3.1 Exogenous serine starvation inhibits proliferation and induces G1 phase cell cycle arrest in ESCC cells
To detect the role of serine in human ESCC cell proliferation, we formulated media analogous to DMEM that contained 0.4 mM serine and 0.4 mM glycine (Com), lacking serine (-Ser), or lacking serine and glycine (-SG)[9]. ESCC cell viability in these media was detected and the results showed that exogenous serine starvation significantly decreased the proliferation and colony formation of ESCC cells (Fig. 1A, S1A -B), but slightly increased the death of the cells (Fig. 1B). Although serine and glycine can be interconverted and either can be used for nucleotide synthesis and one-carbon metabolism, we found that exogenous glycine supplementation did not replace serine to support ESCC cell proliferation (Fig. 1A). Slight suppression of cell proliferation was observed in ESCC cells with glycine starvation (Fig. 1A). This finding is consistent with colon cancer cells[6] and not surprising, as glycine could be replenished by compensatory synthesis from endogenous serine.
Next, we detected the effect of serine concentration on ESCC cell viability and observed significant cell viability inhibition in a serine concentration-dependent manner, especially when the serine concentration was below the 50 µM (Fig. 1C). The maximum growth inhibition of KYSE150 and KYSE180 cells was 40% and 45% at 0 µM serine, respectively. Then, we measured cell proliferation after serine starvation by an EdU incorporation assay. The results showed that the number of EdU-positive cells in KYSE150 and KYSE180 cells was significantly reduced by 95% and 73% compared with the control, respectively, suggesting that the lack of exogenous serine dramatically inhibited DNA synthesis (Fig. 1D).
Then, we analyzed the cell cycle by flow cytometry after serine starvation. After synchronization and arrest in G0/G1 phase by a double-thymidine block, the ESCC cells were simultaneously released into complete or serine-free media for a period of time, followed by cell cycle analysis. Compared with the complete medium, KYSE150 cells had significant G1 phase cell cycle arrest under serine starvation, which was moderate in KYSE180 cells, and both of them eventually recover integral cell cycle (Fig. 1E, S1C). Collectively, exogenous serine starvation inhibits proliferation and induces G1 phase cell cycle arrest in ESCC cells, but did not induce cell death.
3.2 Exogenous serine starvation suppresses purine nucleotide synthesis via one- carbon metabolism
To explore the possible mechanism by which serine deprivation inhibited cell proliferation and induced G1 phase cell cycle arrest, we performed a metabolomics assay on KYSE150 cells after serine starvation for 24 h. A total of 1,315 metabolites were detected in both the positive and negative modes, and 216 metabolites were identified as significantly altered in aqueous solution (Fig. S2A). In agreement with a previous study in colon cancer cells[6], our results showed that exogenous serine starvation significantly inhibited serine-derived one-carbon metabolic pathways, including purine metabolism, arachidonic acid metabolism, and cysteine and methionine metabolism (Fig. S2B). Remarkably, purine nucleotides were markedly depleted, and glycinamide ribonucleotide (GAR) and AICAR, intermediate metabolites of the de novo purine nucleotide synthesis pathway, accumulated after serine deprivation, suggesting that 10-formyl-THF deficiency induced by serine deprivation resulted in the accumulation of metabolic intermediates (Fig. 2A). Unexpectedly, the intermediate metabolites of pyrimidine synthesis, such as UMP, dTMP and dTTP, were maintained at relatively constant levels under all conditions, suggesting that exogenous serine starvation did not affect pyrimidine synthesis in ESCC cells (Fig. S2C). Together, these data suggest that exogenous serine starvation significantly inhibits de novo purine nucleotide synthesis and results in the accumulation of GAR and AICAR.
Formate (FA) contains a single carbon atom and reacts directly with THF, making 10-formyl-THF a one-carbon donor for purine synthesis[26]. To detect whether serine-derived one-carbon unit deficiency was responsible for the inhibition of DNA synthesis, we replenished the one-carbon pool directly by adding 1 mM FA into the medium without serine and then performed MTT assays, cell cycle assays and EdU incorporation assays. Interestingly, FA completely rescued the inhibition of cell proliferation caused by serine starvation but not by serine and glycine starvation (Fig. S3A-B). As expected, FA also rescued G1 phase cell cycle arrest (Fig. S3C) and inhibition of DNA synthesis (Fig. S3D) caused by serine starvation. These results suggest that serine supports DNA synthesis and the cell cycle via one-carbon metabolism in ECSS cells.
