Glycine decarboxylase (GLDC) is a key enzyme of glycine cleavage system that converts glycine into one-carbon units. GLDC is commonly up-regulated and plays important roles in many human cancers. Whether and how GLDC is regulated by post-translational modifications is unknown. Here we report that mechanistic target of rapamycin complex 1 (mTORC1) signal inhibits GLDC acetylation at lysine (K) 514 by inducing transcription of the deacetylase sirtuin 3 (SIRT3). Upon inhibition of mTORC1, the acetyltransferase acetyl-CoA acetyltransferase 1 (ACAT1) catalyzes GLDC K514 acetylation. This acetylation of GLDC impairs its enzymatic activity. In addition, this acetylation of GLDC primes for its K33-linked polyubiquitination at K544 by the ubiquitin ligase NF-X1, leading to its degradation by the proteasomal pathway. Finally, we find that GLDC K514 acetylation inhibits glycine catabolism, pyrimidines synthesis and glioma tumorigenesis. Our finding reveals critical roles of post-translational modifications of GLDC in regulation of its enzymatic activity, glycine metabolism and tumorigenesis, and provides potential targets for therapeutics of cancers such as glioma.
Mechanistic target of rapamycin (mTOR) is a conserved serine–threonine kinase in the phosphoinositide-3 kinase-related kinase family, which integrates a wide array of extracellular and intracellular signals to regulate cell growth, metabolism, translation, and autophagy1,2,3. mTOR is the catalytic subunit of two distinct protein complexes, known as mTOR Complex 1 (mTORC1) and mTOR Complex 2 (mTORC2). The rapamycin-FKBP12 complex directly inhibits mTORC1, whereas mTORC2 is insensitive to acute rapamycin treatment1. mTORC1 is defined by its three core components: mTOR, regulatory protein associated with mTOR (RPTOR), and mammalian lethal with Sec13 protein 8 (mLST8)4. Dysregulation of mTORC1 activity induces highly active cell metabolism and proliferation states, which is commonly observed in many human cancers, including glioblastoma (GBM)5,6.
Glycine is a nonessential amino acid, which is an important residue of many proteins. Glycine can also be cleaved to yield one-carbon units, which are used for nucleotide synthesis via the tetrahydrofolate (THF) cycle5,7. In addition, high levels of glycine are toxic by its conversion to metabolites, such as aminoacetone and methylglyoxal8. The glycine cleavage system controls glycine catabolism through multi-step reactions, generating CO2, NH3, NADH, and 5,10-methylene-THF9,10. The glycine cleavage system is a multi-enzyme complex, consisting of glycine decarboxylase (GLDC, also called P protein), amino-methyltransferase (T protein), dihydrolipoamide dehydrogenase (L protein), and the hydrogen carrier protein (H protein)11. Various studies have demonstrated that glycine metabolism is essential for tumorigenesis7,8,12.
GLDC is a mitochondrial pyridoxal 5’-phosphate (PLP)-dependent enzyme that catalyzes the first and rate-limiting step in glycine catabolism11. GLDC binds glycine through its PLP cofactor to form an external aldimine that loses the carboxyl group as CO2 and donates the remaining aminomethylene moiety to the oxidized lipoamide arm of H protein9. Mutations in GLDC gene cause glycine accumulation, leading to neural tube defect and glycine encephalopathy (also known as nonketotic hyperglycinemia)13,14. It has also been demonstrated that GLDC is hyperactive in different types of cancer cells and plays a fundamental role in tumor growth. For example, increased expression of GLDC in non-small cell lung cancer-initiating cells is essential for tumorigenesis by promoting pyrimidine biosynthesis, glycolysis, and sarcosine production15. GLDC expression is markedly increased in MYCN-amplified neuroblastomas, which is required for neuroblastoma cell proliferation and tumorigenicity16.
In this study, we found that GLDC is acetylated at K514 by ACAT1 following mTORC1 inhibition. GLDC K514 acetylation inhibited its enzymatic activity, promoted its K33-linked polyubiquitination at K544 by NF-X1 and proteasomal degradation, and suppressed glioma tumor growth. Our findings suggest that GLDC activity is regulated by sequential posttranslational modifications, including acetylation and polyubiquitination, and reveal critical regulatory mechanisms of glycine metabolism and tumorigenesis.
Inhibition of mTORC1 suppresses GLDC activity by promoting its acetylation at K514
It has been shown that mTORC1 regulates certain amino acid metabolism and tumorigenesis. However, whether it regulates GLDC-mediated glycine metabolism is unknown. We generated U251 glioma cells stably expressing Flag-tagged GLDC and treated the cells with the mTORC1 inhibitor Rapamycin or left them untreated. We then purified Flag-GLDC by anti-Flag immunoaffinity beads and measured its enzymatic activity. The results indicated that Rapamycin treatment suppressed GLDC enzymatic activity (Fig. 1a). We next explored whether GLDC activity is regulated by posttranslational modifications. We found that Rapamycin treatment inhibited phosphorylation of S6K and 4EBP1 (hallmarks of mTORC1 activation) but did not affect serine/threonine or tyrosine phosphorylation of GLDC (Supplementary Fig. 1a). Interestingly, immunoblotting analysis indicated that Rapamycin treatment increased GLDC acetylation (Fig. 1b). RPTOR is a core component of mTORC1. In RPTOR-deficient cells, the basal acetylation of GLDC was increased and Rapamycin treatment did not further increase its acetylation (Fig. 1c). These results suggest that mTORC1 signal inhibits GLDC acetylation.
