia patients revealed elevated glycine and homocysteine, but low normal methionine concentrations (26). However, no one has previously systematically examined the effects of cellular glycine levels on folate-dependent one-carbon metabolism.
To study the effects of cellular glycine on cSHMT activity and homocysteine remethylation, the cDNA encoding a glycine transporter, GlyT, was transfected into human MCF-7 cells. The transporter is Na ϩ -and Cl Ϫ -dependent and functions to clear glycine from synaptic clefts. Because it is a reversible transporter, GlyT equalizes intra-and extracellular glycine concentrations, thereby allowing precise control of intracellular glycine through culture conditions. MCF-7 cells expressing the glycine transporter accumulate ϳ10-fold more [ 3 H]glycine when cultured in the presence of NaCl, whereas nontransfected MCF-7 cells do not display any increased glycine accumulation in the presence of NaCl (Table I). Additionally, the transporter can deplete intracellular glycine in cells cultured in the absence of glycine: the concentration of free cellular glycine in MCF-7 cells that express the transporter is reduced by 85% compared with untransfected cells when cultured in the absence of glycine (Table II).

Effects of Glycine on the Cellular Distribution of Folate Derivatives-
The effect of intracellular glycine on cellular folate one-carbon distribution was determined in MCF-7 and MCF-GlyT cells (Fig. 2). 5-methylTHF levels respond to changes in medium glycine concentrations in all cells ( Fig. 2A) and are highest at 0 mM glycine and decrease with increasing concentrations of medium glycine. For MCF-7 cells, 5-methylTHF comprises about 40% of total cellular folate when cultured without glycine and decreases to about 25% of total folate at 1 mM glycine in the culture medium. The 5-methylTHF levels are not affected by increases in medium glycine from 1 to 10 mM in these cells. In contrast, 5-methylTHF accounts for up to 55% of total cellular folate in MCFGlyT cells when cultured without glycine, and 5-methylTHF levels decrease nearly linearly as a function of increasing medium glycine from 0 to 10 mM. All decreases in 5-methylTHF levels are accompanied by increases in 10-formylTHF relative concentrations of similar magnitude (Fig. 2B). The relative concentration of unsubstituted THF did not change with exogenous glycine and was not different between the transfected and nontransfected cells (data not shown). These observations indicate that glycine alters the distribution of cellular one-carbon substituted folate derivatives but does not limit the availability of one-carbon units for cytoplasmic metabolism.
The glycine-induced decrease in cellular 5-methylTHF levels indicates that glycine increases cSHMT-catalyzed serine synthesis and thereby increases the enzyme's effectiveness in competing with MTHFR for one-carbon units in the form of 5,10-methyleneTHF ( Fig. 1). Relative 5-methylTHF levels reach a minimum threshold in wild type cells between 1 and 2 mM exogenous glycine, whereas 5-methylTHF levels decrease nearly linearly in cells expressing GlyT over the range of glycine concentrations tested. This indicates the MCF-7 cells are capable of sequestering, accumulating, and regulating cellular glycine and that the expression of GlyT disrupts the ability of these cells to concentrate glycine. In both MCF-7 and GlyTexpressing cell lines, the highest relative 5-methylTHF levels were observed at 0 mM glycine, further indicating that cSHMT activity is influenced by mass action and that glycine is critical in preventing the accumulation of folate as 5-methylTHF. At the pharmacological concentration of 50 mM exogenous glycine, 5-methylTHF levels were reduced to 2.5-3.0% of total folate in all cell lines (data not shown). This relationship suggests that the regulation of cellular glycine can be disrupted when cells are exposed to superphysiological concentrations of glycine. If relative 5-methylTHF levels are reflective of cellular glycine concentrations and cells expressing GlyT do precisely equalize cellular and medium glycine concentrations, then Fig. 2A indicates that MCF-7 cells maintain intracellular glycine levels between 8 and 10 mM when cultured between 1.0 and 10 mM glycine.
