HIF1α/ATF3 partake in PGK1 K191/K192 succinylation by modulating P4HA1/succinate signaling in glioblastoma

Abstract

Background

Hypoxia is a pathological hallmark in most cancers, including glioblastoma (GBM). Hypoxic signaling activation and post-translational modification (PTM) of oncogenic proteins are well-studied in cancers. Accumulating studies indicate glycolytic enzyme PGK1 plays a crucial role in tumorigenesis, yet the underlying mechanisms remain unknown.

Methods

We first used ChIP assays to uncover the crosstalk between HIF1α and ATF3 and their roles in P4HA1 regulation. Protein degradation analysis, LC-MS/MS, and in vitro succinate production assays were performed to examine the effect of protein succinylation on GBM pathology. Seahorse assay measured the effects of PGK1 succinylation at K191/K192 or its mutants on glucose metabolism. We utilized an in vivo intracranial mouse model for biochemical studies to elucidate the impact of ATF3 and P4HA1 on aerobic glycolysis and the tumor immune microenvironment.

Results

We demonstrated that HIF1α and ATF3 positively and negatively regulate the transcription of P4HA1, respectively, leading to an increased succinate production and increased activation of HIF1α signaling. P4HA1 expression elevated the succinate concentration, resulting in the enhanced succinylation of PGK1 at the K191 and K192 sites. Inhibition of proteasomal degradation of PGK1 by succinylation significantly increased aerobic glycolysis to generate lactate. Furthermore, ATF3 overexpression and P4HA1 knockdown reduced succinate and lactate levels in GBM cells, inhibiting immune responses and tumor growth.

Conclusions

Together, our study demonstrates that HIF1α/ATF3 participated in P4HA1/succinate signaling, which is the major regulator of succinate biosynthesis and PGK1 succinylation at K191 and K192 sites in GBM. The P4HA1/succinate pathway might be a novel and promising target for aerobic glycolysis in GBM.

Graphical Abstract

Open in new tabDownload slide

Glioblastoma (GBM), a lethal primary brain malignancy, alters cancer cells’ metabolic reprogramming, promoting tumorigenesis and cancer progression.¹ Tumor cells often exploit the hypoxic condition in the tumor microenvironment (TME) triggering a switch in energy metabolism from mitochondrial oxidative phosphorylation (OXPHOS) to glycolysis. Notably, the hypoxia-inducible factor 1 alpha (HIF1α), a master regulator of hypoxic responses, orchestrates this hypoxia-associated metabolic reprogramming in tumor cells.² Importantly, the aerobic glycolysis (Warburg effect) is activated by the HIF1α signaling to support high-energy metabolic processes in cancer cells. Recent studies demonstrate that altered aerobic glycolysis is one of the key determinants of chemoresistance, immune evasion, and metastasis in several cancers.^3–5

Proline 4-hydroxylase (P4H) is an α2β2 tetrameric enzyme that promotes collagen biogenesis by catalyzing proline hydroxylation within collagen and converting α-ketoglutarate (α-KG) into succinate.⁶ The P4H α subunit (P4HA1) is a predominant peptide binding and catalytically active domain of this dioxygenase,⁷ which is located in the endoplasmic reticulum (ER), vesicles, and mitochondria. During hypoxia, HIF1α induces P4HA1 expression, thereby enhancing the extracellular matrix (ECM) remodeling in the TME.⁸ On the other hand, P4HA1 inhibits proline hydroxylation on HIF1α to adjust oncometabolite production, as well as inhibit HIF1α ubiquitination and degradation.⁹ The phenomenon of oncometabolite (such as succinate, fumarate, 2-hydroxyglutarate (2HG), and lactate)-mediated activation of the HIF1α signaling is called pseudohypoxia.¹⁰ It has been confirmed that altered succinate level polarizes normal macrophages into tumor-associated macrophages (TAM),¹¹ remodels the epigenome and modulates the gene expression.¹² Furthermore, succinate can post-translationally modify proteins.¹³ These findings suggest that P4HA1 regulates tumors through multiple pathways, however, the underlying mechanisms remain enigmatic.

The glycolytic enzyme phosphoglycerate kinase 1 (PGK1) participates in the first step of ATP synthesis during glycolysis and maintains energy homeostasis in tumor cells.¹⁴ Additionally, PGK1 acts as the protein kinase modulating mitochondrial translocation and autophagy.^15,16 PGK1 activity is regulated by diverse protein modifications. For example, PGK1 acetylation at K323 enhances its enzymatic activity, promoting liver cancer carcinogenesis,¹⁷ while its phosphorylation at S203 or Y324 supports the glycolytic pathway in brain tumors.^15,18 Moreover, PGK1 acetylation at K388 enhances its kinase activity toward Beclin1 to initiate glutamine deprivation-induced autophagy.¹⁶ Further, PGK1 O-GlcNAcylation at T255 enhances lactate production promoting colon tumorigenesis.¹⁹ Post-translational modifications of PGK1 play 2 major roles in carcinogenesis. First, this modification of PGK1 is crucial for regulating pathogenic mechanisms in the hypoxic TME. Second, post-transcriptionally modified PGK1 modulates the metabolic preference for glycolysis in tumor cells. Succinate is required for the protein succinylation at lysine residues. Since P4HA1 regulates succinate accumulation in GBM, we infer that PGK1 most likely undergoes succinylation in glioma cells. However, the physiological implication of PGK1 post-transcriptional modification in GBM cells remains unknown to date.

