PCK2 as a potential therapeutic target for aggressive MYC-amplified non-WNT/non-SHH medulloblastoma based on tumor continuum Authors: Shuaishuai Xue, Zhengming Zhan, Yonghua Cai, Wei Chen, Ruizhe Chen, Yangqi Mao, Cheng Xie, Jun Liu, Ziyi Luo, Siyuan Liu, Xian Zhang, Peng Li, Hao Long, Ye Song Published: 2026 ABSTRACT AbstractMedulloblastoma, the most common malignant pediatric posterior fossa tumor, exhibits metabolic reprogramming and tumor immune microenvironment heterogeneity in non-WNT/non-SHH subgroups, yet the interplay between these features remains poorly defined. Revealing potential molecular features and hidden therapeutic target, we integrated the transcriptomic data of non-WNT/non-SHH medulloblastoma from GEO and EMBL-EBI databases. Through examination of the Grp3/4 transcriptional continuum, we identified PCK2 (a key gene in the TCA cycle with the highest correlation with continuum score). We validated PCK2 as a driver of proliferation, migration, invasion, glycolysis, and M2 macrophage polarization in MYC-amplified MB cells through short hairpin RNA knockdown experiments. Together, our findings establish metabolic-TIME crosstalk as a prognostic determinant and proffer PCK2 as a therapeutic target for aggressive MB subtypes, offering insights into metabolic subtyping and precision therapy strategies to improve clinical outcomes.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s40478-026-02304-z.Keywords: Medulloblastoma, non-WNT/non-SHH, Metabolism, Immune Microenvironment, PCK2 Introduction Medulloblastoma (MB), a common malignant childhood brain tumor, consists of four molecular groups: WNT-activated, SHH-activated with TP53-wildtype, SHH-activated with TP53-mutant, and non-WNT/non-SHH [1]. The non-WNT/non-SHH group, also known as group 3 and group 4 MB, is particularly aggressive, often metastasizes, and has limited treatment options [2]. Despite advances in molecular subtyping for risk assessment, outcomes remain poor, especially for patients with MYC amplification [3]. Recent studies highlight the importance of metabolic reprogramming in MB, where tumor cells exploit pathways like glycolysis, and lipid synthesis for growth and treatment resistance [4]. There is growing interest in targeting metabolic reprogramming and the tumor immune microenvironment (TIME) in non-WNT/non-SHH MB as potential therapeutic strategies. Metabolic reprogramming in medulloblastoma (MB) not only supports energy production but also drives tumor growth and resistance to therapy [5]. In Group 3 MB, increased glycolysis and mitochondrial activity are associated with MYC-driven cancer, while altered fatty acid metabolism enhances stemness and invasion in metastatic types [6, 7]. Oncometabolites like lactate boost tumor survival and create an immunosuppressive environment by affecting immune cells [8]. The complexity of non-WNT/non-SHH MB remains underexplored, necessitating further investigation into its metabolic and immune characteristics using advanced methods like RNA-seq and scRNA-seq. The TIME in medulloblastoma (MB) shows significant spatial and molecular diversity, influencing clinical outcomes [9]. SHH-subgroup tumors often have immune-rich environments with CD8 + T-cell infiltration, while non-WNT/non-SHH MB are usually “cold”, with M2 macrophages and regulatory T cells [10, 11]. The tumor is a complex mix of neural progenitor, malignant, immune, and stromal cells [12]. Tumor cells compete for nutrients like glucose and arginine, impairing immune cell function [13]. Interestingly, metabolic diversity in non-WNT/non-SHH MB may lead to unique immune features, affecting responses to checkpoint inhibitors [14]. Understanding the interplay between tumor cells and TIME is crucial. We use multi-omics and functional assays to study the metabolic and immune diversity in non-WNT/non-SHH medulloblastoma. A continuum score-based prognostic model was constructed using metagene expression profiles, demonstrating superior predictive power for patient outcomes. Our research provides new insights into the metabolic traits of cancer cells and the complex TIME in non-WNT/non-SHH medulloblastoma, enhancing understanding of MB's heterogeneity and progression. The TCA cycle is crucial for cellular metabolism, acting as a central hub for catabolic and anabolic processes [15]. PCK2 genes are important in the TCA cycle, glycolysis, and gluconeogenesis, but their specific effects on Group 3 MB progression remain unexplored [16]. We evaluated PCK2 as a biomarker and therapeutic target in aggressive