AMPK-HIF-1α signaling enhances glucose-derived de novo serine biosynthesis to promote glioblastoma growth

Abstract

Background: Cancer cells undergo cellular adaptation through metabolic reprogramming to sustain survival and rapid growth under various stress conditions. However, how brain tumors modulate their metabolic flexibility in the naturally serine/glycine (S/G)-deficient brain microenvironment remain unknown.

Methods: We used a range of primary/stem-like and established glioblastoma (GBM) cell models in vitro and in vivo. To identify the regulatory mechanisms of S/G deprivation-induced metabolic flexibility, we employed high-throughput RNA-sequencing, transcriptomic analysis, metabolic flux analysis, metabolites analysis, chromatin immunoprecipitation (ChIP), luciferase reporter, nuclear fractionation, cycloheximide-chase, and glucose consumption. The clinical significances were analyzed in the genomic database (GSE4290) and in human GBM specimens.

Results: The high-throughput RNA-sequencing and transcriptomic analysis demonstrate that the de novo serine synthesis pathway (SSP) and glycolysis are highly activated in GBM cells under S/G deprivation conditions. Mechanistically, S/G deprivation rapidly induces reactive oxygen species (ROS)-mediated AMP-activated protein kinase (AMPK) activation and AMPK-dependent hypoxia-inducible factor (HIF)-1α stabilization and transactivation. Activated HIF-1α in turn promotes the expression of SSP enzymes phosphoglycerate dehydrogenase (PHGDH), phosphoserine aminotransferase 1 (PSAT1), and phosphoserine phosphatase (PSPH). In addition, the HIF-1α-induced expression of glycolytic genes (GLUT1, GLUT3, HK2, and PFKFB2) promotes glucose uptake, glycolysis, and glycolytic flux to fuel SSP, leading to elevated de novo serine and glycine biosynthesis, NADPH/NADP⁺ ratio, and the proliferation and survival of GBM cells. Analyses of human GBM specimens reveal that the levels of overexpressed PHGDH, PSAT1, and PSPH are positively correlated with levels of AMPK T172 phosphorylation and HIF-1α expression and the poor prognosis of GBM patients.

Conclusion: Our findings reveal that metabolic stress-enhanced glucose-derived de novo serine biosynthesis is a critical metabolic feature of GBM cells, and highlight the potential to target SSP for treating human GBM.

Keywords: AMPK; De novo serine synthesis; Glycine; HIF-1α; Serine.

Fig. 1

Genes in glycolysis and de novo serine biosynthesis are highly expressed in GBM cells upon serine/glycine deprivation. A U87MG cells were cultured with serine/glycine-containing complete media (S/G( +)) or equivalent media lacking serine/glycine (S/G(-)) for 24 h. B Heatmap presentation of 1,173 genes (fold change ≥ 1.3 and ≤  − 1.3; P < 0.05) in U87MG cells cultured as in (A). C Bubble plot for ontology-based pathway enrichment analysis for DEGs of U87MG cells in S/G-deprived U87MG cells, compared to control. Enrichment score is calculated with odds ratio and P-value and used to rank the pathways. Bubble size indicates odds ratio and is used to assess the strength of association between DEGs and the pathway. Color indicates − log10 (P-value) for determination of the representation significance. THF, tetrahydrofolate; CRF, Corticotropin releasing factor; CCKR, cholecystokinin receptor. D Analysis of trimmed DEGs network with biological functions, including cell proliferation, cell survival, L-serine biosynthetic process, glycolytic process, and glucose import, using IPA in S/G-deprived U87MG cells, compared to control. The analysis involved a fold change cut-off value of ± 1.3. Red and green colors indicate genes that were upregulated and downregulated, respectively, compared to the control. Figure S1A of the SI provides details of shape, which originate from Ingenuity Systems (http://www.ingenuity.com). E Heatmap presentation for gene expression profiles in the L-serine biosynthetic process, glycolytic process, and glucose import. F Schematic of the glucose-derived serine synthesis pathway

Fig. 2

Serine/glycine deprivation induces glucose uptake, glycolytic flux, and de novo serine biosynthesis. A Indicated cells were cultured with or without S/G(-) media for 24 h. Quantitative real-time PCR analyses were performed with the indicated primers. B Indicated cells were cultured with or without S/G(-) media for 48 h. Glucose consumption was measured. C Schematic of [U-¹³C₆]-glucose into glycolysis and serine synthesis pathway. Glucose (M + 6) is converted into 3-PG (M + 3), a three-carbon sugar, through several steps. Serine (M + 3) is synthesized from the glycolytic intermediate 3-PG, and then glycine (M + 2) is generated form the serine. D Isotopomer distributions of G6P, F6P, F1,6BP, 3-PG from [U-¹³C₆]-glucose in U87MG cells cultured with or without S/G(-) media, n = 3 biologically independent samples. E and F Indicated cells were cultured without S/G(-) media for the indicated periods of time. Quantitative real-time PCR (E) and Immunoblotting (F) analyses were performed with the indicated antibodies and primers, respectively. G Isotopomer distributions of 3-PS, serine, and glycine from [U-¹³C₆]-glucose in U87MG cells cultured with or without S/G(-) media, n = 3 biologically independent samples. The data represent the mean ± s.d. of three independent experiments (A, B, D, E, G). *P < 0.05; **P < 0.01; ***P < 0.001, based on the Student’s t-test