Here, we report that HIF1α/ATF3 positively and negatively modulate the transcription of P4HA1, respectively. Furthermore, succinate, an oncometabolite generated by P4HA1, has the ability to activate the HIF1α pathway, which in turn suppresses immune responses and promotes tumor growth in GBM. That is, P4HA1/succinate enhances aerobic glycolysis through PGK1 succinylation at K191 and K192 sites in GBM. Mechanistically, succinylation of those 2 lysine residues increases PGK1 stability by inhibiting its ubiquitination-mediated degradation through the ubiquitin-proteasome system. Furthermore, ATF3 unregulation and P4HA1 depletion together effectively decreased the oncometabolite level and consequently inhibited tumor development.

Materials and Methods

Liquid Chromatography–Tandem Mass Spectrometry (LC-MS/MS) Analysis

For data-independent acquisition (DIA) proteomics, which was performed by PROTEINT (Tianjin, China), and the DIA data were compared to the human UniProt database (20 365 sequences) using DIA-NN (v1.8.1) with the default settings, trypsin/P digest rule, high protein and peptide confidential level, and FDR of 0.01. See Supplementary Materials for details.

Chromatin Immunoprecipitation

Chromatin immunoprecipitation (ChIP) assay was performed using the Magna ChIP™ A/G Chromatin Immunoprecipitation kit (Cat# 17-10085). Purified DNA samples were quantified by the qRT-PCR. The primer sequences used are listed in Supplementary Table 4.

Co-Immunoprecipitation (Co-IP)

IP lysis buffer (Beyotime) was used to lyse cells. Then cell lysates were incubated with 40 μL of protein A/G magnetic beads (Bimake) and 5 μg of primary target antibody overnight at 4°C. After washing the samples with IP lysis buffer 5×, co-IP’ed proteins were detected by western blotting analysis.

Seahorse Assay XFe24 Extracellular Flux Analysis

Seahorse extracellular flux analyzer (Seahorse Bioscience) was used to measure the extracellular acidification rate (ECAR) and cellular oxygen-consumption rate (OCR), according to the manufacturer’s instructions.^20,21 Data were analyzed using the Seahorse Wave Controller software, and the results were normalized to the number of cells.

Succinylated Peptide Enrichment and LC-MS/MS Analysis

Whole-cell lysates were prepared in 1.4 mL of chilled IAP buffer, followed by target protein pool down with anti-succinyl-lysine antibody beads (PTMScan® Succinyl-Lysine Motif Kit, Cell Signaling Technology), according to the manufacturer’s instructions. A Q Exactive HF/HF-X MS (Thermo Scientific) connected to an Easy nLC (Thermo Fisher Scientific) system was used for the LC-MS/MS analysis. The MaxQuant software was used for the analysis of MS raw data, and protein motifs were analyzed by MeMe (https://meme-suite.org/meme/). We extracted the amino acid sequences containing post-transcriptionally modified residues, along with 6 upstream and downstream amino acids flanking each modified residue (13 amino acid residues in total). These sequences were used for motif prediction (parameters: width:13, occurrences:20, background: species) in this study. See Supplementary Materials for details.

In Vivo Studies

Orthotopic animal survival studies were performed, as described elsewhere.²⁰ All animal experiments were performed by protocols approved by the Animal Ethics and Welfare Committee (AEWC) of Tianjin Medical University (No. IRB2022-DWFL-069). For detailed methodology, refer to Supplementary Materials.

For detailed methodologies of western blotting (WB), confocal laser scanning microscope (CLSM), succinate/lactate quantification assay, enzyme-linked immunosorbent assay (ELISA), and flow cytometry (FC), see Supplementary Materials.

Datasets

GBM datasets were downloaded from both CGGA (http://www.cgga.org.cn/), TCGA (https://www.cancer.gov/), and UCSC (https://genome.ucsc.edu/).

Statistical Analysis

Statistical analysis was performed using GraphPad Prism 8.0 software. All data are presented as mean ± SEM. The paired Student’s t-test was used to compare between the experimental and control groups. For multiple experimental groups, one-way or two-way analysis of variance (ANOVA) was used. Significance was defined as ^*P < .05, ^**P < .01, ^***P < .001, ^****P < .0001, or n.s. = non-significant.