Analysis of proteomic databases indicates that 15 lysine residues of GLDC are potentially acetylated (https://www.phosphosite.org/). To test whether these residues are primary acetylation sites regulated by mTORC1, we generated Arg (R) (to mimic deacetylated lysine) substitution mutants of each lysine residue. We found that Rapamycin treatment induced increase of acetylation of wild-type GLDC and its 14 mutants but not the GLDCK514R mutant, which had basal acetylation similar as wild-type GLDC without Rapamycin treatment (Fig. 1d). These results suggest that, while GLDC is basally acetylated at other lysines, Rapamycin induces acetylation of GLDC at K514. Human GLDC K514 is conserved across species (Supplementary Fig. 1b). Previous studies have demonstrated that K514 is localized in the catalytic pocket of GLDC9,17. We generated an antibody specifically recognizing K514-acetylated GLDC (anti-Ac-K514-GLDC; Supplementary Fig. 1c). Immunoblotting analysis with this antibody confirmed that GLDC K514 was acetylated and this was increased following Rapamycin treatment (Fig. 1e, f). These results suggest that K514 is the major acetylation site negatively regulated by mTORC1. To investigate the effects of K514 acetylation of GLDC, we generated GLDCK514Q mutant, which mimics its K514 acetylation. We transfected Flag-tagged wild-type GLDC, GLDCK514R, and GLDCK514Q into HEK293 cells and purified these proteins by anti-Flag immunoaffinity beads. We found that the enzymatic activity of wild-type GLDC was inhibited in Rapamycin-treated cells. GLDCK514R had slightly reduced activity in comparison to wild-type GLDC, but its activity was not inhibited following rapamycin treatment. However, the enzymatic activity of GLDCK514Q was impaired in untreated and Rapamycin-treated cells (Fig. 1g). Taken together, these results suggest that inhibition of mTORC1 signal suppresses GLDC enzymatic activity by promoting its K514 acetylation.
ACAT1 mediates GLDC K514 acetylation
We next investigate the enzymes that catalyze GLDC K514 acetylation. Previously, two mitochondrial acetyltransferases, including acetyl-CoA acetyltransferase 1 (ACAT1) and general control nonrepressed 5-like 1 (GCN5L1), have been identified18,19. Co-immunoprecipitation experiments indicated that ACAT1 but not GCN5L1 was associated with GLDC (Fig. 2a). Overexpression of ACAT1 but not GCN5L1 markedly increased acetylation of wild-type GLDC but not GLDCK514R (Fig. 2b). In vitro acetylation assays further showed that ACAT1 catalyzed acetylation of wild-type GLDC, but not GLDCK514R (Fig. 2c). Endogenous GLDC was constitutively associated with ACAT1, and knockout of ACAT1 by the CRISPR-Cas9 system impaired basal and Rapamycin-induced GLDC K514 acetylation in U251 or U87 cells (Fig. 2d). Consistently, overexpression of ACAT1 inhibited enzymatic activity of wild-type GLDC but not GLDCK514R, whereas knockout of ACAT1 increased the enzymatic activity of wild-type GLDC but not GLDCK514Q (Fig. 2e). These results suggest that the acetyltransferase ACAT1 catalyzes GLDC K514 acetylation and inhibits its enzymatic activity.
SIRT3 deacetylases GLDC at K514
Lysine deacetylation is mediated by two protein families, including the nicotinamide adenine dinucleotide (NAD+)-dependent deacetylases (sirtuins (SIRTs)) and the zinc-dependent histone deacetylases (HDACs)20. We found that treatment of U251 or U87 cells with nicotinamide (NAM, an inhibitor of SIRTs) but not Trichostatin A (TSA, an inhibitor of HDACs) markedly increased GLDC K514 acetylation (Fig. 3a). Overexpression of SIRT3 but not the other 6 SIRTs reduced GLDC pan-acetylation (Fig. 3b). The enzymatic inactive mutant of SIRT3, SIRT3H248Y, also lost its ability to inhibit GLDC K514 acetylation (Fig. 3c). In vitro assays confirmed that SIRT3 but not SIRT3H248Y mediated GLDC K514 deacetylation (Fig. 3d). Knockout of SIRT3 increased GLDC K514 acetylation, which was reversed by complement with wild-type SIRT3 but not SIRT3H248Y (Fig. 3e). Rapamycin failed to further increase GLDC K514 acetylation in SIRT3-deficient cells (Fig. 3f). Consistently, overexpression of SIRT3 but not SIRT3H248Y increased wild-type GLDC but not GLDCK514Q enzymatic activity, whereas knockout of SIRT3 inhibited it (Fig. 3g). These results suggest that SIRT3 mediates deacetylation of GLDC at K514 and increases its enzymatic activity.
mTORC1 signal upregulates SIRT3
In our early experiments, we found that Rapamycin had no marked effects on the protein levels of ACAT1 as well as its association with GLDC (Fig. 2d). Rapamycin treatment also did not affect the ability of ACAT1 to acetylate GLDC (Fig. 4a). These results suggest that mTORC1 signal does not affect ACAT1 activity. In addition, Rapamycin also did not induce interaction between SIRT3 and GLDC (Fig. 4b). In contrast, we found that the protein levels of SIRT3 and GLDC in U251 or U87 cells were markedly downregulated following Rapamycin treatment (Fig. 4c). Treatment of U251 cells with the protein synthesis inhibitor cycloheximide (CHX) did not affect Rapamycin-induced downregulation of SIRT3 (Fig. 4d), suggesting that regulation of SIRT3 level by Rapamycin is independent on posttranslational mechanisms. Quantitative PCR (qPCR) analysis indicated that Rapamycin inhibited SIRT3 mRNA levels in U125 or U87 cells (Fig. 4e). Previous studies suggest that peroxisome proliferator-activated receptor gamma coactivator 1-α (PGC1α) is a crucial transcription factor for Sirt3 gene21,22. It has also been shown that mTORC1 activity induces transcription of PGC1α, which is mediated by the transcription factor YY123. Consistently, knockout of PGC1α in U251 or U87 cells inhibited SIRT3 mRNA level, which was not further inhibited by Rapamycin treatment (Fig. 4f). In addition, knockout of PGC1α increased the acetylation of GLDC at K514 and decreased its abundance (Fig. 4g). Together, these results suggest that mTORC1 signal transcriptionally induces SIRT3 levels through PGC1α.