3.3 Exogenous serine starvation induces AMPKα activation to inhibit mTORC1 signaling pathway
As we found that serine starvation led to significant accumulation of AICAR, which is an AMPK activator (Fig. 2A), the phosphorylation level of AMPKα was measured after serine starvation by Western blot analysis. Notably, phosphorylation of AMPKα at Thr172 was significantly upregulated in KYSE150 and KYSE180 cells after serine starvation (Fig. 2B). Since serine can maintain mTORC1 activity in lung cancer cells[25], we evaluated mTORC1 activity after serine starvation. The levels of phospho-S6 kinase (p-p70S6K) and phospho-S6 (p-S6), downstream effectors of mTORC1, were clearly downregulated after serine starvation in KYSE150 and KYSE180 cells (Fig. 2B). Furthermore, the phosphorylation of Raptor at Ser792 was increased, but mTOR was not phosphorylated at Ser2448 (Fig. 2B). As expected, FA attenuated the upregulation of p-AMPKα (Thr172) and p-Raptor (Ser792) and restored the downregulation of p-p70S6K and p-S6 induced by serine starvation (Fig. 2C).
To further confirm that serine starvation-induced inhibition of mTORC1 activity is directly mediated by AMPKα, we silenced PRKAA1 (encoding AMPKα protein) using one siRNA numbered #1 (> 90% depletion compared with the control; Fig. 2D) or inhibited AMPKα by compound C. As expected, both compound C and siPRKAA1 rescued the downregulation of p-p70S6K and p-S6 and attenuated the upregulation of p-AMPKα (Thr172) and p-Raptor (Ser792) induced by serine deprivation (Fig. 2D, E). These results suggest that serine starvation induces AMPKα/Raptor activation to suppress mTORC1 signaling pathway.
3.4 Exogenous serine starvation inhibits NADPH synthesis and induces DNA oxidative damage by inhibiting one-carbon metabolism
In addition to the oxidative pentose phosphate pathway from glucose, NADPH is also derived from serine-driven one-carbon metabolism (Fig. S3E)[11]. The NADPH/NADP+ ratio was measured after serine starvation in ESCC cells. The NADPH/NADP+ ratio was decreased by 42% and 41% in serine-starved KYSE150 and KYSE180 cells, respectively (Fig. 3A). Moreover, a decreased GSH/GSSG ratio was also observed after serine deprivation (Fig. 3B). Since the hydrogen atom of NADPH derived from serine via one-carbon metabolism can be transferred into proline directly[11, 27], proline was further detected after serine starvation. As expected, proline was decreased by 54% in KYSE150 cells, which further demonstrated NADPH reduction after serine starvation (Fig. 3C). Collectively, the results suggest that serine starvation significantly inhibits NADPH synthesis and the GSH/GSSG ratio in ESCC cells.
The decreases in NADPH/NADP+ and the GSH/GSSG ratio suggested that the antioxidant ability of ESCC cells was decreased after serine starvation, which may lead to ROS accumulation to damage cellular components, promoting cytotoxicity and inducing cell death. Thus, we examined the antioxidant ability of ESCC cells to hydrogen peroxide (H2O2)-induced oxidative stress after serine starvation for 24 h by measuring the mortality of ESCC cells using trypan blue staining. The results showed that H2O2 treatment significantly increased the mortality of KYSE150 and KYSE180 cells in serine-deficient medium for 24 h, suggesting that serine deprivation decreases the antioxidant ability of ESCC cells (Fig. 3D). Furthermore, we detected intracellular ROS accumulation after serine starvation by the oxidation-activated fluorescent dye DCFH-DA. As expected, the results showed that serine starvation for 24 h significantly increased the intracellular ROS level (Fig. 3E). Since excessive ROS cause DNA oxidative damage by oxidizing guanine in DNA to 8-oxoguanine (8-OxoG), the DNA damage of ESCC cells was evaluated after serine starvation by immunofluorescence assay. Consistently, the 8-OxoG-positive cells were significantly increased in ESCC cells after serine starvation for 24 h (Fig. 3F). Taken together, these results suggest that serine starvation induces cell oxidative stress and DNA oxidative damage.
To further confirm that the decrease in NADPH was induced by serine-driven one-carbon metabolism, 1 mM FA was added to the medium without serine followed by H2O2 treatment for 24 h. The results showed that FA significantly rescued H2O2-induced oxidative stress (Fig. S4A). Furthermore, 1 mM FA partially rescued the GSH/GSSG ratio (Fig. S4B) and clearly decreased the intracellular ROS (Fig. S4C) and DNA oxidative damage induced by serine starvation (Fig. S4D). These results suggest that serine-derived one-carbon units support NADPH synthesis and maintain redox homeostasis in ESCC cells.
