Quantitative Phosphoproteomics Reveal mTORC1 Activates de Novo Pyrimidine Synthesis

See allHide authors and affiliations

Science  15 Mar 2013:
Vol. 339, Issue 6125, pp. 1320-1323
DOI: 10.1126/science.1228771

Coordinating Metabolism

Growth factors help to coordinate metabolism with growth in part by stimulating the activity of the protein kinase mTORC1 (mechanistic target of rapamycin complex 1). Ben-Sahra et al. (p. 1323, published online 21 February) and Robitaille et al. (p. 1320, published online 21 February) independently identified a key target of mTORC1—carbamolyl-phosphate synthase 2, or CAD, the rate-limiting enzyme for de novo synthesis of pyrimidines. Metabolomic profiling and phosphoproteomic analyses of normal cells and cells lacking signaling by mTORC1 converged on CAD as a key point at which growth-promoting signals also ramp up production of nucleic acids.


The Ser-Thr kinase mammalian target of rapamycin (mTOR) controls cell growth and metabolism by stimulating glycolysis and synthesis of proteins and lipids. To further understand the central role of mTOR in cell physiology, we used quantitative phosphoproteomics to identify substrates or downstream effectors of the two mTOR complexes. mTOR controlled the phosphorylation of 335 proteins, including CAD (carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase). CAD catalyzes the first three steps in de novo pyrimidine synthesis. mTORC1 indirectly phosphorylated CAD-S1859 through S6 kinase (S6K). CAD-S1859 phosphorylation promoted CAD oligomerization and thereby stimulated de novo synthesis of pyrimidines and progression through S phase of the cell cycle in mammalian cells. Thus, mTORC1 also stimulates the synthesis of nucleotides to control cell proliferation.

The conserved Ser-Thr kinase target of rapamycin (TOR) controls growth and metabolism by regulating several anabolic and catabolic processes (1, 2). Deregulation of the mammalian TOR (mTOR) signaling network is associated with aging and various disorders, including cancer, diabetes, obesity, cardiovascular disease, inflammation, and neurodegeneration (3). Allosteric mTOR inhibitors such as rapamycin are clinically approved for treatment of allograft rejection, cancer, and cardiovascular disease; and a new generation of adenosine triphosphate (ATP) competitive mTOR inhibitors are now in development as anticancer drugs (4). The identification of new mTOR targets may reveal further therapeutic strategies.

TOR forms two structurally and functionally distinct multiprotein complexes, TORC1 and TORC2, that are conserved from yeast to human (5). Rapamycin acutely inhibits TORC1. In mammals, nutrients (amino acids), growth factors (insulin), and cellular energy [high ATP:adenosine monophosphate ratio] activate mTORC1, whereas growth factors alone activate mTORC2, through phosphoinositide 3-kinase–dependent association of mTORC2 with the ribosome (6, 7). mTORC1 and mTORC2 directly phosphorylate several members of the AGC kinase family, including S6 kinase (S6K), protein kinase B (PKB) (also called Akt), serum and glucocorticoid regulated kinase, and canonical protein kinase C. Although mTOR direct substrates and downstream effectors are known, more are expected because of the broad role of mTOR in cell physiology.

To identify mTOR targets, we determined the phosphoproteome of mouse embryonic fibroblasts (MEFs) lacking raptor or rictor (fig. S1A). Raptor and rictor are specific and essential components of mTORC1 and mTORC2, respectively. Floxed raptor or rictor genes in MEFs, referred to as iRapKO and iRicKO, respectively, were deleted by tamoxifen (4-OHT)–induced activation of CreERT2 recombinase (fig. S1B) (8). In total, we detected and quantified 4584 phosphorylation sites on 1398 proteins (fig. S1, C to E, and table S1), including 18 validated phosphorylation sites in nine known mTORC1 or mTORC2 target proteins (table S2). We also detected mTOR-regulated phosphorylation of 326 proteins that had not been previously identified as mTOR targets (table S3). Two recent studies described the phosphoproteome of cells treated with an ATP-competitive mTOR inhibitor (9, 10). A comparison of the three mTOR phosphoproteomes revealed only ~30% overlap between any two data sets with regard to both validated and previously unknown mTOR-regulated proteins, although the three data sets are similar in size (fig. S1F and tables S4 and 5). Taken together, the three independent studies suggest that mTOR directly or indirectly regulates the phosphorylation of at least 803 distinct proteins. This corresponds to 3.9% of all mammalian genes, consistent with the broad role of mTOR in regulating cell physiology.