Effects of Glycine on Cellular SAM and SAH-The determination of SAM and SAH concentrations provides direct meas-   11.2% (v/v) dialyzed fetal bovine serum (HyClone), 0.1 mM methionine lacking ribo-and deoxyribonucleosides, hypoxanthine, thymidine, glycine, and serine. ␣MEM contains 145 mM Na ϩ and 106 mM Cl Ϫ . Free amino acids were isolated and quantified as described elsewhere (18) and normalized to cellular valine concentrations. All values are expressed as the mean and standard deviation of three independent experiments.
urement of the methylation capacity of the cell. The effects of increasing exogenous glycine on the cellular concentrations of SAM and SAH are presented for MCF-7 and GlyT-expressing cells (Fig. 3, A and B). SAM concentrations in MCF-7 cells decreased by 50% between 0 and 10 mM medium glycine; similarly, there is more than a 3-fold decline in SAM in GlyTexpressing cells from 2 to 10 mM glycine. The lack of SAM at 0 mM glycine in GlyT-expressing cells may be explained by a shortage of adenosine, because adenosine synthesis requires glycine as a precursor. Although there appears to be a slight increase in SAH concentrations as medium glycine is increased (Fig. 3B), the magnitude of the change is dramatically less than that observed for SAM. This difference may be explained by the activity of SAH hydrolase, which catabolizes SAH to adenosine and homocysteine, and by efficient efflux of homocysteine from these cells. Kinetic data have demonstrated that SAH hydrolase activity favors the formation of SAH, but in the cellular environment, where homocysteine and adenosine are removed, the equilibrium favors hydrolysis (27). Fig. 3 (A and B) indicates that the glycine-induced decrease in 5-methylTHF levels affects the methionine cycle primarily through depletion of available methyl groups. SAM levels decrease in MCF-7 cells over the range of 2-10 mM medium glycine even though 5-methylTHF levels are constant over this range. This continued reduction in SAM may reflect the activity of glycine N-methyltransferase (GNMT), a SAM-dependent enzyme that converts glycine to sarcosine. GNMT activity is stimulated by increasing glycine concentrations and is inhibited by 5-methylTHF (2, 10). However, GNMT expression is tissue-specific, and its activity has been detected in liver and pancreas only (2). A more likely explanation involves feedback regulation of MTHFR activity by SAM, an allosteric inhibitor of MTHFR (Fig. 1) (4). As SAM concentrations continue to decrease as a function of increasing glycine levels, MTHFR activity increases such that 5-methylTHF synthesis is stimulated. In this manner, SAM's effect on MTHFR activity may buffer cytosolic 5-methylTHF levels even in the presence of increasing glycine concentrations. This buffering of 5-methylTHF concentrations may also serve to ensure cSHMT is inhibited during SAM depletion. In summary, these data demonstrate that glycine influences the distribution of folate derivatives and the availability of methyl groups for SAM-dependent methylation reactions in MCF-7-and GlyT-expressing cells. These observations support the proposition that serine and methionine syntheses are competitive pathways.
Effects of cSHMT Expression on the Distribution of Folate Derivatives and Cellular Methylation Capacity-The cSHMT protein is known to sequester 5-methylTHF as a tight-binding inhibitor (28). Therefore, it is predicted that increased expression of cSHMT would affect the relative level of 5-methylTHF in cultured mammalian cells. We have previously reported an MCFHCF2 cell line and reported that expression of rat heavy chain ferritin (HCF) results in a 5-to 10-fold increase in cSHMT protein concentrations (16). We have also developed MCFcSHMT21 cells, which express the human cSHMT cDNA and contain 2-to 4-fold increased cSHMT protein levels (16). 5-methylTHF levels were determined in all cell lines as a function of medium glycine concentrations (Fig. 4). Compared with MCF-7 cells, cellular 5-methylTHF levels are elevated nearly 2-fold for all cell lines with elevated cSHMT expression when cultured in medium lacking glycine (Fig. 4). Because nearly 50% of total cellular folate is localized in mitochondria, which do not accumulate 5-methylTHF (3,29), it is likely that nearly all cytoplasmic folate accumulates as 5-methylTHF in MCF-7 cells expressing the cSHMT or HCF cDNA. Increasing medium glycine above 1 mM decreases 5-methylTHF levels in HCF-expressing cells to levels found in MCF-7 cells cultured in the same medium, whereas cells expressing the cSHMT cDNA retain elevated 5-methylTHF levels when medium glycine concentrations are maintained between 0 and 2.0 mM.