3.5 p53-mediated DNA damage response protects cells survival during serine starvation
To further gain insight into the mechanism by which serine starvation induced DNA damage response, we applied mRNA sequencing technology (mRNA-seq) to explore the gene expression profiles in KYSE150 cells after serine starvation. Through enrichment analysis of the significantly changed genes with the Gene Ontology Consortium and KEGG pathway databases, the altered genes were closely associated with the cell cycle, nuclear division, nuclear chromosome segregation, p53 signaling pathway and purine metabolism (Fig. 4A, B). We then performed qPCR assays to verify the representative genes obtained from mRNA-seq, especially genes involved in cell cycle regulation, given that we found that serine deprivation induced cell cycle arrest. The results showed that TP53 as well as its downstream target genes, CDKN1A and GADD45A, were significantly upregulated (Fig. 4C). DNA damage causes ATM to phosphorylate p53 Ser15, which promotes the accumulation of p53 and thereby transcriptionally increases the expression of CDKN1A[28]. Meanwhile, the CCNA2 responsible for S phase and CCNB2 responsible for G2/M DNA damage detection were significantly decreased after serine starvation (Fig. 4C). These data were consistent with the results of mRNA-seq assays. Western blot analysis was performed to detect the protein levels, and the results showed that serine starvation clearly upregulated the expression of p-p53 (Ser15) and p21 in KYSE150 and KYSE180 cells, and a moderate increase in level of p53 (Fig. 4D).
To further confirm that G1 phase cell cycle arrest was induced by p53-p21, we silenced P53 using two siRNAs numbered #1 and #2 (> 90% depletion compared with the control, Fig. 4E). As expected, p53 knockdown attenuated the upregulation of p21 induced by serine deprivation (Fig. 4E). The results suggest that DNA oxidative damage induced by serine starvation induced p53-p21-mediated G1 phase cell cycle arrest in ESCC cells. Meanwhile, the role of p53 during serine starvation is evaluated. The result shown that p53 knockdown increased modestly sensitivity to serine starvation in p53 mutant KYSE150 (p53R248Q) cells and KYSE180 (p53I195T) cells (Fig. 4F). This result may be due to the fact that mutated p53-mediated G1 arrest in response to serine starvation could increase time for DNA repair and endogenous serine synthesis, and/or allowing the limited levels of de novo serine to be channelled to GSH production to counter oxidative stress.[9]
Since FA could rescue DNA synthesis, cell cycle arrest and DNA oxidative damage induced by serine starvation, the levels of cell cycle regulatory genes were measured by qPCR in KYSE150 and KYSE180 cells in serine-deficient medium containing 1 mM FA. The results showed that FA treatment significantly restored the levels of the above genes (Fig. 4G). Meanwhile, Western blot analysis found that FA clearly attenuated the expression of p-p53 (Ser15) induced by serine starvation in KYSE150 and KYSE180 cells (Fig. 4H). These results were consistent with the rescued effect of FA on DNA synthesis and oxidative damage induced by serine starvation. Collectively, the results suggest that serine starvation triggers p53-p21-mediated G1 phase cell cycle arrest by targeting one-carbon metabolism and partly protects ESCC cells from DNA oxidative damage upon serine deprivation.
3.6 Exogenous serine starvation compensatorily activates metabolic enzymes of the serine synthesis pathway (SSP) and increases endogenous serine synthesis
Mammalian cells synthesize endogenous serine via SSP enzymes, including phosphoglycerate dehydrogenase (PHGDH), PSAT1 and phosphoserine phosphatase (PSPH), by channeling the glycolytic intermediate 3-phosphoglycerate into the ‘phosphorylated pathway’ of serine synthesis. As exogenous serine starvation may motivate the endogenous synthesis pathway, we monitored the serine and glycine levels in ESCC cells during serine starvation. After exogenous serine deprivation, intracellular serine decreased significantly, especially within 6 h of serine starvation in KYSE150 cells (Fig. 5A). Interestingly, we observed that serine levels showed a clear upward trend, and recovered to 80% of the control 24 h after serine starvation (Fig. 5A). Glycine levels showed only a decrease in the short period (3 h) and rapidly recovered to normal levels after 6 h of serine deprivation (Fig. 5B). Consistently, we found that serine starvation significantly induced transcriptional upregulation of the PHGDH, PSAT1 and PSPH genes, suggesting activation of the endogenous SSP pathway to generate endogenous serine to maintain cancer cell growth (Fig. 5C, D).