Phosphorylation motif analysis (fig. S2A) and immunoblotting (fig. S2B) suggested that the mTOR-regulated proteins we identified include both direct substrates and indirect effectors. Peptide array in vitro kinase assays indicated that mTOR directly phosphorylated 26 newly identified sites, corresponding to 21 potential mTOR substrate proteins (fig. S2, C to E, and tables S6 and S7), and confirmed that mTOR phosphorylates mainly but not exclusively S/T-P (where S is Ser; T, Thr; and P, Pro) and hydrophobic motif (Φ-x-x-Φ-S/T-Φ) sites. Thus, mTOR appears to be a promiscuous kinase that phosphorylates diverse target sites.

Gene Ontology (GO) analysis of mTOR-regulated proteins revealed enrichment of KEGG-annotated insulin signaling, mTOR signaling, cancer, and ribosome biogenesis pathways (fig. S3A). GO analysis also identified RNA metabolism and DNA replication as mTOR-regulated processes, yet few mTOR targets involved in these processes are known. We therefore investigated CAD (carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase) as a potential mTORC1 effector. CAD functions in pyrimidine synthesis, a conserved metabolic pathway essential for S phase progression (fig. S3B). In mammals, de novo pyrimidine synthesis is activated by growth factors (11, 12). However, despite the central importance of pyrimidine synthesis, its regulation by growth cues is incompletely understood. Our phosphoproteomic analysis indicated that mTORC1 mediates phosphorylation of CAD at S1859 (Fig. 1A). To confirm that mTORC1 controls CAD-S1859 phosphorylation, we generated a phosphospecific antibody to CAD-pS1859. The antibody failed to recognize a S1859→A1859 (S1859A, where A is Ala) mutant version of CAD (fig. S3C). Western immunoblotting showed that CAD-S1859 phosphorylation was stimulated by growth factors (dialyzed serum) and amino acids in a rapamycin-sensitive manner in HeLa and U2OS cells (Fig. 1B and fig. S3, D and E). CAD-S1859 phosphorylation was inhibited in MEFs depleted of raptor but not those depleted of rictor (Fig. 1C). We examined CAD phosphorylation in the livers of liver-specific tuberous sclerosis 1 (TSC1) knockout mice (L-TSC1 KO). Deletion of the tumor suppressor TSC1 hyperactivates mTORC1 (13, 14). CAD-S1859 phosphorylation was increased in TSC1-deficient liver in a rapamycin-sensitive manner (Fig. 1D). CAD-S1859 is part of a sequence, R-I-H-R-A-S1859 (where R is Arg; I, Ile, and H, His), that weakly resembles a consensus target site for AGC protein kinases, suggesting that mTORC1 may phosphorylate S1859 through S6K (15). Pharmacological inhibition of S6K with PF-4708671 (fig. S3F) or depletion of S6K1 and S6K2 prevented growth factor and amino acid–stimulated CAD-S1859 phosphorylation (Fig. 1E). Furthermore, recombinant S6K directly phosphorylated immunopurified CAD at S1859 (fig. S3G). Thus, mTORC1 phosphorylates CAD-S1859 through S6K.

Fig. 1

CAD as a target of mTORC1. (A) Diagram of CAD protein, including CPS (carbamoyl-phosphate synthetase), ATC (aspartate transcarbamylase), and DHO (dihydroorotase) enzymatic domains. (B) HeLa cells were deprived of serum for 16 hours in Dulbecco's minimum essential medium (DMEM) and incubated for 15 min without amino acids (AA) in 1× phosphate-buffered saline. Cells were then stimulated in DMEM with 10% dialyzed fetal calf serum (D-FCS) and 2× AA for 1 hour with or without rapamycin. (C) iRapKO and iRicKO MEFs were treated as in fig. S1B. The asterisk indicates a nonspecific band. (D) Mouse liver extract from wild-type (WT) (TSC1-fl/fl) or knockout (KO) (TSC1-fl/fl; Albumin-Cre) mice. Twelve-week-old littermates were starved overnight and treated with rapamycin or sham (0.9% NaCl) for 6 hours before killing. (E) WT (control) and DKO (S6K1-S6K2 double knockout) MEFs were deprived of serum for 2 hours and then stimulated for 1 hour in modified Hank's buffered salt solution plus 10% D-FCS, glucose, vitamins, and 2× AA, with or without rapamycin or PF-4708671.

