Background
Metabolic reprogramming, one of the crucial hallmarks of cancer, provides advantages in cell proliferation and survival for tumor growth [
1,
2]. It is well known that the Warburg effect, characterized by enhanced glycolytic flux even in aerobic conditions, is a unique metabolic phenotype in many types of cancer cells [
3,
4]. An increased glucose uptake resulting from the Warburg effect supplies cancer cells with energetic and biosynthetic pathways for rapid proliferation or overcoming numerous stresses [
2]. Including the Warburg effect, many metabolic pathways are altered or reprogrammed in cancer cells, which are influenced by both genetic alterations (e.g., amplification of metabolic enzymes, activation of putative oncogenes, or loss of tumor suppressor genes) and microenvironmental changes (e.g., altered nutrition, hypoxia, growth factors/cytokines, or infiltrated immune/stromal cells) [
5]. In particular, cancer cells adapt their metabolism under the tissue- or organ-specific tumor microenvironment through acquiring metabolic flexibility to support survival and growth, leading to various dependencies and vulnerabilities that could be targeted for therapy [
6‐
8].
The metabolism of serine and glycine is upregulated in many types of human cancers and plays important roles in cancer proliferation and tumor growth [
9‐
16]. As important one-carbon donors to one-carbon metabolism (which comprises two interconnected metabolic cycles, the folate cycle and the methionine cycle), the non-essential amino acids serine and glycine contribute to the one-carbon metabolism-mediated production of cellular building blocks, including nucleic acids, proteins, and lipids, as well as modulation of the NADPH/NADP
+ ratio and glutathione production to maintain cellular redox balance [
17,
18]. Thus, serine and glycine play critical roles in the cellular biosynthesis, metabolic processes, and homeostasis for rapidly proliferating cancer cells. Serine and glycine (which is directly converted from serine by the serine hydroxymethyltransferase (SHMT) 1/2 reaction) can be taken up into cells using a number of different transporters from the extracellular environment or can be synthesized de novo by cells using the serine synthesis pathway (SSP) [
17,
18]. The SSP is one of the branches of glycolysis, allowing glucose-derived carbons to be diverted into synthesized serine. The glycolytic intermediate, 3-phosphoglycerate (3-PG) can be converted to serine through three consecutive enzymatic reactions of SSP [
17,
18]; Phosphoglycerate dehydrogenase (PHGDH) catalyzes the first step of NAD
+-dependent oxidation of 3-PG to 3-phosphohydroxypyruvate (3-PHP); subsequently, phosphoserine aminotransferase 1 (PSAT1) converts 3-PHP into 3-phosphoserine (3-PS) in a glutamate-dependent transamination reaction; and finally, serine is then generated from phosphoserine through phosphoserine phosphatase (PSPH)-mediated dephosphorylation. Recent studies have shown that the inhibition of serine biosynthesis and/or exogenous serine supply significantly inhibited the growth of some cancer cells in vitro and in vivo [
13,
19], indicating the dependency of rapid proliferating cells on serine. An increased serine biosynthesis is observed in many types of cancer, and has been shown to be required for tumor development. Indeed, PHGDH is frequently overexpressed in estrogen receptor-negative breast cancer, lung adenocarcinoma, and melanoma [
9,
10,
20]. Loss of PHGDH or direct blockage of serine biosynthesis via the pharmacological inhibition of PHGDH can inhibit cancer cell proliferation and attenuate tumor growth and metastasis [
10,
11,
21,
22]. Hence, serine biosynthesis has emerged as a promising target for therapeutic intervention against cancer.
Differences in amino acid concentrations in the brain microenvironment relative to plasma have been observed. With the exception of glutamine, most amino acid levels are dramatically lower (maximum 100-fold lower) in both the brain interstitial fluid (ISF) and cerebrospinal fluid (CSF), which buffers the brain, than in plasma [
21,
23]. In particular, both serine and glycine are among the most depleted amino acids in the brain ISF and the CSF, indicating that the brain is a serine- and glycine-limited microenvironment. Given the brain microenvironment, the growing glioblastoma (GBM) cells, which is a highly malignant primary adult brain tumor with dismal prognosis [
24], encounter the limited exogenous serine and glycine, and the cells have to adapt their metabolism to survive and proliferate. However, how GBM cells reprogram their metabolic flexibility to adapt to and overcome these environmental stress conditions for their survival and continuous growth remains unknown.