Results

P4HA1 Requires HIF1α for its Transcriptional Activation in GBM

Compared with the control group, the World Health Organization (WHO) grade IV glioma group exhibited a higher expression of HIF1α (Supplementary Figure 1A), implying that hypoxic stress could be responsible for the GBM progression. To investigate the influence of hypoxia on GBM pathogenesis, MS-based proteomic analysis was performed using glioma cells under either hypoxic (1% O₂) and normoxia (21% O₂) conditions. All protein expressions are presented in the heatmap (Supplementary Figure 1B). In total, 5 320 proteins were identified as differentially regulated in the GBM group, of which 27 proteins were upregulated and 86 proteins were downregulated (Figure 1A). To better understand associated biological processes, we performed Gene Ontology (GO) enrichment analysis for biological functions between the hypoxia and normoxia groups, showing that RNA processing was significantly affected in response to hypoxic shocks, which was in agreement with the previous finding (Supplementary Figure 1C). The 27 upregulated proteins are shown in Figure 1B. P4HA1 is shown to facilitate GBM cell migration and invasion by triggering the epithelial-to-mesenchymal transition (EMT) during hypoxia.²² P4HA1 showed the highest expression level in the WHO grade IV glioma group (Supplementary Figure 1D), which was consistent with that of HIF1α. Our findings also indicated a correlation relationship between the HIF1α and P4HA1 expressions (Supplementary Figure 1E). Compared to the unstressed condition, hypoxic stress for 48 h significantly increased P4HA1 mRNA levels to nearly 8-fold in the human GBM cell lines U87MG and TBD0220 and the nearly 3-fold murine GBM cell lines CT2A. (Figure 1C, Supplementary Figure 1G). Since HIF1α is a transcription factor, we identified potential binding sites of HIF1α on the P4HA1 promoter region (Supplementary Figure 1F) using ChIP and qRT-PCR, demonstrating hypoxia-induced enhanced binding of HIF1α at P4HA1 promoter in GBM cells (Figure 1D–E, Supplementary Figure 1H). Next, we performed a co-IP assay to examine the interaction between HIF1α and its co-factor p300/CBP in GBM cells under hypoxia, revealing a significant increase in interaction between these 2 factors, which might mediate HIF1α’s binding to hypoxia-response elements (HREs) in the P4HA1 promotor (Figure 1F). Collectively, these results suggest that hypoxia-induced HIF1α activation regulates the transcriptional activation of P4HA1 in GBM.

ATF3 Inhibits P4HA1 Overexpression Under Hypoxia in GBM

Previously, we have shown that activating transcription factor 3 (ATF3) acts as a transcriptional repressor for NF-κB and succinate dehydrogenase A (SDHA), thereby modulating the expression of its downstream target genes.^23,24 Even when ATF3 is expressed basally, at a very low level, it binds a significant portion of active enhancers that are characterized by p300 binding.²⁵ Moreover, there exists potential binding sites of ATF3 on the P4HA1 promoter region (Supplementary Figure 1F). Therefore, we speculated that increased expression of ATF3 might affect HIF1α-mediated transcriptional regulation under hypoxia conditions. To verify this hypothesis, we transduced doxycycline (Dox) inducible lentiviral ATF3 plasmid into U87MG and TBD0220 cell lines. We demonstrated that Dox-induced ATF3 overexpression inhibited HIF1α’s binding to p300/CBP under hypoxic conditions, suggesting that ATF3 might have a higher affinity for p300/CBP in the promoter region of P4HA1 (Figure 1G). Further, qRT-PCR assay showed a significantly reduced P4HA1 mRNA level under ATF3 overexpression (Figure 1H, Supplementary Figure 1I). Also, ChIP analysis confirmed that ATF3 induction could negatively interfere with HIF1α’s regulatory role in P4HA1 expression by enhancing its binding to the P4HA1 promoter in GBM cells (Figure 1I and J, Supplementary Figure 1J and K). Together, our data support that hypoxic GBM cells release cues to induce P4HA1, which could be prevented by inducing ATF3 expression (Figure 1K).

P4HA1 Is Protected From Lysosomal Degradation by PTRF

In GBM cells, polymerase I and transcript release factor (PTRF) are highly expressed and are directly correlated with the patient outcome.²⁶ Interestingly, a comparison of proteomic profiles between the PTRF overexpression (OE) and the PTRF knockdown (KD) groups revealed that P4HA1 level had a linear correlation with PTRF expression in GBM (Figure 2A). Thus, we hypothesized that PTRF might be a critical regulator of P4HA1 level in GBM cells. We also noticed that P4HA1 protein level was very consistent with the degree of PTRF expression in glioma cell lines (Figure 2B, Supplementary Figure 2A). To uncover the underlying mechanism of PTRF-mediated regulation of P4HA1 expression, we first performed qRT-PCR assay showing that PTRF overexpression had no significant effect on P4HA1 mRNA level (Figure 2C, Supplementary Figure 2B). Then, we speculated that PTRF might regulate the P4HA1 protein level by preventing its degradation. Given that autophagy and the ubiquitin-proteasome system (UPS) are the primary proteolytic systems, we sought to identify the 1 that might contribute to the regulation of P4HA1 protein level. These cells were treated with either lysosomal inhibitor chloroquine (CQ) or proteasome inhibitor MG132, which showed that only CQ could elevate the P4HA1 level in eGFP-PTRF-positive cells, and restored P4HA1 levels in PTRF-depleted cells (Figure 2D, Supplementary Figure 2C), suggesting that PTRF could retain the P4HA1 protein level by inhibiting its lysosomal degradation. Subsequent confocal immunofluorescence (IF) analysis also confirmed that P4HA1 colocalized with lysosome-associated membrane protein 1 (LAMP1) in the cytoplasm of GBM cells (Figure 2E). We next examined the stability of P4HA1 using cycloheximide (CHX) chase assay. We found that PTRF upregulation retarded the P4HA1 protein degradation and also increased P4HA1’s half-life in U87MG cells (Figure 2F–G) and TBD0220 cells (Figure 2H and I). PTRF downregulation accelerated the degradation of the P4HA1 protein in CT2A cells (Supplementary Figure 2D and E). Notably, irrespective of PTRF expression in GBM cells, the rate of P4HA1 protein breakdown significantly reduced in CQ-treated cells (Figure 2F–I, Supplementary Figure 2D and E), suggesting that PTRF induction may prevent lysosomal degradation of P4HA1 in glioma cells. To figure out how PTRF regulates P4HA1 in the lysosomal degradation pathway, we characterized key genes that were known to play important roles in the flow of autophagy. We observed that the total protein levels of these genes reduced after PTRF overexpression and increased after PTRF knockdown (Supplementary Figure 2F), which verified that PTRF modulates autophagic flow to regulates P4HA1 protein levels.