CAD is a 250-kD protein containing three distinct enzymatic activities that catalyze the initial three steps in de novo pyrimidine synthesis, including the rate-limiting first step. In particular, CAD uses glutamine, bicarbonate, and aspartic acid to form a pyrimidine ring that is subsequently attached to ribose to generate a pyrimidine nucleotide. To measure de novo pyrimidine synthesis, we labeled HeLa cells with 15N-amide–labeled glutamine and measured 15N incorporation into uridine diphosphate (UDP) and uridine triphosphate (UTP) (16) (Fig. 2A and fig. S4, A and B). Growth conditions did not affect glutamine uptake into cells, and 15N incorporation was not detected in aspartic acid or the pentose phosphate pathway intermediate 6-phospho-gluconate (fig. S4C). Furthermore, we did not detect 15N incorporation into UDP or UTP in G9c cells, a Chinese hamster ovary cell line lacking endogenous CAD activity (fig. S4D). The above results indicate that the analysis was specific for CAD-dependent de novo pyrimidine synthesis. HeLa cells deprived of nutrients exhibited a low basal rate of de novo pyrimidine synthesis, whereas growth factors and amino acids stimulated synthesis of UDP and UTP. Rapamycin or PF-4708671 treatment inhibited growth factor–stimulated de novo pyrimidine synthesis, decreasing the absolute synthesis of UDP and UTP by about 60% (Fig. 2B and fig. S4, E and F). Rapamycin also inhibited growth factor–stimulated increase in the intracellular concentrations of dihydroorotate (DHOA), the final CAD-synthesized metabolite, and the downstream products orotate (OA) and UTP (Fig. 2C). In contrast, rapamycin had no effect on purine synthesis as measured by guanosine diphosphate (GDP) and guanosine triphosphate (GTP) concentrations (Fig. 2D). Thus, mTORC1 activates de novo pyrimidine synthesis in response to growth factors and amino acids.

Fig. 2

Activation of de novo pyrimidine synthesis by mTORC1. (A) Diagram of de novo pyrimidine synthesis pathway. CAP, carbamoyl phosphate; CAA, carbamoyl aspartic acid; OMP, orotidine monophosphate. (B) HeLa cells were metabolically labeled with 4 mM 15N-amide glutamine. Metabolites were measured with targeted ultra high-performance liquid chromatography–tandem mass spectrometry. Values are expressed as mean ± SD. Asterisks indicate a statistical difference between stimulated and rapamycin treatment: *P < 0.05, **P < 0.01, ***P < 0.001 (Student's t test) n = 3 to 6. (C) Inhibition of growth factor–stimulated increase in DHOA, OA, and UTP cellular concentrations by rapamcyin. (D) Rapamycin did not inhibit growth factor–stimulated increase in GDP or GTP cellular concentrations.

To test the role of CAD-S1859 phosphorylation, we examined de novo pyrimidine synthesis in CAD-deficient G9c cells reconstituted with CAD-S1859A or wild-type CAD. CAD-S1859A–expressing cells exhibited reduced incorporation of 15N into UDP and UTP (fig. S4, G and H) compared with cells expressing wild-type CAD. The small effect of CAD-S1859A likely reflects a high basal rate of de novo pyrimidine synthesis in the reconstituted G9c cells that overexpress wild-type and mutant CAD about four- to fivefold. Thus, we conclude that mTORC1 activates de novo pyrimidine synthesis through phosphorylation of CAD-S1859.

mTORC1 transcriptionally enhances the pentose phosphate pathway (17) which produces 5-phosphoribosyl-1-pyrophosphate (PRPP). PRPP in turn is required for a late step in de novo pyrimidine synthesis (Fig. 2A and fig. S4B). Thus, mTORC1 may regulate de novo pyrimidine synthesis by acute stimulation of CAD and by delayed transcriptional activation of the pentose phosphate pathway. One-hour treatment of HeLa cells with growth factors increased cellular concentrations of DHOA and OA, metabolites synthesized before the PRPP requirement in pyrimidine synthesis, whereas the cellular concentration of ribose 5-phosphate (R5P), a PRPP precursor, remained unchanged (fig. S5A). Furthermore, 1-hour rapamycin treatment inhibited CAD phosphorylation and de novo pyrimidine synthesis before changes in expression of pentose phosphate pathway genes were observed (fig. S5B and table S8). Lastly, TSC1-deficient liver displayed a rapamycin-sensitive increase in the amount of OA and increased transcriptional expression of glucose-6-phosphate dehydrogenase, a pentose phosphate pathway enzyme (fig. S5, C and D), indicating that mTORC1 activates de novo pyrimidine synthesis by both acute stimulation of CAD activity and delayed activation of genes encoding enzymes of the pentose phosphate pathway.