In this study, we demonstrated that glycolytic metabolism and serine biosynthesis are mainly reprogrammed and activated in GBM cells in response to limited serine and glycine conditions. ROS-AMPK-HIF-1α signaling, activated by serine and glycine deprivation, induces the expression of genes for glucose uptake, glycolysis, and serine synthesis pathway, leading to enforced glucose-derived de novo serine and glycine biosynthesis, which is required for the proliferation and survival of GBM cells and brain tumor growth.
Materials and methods
Materials
Rabbit monoclonal antibodies for PHGDH (#66350S, 1:1,000 for immunoblotting, 1:100 for immunohistochemistry), AMPK (#5831S, 1:1,000 for immunoblotting), and p-AMPK (T172, #2535S, 1:1,000 for immunoblotting, 1:100 for immunohistochemistry) were purchased from Cell Signaling Technology (Danvers, MA). Rabbit monoclonal antibodies for PSAT1 (#ab96136, 1:1,000 for immunoblotting, 1:100 for immunohistochemistry) and HIF-1α (#ab51608, 1:1,000 for immunoblotting, 1:100 for immunohistochemistry) were purchased from Abcam (Cambridge, MA). Rabbit monoclonal antibody for PSPH (#13,503-R001, 1:1,000 for immunoblotting, 1:100 for immunohistochemistry) was purchased from Sino Biological (China). Mouse monoclonal antibody for tubulin (#T4026-100UL, 1:1,000 for immunoblotting) was purchased from Sigma-Aldrich (St. Louis, MO). Mouse monoclonal antibodies for lamin B1 (#sc-374015, 1:1,000 for immunoblotting) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). PX-478 (#S7612) was purchased from Selleck Chemicals (Houston, TX). Cycloheximide (CHX; S1988), Compound C (#171,260), 2-Deoxy-D-glucose (2-DG) (#D6134), Cobalt (II) chloride (CoCl2) (#60,818), L-Serine (#S4311), Glycine (#G8790), and Sarcosine (glycine transporter I inhibitor, #131,776) were purchased from Sigma Aldrich (St. Louis, MO). 4-Fluoro-L-2-phenylglycine (D-Serine transport inhibitor, #F0862) was purchased from Tokyo Chemical Industry (Japan). [U-13C6] glucose (#110,187–42-3) was purchased from Cambridge Isotope Laboratory (Tewksbury, MA).
Cell culture
The U87MG and U373MG GBM cells were purchased from the Korean Cell Line Bank (KCLB; Seoul, Republic of Korea). Normal human astrocytes (NHA) and GBM cells, including LN18, T98G, A172, and LN229, were kindly provided by Dr. Hyunggee Kim (Korea University, Seoul, Republic of Korea). The U251 GBM cells were kindly provided by Dr. In Ah Kim (Seoul National University, Seoul, Republic of Korea). All cells were authenticated and routinely tested for mycoplasma. NHA were maintained in Astrocyte medium (#1801, Scien-Cell, Carlsbad, CA). GBM cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (#LM001-05, Welgene, Republic of Korea) supplemented with 10% fetal bovine serum (#S001-04, Welgene, Republic of Korea) and 1% Penicillin/Streptomycin (#PS-B, Capricorn Scientific, Germany). For all serine and glycine-deprivation experiments, cells were cultured in customized serine/glycine-deprived medium (#LM001-229, Welgene, Republic of Korea) supplemented with or without dialyzed fetal bovine serum (#26,400,044, Gibco, Gaithersburg, MD). Glioma stem cells (GSCs) originally isolated from human GBM specimens of patients undergoing surgery [
25,
26] have been studied by several other groups [
27,
28]. They were kindly provided by Dr. Yong Tae Kwon’s group (Seoul National University, Seoul, Republic of Korea). GSCs were maintained in Neurobasal Plus Medium (#A3582901, ThermoFisher; Pittsburgh, PA) supplemented with 2% B-27 (minus vitamin A; #12,857,010, ThermoFisher, Pittsburgh, PA), EGF (20 ng/mL) (#AF100-15, PeproTech, Seoul, Republic of Korea), and FGF (20 ng/mL) (#100-18B, PeproTech, Seoul, Republic of Korea).