The HIF1α Signaling Pathway is Composed of PTRF and P4HA1 and is Inhibited by ATF3

Earlier, we have demonstrated a regulatory axis between HIF1α and PTRF.²⁷ Thus, we confirmed the binding of HIF1α on the PTRF promoter region using the ChIP and qRT-PCR assay in glioma cells (Figure 3A and B, Supplementary Figure 3A). Also, we confirmed that ATF3 could also negatively interfere with HIF1α’s upregulation of PTRF (Figure 3C and D, Supplementary Figure 3B and C). P4HA1 activates the HIF1α pathway by inhibiting its degradation in breast cancer cell lines.⁹ Thus, we transduced glioma cells (U87MG and TBD0220) with P4HA1 by lentivirus exhibiting increased HIF1α protein levels. Importantly, HIF1α ubiquitination was also reduced in P4HA1 overexpressing cells (Figure 3E). shP4HA1#1 was selected as the most effective 1 in CT2A cells, and the P4HA1-depleted CT2A cells showed that HIF1α ubiquitination was also elevated (Supplementary Figure 3D and E). We previously showed that hypoxia could induce upregulation of P4HA1 mRNA level and PTRF could prevent lysosomal degradation of P4HA1 protein. Thus, we performed a series of KD and recovery experiments on HIF1α, PTRF, and P4HA1 to unveil their mechanistic crosstalks. We first confirmed that both the hypoxia and PTRF-OE resulted in the elevated P4HA1 protein level. Compared to control cells, glioma cells with PTRF overexpression exhibited higher levels of HIF1α protein. Hypoxic stress also induced PTRF protein expression in GBM cells. (Figure 3F, Supplementary Figure 3F). Collectively, these results suggest that these 3 factors could be involved in the same regulatory axis. We respectively compared human PTRF-KD and P4HA1 KD efficiencies by indicated siRNAs and selected siPTRF#1 and siP4HA1#2 as the most effective ones (Supplementary Figure 3G). Additionally, the glioma cells with PTRF depletion were exposured to a hypoxic condition, and results showed that compared to the depleted-PTRF group, P4HA1 only slightly improved in the hypoxic group, and the protein level was far inferior to the untreated group (Figure 3G, Supplementary Figure 3H). This suggested that PTRF is necessary for the stability of the P4HA1 protein even in the presence of hypoxic stress. Next, we downregulated P4HA1 in PTRF overexpressing glioma cells to examine whether PTRF maintained HIF1α protein level through P4HA1. We found that, compared to the PTRF overexpression alone group, the P4HA1 KD group had a similar level of HIF1α to the control group (Figure 3H, Supplementary Figure 3H), indicating that PTRF regulated HIF1α through P4HA1 and P4HA1 might be the key regulator for HIF1α pathway activation. Furthermore, we showed by qRT-PCR and WB assays that PTRF and P4HA1 are unable to change ATF3 and p300/CBP transcription and protein levels (Supplementary Figure 3I–L). Collectively, ATF3 could disturb the HIF1α pathway by modulating P4HA1 and PTRF. And P4HA1 contributed to the HIF1α pathway activation (Figure 3I).

P4HA1/Succinate Participates in the Aerobic Glycolysis in Glioma Cells

Emerging studies suggest that HIF1α is the primary transcription factor that promotes Warburg-like metabolism (high glycolysis and low OXPHOS).^2,28 Here, we investigated whether PTRF and P4HA1 played similar roles in GBM metabolism. First, we explored the effect of PTRF and P4HA1 on aerobic glycolysis using seahorse assay and found higher glycolysis rates in PTRF and P4HA1 overexpressing U87MG cells (Supplementary Figure 4A and B), and a pronounced inhibition of glycolysis in PTRF and P4HA1 silenced TBD0220 cells (Figure 4A, Supplementary Figure 4C–E). Expression of PTRF and P4HA1 significantly increased the basal glycolysis rate and glycolytic capacity, while depletion of these factors had a reverse effect on the glycolysis pathway (Supplementary Figures 4B and 4D–E). P4HA1 expression alters the level of succinate and α-KG, thereby reducing proline hydroxylation on HIF1α and improving its stability.⁹ Thus, we performed succinate and α-KG assays to determine whether changes in PTRF and P4HA1 levels affected the amount of succinate and α-KG in glioma cells. The results revealed that the succinic acid content was significantly increased by PTRF or P4HA1 overexpression, and reduced by PTRF or P4HA1 depletion, and the change in α-KG content was in contrast (Figure 4B, Supplementary Figure 4F and G). Next, we performed an ECAR assay by treating U87MG cells with different doses of succinate, finding a gradual increase in the glycolysis rate with an increasing concentration of succinate (Figure 4C), especially basal glycolysis and glycolytic capacity (Supplementary Figure 4H). Likewise, succinate supplementation to PTRF or P4HA1-depleted TBD0220 cells could consistently restore their glycolysis states (Figure 4A, Supplementary Figure 4C–E). Additionally, succinate supplementation to P4HA1-depleted cells reverted the glycolysis status completely. Since the final product of the glycolytic pathway is lactate, monitoring the amount of lactate present within tumor cells can provide insights into cellular glycolytic activities. Consequently, we measured the lactate levels in glioma cells following indicated treatments, demonstrating that increasing levels of PTRF or P4HA1 elevated cellular lactate content, which could also be achieved through succinate supplementation. Furthermore, succinate supplementation could reverse the decrease in cellular lactate content which was increased by PTRF or P4HA1 silencing (Supplementary Figure 4I and J). These results suggest that physiological levels of P4HA1 and succinate are crucial to aerobic glycolysis in GBM cells (Figure 4D).