We examined the cellular distribution of endogenous CAD in HeLa cells. As visualized by immunofluorescence with antibody to CAD, growth factors and amino acids stimulated the formation of intracellular punctate structures in a rapamycin-sensitive manner (Fig. 3A and fig. S6, A and B). Growth factors and amino acids failed to stimulate formation of puncta in MEFs depleted for raptor (fig. S5C) or S6K1 and S6K2 (fig. S6, D and E). The puncta did not colocalize with the nucleus, endosomes, lysosomes, peroxisomes, or mitochondria (fig. S6F). Visualization of CAD in stimulated HeLa cells by immunogold electron microscopy revealed cytoplasmic clusters containing six to eight particles (fig. S6, G and H). CAD forms higher-order oligomers in vitro (18, 19). Thus, the puncta may represent CAD oligomers in vivo.

Fig. 3

Effects of mTORC1 on CAD localization and oligomerization. (A) HeLa cells were treated as in Fig. 1B. Endogenous CAD was visualized by immunocytochemistry. (B) G9c cells were transfected with plasmid-encoding, wild-type or mutant CAD. CAD oligomers were detected by sedimentation in 10 to 35% glycerol gradients. Fractions 2 (*) and 10 (**) correspond to 250-kD and 1.5-MD size standards, respectively. IB, immunoblot. (C) HeLa cells were treated as in Fig. 1B, G9c cells were treated as in (B), and iRapKO cells were treated as in fig. S1B. Oligomer is fraction 10 (**).

We investigated a possible role of mTORC1-mediated CAD-S1859 phosphorylation in CAD oligomerization. CAD oligomerization was assayed by sedimentation of a cell extract through a glycerol gradient and subsequent immunoblotting of gradient fractions (Fig. 3B and fig. S6I). Treatment of HeLa cells with growth factors and amino acids stimulated the formation of CAD oligomers in a rapamycin-sensitive manner. CAD oligomerization was reduced in MEFs depleted for raptor and in CAD-deficient G9c cells reconstituted with CAD-S1859A (Fig. 3C). Thus, mTORC1-mediated CAD-S1859 phosphorylation appears to promote CAD oligomerization.

CAD is essential for progression through S phase of the cell cycle because of the increased requirement for pyrimidines during DNA synthesis (20). We examined whether mTORC1-mediated CAD activation is also important for cell cycle progression. HeLa cells were synchronized in early S phase by using a double-thymidine block, and cell-cycle progression of released cells was monitored by flow cytometry. Similar to cells deprived of serum, cell treated with rapamycin showed a 32% delay in S phase progression upon release from the thymidine block (Fig. 4A, top). As cells entered G2 and M phase, the delay in cell cycle progression increased to 52%, consistent with an additional requirement for mTORC1 in G2-M (21). The rapamycin-induced delay in cell-cycle progression was suppressed by addition of exogenous uridine (Fig. 4A, bottom). Inhibition of S6K with PF-4708671 delayed cell-cycle progression (fig. S7A). Overexpression of wild-type CAD restored proliferation in G9c cells in the absence of exogenous uridine. In contrast, overexpression of CAD-S1859A only weakly restored cell proliferation (Fig. 4, B and C, and fig. S7, B and C), although CAD-S1859A was slightly less expressed than wild-type CAD in the corresponding reconstituted G9c cells (figs. S4G and S7C). Thus, mTORC1 appears to promote pyrimidine synthesis and S phase progression through S6K-mediated phosphorylation and stimulation of CAD (fig. S7D), consistent with reports that rapamycin inhibits DNA synthesis and cell proliferation independently of eukaryotic initiation factor 4E–binding proteins (22, 23).

Fig. 4

Promotion of S phase progression by mTORC1. (A) HeLa cells were synchronized in early S phase by using a double-thymidine block. Cells were then released in DMEM (starved) (red); DMEM, 10% D-FCS, and 2× AA (stimulated) (blue); or DMEM, 10% D-FCS, 2× AA, and 100 nM rapamycin (rapamycin) (green); in the absence (top) or presence (bottom) of 30 μM uridine. DNA content was analyzed by flow cytometry. (B and C) G9c cells were transfected with equal amounts of plasmid-encoding green fluorescent protein (GFP), WT, or mutant CAD. Cells were visualized with crystal violet 5 days after transfection (B), or proliferation was assayed via soft agar colony formation (C). Colonies were visualized after 8 days growth in the absence of uridine.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S7

Tables S1 to S8

References (2442)

  • Corresponding author. E-mail: m.hall{at}

References and Notes

  1. Acknowledgments: We acknowledge support from the Werner Siemens Foundation (A.M.R.), the Société Francophone du Diabète–Association de Langue Française pour L'Etude du Diabète et des Maladies Metaboliques (M.C.), the Swiss Cancer League, the Louis Jeantet Foundation, the Swiss National Science Foundation, and the project YeastX. We declare no conflicts of interest. The phosphoproteomic data, including UniProt accession numbers reported in this paper, are deposited in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article