GLDC K514 acetylation promotes its K33-linked polyubiquitination and proteasomal degradation
In our above experiments, we found that both SIRT3 and GLDC were downregulated following Rapamycin treatment of U251 or U87 cells (Fig. 4c). However, Rapamycin did not affect the mRNA levels of GLDC gene in U251 or U87 cells (Supplementary Fig. 2a). In light of these and other results, we hypothesized that GLDC K514 acetylation following SIRT3 downregulation promotes its degradation. Consistently, we found that SIRT3 deficiency caused downregulation of GLDC protein levels, which was reversed by reconstitution with wild-type SIRT3 but not SIRT3H248Y in U251 or U87 cells (Fig. 5a). SIRT3 deficiency did not affect GLDC mRNA levels in U251 or U87 cells (Supplementary Fig. 2b). GLDCK514Q, a mutant that mimics K514 acetylation, had a shorter half-life than wild-type GLDC (Fig. 5b). These results suggest that GLDC K514 acetylation following SIRT3 downregulation promotes its degradation.
To further investigate the mechanisms responsible for acetylation-primed degradation of GLDC, we treated U251 cells with inhibitors for various protein degradation pathways. The proteasome inhibitor MG132, but not the lysosome inhibitor ammonium chloride (NH4Cl) or autophagosome inhibitor 3-methyladenine (3-MA), markedly inhibited GLDC degradation after termination of protein synthesis by CHX (Fig. 5c). Mutation of K514 to Q caused downregulation of GLDC, which was inhibited by MG132 but not NH4Cl or 3-MA (Fig. 5d). MG132 also inhibited Rapamycin-induced degradation of GLDC (Fig. 5e). These results suggest that GLDC K514 acetylation primes it for proteasomal degradation.
We further investigated the mechanisms leading to GLDC proteasomal degradation. We found that Rapamycin treatment as well as SIRT3 deficiency induced polyubiquitination of GLDC (Fig. 5f). Utilizing ubiquitin mutants in which one or six lysine residues are replaced with Arginine (R), we found that SIRT3 deficiency increased K33-linked polyubiquitination but not other lysine residue-linked polyubiquitination of GLDC (Fig. 5g and Supplementary Fig. 2c). Consistently, Rapamycin treatment induced K33-linked polyubiquitination of wild-type GLDC but not GLDCK514R, whereas GLDCK514Q was constitutively modified by K33-linked polyubiquitination (Fig. 5h). These results suggest that K514 acetylation of GLDC promotes its K33-linked polyubiquitination and proteasomal degradation. Consistently, ectopic expression of SIRT3 abolished K33-linked polyubiquitination of GLDC induced by Rapamycin (Fig. 5i).
NF-X1 mediates K33-linked polyubiquitination of GLDC
We next attempted to identify the E3 ubiquitin ligases that catalyze K33-linked polyubiquitination of GLDC. We expressed HA-tagged GLDC in HEK293 cells, and then GLDC-bound proteins were immunoprecipitated with anti-HA and analyzed by mass spectrometry (MS). Among the 279 proteins identified, 9 are E3 ubiquitin ligases (Supplementary Data 1). Overexpression experiments indicated that only NF-X1 promoted K33-linked polyubiquitination of GLDC. NF-X1 contains a RING-type zinc finger domain and has been shown to be a potential E3 ubiquitin ligase24. Endogenous co-immunoprecipitation experiments indicated that GLDC was basally associated with NF-X1 and Rapamycin treatment promoted their association (Fig. 6a). In addition, GLDCK514Q interacted with NF-X1 better than wild-type GLDC (Fig. 6b). These results suggest that Rapamycin-induced GLDC K514 acetylation promoted its association with NF-X1. Overexpression of NF-X1 but not its inactive mutants (C358 or C361 mutated to serine) promoted K33-linked polyubiquitination of GLDC (Fig. 6c). NF-X1 deficiency impaired K33-linked polyubiquitination of GLDC induced by Rapamycin treatment (Fig. 6d). These results suggest that NF-X1 catalyzes K33-linked polyubiquitination of GLDC following its K514 acetylation induced by mTORC1 inhibition.
We then attempted to identify the residues of GLDC that are conjugated with K33-linked polyubiquitin chains by NF-X1. We first examined which parts of GLDC were polyubiquitinated by NF-X1. The results indicated that NF-X1 catalyzed polyubiquitination of GLDC(489–740) but not GLDC(1–488) or GLDC(741–1020) (Supplementary Fig. 3a). We then individually mutated each of the 13 lysine residues within aa489–740 of GLDC to arginine and examined whether the mutants could be modified by K33-linked polyubiquitination. The results indicated that mutation of K544 but not the other 12 lysine residues of GLDC to arginine dramatically reduced its K33-linked polyubiquitination (Supplementary Fig. 3b). Both overexpression of NF-X1 and SIRT3 deficiency increased K33-linked polyubiquitination of wild-type GLDC and GLDCK539R but not GLDCK544R (Fig. 6e, f). In addition, overexpression of NF-X1 caused downregulation of wild-type GLDC and all the mutants except for GLDCK544R (Supplementary Fig. 3c). Sequence analysis showed that K544 of GLDC was conserved in various vertebrate species (Supplementary Fig. 3d). These results suggest that NF-X1 targets GLDC K544 for K33-linked polyubiquitination. Consistently, Rapamycin induced downregulation of wild-type GLDC but not GLDCK544R (Fig. 6g), whereas NF-X1 deficiency prevented Rapamycin-induced GLDC degradation in U251 (Fig. 6h) or U87 (Supplementary Fig. 3e) cells. Cellular fractionation experiments indicated that NF-X1 was located in the cytoplasm but not mitochondria, and Rapamycin treatment induced increase of GLDC in the cytoplasm and its decrease in the mitochondrion fraction of U251 and U87 cells (Fig. 7a). In addition, the results also showed that Rapamycin induced K33-linked polyubiquitination of GLDC in the cytoplasm but not in the mitochondria (Fig. 7a). Confocal microscopy further confirmed that Rapamycin treatment induced increase of GLDC in the cytoplasm (Fig. 7b). These results suggest that acetylated GLDC may be translocated from mitochondria to the cytoplasm for NF-X1-mediated polyubiquitination and degradation.