RNA sequencing
The 5 × 106 U87MG cells were seeded in 10 cm dish. After culturing serine/glycine-containing complete media or equivalent media lacking serine/glycine for 24 h, total RNA was extracted. RNA sequencing and library construction were performed according to the QuantSeq 3’ mRNA-Seq Library Prep Kit (#015.2 × 96, Lexogen, Inc., Austria).
Gene Ontology (GO) and pathway analyses
Enrichr was utilized for functional enrichment analysis of GO annotations database [
29], while PANTHER Pathway was used for ontology-based pathway analysis to identify significant biological pathways [
30]. Ingenuity Pathway Analysis (IPA) bioinformatics software (Qiagen, Valencia, CA, USA) was employed for network analysis to investigate the relationship between the differentially expressed genes (DEGs) and biological functions [
31]. A 1.3-fold change with
P-value less than 0.05 for the differences in gene expression was used as the cut-off value for determining significant changes in gene expression levels.
Quantitative real-time PCR analysis
Total RNA was isolated with TRIsure reagent (#BIO-38033, Bioline, UK), according to the manufacturer’s instructions. Equal amount of total RNA was then used for cDNA synthesis using PrimeScript™ RT Master Mix (Perfect Real Time) (#RR036A, Takara, Japan). Real-time PCR was performed on an ABI Prism 7500 sequence detection system using a SYBR® Green PCR Master Mix (#43–091-55, Applied Biosystems; Foster City, CA), following the manufacturer’s protocols. The ABI 7500 sequence detector was programmed with the following PCR conditions: 40 cycles of 15 s denaturation at 95°C, and 1 min amplification at 60°C. The relative differences of PCR results was evaluated using the comparative cycle threshold (CT) method [
32]. All reactions were run in triplicate, and normalized to the housekeeping gene
GAPDH. The following primer pairs were used for quantitative real time-PCR:
PHGDH, 5’- GCCCTTACCAGTGCCTTCTC -3’ (forward) and 5’-GACAATGACTGCGGGGCTTA -3’ (reverse);
PSAT1, 5’-GGCCAGTTCAGTGCTGTCC -3’ (forward) and 5’- GCTCCTGTCACCACATAGTCA -3’ (reverse);
PSPH, 5’-AAATCTGTGGCGTTGAGGAC-3’ (forward) and 5’-ACCTGAACATTTCGCTCCTG-3’ (reverse);
GLUT1, 5’-TCAACACGGCCTTCACTG-3’ (forward) and 5’-CACGATGCTCAGATAGGACATC-3’ (reverse);
GLUT3, 5’- TCCCCTCCGCTGCTCACTATTT-3’ (forward) and 5’-ATCTCCATGACGCCGTCCTTTC-3’ (reverse);
HK2, 5’-AACAGCCTGGACGAGAGCATC-3’ (forward) and 5’-AGGTCAAACTCCTCTCGCCG-3’ (reverse);
PFKFB2, 5’-AGTCCTACGACTTCTTTCGGC-3’ (forward) and 5’-TCTCCTCAGTGAGATACGCCT-3’ (reverse);
HIF-1α, 5’-CATAAAGTCTGCAACATGGAAGGT-3’ (forward) and 5’-ATTTGATGGGTGAGGAATGGGTT-3’ (reverse);
GAPDH, 5’-GCATCTTCTTTTGCGTCG-3’ (forward) and 5’-TGTAAACCATTGTAGTTGAGGT-3’ (reverse).
Immunoblot analysis
Extraction of proteins from cultured cells was performed using a lysis buffer (50 mM Tris–HCl [pH 7.5], 0.1% sodium dodecyl sulfate (SDS), 1% Triton X-100, 150 mM NaCl, 1 mM DTT, 0.5 mM EDTA, 100 µM sodium orthovanadate, 100 µM sodium pyrophosphate, 1 mM sodium fluoride, and proteinase inhibitor cocktail). The cell extracts were centrifuged at 15,000 rpm (at 4℃ for 15 min), and protein concentrations of cell lysates were determined using the DC protein assay Kit (#5,000,112, Bio-Rad, Hercules, CA). Equal amounts of lysates were resolved by SDS- polyacrylamide gel electrophoresis (PAGE), and were then transferred to a nitrocellulose membrane. The membrane was blocked with 5% skim milk in TBST at room temperature (RT) for 30 min, and next incubated with the indicated antibodies at 4℃ overnight. The blots were then incubated with horseradish peroxidase-conjugated secondary antibodies (anti-rabbit (#NA934V, Sigma Aldrich, St. Louis, MO) or anti-mouse (#NA931V, Sigma Aldrich, MO) at RT for 2 h. Band intensity was quantified using ImageJ 1.53e software (National Institutes of Health). Each experiment was repeated at least three times. Figure S
7 of the SI displays full scans of the original immunoblots.