Fig. 3

Serine synthesis pathway genes are overexpressed in gliomas and required for brain tumor growth. A Microarrays of human gliomas and normal brain tissue were immunostained with the indicated antibodies. Representative images are shown (left panel). Data represent the mean ± s.d. (right panel). Scale bar, 100 μm. NS, not significant. B The mRNA and protein expression levels of PHGDH, PSAT1, and PSPH in NHA, GSCs and the indicated GBM cells were determined by quantitative real-time PCR (top panel) and immunoblotting (bottom panel) analyses with the indicated primers and antibodies, respectively. C U87MG cells with or without the depletion of SSP genes were cultured with or without S/G(-) media for the indicated periods of time, and harvested for cell counting. D U87MG cells with or without the depletion of SSP genes were cultured with or without S/G(-) media for 4 d, and stained with Annexin V. Representative staining (top panel) and quantification of the staining (bottom panel) are shown. Scale bar, 20 μm. E A total of 5 × 10⁵ control GSCs or SSP-depleted GSCs were intracranially injected into athymic nude mice. After 21 d, the mice were euthanized and examined for tumor growth. Hematoxylin-and-eosin–stained coronal brain sections show representative tumor xenografts (left panel). Tumor volumes were measured using the length (a) and width (b) and calculated using the equation V = ab²/2. Data represent the mean ± s.d. of 5 mice (right panel). Note that the scores of some samples overlap. Scale bar, 2 mm. F IHC analyses of the tumor tissues were performed with anti-PHGDH, anti-PSAT1, anti-PSPH, and anti-Ki-67 antibodies and TUNEL. Representative staining (left panel) and quantification of the staining (right panel) are shown. Scale bar, 100 μm. The data represent the mean ± s.d. of three independent experiments (A-D, F). *P < 0.05; **P < 0.01; ***P < 0.001, based on the Student’s t-test

Fig. 4

HIF-1α induces SSP gene expression in response to serine/glycine deprivation. A Indicated cells were cultured with or without S/G(-) media for the indicated periods of time. Whole cell lysates (WCL) and nuclear fractions were prepared, and immunoblotting analyses were then performed with the indicated antibodies. B HRE luciferase activities were measured in the indicated cells culturing with or without S/G(-) media for the indicated periods of time. C Indicated cells were cultured with or without S/G(-) media for the indicated periods of time. Quantitative real-time PCR analyses were performed with the indicated primers. D Indicated cells were cultured with or without S/G(-) media for the indicated periods of time. Immunoblotting analyses were performed with the indicated antibodies. E Indicated cells were cultured with or without S/G(-) media for 24 h, and then treated with CHX (100 μg⋅mL⁻¹) for the indicated periods of time. Immunoblotting analyses were performed with the indicated antibodies. Quantification of HIF-1α levels relative to tubulin is shown. Data represent the mean ± s.d. of three independent experiments. F Indicated cells were cultured with S/G(-) media in the presence or absence of PX-478 (10 μM) for 24 h. Quantitative real-time PCR analyses were performed with the indicated primers. G Schematics of the putative HIF-1α binding site on the PHGDH, PSAT1, and PSPH promoter regions, respectively. H Indicated cells were cultured with or without S/G(-) media for the indicated periods of time. ChIP assays were performed with anti-HIF-1α antibody, and quantitative real-time PCR analyses were performed with primers against the PHGDH, PSAT1, and PSPH promoters. I and J Indicated cells with or without the depletion of HIF-1α were cultured with or without S/G(-) media for the indicated periods of time. Quantitative real-time PCR (I) and immunoblotting (J) analyses were performed with the indicated primers and antibodies, respectively. K HRE luciferase activities were measured in the indicated cells cultured in complete media with or without CoCl₂ for 24 h. L and M Indicated cells were cultured in complete media with or without CoCl₂ for 24 h. Immunoblotting (L) and quantitative real-time PCR (M) analyses were performed with the indicated antibodies and primers, respectively. The data represent the mean ± s.d. of three independent experiments (B, C, E, F, H, I, K, M). *P < 0.05; **P < 0.01; ***P < 0.001, based on the Student’s t-test