P4HA1/Succinate Pathway Contributes to Tumor Growth and Inhibit Immune Responses

To explore the antitumor efficacy of the P4HA1 depletion in vivo, an intracranial GBM model was established. Firstly, we performed blood tests, as well as liver and kidney function tests, on mice after 3 weeks of succinate or water administration. H&E staining of mice’s major organs was performed 3 weeks after treatment to assess the biosafety of repeated chemical administrations. The results showed that repeated succinate doses had no significant adverse effects on normal tissues or cells (Supplementary Tables 1–3, Supplementary Figure 4K). we found no discernible difference in body weight between the 2 treatment groups (Supplementary Figure 4L). Succinate supplementation (2 mg/ml) in drinking water was given to shP4HA1 transduced nude mice. Compared to the control group, P4HA1 depletion significantly reduced the tumor growth (Figure 4E and F) and increased survival rate (Figure 4G). However, the succinate supplementation group, in contrast to the P4HA1 group, could reverse the tumor growth inhibition caused by P4HA1 KD (Figure 4E and F), concomitant with an obvious decrease in the survival rate (Figure 4G), thus showing a similar trend to the control group. In the survival time analysis, compared to the mice with a median survival of 30 days in the ctrl group and shP4HA1 with succinate group, shP4HA1 achieved a median survival of 52 days (Figure 4G). We further confirmed these findings through IHC staining using an anti-Ki-67 antibody (Supplementary Figure 4M).

HIF1α enhances tumor-associated signaling and inhibits immune cell function, allowing tumor immune escape.²⁹ Moreover, P4HA1 and the oncometabolites (succinate and lactate) produced by P4HA1 and glycolysis also contributed to tumor growth by reshaping the TME.³⁰ We then established a CT2A intracranial tumor model in C57BL/6J mice to investigate the role of P4HA1 in TME (Figure 4H). Compared with the ctrl group, the P4HA1 downregulated group exhibited a longer survival, while the succinate supplementation further shortened the survival of P4HA1 KD mice (Figure 4I). And, we noticed a great survival benefit (1.5-fold increase in median survival) and decreased tumor burden (H&E) in the shP4HA1 group (Figure 4J). Moreover, the tumor growth inhibition was validated by Ki-67 staining (Figure 4K). Importantly, P4HA1 silencing-mediated tumor growth inhibition could be reverted by succinate supplementation in vivo. We also verified a decrease in P4HA1 expression by IHC staining (Figure 4K). Specifically, IF analysis of CD8α and inducible nitric oxide synthase (iNOS, a marker of M1 macrophages) in tumor tissues showed increased infiltration of inflammatory cells into tumors of mice with P4HA1 KD, while the succinate-supplemented group had no difference compared to the control group (Figure 4L and M). Flow cytometry analysis revealed that P4HA1 silencing further enhanced CD8 + infiltration into tumors (Figure 4N). Additionally, levels of IFN-γ and Granzyme B(GB) in the tumors were evaluated to assess the capability of P4HA1 to influence immune responses. P4HA1 deficiency significantly elevated IFN-γ and GB levels (Figure 4O). However, P4HA1 KD elevated immune responses were reversed by succinate administration, indicating that succinate, as a metabolite, also participated in the tumor cell immune response pathway. Further, we analyzed plasma samples from the 3 groups of mice at 2 weeks post-treatment, which revealed that both succinic acid and lactic acid had reduced levels in the serum of P4HA1 deficient mice compared to those in sham mice. Additionally, succinate supplementation elevated their serum levels (Supplementary Figure 4N). Collectively, the oncogenic pathway of P4HA1/succinate can boost tumor growth and inhibit immune responses in GBM.