The effect of increased cSHMT expression on 5-methylTHF levels is predicted by the metabolic scheme presented in Fig. 1. The cSHMT protein binds 5-methylTHF tightly, so cells with elevated cSHMT protein are expected to have a greater capacity to sequester and concentrate 5-methylTHF. In contrast to the effects of increased cSHMT levels on 5-methylTHF levels, variations in cSHMT concentration display an inverse relationship with SAM concentrations (Table III). At all medium glycine concentrations, cellular SAM concentrations in MCFHCF2 and MCFcSHMT21 cells are dramatically decreased compared with MCF-7 cells. Decreases in SAM concentrations may result from the sequestration of 5-methylTHF and subsequent inhibition of homocysteine remethylation and may be equivalent to 5-methyl-THF trapping that results from vitamin B12 deficiency.
Effect of Glycine on Thymidylate Synthesis-Previous studies from our laboratory have demonstrated that increased cSHMT expression stimulates de novo synthesis of thymidine precursor, presumably by supplying methyleneTHF for this pathway (16) and that cSHMT activity is rate-limiting for de novo thymidine biosynthesis. The effect of medium glycine on thymidylate biosynthesis was determined for MCF-7 cells and MCF-7 cells that express GlyT (Fig. 5). Over a physiological range of medium glycine levels, we did not observe a decreasing trend in de novo thymidylate synthesis for either cell line. In fact, thymidylate synthase was stimulated when medium glycine concentrations were increased from 0.2 to 0.5 mM. This indicates that the stimulation of thymidylate biosynthesis by cSHMT (16) is not inhibited by glycine.

Effects of Glycine and cSHMT on the Flux of Formate between the Thymidylate and Methionine Biosynthetic Pathways-The
incorporation of exogenously supplied [ 13 C]formate into methionine and serine (present in cellular protein) and purines (present in nuclear DNA) was determined in MCF-7 cells as a function of cSHMT and glycine concentration (Table IVA). The one-carbon precursor pools used for synthesis of these metabolites were derived from the exogenous labeled formate or from endogenously synthesized unlabeled serine metabolites. Increasing medium glycine concentrations from 0 to 2 mM enhances exogenous [ 13 C]formate incorporation into methionine and serine in MCF-7 cells (Table IVA). The increase in [ 13 C]formate incorporation into methionine indicates endogenous formate synthesis is depressed when glycine is elevated, presumably by shifting the equilibrium of the mSHMT reaction toward serine synthesis (Fig. 1). [ 13 C]formate incorporation into serine is also increased because of increased serine synthesis by the mSHMT and/or cSHMT enzymes (Fig. 1). Table IVB demonstrates that glycine increases the ability of exogenous formate to enrich the 10-formylTHF precursor pool used for purine synthesis, confirming that glycine inhibits the endogenous synthesis of formate in MCF-7 cells. The effect of glycine on formate production is most pronounced in the MCFGlyT cells. When these cells are cultured without glycine, only 20% of one-carbon units that are incorporated into purines are derived from [ 13 C]formate. At 2 mM medium glycine, 40% are derived from [ 13 C]formate; at 10 mM medium glycine, 50% are derived from [ 13 C]formate. Therefore, elevations of cellular glycine in MCF-7 cells have a general effect of depleting the supply of formate that is available for DNA precursor and SAM synthesis.
Elevated cSHMT expression, as seen in MCFcSHMT21 and MCFHCF2 cells, stimulates [ 13 C]formate incorporation into methionine but has no effect on the incorporation of [ 13 C]formate into purines (Table IV). Previously, we demonstrated that increased cSHMT expression stimulates thymidylate biosynthesis (16). Because TS and MTHFR compete for methyl-eneTHF, the cSHMT-mediated stimulation of [ 13 C]formate incorporation into methionine probably results from decreased demand of [ 13 C]formate-derived methyleneTHF for thymidine precursor synthesis, because TS seems to prefer methyl-eneTHF supplied from the cSHMT enzyme (Fig. 1). The data in Table IV demonstrate that the stimulation of [ 13 C]formate incorporation into methionine by elevations in cSHMT is glycinedependent. For MCF-7 cells with increased cSHMT expression, [ 13 C]formate incorporation into methionine is increased by about 70% compared with untransfected cells when cultured without glycine but by only 30% or less when cells are cultured with 2 mM glycine. Elevated medium glycine diminishes the ability of cSHMT-derived one-carbon units to compete with exogenously supplied [ 13 C]formate (Fig. 1).