GLDC K514 acetylation inhibits glycine catabolism and pyrimidine synthesis
To investigate the functional significance of GLDC K514 acetylation, we reconstituted GLDC-knockout U251 or U87 cells with Flag-tagged wild-type GLDC or GLDCK514Q (Fig. 8a and Supplementary Fig. 4a). We found that GLDC deficiency increased glycine levels, and GLDCK514Q-reconstituted cells had increased glycine levels in comparison to wild-type GLDC-reconstituted U251 (Fig. 8b) or U87 (Supplementary Fig. 4b) cells. Consistently, the levels of glycine-related metabolites, such as sarcosine (N-methylglycine), betaine aldehyde, and pyrimidines, including thymidine, deoxyuridine, thymine, cytosine, and uracil, were decreased in GLDC-deficient or GLDCK514Q-reconstituted compared to wild-type U251(Fig. 8b, c) or U87 (Supplementary Fig. 4b, c) cells. SIRT3 deficiency promoted glycine accumulation and suppressed production of its metabolites, whereas ACAT1 deficiency and NF-X1 deficiency had the opposite effects in these cells (Fig. 8d and Supplementary Fig. 4d). The GLDC-catalyzed reaction converts glycine into CH2-THF, which is a substrate for thymidylate synthesis10,15,25. We further investigated the role of GLDC K514 acetylation in nucleotide synthesis utilizing 13C-2-glycine isotope tracers. U87 or U251 cells were cultured in 50% extracellular 13C-2-glycine (0.4 mM glycine + 0.4 mM 13C-2-glycine)-containing medium for 24 h. Tracer analysis showed that approximately a quarter of intracellular glycine was 13C-labeled and the consumed glycine was incorporated into nucleotides (Supplementary Fig. 5a, b). The levels of labeled thymidine and thymine were impaired in GLDC-deficient compared to wild-type cells (Supplementary Fig. 5b). Reconstitution of GLDC-deficient cells with wild-type GLDC but not GLDCK514Q rescued nucleotide synthesis (Supplementary Fig. 5b). These results suggest that GLDC K514 acetylation inhibits glycine catabolism and pyrimidine synthesis.
GLDC K514 acetylation suppresses tumorigenesis
We next investigated the biological functions of GLDC K514 acetylation. Cell proliferation assays showed that GLDC deficiency dramatically inhibited U251 or U87 cell proliferation, which was reversed by reconstitution with wild-type GLDC but not GLDCK514Q (Fig. 9a). To test whether inhibition of cell proliferation by GLDC deficiency are mediated through depletion of the one-carbon pool, we replenished the one-carbon pool by adding formate. Formate contains a single carbon atom and reacts directly with THF, making formyl-THF a one-carbon donor for purine synthesis25,26. The results showed that formate treatment only partially rescued the proliferation of GLDC deficiency or GLDCK514Q cells (Supplementary Fig. 6a). These results suggest that reduced purine biosynthesis partially contributes to the inhibition of cell proliferation caused by GLDC deficiency. In addition, tumor growth in nude mice intracranially injected with wild-type GLDC-reconstituted U251 or U87 glioma cells was markedly faster than those injected with GLDC-deficient or GLDCK514Q-reconstituted cells (Supplementary Fig. 6b). These results suggest that GLDC K514 acetylation suppresses cell proliferation and tumor development.
To further investigate the clinical significance of GLDC K514 acetylation, we collected 18 non-tumor (N) and 18 GBM (T) tissues from patients and performed quantitative immunoblotting analysis. The results indicated that GBM tissues had increased levels of GLDC, SIRT3, and pS6K (a hallmark of mTORC1 activity) and decreased levels of ACAT1, NF-X1, and GLDC K514 acetylation in comparison to non-tumor tissues (Fig. 9b and Supplementary Fig. 7). Furthermore, GLDC K514 acetylation levels were negatively correlated with GLDC and SIRT3 levels and positively correlated with ACAT1 levels (Fig. 9c). GLDC levels were negatively correlated with ACAT1 and NF-X1 levels but positively correlated with SIRT3 levels (Fig. 9d). Finally, mTORC1 activities were positively correlated with GLDC and SIRT3 levels but negatively correlated with levels of GLDC K514 acetylation (Fig. 9e). Taken together, these data suggest that GLDC K514 acetylation plays a crucial role in regulation of tumorigenesis of GBM.
Posttranslational modifications such as phosphorylation, sumoylation, acetylation, and polyubiquitination are known to be critical for regulation of protein functions27,28,29,30. Although the abnormal expression of GLDC has been shown to promote tumorigenesis, it is unknown whether and how GLDC is regulated by posttranslational modifications and how its posttranslational modifications contribute to cancer progression. In this study, we demonstrated that acetylation of GLDC K514 impairs its enzymatic activity. In addition, this acetylation of GLDC also primes for its K33-linked polyubiquitination and proteasomal degradation, leading to suppression of glycine metabolism and tumorigenesis. Our findings underscore the essential roles of GLDC posttranslational modifications in regulation of glycine metabolism and tumorigenesis.
mTORC1 is a central regulator of cell growth conserved from yeast to mammals, which integrates intracellular and environmental signals to regulate cell growth, autophagy, translation, and metabolism1. In our experiments, we found that constitutive mTORC1 activity in cells induced transcription of the deacetylase SIRT3. Overexpression of SIRT3 but not its inactive mutant SIRT3H248Y deacetylated GLDC. Knockout of SIRT3 increased GLDC acetylation, which was reversed by complement with wild-type SIRT3 but not SIRT3H248Y. Inhibition of mTORC1 by Rapamycin increased GLDC acetylation but failed to further increase GLDC acetylation in SIRT3-deficient cells. These results suggest that mTORC1 activity in cells leads to deacetylation of GLDC by SIRT3.