Succinylation at Lysine 191 and Lysine 192 Modulates PGK1 Activity

Succinylation is a distinct post-translational modification that utilizes metabolically derived succinyl-CoA to modify lysine residues on the target protein.^31–33 Cytosolic succinate is finally converted back to succinyl-CoA by succinyl-CoA ligase.³⁴ Hence, protein succinylation is stimulated by additional succinate.³³ We also found consistent results in the glioma cells (Figure 5A). Moreover, the intracellular succinyl-CoA level was raised in tandem (Figure 5B). In summary, we hypothesized that P4HA1 might modulate the succinylation modification of lysine residues via regulating succinate synthesis in cancer cells. To identify succinylated lysine residues of glycolytic enzymes, we performed LC-MS/MS analysis of co-IP’ed proteins from TBD0220 cells transduced with lentiviral shP4HA1 or shcontrol plasmid, revealing that levels of succinylated K191 and K192 residues on PGK1 decreased significantly due to P4HA1 KD (Supplementary Figure 5A). The dual-succinylation of PGK1 at K191 and K192 was confirmed by LC-MS/MS analysis (Figure 5C). Notably, amino acids flanking K191/K192 were found to be remarkably conserved across a wide range of species (Figure 5D, Supplementary Figure 5B). To understand the regulatory role of P4HA1 in protein succinylation, we analyzed succinylated-amino acids levels of PGK1 in GBM cells by co-IP, demonstrating that P4HA1 overexpression increased lysine-succinylation on PGK1, P4HA1 knockdown was opposite (Figure 5E). To investigate the role of K191/K192 succinylation on PGK1, we double-mutated K191/K192 to either K191R/K192R(2KR) or K191Q/K192Q(2KQ), which influenced their charges and influenced P4HA1-mediated succinylation. The co-IP results showed that either type of double mutation abolished succinylation level, compared with the wild-type (WT) (Figure 5F). Seahorse assays revealed that both WT and mutant PGK1 could increase glycolytic rate when compared to the PGK1 vector (Figure 5G and H, Supplementary Figure 5C and D). And 2KR or 2KQ both had a stronger capacity for glycolysis than the WT. The lactate production assay revealed a similar trend in these groups (Supplementary Figure 5E). Additionally, the PGK1 mutant (2KR,2KQ) exhibited enhanced mitochondrial respiration in GBM cells, as evidenced by OCR, and ATP production rates (Supplementary Figure 5F and G). Furthermore, the PGK1-WT cells treated with succinate showed an increase in glycolytic rate, while the PGK1 mutant (2KR,2KQ) with succinate supplement exhibited no difference in glycolytic rate (Figure 5I, Supplementary Figure 5H–J). Consistently, glioma cells (Flag-WT, Flag-2KR, Flag-2KQ) treated with succinate all showed a similar lactate production trend as above, and the PGK1-WT with succinate displayed a higher lactate production than mutants (Supplementary Figure 5K). We also conducted protein immunoprecipitation experiments targeting succinylation, showing that the WT group’s succinylation modification level can be effectively raised by adding succinate (Supplementary Figure 5L). However, following supplementation with succinate, the mutant group without the 2 key lysine residues did not attain an increase in succinylation modification level relative to the WT group (Supplementary Figure 5L). Collectively, these data demonstrate that K191 and K192 are crucial succinylation sites on PGK1, and the succinylation at K191 and K192 increases PGK1 activity to support glycolysis.

Succinylation Modification Decreases the Affinity of PGK1 to Ubiquitin

Previous studies show that PGK1 activity is regulated by post-transcriptional modifications, such as phosphorylation at S203, O-GlcNAcylation at T255, and acetylation at K388.^15,16,19 We first identified that succinylation, a bulky modification of the lysine side chain, occurred in PGK1. And ubiquitination modification also happened at the lysine amino acid residues of PGK1.^35,36 Our results support the hypothesis that succinylation competes with ubiquitination at K191/K192 sites on PGK1 and that this retards its UPS-mediated degradation. PGK1 is shown to undergo proteasome-mediated degradation in thyroid cancer cells.³⁷ Consistently, we found higher PGK1 levels in GBM cells treated with MG132 but not with CQ (Figure 5J). As expected, when MG132 was administered, the succinate supplementation increased the PGK1 protein level more noticeably (Figure 5K). Co-IP assays confirmed that succinate supplementation increased the lysine-succinylation modification levels of PGK1 while concurrently lowering the ubiquitin level in PGK1 lysine residues (Supplementary Figure 5M), further inhibiting PGK1 protein degradation by the ubiquitin-proteasome system. In a similar line, CHX-mediated inhibition of protein synthesis resulted in the accelerated degradation of WT PGK1 than mutant variants (2KR,2KQ) in U87MG and TBD0220 cells (Figure 5L). Furthermore, either K to R or K to Q mutations at K191/192 sites dramatically attenuated the PGK1 ubiquitination compared to that of the WT variant (Figure 5M). Therefore, these findings suggest that succinylation at K191/K192 residues increases the stability of PGK1 protein by inhibiting its proteasome-dependent degradation.