The data in Table 4C indicate that most of the serine that is used for protein synthesis (and presumably most of the serine in the cell) is synthesized from folate-dependent one-carbon metabolism in these cells. Assuming that the 10-formylTHF one-carbon pool is in equilibrium with the methyleneTHF pool, the mass isotopomer distribution (determined from the ratio of the Mϩ1 and Mϩ2 isomers of dA and dG) can be used to calculate the enrichment of formate into serine and methionine (Table IVC). Note that this calculation includes one-carbon units that are derived from unlabeled serine that become incorporated into the one-carbon pool. These data indicate that 69% of serine used for protein synthesis in MCF-7 cells was derived from the one-carbon pool in cells cultured without glycine, and this value increases to 97% in the presence of 2 mM medium glycine. The methionine enrichment reflects competition between methionine derived from the methyleneTHF pool and exogenous methionine present in the culture medium. Consequently, the extent of folate-dependent methionine synthesis can be calculated from the met/mass isotopomer distribution ratio. This calculation indicates that 19 -21% of methionine in MCF-7 cells is remethylated independently of medium glycine concentrations, whereas cells with increased expression of cSHMT displayed decreased homocysteine remethylation when medium glycine concentrations were increased (Table IVC).
Cells were also labeled with L-[2,3,3-2 H 3 ]serine to establish the contribution of cSHMT-derived one-carbon units into the thymidine and methionine pools (Fig. 1). The data in Table V quantify the direct incorporation of the hydroxymethyl group of L-[2,3,3-2 H 3 ]serine into methionine or thymidine by cSHMT as a percentage of all one-carbon units derived from exogenous L-[2,3,3-2 H 3 ]serine. MethyleneTHF that is supplied by cSHMT and incorporated directly into methionine or thymidine is expected to retain the two deuterium atoms (CD 2 ) that are present on the hydroxymethyl group of serine. Alternatively, if the L-[2,3,3-2 H 3 ]serine enters the mitochondria and the hydroxymethyl group is released from the mitochondria as formate, it should contain a single deuterium atom (CD 1 ). Previous studies have shown that up to 90% of one-carbon units used for cytosolic one-carbon metabolism in mammalian cells are derived from formate generated by mitochondrial serine metabolism. 2 During derivatization of serine for GC analysis, most of the serine is converted to dehydroalanine (DHA), with loss of the proton on the C2 position. Consequently, the m/z distribution of DHA isotopes gives the isotopic distributions at the C3 position of serine. Table V shows that about 70% of the isotopically labeled cellular serine pool that provided serine for protein synthesis retained both deuterium atoms on the C3 carbon (as assessed by DHA labeling), and this percentage did not differ significantly under the different culture conditions. The ϳ30% of serine that contained one deuterium in this position is due to serine resynthesized from glycine using a CD 1 -THF pool. If the methionine and dTMP one-carbon units were derived directly from cSHMT only, the mass ϩ2 species of these metabolites should also be about 70% of the labeled species. The data in Table V show that the proportions are much lower, indicating 2 S. Myong, S. Cho, and B. Shane, unpublished data. that most of the one-carbon units used for cytosolic methionine and dTMP synthesis were derived from mitochondrial serine metabolism.