Fig. 5

AMPK activation is required for the HIF-1α-induced SSP gene expression in response to serine/glycine deprivation. A-C Indicated cells were cultured with or without S/G(-) media for the indicated periods of time. Intracellular ATP and AMP levels (A), NADPH and NADP⁺ levels (B), and ROS levels (C) were measured. NS, not significant. D Indicated cells were cultured with or without S/G(-) media in the presence or absence of 2-DG (10 mM) for the indicated periods of time. Immunoblotting analyses were performed with the indicated antibodies. E Indicated cells were cultured with or without S/G(-) media for the indicated periods of time in the presence or absence of NAC (2 mM). Immunoblotting analyses were performed with the indicated antibodies. F Indicated cells were cultured with S/G(-) media for 24 h in the presence or absence of compound C (5 μM), and then treated with CHX (100 μg⋅mL⁻¹) for the indicated periods of time. Immunoblotting analyses were performed with the indicated antibodies. Quantification of HIF-1α levels relative to tubulin is shown. Data represent the mean ± s.d. of three independent experiments. G Indicated cells with or without the depletion of AMPK were cultured with or without S/G(-) media for 24 h. Immunoblotting analyses were performed with the indicated antibodies. H HRE luciferase activities were measured in the indicated cells with or without the depletion of AMPK culturing with or without S/G(-) media for the indicated periods of time. NS, not significant. I and J Indicated cells with or without the depletion of AMPK were cultured with or without S/G(-) media for 24 h. Quantitative real-time PCR (I) and immunoblotting (J) analyses were performed with the indicated primers and antibodies, respectively. K Indicated cells were cultured with S/G(-) media in the presence or absence of compound C (10 μM) for 24 h. Quantitative real-time PCR analyses were performed with the indicated primers. The data represent the mean ± s.d. of three independent experiments (A-C, F, H, I, K). *P < 0.05; **P < 0.01; ***P < 0.001, based on the Student’s t-test

Fig. 6

AMPK-HIF-1α signaling promotes de novo serine biosynthesis, proliferation, and survival of GBM cells upon serine/glycine deprivation. A Isotopomer distributions of G6P, F6P, F1,6BP, 3-PG, 3-PS, serine, and glycine from [U-¹³C₆]-glucose in U87MG cells cultured with S/G(-) media with or without the depletion of AMPK or HIF-1α, n = 3 biologically independent samples. B and C Indicated cells with or without the depletion of AMPK or HIF-1α were cultured with S/G(-) media for 2 d (to measure glucose consumption), or 5 d (to measure intracellular serine/glycine levels). Glucose consumption (B) and intracellular serine/glycine levels (C) were measured. D Indicated cells with or without the depletion of AMPK or HIF-1α and with or without the reconstituted expression of SSP genes were cultured with S/G(-) media for 5 d. Intracellular NADPH and NADP⁺ levels (top panel) and ROS levels (bottom panel) were measured. E Indicated cells with or without the depletion of AMPK or HIF-1α were cultured with or without S/G(-) media for the indicated periods of time, and harvested for cell counting. NS, not significant. F U87MG cells with or without the depletion of AMPK or HIF-1α were cultured with or without S/G(-) media for 4 d, and stained with Annexin V. Representative staining (top panel) and quantification of the staining (bottom panel) are shown. Scale bar, 20 μm. G and H Indicated cells with or without the depletion of AMPK or HIF-1α, and with or without the reconstituted expression of SSP genes, were cultured with S/G(-) media for 4 d (for Annexin V staining) or 6 d (for cell counting). The data represent the mean ± s.d. of three independent experiments (A-H). *P < 0.05; **P < 0.01; ***P < 0.001, based on the Student’s t-test or one-way ANOVA with Tukey’s post hoc test

Fig. 7

The expression levels of SSP enzymes are positively correlated with levels of AMPK T172 phosphorylation and HIF-1α expression in human GBM specimens and poor GBM patient prognosis. A IHC staining of human GBM specimens was performed with the indicated antibodies (n = 50). Representative images from the staining of four different specimens are shown. Scale bar, 100 μm. B The IHC stains were scored, and correlation analyses were performed. The Pearson correlation test was used. Note that the scores of some samples overlapped. C Kaplan–Meier analyses for GBM patients with high or low level of SSP. D Network with DEGs, biological functions, glycolytic metabolism, and SSP using IPA in S/G-deprived U87MG cells, compared to control. The analysis involved a fold change cut-off value of ± 1.3. Red and green colors indicate genes that were upregulated and downregulated, respectively, compared to the control. Reactive oxygen species (ROS), AMPK, and HIF-1α are added for analysis of the relationships among the genes for the biological functions, glycolytic metabolism, and SSP. Figure S1A of the SI provides the details of shape, which originate from Ingenuity Systems (http://www.ingenuity.com). E Schematic of the regulation of Brain Tumor Serine metabolism (BTS); AMPK-HIF-1α signaling-enhanced glucose-derived de novo serine biosynthesis to promote the proliferation and survival of GBM and brain tumor growth under the brain microenvironment of limited serine/glycine