ATF3 Upregulation and P4HA1 Depletion Modulate Immunosuppression and Suppress Tumor Growth

After understanding the mechanism by which P4HA1/succinate signaling promotes glucose metabolism, how to more comprehensively inhibit tumor growth caused by P4HA1 is a problem that needs to be resolved. The elevated expression of ATF3 inhibited PTRF and P4HA1 transcription, substantially blocking the progression of the HIF1α pathway. We first identified the induction of ATF3 and the knockdown of P4HA1(Supplementary Figure 6A). The protein level of PGK1 dropped when ATF3 was highly expressed and became more noticeable after the combined knockdown of P4HA1 (Supplementary Figure 6B). The reduction in PGK1 protein levels is due to a decrease in PGK1 succinylation (Figure 6A) and an increase in ubiquitination (Supplementary Figure 6C). Next, a lactate production assay was performed to measure the glycolysis ability, the results demonstrated that lactate production was reduced following the ATF3 overexpression and decreased more significantly in the combination of ATF3 overexpression and P4HA1 depletion in glioma cells (Figure 6B). The difference in rates of glycolysis and glycolytic capacity among the 3 groups concurred with the lactate production data (Figure 6C–F, Supplementary Figure 6D). These results suggest that ATF3 plays a key role in inhibiting aerobic glycolysis, and the combination with P4HA1 silencing further boosts glycolysis inhibition in GBM cells.

Next, we constructed a subcutaneous nude mouse model using the TBD0220 cell line, and we could observe that the mice in the ATF3 induction combined with P4HA1 depletion group had a smaller tumor burden compared to the ATF3 overexpression alone group (Figure 6G and H). In the survival time analysis, compared to the mice with a median survival of 22 days in the ctrl group and 30 days in ATF3 group, ATF3 with shP4HA1 group achieved a median survival of 51.5 days (Figure 6I). IHC experiments on tumor sections showed that the combination group expressed less Ki-67 than ATF3 group and control group (Figure 6J). CT2A intracranial tumor models were established to look into the function of ATF3 in TME. Similarly, ATF3-upregulated groups had a longer survival period than the ctrl group, and this survival period was further prolonged by the combination with shP4HA1(Figure 6K). Furthermore, we observed a significant improvement in survival (a 0.72-fold increase in median survival) and a reduction in tumor burden (H&E) in the combination group (Figure 6L). Furthermore, Ki-67 staining confirmed the inhibition of tumor growth, and we also verified an increase in ATF3 and a decrease in P4HA1 and PGK1 expression by IHC staining (Figure 6M). IF analysis of iNOS and CD8α in tumor tissues showed increased infiltration of inflammatory cells into tumors of mice with ATF3 upregulated, which was lower than the combination group (Figure 6N and O). Additionally, ATF3 upregulation in conjunction with the P4HA1 depletion group demonstrated a higher level of IFN-γ and granzyme B (GB), and ATF3 overexpression increased IFN-γ and GB levels (Figure 6P). Collectively, our results demonstrated that ATF3 overexpression and P4HA1 depletion together effectively decreased the PGK1 succinylation modification level and consequently disrupted tumor development.

Discussion

Despite significant research efforts, GBM, a primary brain malignancy with high invasiveness and resistance, has poor clinical outcomes.³⁸ Here, we showed HIF1α/ATF3 participated in mediating the transcriptional regulation of PTRF and P4HA1. The production of succinate mediated by P4HA1 was significantly inhibited by the upregulation of ATF3, which would lower the total amount of succinylation in cells. Further, due to the decrease in P4HA1-associated succinylation, the protein levels of PGK1 and aerobic glycolysis were decreased in tumors. Ultimately, the absence of the oncometabolites lactate and succinate contributed to the immune response and hampered the growth of the tumor (Figure 6Q). In short, the tumor transcription factor ATF3 inhibits PGK1 succinylation by regulating the P4HA1/succinate signaling pathway, hindering glycolysis and tumor growth.

Hypoxia, resulting in the HIF1α signaling activation, serves as a hallmark in many diseases. However, some emerging studies have unfolded hypoxia-independent mechanisms for the HIF1α signaling activation.^39,40 Even when a sufficient level of oxygen is present, this mechanism can create pseudohypoxic conditions. In this study, we identified that HIF1α regulated the transcription of P4HA1 and PTRF, and showed that PTRF or P4HA1 overexpression resulted in a higher protein level of HIF1α (Figure 3E and F) and a stronger glycolytic rate (Supplementary Figure 4A). Our results indicated that P4HA1-associated elevated succinate level could be responsible for activating the HIF1α signaling in GBM, which was consistent with the findings of a previous study.⁴⁰ Moreover, the active aerobic glycolysis facilitated by PGK1 succinaylation also resulted from P4HA1/succinate signaling (Figure 5I). Thus, the P4HA1/succinate signaling has shed light on the underlying mechanism of pseudohypoxia, advancing our understanding of the Warburg effect under normoxic conditions.

Given the importance of HIF1α in cancer pathobiology, we demonstrated that ATF3 overexpression prevents HIF1α from activating P4HA1 and PTRF transcription (Supplementary Figure 1K, Figure 3D), however, the underlying molecular basis of interaction is unclear. ATF3, a member of the ATF/cAMP responsive element binding (CREB) protein family, recognizes the consensus binding site TGACGTCA.²⁵ CBP and p300 are paralogous multidomain proteins that function as transcriptional coactivators by binding the transactivation domains of transcription factors.⁴¹ The C-terminal activation domain (CAD) of HIF1α binds to the CH1 domain of p300 and the kinase-inducible activation (KIX) domain of CBP forms a complex with the phosphorylated KID-activation domain of CREB.^42,43 Although the KIX domain is structurally unrelated to the CH1 domain, its CREB-binding partner contains 2 amphipathic helices and additional acidic residues (including a phosphorylated serine) that make complementary electrostatic interactions with basic residues in the KIX domain.⁴² Moreover, ATF3 binds a wide range of active enhancers.²⁵ ATF3 interacts with the KIX domain, changing the spatial conformation of CBP/p300, thereby reducing the binding of HIF1α and p300, and limiting the transcription of downstream target genes.