Cells with increased cSHMT expression show increased incorporation of CD2 into both methionine and dTMP, indicating that cSHMT can deliver single carbons to both pathways (Table  V). However, the data also show that cSHMT preferentially directs methyleneTHF to dTMP synthesis, as evidenced by the higher percentage of CD2 in dTMP compared with methionine in MCF-7 cells. This indicates that there are two methyl-eneTHF pools in the cytoplasm: one generated from serine through cSHMT (CD 2 ), the other generated from mitochondriaderived formate (CD 1 ) (Fig. 1). MethyleneTHF derived from cSHMT is preferentially directed toward dTMP synthesis, but can also be incorporated into methionine, indicating that the two methyleneTHF pools are in equilibrium. Increasing the medium glycine concentration from 0 to 2 mM enhanced the incorporation of CD 2 into both methionine and thymidine, consistent with glycine inhibiting formate production in mitochondria (Table IV). In MCF-7 cells, CD2 incorporation into dTMP was further enhanced when the concentration of glycine in the culture medium was increased to 10 mM glycine, whereas incorporation of CD 2 into methionine was depressed by greater than 40% when medium glycine concentration was increased from 2 to 10 mM. These data indicate that cSHMT contributes one-carbon units to both methionine and thymidine biosynthesis but that the incorporation of CD 2 into methionine is inhibited by elevated glycine concentrations, whereas glycine does not impair the incorporation of CD 2 into thymidine, consistent with the deoxyuridine suppression data (Fig. 4). The stimulation of CD 2 incorporation, catalyzed by cSHMT, into both methionine and thymidine biosynthesis at 0 and 2 mM exogenous glycine may also result from glycine-induced reductions of cellular 5-methyl-THF, which is an inhibitor of cSHMT catalytic activity (Fig. 2).
Increased cSHMT expression also enhanced the incorporation of cSHMT-derived one-carbon units into methionine and thymidylate at 0 and 2 mM exogenous glycine, and increasing the medium glycine concentration from 0 to 2 mM stimulated the incorporation of CD 2 into thymidine and methionine in all cells lines. It should be noted that the CD2 proportions shown     26.5 (160) in Table V underestimate the direct contribution of cSHMTderived one-carbon units to methionine and dTMP synthesis by about 50%, as the CD2 proportion of the labeled serine precursor was about 70%. DISCUSSION Several studies have investigated the competition among folate-dependent enzymes. Based on observations of monkeys rendered vitamin B 12 -deficient through long term nitrous oxide administration, Scott et al. (13) proposed that limited methyl group availability, caused by either folate or methionine deficiency, shifts the flux of one-carbon units among folate-dependent pathways such that folate cofactors are preferentially shuttled to the methionine cycle to protect methylation reactions and thereby suppress DNA synthesis. Similarly, Green et al. (27) made use of the known affinities of several relevant enzymes that utilize 5,10-methyleneTHF and predicted that folate coenzymes are preferentially directed toward SAMdependent methylation reactions at low cellular folate concentrations. Additionally, these authors projected that MTHFR enzyme would be insensitive to changes in methyl-eneTHF availability, whereas TS activity would be highly dependent on them. This model assumed that these two enzymes directly compete for a common cellular pool of methyleneTHF. Neither of these studies directly measured the utilization of one-carbon units through these competing pathways.
More recent studies have included some measurements of the flow of folate one-carbon units through these pathways, and this work has confirmed that changes in a single biosynthetic pathway influence the shuttling of folate cofactors through other pathways. Fell and Selhub (30) demonstrated that Raji cells cultured in high methionine (7 mM) or homocystine (0.7 mM) display decreased de novo thymidylate synthesis and increased utilization of 5,10-methyleneTHF for serine synthesis. Furthermore, evidence from studies of cultured fibroblasts from patients suffering from either MS or MTHFR genetic deficiencies also supports competition for 5,10-methyleneTHF between MTHFR and cSHMT (15). MTHFR-deficient cells have reduced concentrations of 5-methylTHF (31, 32) and increased serine production from radioactively labeled formate, whereas serine production was decreased in cultures from MS-deficient patients.