Nuclear fractionation
Lysis of membrane from cultured cells was performed using a 0.1% NP-40. The cell suspension was centrifuged at 1,000 × g (at 4℃ for 5 min). Extraction of nuclear contents from pellet was performed using a radioimmunoprecipitation assay (RIPA) buffer (25 mM Tris–HCl [pH 7.4], 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% NP-40, proteinase inhibitor cocktail). Nuclear concentrations of cell lysates were determined using the DC protein assay Kit (#5,000,112, Bio-Rad, Hercules, CA). Nuclear and cytosolic proteins were used in immunoblot analyses. Each experiment was repeated at least three times.
Measurements of glucose consumption, and ATP and AMP levels
The culture medium was collected for measurement of glucose concentrations [
33], and the remained cells were measured for adenosine triphosphate (ATP) or adenosine monophosphate (AMP) levels. Glucose levels were determined using a Glucose (GO) Assay Kit (#GAGO20, Sigma, St. Louis, MO). Glucose consumption was the difference in glucose concentration from the collected culture medium between experimental conditions. Absorbance was recorded at 540 nm at RT in a 96-well plate. All results were normalized to the final cell number. The ATP or AMP levels were assessed using an ATP Colorimetric/Fluorometric Assay Kit (#354, BioVision, Milpitas, CA) or an AMP Colorimetric Assay Kit, respectively (#229, BioVision, Milpitas, CA). The reaction was performed using a cell lysate in 100 µL of reaction buffer, which was prepared according to the corresponding assay kit instructions. Absorbance was recorded at 570 nm at RT in a 96-well plate.
Measurements of intracellular serine and glycine levels
Serine or glycine levels were assessed using a DL-Serine Assay Kit or a Glycine Assay Kit (BioVision, Milpitas, CA), respectively. The reaction was performed using a cell lysate in 100 µL of reaction buffer, which was prepared according to the corresponding assay kit instructions, and the fluorescence was measured at Ex/Em = 535/587 nm in endpoint mode.
Measurement of the NADPH/NADP+ ratio and ROS levels
The NADPH/NADP+ ratio was assessed using a NADP/NADPH Quantification Colorimetric Kit (#K347, BioVision, Milpitas, CA). The reaction was performed using a cell lysate in 800 μL of reaction buffer, which was prepared according to the NADP/NADPH Quantification Colorimetric Kit instructions. Absorbance was recorded at 450 nm at RT in a 96-well plate. ROS levels were assessed using a Reactive Oxygen Species (ROS) Detection Assay Kit (#K936, BioVision, Milpitas, CA). The reaction was performed according to the assay kit instructions.
siRNA and shRNA transfection
AMPKα1/2 siRNA (#sc-45312) was purchased from Santa Cruz (Dallas, TX). HIF-1α siRNA (#104,730,345) was purchased from Integrated DNA Technologies (Coralville, IA). The pLKO.1-puro lentiviral vectors encoding shRNA targeting GFP, PHGDH (#TRCN0000028532), PSAT1 (#TRCN0000035268), PSPH (#TRCN000002795), or HIF-1α were purchased from Sigma Aldrich (St. Louis, MO). Cells were plated at a density of 4 × 105 per 60 mm dish 18 h before transfection. Transfection of siRNAs was performed using Lipofectamine™ RNAiMAX transfection reagent (#13,778,150, ThermoFisher, Pittsburgh, PA), according to the manufacturer’s instructions. Transfection of shRNAs was performed using Lipofectamine2000 transfection reagent (#11,668,027, ThermoFisher, Pittsburgh, PA), according to the manufacturer’s instructions.
Luciferase reporter assay
The tumor cells were co-transfected with a pGL3 empty vector or a pGL3-HRE-luciferase plasmid containing five copies of HREs (as an inner control that contained Renilla luciferase sequences (#E2231, Promega, Madison, WI)) using Lipofectamine
2000 transfection reagent according to the manufacturer’s instructions, and then grown under different experimental conditions. After incubation, firefly and Renilla luciferase activities were measured using a Dual-Luciferase® Reporter Assay System (#E1910, Promega; Madison, WI), and the ratio of firefly/Renilla luciferase was determined [
34,
35].