The most common post-translational modifications, including methylation, acetylation, biotinylation, ubiquitination, ubiquitin-like modifications, propionylation, butyrylation, and succinylation, occur on lysine residues that remarkably alter protein functionality and regulatory roles.³³ We found that K191 and K192 of PGK1 underwent succinylation, which increased the half-life of the protein, fueling the glycolysis and lactate synthesis pathways. It’s shown that PGK1 phosphorylation at S203 causes its relocation into mitochondria, where it acts as a protein kinase to phosphorylate and activate pyruvate dehydrogenase kinase 1 (PDHK1) that inhibits mitochondrial pyruvate metabolism and enhance the Warburg effect.¹⁵ When given enough succinic acid, the wild-type showed higher levels of glucose metabolism than mutants (Supplementary Figure 5K). Here, we confirmed that K191 and K192 were the key succinylation modification sites. By altering the net charge on these amino acids, succinylation cloud reduced the affinity of PGK1 toward ubiquitin, and retarded UPS-mediated degradation of PGK1, which in turn influences the glycolysis rate. This altered modification has increased mitochondrial respiration in GBM cells (Supplementary Figure 5F and G). PGK1’s phosphorylation at S203 supports glucose metabolism but inhibits mitochondrial respiration in brain tumors. Given the proximity of K191, K192, and S203 sites, we speculated that this succinylation modification might play a stabilizing role for PGK1 through phosphorylation at the S203 site. Therefore, mutation of these 2 lysine residues significantly affected the stability of the S203 phosphorylation domain, and subsequent kinase activity of PGK1 but enhanced the overall mitochondrial OXPHOS activity. In addition to K191 and K192 sites, MS analysis revealed additional (K131, K133, and K323) residues that had higher susceptibilities toward succinylation (Supplementary Figure 5A). Considering that the K323 residue has been proven to be acetylated in liver cancers, it might also be succinylated to activate the kinase function of PGK1 in GBM. We will further explore the roles of succinylation of PGK1 in more detail in future studies.

The metabolic states in the TME displayed elevated levels of lactate and other oncometabolites but low levels of nutrients. The extracellular space invaded by lactate and succinate secreted by cancer cells reshapes the TME and accelerates tumor growth.³⁰ Because the accumulation of lactate in TME maintains extracellular pH between 6.0 and 6.5, tumor-specific CD8⁺T cells can enter an anergic state, which lowers the cytolytic activity and cytokine production.⁴⁴ At low pH, iNOS expression in M1 macrophages decreases while M2 macrophage markers increase.⁴⁵ Additionally, human CD8⁺T cells exposed to tumor cell-secreted succinate inhibit degranulation and cytokine secretion, including IFN-γ.⁴⁶ And succinate has also been shown to serve as a proinflammatory oncometabolite that accumulates in lipopolysaccharide (LPS)-induced M1 macrophages.⁴⁷ Our research supports this finding as shown in Figure 4, that the depletion of P4HA1 decreased immunosuppression, while supplementation with succinate reversed the antitumor effect. Although some studies indicate that succinate secretion allows CD8⁺ T cells to retain cytotoxicity.⁴⁴ The tumor-derived oncometabolites were shown to modulate the antitumor capacity of immune cells in TME.

In conclusion, our data present evidence for the activation of HIF1α pathway induced by the P4HA1/succinate signaling, which increases aerobic glycolysis. Furthermore, we have identified a novel post-translational modification in glioma cells: the succinylation of PGK1 at K191 and K192 sites. Our research has identified that the combination of ATF3 overexpression and P4HA1 silencing hampered the development of GBM. We believe that this work will further elucidate the key players that contribute to tumor growth and malignancy in GBM and accelerate the discovery of more effective treatments.

Supplementary material

Supplementary material is available online at Neuro-Oncology (https://dbpia.nl.go.kr/neuro-oncology).

Funding

This work was supported by the National Key R&D Program of China, MOST (no. 2023YFC2510000), the National Natural Science Foundation of China (NSFC, no. 82272893), the Key-Area Research and Development Program of Guangdong Province (no. 2023B1111020008), and the Tianjin Medical University General Hospital Clinical Research Program (no. 22ZYYLCZD05).

Conflict of interest statement

None declared.

Authorship statement

Conceived the concept and designed the study: C.S.K. and S.X.Y. Carried out the experiments: S.X.Y., Q.Z., D.Y.S., X.T.C., J.X.Z., Q.X.W., B.H., J.S.J., C.C.C., and E.Y.Y. Analyzed the data, prepared the figures and wrote the manuscript: S.X.Y., Q.Z., and D.Y.S. Financially supported and supervised this work: C.S.K. Approved the final manuscript: All authors.

Data Availability

All data generated or analyzed during this study are included in this article, and all data supporting the findings of this study are available from the corresponding author on reasonable request

References

Author notes