The results presented in this study strongly indicate that the cSHMT enzyme mediates the competition between TS and MTHFR for one-carbon units. We have previously reported that increased cSHMT expression results in increased rates of de novo thymidine synthesis, indicating that cSHMT shuttles methyleneTHF to TS through production of 5,10-methyl-eneTHF from serine (16). We anticipated that increased cellular glycine concentrations would impair de novo thymidine synthesis by reversing the directionality of the cSHMT enzyme, thereby depleting 5,10-methyleneTHF pools. However, the results from deoxyuridine suppression assays (Fig. 5) and metabolic tracer studies (Tables IV and V) revealed that neither thymidine biosynthesis, nor the supply of cSHMT-derived onecarbon units for thymidine biosynthesis, is inhibited by glycine. In contrast, increases in cSHMT expression and/or elevated cellular glycine concentrations decrease the availability of folate cofactors for homocysteine remethylation. Increased cSHMT expression results in a 5-methylTHF trap and depressed SAM concentrations. This event is glycine-independent, considering that glycine binds synergistically with 5-methylTHF polyglutamates with a K d of 40 M (28). Therefore, our results, together with those of Oppenheim et al. (16), suggest that long term regulation of one-carbon flux may be achieved through changes in cSHMT protein expression. A short term, glycine-dependent regulatory mechanism whereby cSHMT inhibits SAM synthesis was also identified. Changes in cellular glycine concentrations allow control of one-carbon flux through the cSHMT-catalyzed conversion of glycine to serine and by inhibiting formate production in mitochondria. Depletion of cellular glycine concentrations elevated relative 5-meth-ylTHF levels, and we propose that low cellular glycine concentrations impair the ability of cSHMT to compete for 5,10-methyleneTHF cofactors, thereby facilitating 5-methylTHF synthesis. Increases in cellular glycine result in decreased steady-state levels of 5-methylTHF (Fig. 2) and decreased steady-state levels of SAM ( Fig. 3 and Table III).
The present study clearly supports the hypothesis that cSHMT regulates both thymidylate and methionine biosynthesis and allows us to develop a model that addresses the two mechanisms whereby cSHMT activity mediates this competition (Fig. 1). We propose that there exist two distinct pools of cSHMT protein in MCF-7 cells. One cSHMT pool is compartmentalized with TS and DHFR protein and aids in thymidylate synthesis by supplying or perhaps channeling methyleneTHF to TS. Because this pool of cSHMT would not be in equilibrium with the bulk phase, it would be insensitive to changes in equilibrium levels of 5,10-methyleneTHF. Data also show that this pool of cSHMT is insensitive to cellular glycine concentrations. Our model further predicts a second pool of cSHMT enzyme that competes with MTHFR for 5,10-methyleneTHF in a glycine-dependent manner and competes with MS for 5-meth-ylTHF in a glycine-independent manner. Our model does not account for the data in Tables IV and V that indicate that cSHMT associated with dTMP synthesis is insensitive to glycine concentrations, whereas the cSHMT pool that competes with MTHFR is sensitive to cellular glycine concentrations. Additional studies are required to account for these data.
Previous studies have indicated that folate metabolism is regulated such that SAM synthesis has metabolic priority over thymidylate biosynthesis (13,27). We propose that, under certain conditions, cSHMT acts as a switch to increase DNA synthesis at the expense of homocysteine remethylation. This regulation is accomplished in two ways. Deoxyribonucleotide biosynthesis is enhanced by providing 5,10-methyleneTHF to TS for thymidylate synthesis as observed by Oppenheim et al. (16) and by increasing the cytoplasmic availability of THF for conversion to 10-formylTHF and use in purine synthesis (Fig.  2B). Simultaneously, cSHMT inhibits homocysteine remethylation by two mechanisms: (1) by decreasing the availability of 5,10-methyleneTHF to MTHFR (at elevated glycine concentrations) and (2) by sequestering 5-methylTHF and depleting cellular levels of SAM. The cSHMT-induced depletion of SAM concentrations may result directly from 5-methylTHF sequestration, but this mechanism was not conclusively demonstrated in this study.
The two most common biomarkers for impaired folate metabolism are elevated tissue and serum homocysteine (4,33) and increased uracil content in DNA (34). These biochemical markers are risk factors for certain cancers, vascular disease, and neural tube defects (4,35). Therefore, the ability of cSHMT to affect these pathways indicates a potential role for this enzyme in the etiology of homocysteine-and uracil-related diseases. Additionally, cSHMT expression and activity is regulated by a number of factors, including retinoic acid (36), ferritin, and developmental stage (16). Further studies are required to determine the effects of altered cSHMT activity on these folate-sensitive biomarkers in animal models and its role in folate-related pathologies.