Chromatin immunoprecipitation (ChIP) assay
A ChIP assay was performed using a SimpleChIP Enzymatic Chromatin IP kit (#9003S, Cell Signaling Technology, Danvers, MS) [
36]. Chromatin prepared from 2 × 10
6 cells (in a 10 cm dish) was used to determine the total DNA input, and then incubated overnight with PHGDH, PSAT1, or PSPH antibodies, or with normal mouse IgG, at 4℃ overnight. Immunoprecipitated chromatin was detected using real-time PCR. The PCR primer sequences for the
HIF-1α promoter were
PHGDH, 5’-GCTTCTGATTCTAGGTGACTT-3’ (forward) and 5’-ACGGGATGTCAGTGTGGTTTA-3’ (reverse);
PSAT1, 5’-GGAGAATCAGCGACTTTAAAGG-3’ (forward) and 5’-TTGGAAGCGCAGGATGAAGAA-3’ (reverse),
PSPH, 5’-GCACTCAGCATCGTTTCCTTT-3’ (forward) and 5’-TACATCTTCATGGTGCCCTTG-3’ (reverse).
Two million U87MG cells were seeded in 10 cm plates in triplicate and cultured in experimental medium with [U-13C6]-glucose for 12 h. Samples were thawed in an ice bath to reduce sample degradation before processing, and then 400 μL of 80% methanol solution was added to each tube of cell samples, and the cells sonicated. The samples were centrifuged at 18,000 × g for 15 min at 4°C, and the supernatant was then collected. Samples were reconstituted in 80% methanol solution, awaiting LC–MS analysis. The metabolites were analyzed by ultrahigh-pressure liquid chromatography–triple quadrupole mass spectrometry (ACQUITY UPLC-Xevo TQ-S, Waters Corp., Milford, MA, USA). For data processing, the raw data files generated by UPLC-MS/MS were processed using MassLynx software (v 4.1, Waters Corp., Milford, MA, USA) for the peak extraction, integration, identification, and quantification of each metabolite. R language (v4.1.1) was used for subsequent statistical analysis.
Cell proliferation assay
A total of 4 × 104 cells were plated and counted at day (d) (2, 4, and 6) after seeding in DMEM supplemented with 10% fetal bovine serum, or customized serine/glycine-deprived DMEM supplemented with 10% dialyzed fetal bovine serum. The cells were trypsinized and counted using a hemocytometer.
Annexin V − FITC staining
Apoptosis was determined using an Annexin V staining. Cells were seeded in 12-well plate, stained with Annexin V − FITC (#K101, Biovision, Milpitas, CA) to the manufacturer’s instruction, and then analyzed under a fluorescence microscopy (Nikon, Japan).
Intracranial implantation of GSCs in mice and histologic evaluation
We injected GSCs (XO6) with or without modulation of PHGDH, PSAT1, or PSPH expression, intracranially into 4-week-old male athymic nude mice (five mice/group), as described previously [
37]. The mice were euthanized 21 d after the GSCs were injected. The brain of each mouse was harvested, fixed in 4% formaldehyde in PBS, and embedded in paraffin. After that, histological Sects. (5 μm) were prepared. The sections were stained with Mayer’s hematoxylin (#HK100, Biogenex Laboratories, San Ramon, CA), and subsequently with eosin (H&E) (Biogenex Laboratories, San Ramon, CA). Afterward, the slides were mounted with Universal Mount (Research Genetics Huntsville, AL). Tumor formation and phenotype were determined by histological analysis of H&E-stained sections. Tumor volume was calculated by the formula 0.5 × L × W
2 (L, length; W, width). All the mice were housed in the Dong-A University animal facility, and all experiments were performed in accordance with relevant institutional and national guidelines. All animal procedures and maintenance conditions were approved by the Dong-A University Institutional Animal Care and Use Committee.
Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay of intracranial tumors
Apoptotic cells were determined by the DeadEnd™ Colorimetric TUNEL System (#G7360, Promega, Madison, WI), according to the manufacturer’s instruction. Briefly, the tumor sections described above were treated with the equilibration buffer, and incubated for 10 min, followed by 10 min incubation in 20 µg/mL proteinase K solution. The sections were washed in PBS, and incubated with TdT enzyme at 37°C for 1 h in a humidified chamber to incorporate biotinylated nucleotides at the 3′-OH ends of DNA. The slides were incubated in horseradish peroxidase-labeled streptavidin to bind the biotinylated nucleotides, followed by detection with a chromogen 3,3′-diaminobenzidine (DAB) (DAB Substrate Kit, #SK-4100; Vector Laboratories).
Immunohistochemical (IHC) analysis and scoring
IHC analyses were conducted using paraffin-embedded tissue sections. The expression of PHGDH, PSAT1, PSPH, and Ki-67 was detected with a VECTASTAIN Elite ABC kit (#PK-6200, Vector Laboratories, Burlingame, CA); tissue sections were then incubated with 3,3′-diaminobenzidine (#SK-4100, Vector Laboratories, Burlingame, CA), and the nuclei were stained with hematoxylin. Six randomly chosen fields per slide were analyzed and averaged.
The human GBM samples and clinical information were from the First Affiliated Hospital, Zhejiang University School of Medicine, China. This study was approved by the Ethics Committee of Zhejiang University School of Medicine (China), and written informed consents were obtained from all patients. The tissue sections from 50 paraffin-embedded human GBM specimens were stained with antibodies against p-AMPK (T172), HIF-1α, PHGDH, PSAT1, PSPH or non-specific immunoglobulin as a negative control. We quantitatively scored the tissue sections according to the percentage of positive cells and staining intensity, as previously defined [
37]. We assigned the following proportion scores: (0, 1, 2, 3, 4, or 5) if (0, (0.1 to 1), (1.1 to 10), (11 to 30), (31 to 70), or (71 to 100)) %, respectively, of the tumor cells showing positive staining. We rated the intensity of staining on a scale of (0 to 3): 0, negative; 1, weak; 2, moderate; and 3, strong. We then combined the proportion and intensity scores to obtain a total score (range, (0 − 8)), as previously described [
37].
Genomic data analysis
Genomic datasets were downloaded from The Gene Expression Omnibus (GEO) database (
https://www.ncbi.nlm.nih.gov/geo) and processed using R-package. Correlation analysis between two genes was performed with
Pearson’s correlation analysis.
P-value indicates the significance of correlation.
Tissue microarray analysis
A paraffin-embedded GBM tissue microarray was obtained from US Biomax (#GL722a, Rockville, MD), and all tissues are collected under the highest ethical standards with the donor being completely informed, and with their written consent. IHC analyses of PHGDH, PSAT1, and PSPH expression were performed according to the protocol described above.
Statistical analysis
All quantitative data are presented as the mean ± SD of at least three independent experiments. A 2-group comparison was conducted using the 2-sided, 2-sample Student’s t-test. A simultaneous comparison of > 2 groups was conducted using one-way ANOVA, followed by Tukey’s post hoc tests. The SPSS statistical package (version 12; SPSS Inc., Chicago, IL) was used for the analyses. Values of P < 0.05 were considered to indicate statistically significant differences.
Discussion
Cancer cells acquire distinct metabolic reprogramming to survive diverse stress conditions and to satisfy the anabolic demands of rapid proliferation for tumor growth and metastasis [
6‐
8]. Thus, during tumor development, cancer metabolic reprogramming-based flexibility is commonly observed. Here, we investigated the mechanisms of how GBM cells achieve their distinct rewired metabolic features to contribute to survival and proliferation within the brain microenvironment. We revealed that GBM exhibited metabolic alterations in enforced glucose-derived de novo serine biosynthesis to promote brain tumor growth under the brain microenvironment of limited serine and glycine. ROS-mediated AMPK activation induced HIF-1α expression and transactivation to simultaneously increase the transcriptional expression of the three SSP genes as well as the expression of genes for glucose transporters and glycolytic process, thereby enhancing glucose uptake, glycolytic flux, and de novo serine and glycine biosynthesis to promote the survival and proliferation of GBM cells and brain tumor growth under the brain microenvironment of restricted serine/glycine (Fig.
7D and E). Our findings highlight that under the limited serine and glycine conditions, GBM cells have an increased dependency on glucose-derived carbon for the biosynthesis of serine and glycine in an AMPK/HIF-1α activation-dependent manner.