ATF4 is a transcription factor that responds to stress, such as amino acid deletions and hypoxia. ATF4 transcriptionally activates serine biosynthetic genes in response to serine starvation in NSCLC cells[29]. Interestingly, we found that both the mRNA and protein levels of ATF4 were significantly upregulated in KYSE 150 and KYSE 180 cells under serine starvation (Fig. 5C, D). Based on data mining of the TCGA database, we found that ATF4 expression was positively correlated with the expression of PHGDH, PSAT1 and PSPH in human ESCC tissues (Fig. 5E). Chromatin immunoprecipitation assays revealed that serine starvation induced remarkable recruitment of ATF4 at the promoters of the PHGDH and PSAT1 genes but not at the PSPH promoter region in KYSE150 cells (Fig. 5F). Furthermore, knockdown of ATF4 by two siRNAs numbered #1 and #2 (> 50% depletion compared with the control) significantly attenuated the transcriptional upregulation of the PHGDH and PSAT1 genes induced by serine starvation (Fig. 5G, H) and completely inhibited cell proliferation (Fig. 5I). Therefore, ATF4 mediates the transcriptional upregulation of PHGDH and PSAT1 in ESCC cells after serine starvation, thus promoting compensatory endogenous serine synthesis and cell growth.
Moreover, we analyzed the gene expression of PHGDH, PSAT1 and PSPH in human ESCC tissues by using the GEO and TCGA databases. The results showed that PSAT1 and PSPH were significantly overexpressed in human EC tissues compared with normal esophageal tissues (Fig. S5A, B). Surprisingly, no difference in PHGDH levels was observed between ESCC tissues and normal esophageal tissues. qPCR and Western blot analysis further confirmed the results from database analysis (Fig. S5C, D). These results reveal that PSAT1 upregulation is highly associated with the progression of ESCC. Then, we focused on the role of PSAT1 in ESCC cell proliferation after serine deprivation.
3.6 The combination of PSAT1 knockdown and serine deprivation inhibited ESCC cell growth in vitro and in vivo
To confirm that the robust expression of PSAT1 is critical to sustain the survival of ESCC cells in the absence of serine, we silenced PSAT1 using shRNA in KYSE150 and KYSE180 cells (> 90% depletion compared with the control) and further evaluated cell proliferation by MTT assay (Fig. 6A). Proliferation curves showed that PSAT1 knockdown caused proliferation inhibition in KYSE150 and KYSE180 cells (Fig. 6B), although this inhibition was less than that induced by serine deprivation, suggesting that exogenous serine is very important for ESCC cell proliferation. Notably, PSAT1 knockdown with serine deprivation completely inhibited cell proliferation (Fig. 6B). These results indicate that knockdown of PSAT1 enhanced the cell proliferation inhibition induced by serine deprivation.
We further performed a xenograft experiment to investigate whether serine produced by endogenous synthesis and/or exogenous uptake supports ESCC cell growth in vivo. Nude mice were subcutaneously injected with KYSE150 cells (shPSAT1 right flank, vector control left flank), and fed a diet with or without serine. The results showed that KYSE150 cells rapidly formed tumors in mice fed a diet with serine. Surprisingly, no xenograft tumor difference was observed between mice fed control and serine-deficient diets alone (Fig. 6C-E). However, PSAT1 knockdown resulted in a significant reduction in xenograft tumors by 71% compared with the control (Fig. 6C-E). The combination of PSAT1 knockdown and serine deprivation displayed more obvious growth inhibition of tumors than serine lacking diet or PSAT1 knockdown alone (Fig. 6C-E). Moreover, half of the PSAT1-silenced KYSE150 cells were not able to form tumors in mice fed a serine-deficient diet compared with mice fed a control diet (Fig. 6C-E). Unlike essential amino acids, the chronic deprivation of nonessential amino acids can be tolerated in vivo. The mice tolerated a serine-lacking diet, and the body weight of the mice was not significantly changed (Fig. 6F).
We then measured the serine concentration in the serum of mice and xenograft tumor tissues by HPLC. The serine starvation diet resulted in a slight but no significant decrease in serine levels in tumor tissues compared with the control diet (Fig. 6G). However, PSAT1 knockdown significantly decreased the serine level in tumor tissues, and the combination of PSAT1 knockdown and serine deprivation more obviously decreased the serine level in tumor tissues (Fig. 6G). No significant drop in serum levels of serine was observed (Fig. 6H). Consistently, a diet lacking serine significantly promoted the upregulation of PSAT1 expression, which maintained the serine level in xenograft tumors (Fig. 6I).
These results suggested that serine deprivation in the diet cannot affect tumor growth or serine levels in tumor tissues, suggesting that endogenous serine synthesis might provide enough serine when exogeneous serine deprivation occurs. The combination of endogenous and exogeneous serine inhibition could inhibit tumorigenesis efficiently in vivo.