Our experiments suggest that inhibition of mTORC1 activity causes GLDC acetylation at K514 by the mitochondrial acetyltransferase ACAT1. Both overexpression and in vitro acetylation experiments indicated that ACAT1 increased acetylation of wild-type GLDC but not GLDCK514R. Knockout of ACAT1 impaired basal and Rapamycin-induced GLDC K514 acetylation.
Several lines of evidence suggest that GLDC acetylation at K514 primes it for K33-linked polyubiquitination at K544 and proteasomal degradation. First, inhibition of mTORC1 activity by Rapamycin or SIRT3 deficiency, which increased GLDC acetylation at K514, caused GLDC degradation that was inhibited by the proteasomal inhibitor MG132. Second, the GLDCK514Q mutant, which mimics K514 acetylation, had a shorter half-life than wild-type GLDC. Third, Rapamycin treatment as well as SIRT3 deficiency induced K33-linked polyubiquitination of GLDC. Rapamycin treatment induced K33-linked polyubiquitination of wild-type GLDC but not of GLDCK514R, whereas GLDCK514Q was constitutively modified by K33-linked polyubiquitination.
Our experiments further identified NF-X1 as the E3 ubiquitin ligase responsible for K33-linked polyubiquitination of GLDC at K544. GLDC was constitutively associated with NF-X1, which was increased following Rapamycin treatment. Overexpression of NF-X1 but not its inactive mutants promoted K33-linked polyubiquitination of GLDC, whereas NF-X1 deficiency impaired Rapamycin-induced K33-linked polyubiquitination of GLDC. Our experiments also identified K544 of GLDC as the target residue for K33-linked polyubiquitination by NF-X1. Mutation of K544 of GLDC to arginine impaired its K33-linked polyubiquitination as well as NF-X1-induced degradation. Overexpression of NF-X1 and SIRT3 deficiency failed to catalyze K33-linked polyubiquitination of GLDCK544R. In addition, NF-X1 deficiency or mutation of K544 of GLDC to arginine prevented Rapamycin-induced GLDC degradation.
In addition to acetylation-primed polyubiquitination and proteasomal degradation, our experiments indicated that acetylation of GLDC at K514 itself impairs its enzymatic activity. In vitro experiments with purified proteins indicated that the enzymatic activity of GLDCK514Q, which mimics its K514 acetylation, was impaired. In addition, in vitro experiments with GLDC purified from Rapamycin-treated or SIRT3-deficient cells had reduced enzymatic activities, whereas GLDC purified from ACAT-deficient cells had increased enzymatic activities.
Based on our results, we propose a model on the regulatory mechanisms of GLDC activity. Constitutive mTORC1 activity in cancerous cells induces expression of the deacetylase SIRT3, which maintains GLDC in un-acetylated and enzymatically active state, leading to increased glycine metabolism. Inhibition of mTORC1 activity, such as in untransformed cells, causes acetylation of GLDC at K514, which is localized in its catalytic pocket by the mitochondrial acetyltransferase ACAT1. This acetylation of GLDC inactivates its enzymatic activity by physical hindrance of its substrates or causing its conformational changes. The acetylation of GLDC at K514 also promotes its mitochondrion to cytoplasm translocation and subsequent K33-linked polyubiquitination at K544 by the E3 ubiquitin ligase NF-X1 and proteasomal degradation in the cytoplasm (Fig. 9f). Consistent with this model, our experiments indicated that overexpression of SIRT3 increased GLDC enzymatic activity, whereas overexpression of ACAT1 inhibited it. In comparison to wild-type GLDC, GLDCK514Q showed decreased enzymatic activity and mediated increased glycine accumulation and decreased production of its metabolites. Deficiency of SIRT3 inhibited GLDC activity and caused increased glycine accumulation and decreased production of its metabolites, whereas deficiency of ACAT1 or NX-F1 had opposite effects. GLDC deficiency dramatically inhibited glioma cell proliferation, which was reversed by reconstitution with wild-type GLDC but not GLDCK514Q. Tumor growth in nude mice intracranially injected with wild-type GLDC-reconstituted glioma cells was markedly faster than those injected with GLDC-deficient or GLDCK514Q-reconstituted cells. Analysis of human GBM tissues indicated that levels of GLDC, SIRT3, and pS6K were positively associated with tumorigenesis, whereas levels of ACAT1, NF-X1, and GLDC K514 acetylation were negatively associated with tumorigenesis. In conclusion, our findings reveal that GLDC activity is regulated by coordinated posttranslational modifications including acetylation and polyubiquitination, and this contributes to regulation of glycine metabolism and tumorigenesis. Thus, targeting of GLDC posttranslational modifications may provide a potential strategy for therapeutics of cancers.
U251 cells, U87 cells, and HEK293 cells were obtained from ATCC, cultured in Dulbecco’s modified Eagle’s medium (DMEM; Sigma) supplemented with 10% fetal bovine serum (FBS; Biological Industries) and 1% penicillin–streptomycin (Thermo Fisher Scientific) at 37 °C with 5% CO2. All cells were negative for mycoplasma.
BALB/c nude mice were housed with 5 mice per cage on a 12-h light/dark cycle in a temperature-controlled room (23–25 °C) and relative humidity of 40–70% with free access to water and food. Mice were allowed at least 7 days for acclimation before experimentation. Eight-to-10-week-old and age-matched male mice were used in all experiments. All animals were randomly divided and injected intracranially with 5 × 105 (in 5 μl of DMEM per mouse) U251 or U87 cells expressing the NC, gGLDC, gGLDC + Flag-GLDC, and gGLDC + Flag-GLDCK514Q. Tumor formation and phenotype were determined by histological analysis of hematoxylin and eosin-stained sections. All animal use and experimental protocols were carried out in compliance with the Institutional Animal Care and Use Committee guidelines and approved by the Animal Care and Ethics Committee of Wuhan University Medical Research Institute.