Serine catabolism is essential to produce one-carbon units for the one-carbon metabolism needed for cancer cell survival and proliferation [
17,
18]; thus, recent attention has focused on the role and regulation of serine metabolism in supporting tumor development [
9‐
15,
21,
46]. Although cancer cells can uptake serine from the extracellular environment, many studies have shown that de novo serine synthesis is activated by several factors in many types of cancer [
14,
19,
47‐
49] and thus most cancer cells can adapt to and become resistant to the deficiency of exogenous serine by upregulating the SSP flux. In particular, the growing GBM cells rely on the de novo-synthesized serine to survive and proliferate in the brain microenvironment of restricted serine/glycine. Indeed, we found that the levels of the three SSP enzymes were overexpressed in human GBM specimens and GBM cells, and SSP activation was upregulated in response to the deprivation of serine/glycine. In accordance, our in vitro and in vivo experiments clearly showed that the genetic downregulation of SSP genes suppressed GBM survival and proliferation and brain tumor growth within the limited serine/glycine conditions, implying that serine biosynthesis is a crucial metabolic process for GBM growth. In line with our results, Ngo et al. reported that brain-metastasized tumors exhibited an increase in the expression of PHGDH and PSAT1 genes and serine synthesis relative to extracranial tumors, thereby potentiating brain metastasis [
21]. Our and other studies indicate that de novo serine biosynthesis is hyperactivated in the brain microenvironment of restricted serine/glycine, which plays roles in both brain tumor growth and brain metastasis. Thus, in future studies, attempting therapeutic targeting of SSP enzymes in tumors growing in the brain microenvironment will be worth exploring.
It is well known that glucose is the major source for the biosynthesis of serine and glycine. Nevertheless, it is quite unclear how glucose-derived serine biosynthesis is comprehensively coordinated to support cellular survival and proliferation under stressed conditions, including serine/glycine deficiency. HIF-1α plays a critical role in glycolytic metabolism through upregulating glycolysis-related enzymes in cancer cells [
39]. A highly active glucose metabolism [
50] and overexpression of HIF-1α [
51] have been observed in GBM, and high levels of HIF-1α are associated with advanced cancer progression and poor clinical outcomes in GBM [
51,
52]. Thus, HIF-1α can contribute to supplying glucose-derived precursor 3-phosphoglycerate for de novo serine biosynthesis in GBM cells. Our transcriptomic data showed that the genes for SSP, glucose transporters, and glycolytic process are highly upregulated in response to serine/glycine deprivation. We demonstrate here that activated transcription factor HIF-1α simultaneously controls the expression of GLUT1, GLUT3, HK2, PFKFB2, PHGDH, PSAT1, and PSPH, and then enhanced glucose uptake, glycolytic flux, and de novo serine biosynthesis under the limited serine/glycine conditions. These findings highlight that HIF-1α acts as a master metabolic coordinator of glucose-derived serine biosynthesis in GBM cells.
In the present study, we found that serine/glycine-deficient conditions enhance HIF-1α stability and transactivation in GBM cells. It has been well reported that both the protein expression and the activity of HIF-1α are regulated by O
2-dependent pathways [
39]. However, recent studies showed that HIF-1α is subject to O
2-independent regulations. Interplays between AMPK and HIF-1α have been reported that increased AMPK expression and activity are paralleled by the upregulation of HIF-1α in cancer cells [
53]. AMPK has been proven to positively regulate HIF-1α protein stability and function. Indeed, the inhibition of AMPK attenuated HIF-1α target genes expression via impairing the expression and nuclear accumulation of HIF-1α in an O
2-independent pathway manner under hypoxia or low glucose conditions [
43‐
45]. Consistent with the results, serine/glycine deprivation-activated AMPK was required for HIF-1α stability and transactivation in GBM cells (Fig.
5). AMPK signaling was found to be hyperactivated in human GBM specimens compared with normal brain, which supports tumor bioenergetics, growth, and survival in GBM [
43], indicating the significance of AMPK activity in GBM. Here, we demonstrate that AMPK activation in human GBM cells resulted from increased ROS, not from change in ATP/AMP ratio under serine/glycine-defective brain microenvironmental conditions. Collectively, through comprehensive molecular and pharmacological approaches, we here show a mechanism by which AMPK induces HIF-1α stability and transactivation, resulting in enhanced de novo serine biosynthesis to support GBM proliferation and survival under serine/glycine deprivation conditions.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.