Reagents and antibodies
Reagents and antibodies used in this study were purchased from the following indicated companies: Rapamycin (Cell Signal Technology), NAM (MedChemExpress), TSA (MedChemExpress), Lipofectamine 2000 (InvivoGen), polybrene (Millipore), SYBR (Bio-Rad), digitonin (Sigma), formate (Sigma), and 13C-2-glycine (Sigma). Information on the commercially available antibodies used in this study is provided in Supplementary Table 1. The antibody that specifically recognizes acetylated K514 of GLDC was raised by immunizing rabbits with a synthetic peptide of human GLDC (510GSVF(K-Ac)RTSP518) by ABclonal Technology (Wuhan).
Mammalian expression plasmids for Flag- or HA-tagged GLDC, ACAT1, GCN5L1, SIRT3, and their mutants and Myc-tagged NF-X1 and its mutants were constructed by standard molecular biology techniques. Guide RNA (gRNA) plasmids targeting GLDC, RPTOR, PGC1α, SIRT3, and NF-X1 were constructed into a lentiCRISPR V2 vector, which was provided by Dr. Shu-Wen Wu (Wuhan University).
Recombinant protein purification
The pGEX-6p-1-GST plasmids encoding wild-type GLDC and GLDCK514R (aa36–1020) were transformed into BL21-competent cells. Expression of the proteins was induced with 0.05 mM IPTG at 16 °C for 24 h. The proteins were purified with GST resins and eluted with elution buffer (phosphate-buffered saline (PBS), 100 mM Tris-HCl pH 8.8, 40 mM reduced glutathione). To prepare Flag-ACAT1, Flag-SIRT3, and Flag-SIRT3H248Y, mammalian expression plasmids for Flag-ACAT1, Flag-SIRT3, and Flag-SIRT3H248Y were transfected into HEK293 cells. The cells were lysed 24 h after transfection. Flag antibody-conjugated beads were then used for immunoprecipitation for 4 h at 4 °C. The beads were washed three times with lysis buffer. The Flag-tagged ACAT1, SIRT3, and SIRT3H248Y were eluted with 3×Flag peptides in 250 mM Tris-HCl, pH 8.0. The Flag-tagged ACAT1, SIRT3WT, and SIRT3H248Y were used for acetylation and deacetylation assays in vitro.
Human tissue samples
Human glioma and non-glioma tissues were collected from the Department of Neurosurgery, Renmin Hospital of Wuhan University, Wuhan, China. Non-glioma tissues were collected during surgery of severe traumatic brain injury after informed consent from the patients who needed post-trauma surgery. The clinical glioma specimens were examined and diagnosed by pathologists at Renmin Hospital of Wuhan University. Tissue procurement and use in this study were performed with written patient informed consents and approved by the Institutional Ethics Committee of Renmin Hospital of Wuhan University. The detailed clinicopathologic characteristics of patients is presented in Supplementary Table 2.
Double-stranded oligonucleotides corresponding to the target sequences were cloned into the Lenti-CRISPR-V2 vector, which were co-transfected with the packaging plasmids into HEK293 cells. Two days after transfection, the viruses were harvested, ultra-filtrated (0.45 μm filter, Millipore), and used to infect U251 cells or HEK293 cells in the presence of polybrene (8 μg/ml). The infected cells were selected with puromycin (1 μg/ml) for at least 6 days. The information of gRNA sequences is shown in Supplementary Table 3.
HEK293 cells were transfected by standard calcium phosphate precipitation. U251 and U87 cells were transfected with Lipofectamine 2000. The empty control plasmid was added to ensure that each transfection receives the same amount of total DNA.
Co-immunoprecipitation, ubiquitination, and immunoblotting analysis
Cells were lysed in 1 ml of NP-40 lysis buffer (20 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 1% Triton X-100, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride). For each immunoprecipitation reaction, a 0.4-ml aliquot of lysate was incubated with 0.5–2 μg of the indicated antibody or control IgG and 35 μl of a 1:1 slurry of Protein-G Sepharose (GE Healthcare) at 4 °C for 3 h. The Sepharose beads were washed three times with 1 ml of lysis buffer containing 500 mM NaCl. The precipitates were fractionated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), and immunoblotting analysis was performed with the indicated antibodies. For ubiquitination assays, the immunoprecipitants were re-extracted in NP-40 lysis buffer containing 1% SDS and denatured by heating for 10 min. The supernatants were diluted with regular lysis buffer until the concentration of SDS was decreased to 0.1%, following by re-immunoprecipitation with the indicated antibodies. The immunoprecipitants were analyzed by immunoblotting with the ubiquitin antibody.
Total RNA was isolated for qPCR analysis to measure mRNA abundance of the indicated genes. Data shown are the relative abundance of the indicated mRNAs normalized to that of glyceraldehyde 3-phosphate dehydrogenase. The qPCR data were collected with Bio-Rad CFX96 (Version 3.1) and analyzed with Bio-Rad CFX Manager (Version 3.1). Gene-specific primer sequences are listed in Supplementary Table 4.
HEK293 cells (1 × 108) were transfected with HA-tagged human GLDC. HA-tagged GLDC was immunoprecipitated and desalted. MS analysis was performed by SpecAlly (Wuhan) Life Science and Technology Company as previously described31.
In vitro acetylation and deacetylation assay
The recombinant GST-GLDC and Flag-ACAT1 were purified and mixed in the buffer containing 40 mM Tris-HCl pH 8.0, 75 mM potassium chloride (KCl), and 10 μM acetyl CoA in a final volume of 30 μl. The reaction was incubated at 30 °C for 45 min and terminated by the addition of SDS-PAGE sample buffer and acetylated proteins were analyzed by immunoblots. For deacetylation assay, purified recombinant GST-GLDC and SIRT3 were incubated in 20 μl of deacetylation buffer (50 mM Tris–HCl pH 9.0, 4 mM MgCl2, 50 mM NaCl, 0.5 mM dithiothreitol (DTT), 0.5 μM TSA) with or without 10 μM Ac-CoA, 1 mM NAD+, and NAM at 37 °C for 3 h with gentle agitation. For NAM treatment as a control, reactions were pretreated with 10 mM NAM for 10 min. The reaction was terminated by SDS/PAGE sample buffer, and the protein samples were subjected to immunoblotting analysis with the indicated antibodies.
Liquid chromatography–high-resolution MS (LC-HRMS)
LC-HRMS analysis was performed on a C18 column (Inertsil ODS-3, 4.6 by 250 mm, 5 μM) with the elution gradient of 5–25% methanol–0.15% methanoic acid over 20 min at a flow rate of 0.5 ml/min. LC-HRMS was conducted on an electrospray ionization–ion trap mass spectrometer (Thermo LTQ Obitrap Elite) in a positive mode with drying gas (275 °C, 10 l/ml) and a nebulizer pressure of 30 lb/in2.
For isotope tracing studies, 1 × 107 U87 or U251 cells were cultured in DMEM (0.4 mM unlabeled glycine) supplemented with 10% FBS (Biological Industries) and 0.4 mM 13C-2-glycine (Sigma). Intracellular metabolites were extracted by lysing cells in ice-cold 100% methanol. Samples were centrifuged for 20 min at 13,000 × g, and the supernatant was collected and analyzed by LC-HRMS.
GLDC activity assay
Human GLDC and mutants were expressed in U251 or HEK293 cells and purified by immunoprecipitation. GLDC activity was determined by measuring the exchange of the carboxyl carbon of glycine against 13C-bicarbonate carbon32. The standard assay mixture, in a total volume of 300 μl at 30 °C, contains 100 mM sodium phosphate (pH 6.0), 0.1 mM PLP, 18 mM glycine, 1 mM DTT, 5 μg H protein, 2.5 μg GLDC protein, and 30 mM 13C-NaHCO3. All rates were corrected by control reactions without glycine. All enzyme assays were done in triplicates and repeated at least once with independent protein preparations. Immediately after starting the reactions by the addition of 13C-NaHCO3 for a total time of 20 min, aliquots of 200 μl were mixed with 20 μl of trichloroacetic acid. These samples were dried overnight to remove remaining 13C-NaHCO3; the 13C-glycine was quantified by LC-HRMS.
Cells were treated with Rapamycin for various times and lysed by douncing 40 times in 1 ml of homogenization buffer (10 mmol/l Tris-HCl, pH 7.4, 2 mmol/l MgCl2, 10 mmol/l KCl, 250 mmol/l sucrose). The homogenized samples were centrifuged at 500 × g for 10 min. The supernatants were centrifuged at 5000 × g for 10 min to precipitate crude mitochondria, and the supernatants were saved as cytoplasm fractions33.
U251 cells were treated with dimethyl sulfoxide or Rapamycin (50 nM). Twelve hours after treatment, the cells were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.3% Triton X-100 in PBS for 15 min. The cells were blocked with 5% bovine serum albumin in PBS and stained with the indicated primary and secondary antibodies. The nuclei were strained with 4,6-diamidino-2-phenylindole for 2 min and then washed with PBS for 3 times. The stained cells were observed with a Zeiss LSM880 confocal microscope under a ×63 oil objective.
Cells were seeded in 6-well plates at 5 × 104 and allowed to adhere for 24 h. Triplicate wells were seeded for each experimental condition. Cells were trypsinized, resuspended in DMEM containing 10% FBS, and counted with a Cellometer (Bio Red) every 2 days over a 7-day period.
Statistics and reproducibility
Data were analyzed using a Student’s unpaired t test, multiple t test, or two-way analysis of variance with GraphPad Prism 7. For the correlation study, data were analyzed using a Spearman rank correlation test. The number of asterisks represents the degree of significance with respect to p values, with the latter presented within each figure or figure legend. All the biochemical experiments, particularly immunoblotting analysis, were repeated for at least two times with similar results.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
All the data supporting the findings of this study are available within the article and its supplementary information files or can be obtained from the corresponding author upon reasonable request. A reporting summary for this article is available as a Supplementary Information file. Source data are provided with this paper.
Saxton, R. A. & Sabatini, D. M. mTOR signaling in growth, metabolism, and disease. Cell 168, 960–976 (2017).
Brown, E. J. et al. A mammalian protein targeted by G1-arresting rapamycin-receptor complex. Nature 369, 756–758 (1994).
Sabatini, D. M., Erdjument-Bromage, H., Lui, M., Tempst, P. & Snyder, S. H. RAFT1: a mammalian protein that binds to FKBP12 in a rapamycin-dependent fashion and is homologous to yeast TORs. Cell 78, 35–43 (1994).
Kim, J. & Guan, K. L. Amino acid signaling in TOR activation. Annu Rev. Biochem. 80, 1001–1032 (2011).
Mecca, C., Giambanco, I., Donato, R. & Arcuri, C. Targeting mTOR in glioblastoma: rationale and preclinical/clinical evidence. Dis. Markers 2018, 9230479 (2018).
Saxton, R. A. & Sabatini, D. M. mTOR signaling in growth, metabolism, and disease. Cell 169, 361–371 (2017).
Jain, M. et al. Metabolite profiling identifies a key role for glycine in rapid cancer cell proliferation. Science 336, 1040–1044 (2012).
Kim, D. et al. SHMT2 drives glioma cell survival in ischaemia but imposes a dependence on glycine clearance. Nature 520, 363–367 (2015).
Nakai, T. et al. Structure of P-protein of the glycine cleavage system: implications for nonketotic hyperglycinemia. EMBO J. 24, 1523–1536 (2005).
Kume, A. et al. The glycine cleavage system. Molecular cloning of the chicken and human glycine decarboxylase cDNAs and some characteristics involved in the deduced protein structures. J. Biol. Chem. 266, 3323–3329 (1991).
Hiraga, K. & Kikuchi, G. The mitochondrial glycine cleavage system. Functional association of glycine decarboxylase and aminomethyl carrier protein. J. Biol. Chem. 255, 11671–11676 (1980).
Maddocks, O. D. K. et al. Modulating the therapeutic response of tumours to dietary serine and glycine starvation. Nature 544, 372–376 (2017).
Van Hove, J. L. K. et al. Nonketotic Hyperglycinemia in GeneReviews (eds Adam, M. P. et al.) (University of Washington, Seattle (WA), 1993–2021).
Hellani, A., Sammour, A., Johansson, L. & El-Sheikh, A. Delivery of a normal baby after preimplantation genetic diagnosis for non-ketotic hyperglycinaemia. Reprod. Biomed. Online 16, 893–897 (2008).
Zhang, W. C. et al. Glycine decarboxylase activity drives non-small cell lung cancer tumor-initiating cells and tumorigenesis. Cell 148, 259–272 (2012).
Alptekin, A. et al. Glycine decarboxylase is a transcriptional target of MYCN required for neuroblastoma cell proliferation and tumorigenicity. Oncogene 38, 7504–7520 (2019).
Hasse, D. et al. Structure of the homodimeric glycine decarboxylase P-protein from Synechocystis sp. PCC 6803 suggests a mechanism for redox regulation. J. Biol. Chem. 288, 35333–35345 (2013).
Still, A. J. et al. Quantification of mitochondrial acetylation dynamics highlights prominent sites of metabolic regulation. J. Biol. Chem. 288, 26209–26219 (2013).
Scott, I., Webster, B. R., Li, J. H. & Sack, M. N. Identification of a molecular component of the mitochondrial acetyltransferase programme: a novel role for GCN5L1. Biochem. J. 443, 655–661 (2012).
Shakespear, M. R. et al. Lysine deacetylases and regulated glycolysis in macrophages. Trends Immunol. 39, 473–488 (2018).
Kong, X. et al. Sirtuin 3, a new target of PGC-1alpha, plays an important role in the suppression of ROS and mitochondrial biogenesis. PLoS ONE 5, e11707 (2010).
Hirschey, M. D. et al. SIRT3 deficiency and mitochondrial protein hyperacetylation accelerate the development of the metabolic syndrome. Mol. Cell 44, 177–190 (2011).
Cunningham, J. T. et al. mTOR controls mitochondrial oxidative function through a YY1-PGC-1alpha transcriptional complex. Nature 450, 736–740 (2007).
Lorick, K. L. et al. RING fingers mediate ubiquitin-conjugating enzyme (E2)-dependent ubiquitination. Proc. Natl Acad. Sci. USA 96, 11364–11369 (1999).
Tibbetts, A. S. & Appling, D. R. Compartmentalization of Mammalian folate-mediated one-carbon metabolism. Annu. Rev. Nutr. 30, 57–81 (2010).
Labuschagne, C. F., van den Broek, N. J., Mackay, G. M., Vousden, K. H. & Maddocks, O. D. Serine, but not glycine, supports one-carbon metabolism and proliferation of cancer cells. Cell Rep. 7, 1248–1258 (2014).
Hu, M. M. et al. Sumoylation promotes the stability of the DNA sensor cGAS and the adaptor STING to regulate the kinetics of response to DNA virus. Immunity 45, 555–569 (2016).
Xia, T., Yi, X. M., Wu, X., Shang, J. & Shu, H. B. PTPN1/2-mediated dephosphorylation of MITA/STING promotes its 20S proteasomal degradation and attenuates innate antiviral response. Proc. Natl Acad. Sci. USA 116, 20063–20069 (2019).
Wu, X. et al. Regulation of TRIF-mediated innate immune response by K27-linked polyubiquitination and deubiquitination. Nat. Commun. 10, 4115 (2019).
Song, Z. M. et al. KAT5 acetylates cGAS to promote innate immune response to DNA virus. Proc. Natl Acad. Sci. USA 117, 21568–21575 (2020).
Li, M. & Shu, H. B. Dephosphorylation of cGAS by PPP6C impairs its substrate binding activity and innate antiviral response. Protein Cell 11, 584–599 (2020).
Hasse, D., Mikkat, S., Thrun, H. A., Hagemann, M. & Bauwe, H. Properties of recombinant glycine decarboxylase P- and H-protein subunits from the cyanobacterium Synechocystis sp. strain PCC 6803. FEBS Lett. 581, 1297–1301 (2007).
Zhong, B. et al. The adaptor protein MITA links virus-sensing receptors to IRF3 transcription factor activation. Immunity 29, 538–550 (2008).
We thank Wen-Rui He, Ru Zang, Lu Feng, Deng Gao, Xuan Zhong, Wei Guo, Mi Li, Chen Li, Li Zhong, and Zemin Song for technical help and Dr. Wen-Qing Chen for LC-HRMS analysis. This work was supported by grants from the State Key R&D Program of China (2017YFA0505800), the National Natural Science Foundation of China (31830024, 31630045, 31871415, and 32070775), the CAMS Innovation Fund for Medical Sciences (2019-I2M-5-071), and the Fundamental Research Funds for the Central Universities.
The authors declare no competing interests.
Peer review information Nature Communications thanks Issam Ben-Sahra, Shi-Yuan Cheng, and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Liu, R., Zeng, LW., Gong, R. et al. mTORC1 activity regulates post-translational modifications of glycine decarboxylase to modulate glycine metabolism and tumorigenesis. Nat Commun 12, 4227 (2021). https://doi.org/10.1038/s41467-021-24321-3