MYC-amplified medulloblastoma. Inhibiting PCK2 while downregulating MYC significantly impacts tumor growth and metastasis, suggesting PCK2 inhibitors could enhance treatment for high-risk medulloblastoma patients and improving pediatric patient quality of life. Materials and methods Data collection Medulloblastoma, non-WNT/non-SHH (470 samples from GSE85217 and 223 samples from E-MTAB-10767), was obtained from the Gene Expression Omnibus (GEO) and Array Express database. Gene sets for 82 metabolic pathways and 2022 metabolism-related genes were sourced from the MSigDB (see Additional File Table S2) [17]. Proteomic data in MB come from GEO (https://www.ncbi.nlm.nih.gov/geo/) (GSE109401) [18]. The public MYC ChIP-seq data was from the GEO database (GSE129521, GSE267470). The transcriptomic and metabolomic data are a comparison between the sh-ctrl and sh-PCK2 in D341 cell line, belonging to our self-assessment data (see Additional File Table S3). Clinical samples and cell lines MB tumors and normal brain samples were sourced from Nanfang Hospital, Southern Medical University, China. etc. (Approval number: 2025022012384788). Medulloblastoma cell lines: Daoy, D283, D341, D458 cells were cultured at 37 ℃ and 5% CO2. HEK293T and THP1 were obtained from Procell Life Science. A more detailed methods is available in the supplementary methods. Pathway enrichment analysis Functional analyses were conducted via Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Set Enrichment Analysis (GSEA) [19]. Immune infiltration analysis The ESTIMATE, MCPCOUNTER, and CIBERSORT algorithms were used to infer the immune cell infiltration of MB patients [20]. Continuum score According to methods of Steven C. Clifford [21]. A more detailed methods is available in the supplementary methods. Generation of shPCK2 knockdown and inhibit MYC expression inhibition HEK293T cells were transduced with shRNA vectors (sh: GGTGGCAAGCATGCGTATTAT), following the manufacturer's instructions, and packaging (pMD2.G and pSPAX2 plasmids were from Addgene) [22]. The siRNAs aimed at inhibiting MYC were constructed by Sigma. The c-Myc inhibitor 10,058-F4 (M14382; Luofa Biotechnology, China) was employed to suppress c-Myc activity. Single-cell RNA-sequencing (scRNA-seq) analysis GSE155446 dataset underwent processing and analysis utilizing the established methodologies provided by the R packages 'Seurat v4.0.1', 'infer-CNV v1.24.0', and 'cellchat v2'. The resulting clusters were annotated using well-established cellular markers. Tumor cells were further classified into molecular subgroups based on a continuum score derived from whole-transcriptome profiles, specifically mG3 (score < 0.5) and mG4 (score > 0.5). R version 4.4.2. Cell proliferation, migration, and colony formation assays CCK8 and cloning formation assay were used to detect cell proliferation, scratch assay was used to detect cell migration. Immunofluorescence and immunohistochemistry Immunofluorescence was conducted in accordance with previously established protocols [16]. Alexa fluorescent secondary antibodies (Alexa Fluor 488-conjugated A-11008, 1:5000, Invitrogen, CHINA) were used in Immunofluorescence. The primary antibody employed was anti-IBA1 (ab178846, 1:5000, Abcam, USA). Immunohistochemistry was performed using anti-PCK2 (14,892-1-AP, 1:500, Proteintech, USA) and anti-KI67 (ab16667, 1:5000, Abcam, USA) in accordance with established methodologies. Western blotting Western blotting was performed following the methodology outlined in our previous research [16]. The primary antibodies used included PCK2 (Proteintech, 14,892-1-AP, 1:5000), c-MYC (Proteintech, 10,828-1-AP, 1:1000), phospho-c-MYCSer62 (Cell Signaling Technology, 13,748, 1:1000), GAPDH (Proteintech, 60,004-1-Ig, 1:5000), SOX2 (PTM, ptm-5072, 1:1000), and NEUROD1 (Proteintech, 12,081-1-AP, 1:1000). The secondary antibodies employed were anti-rabbit (1:5000; CoWinBioscience, Cat:CW0103, Beijing, China) or anti-mouse IgG (1:5000; CoWinBioscience, Cat:CW0102, Beijing, China). Medulloblastoma xenografts (intracranial) Animal studies were conducted in compliance with the guidelines and approval of the Ethical Committee at Nanfang Hospital, Southern Medical University. (Approval No.: IACUC-LAC-20240308-002). Orthotopic xenograft models were established in the previous report [16]. 1 × 106 D341-Luc cells, D341-shPCK2-Luc cells, D458-Luc cells, and D458-shPCK2-Luc cells with a total 5 μl volume in PBS were engrafted intracranially into the cerebellum of anesthetized BALB/c nude mice. Survival analysis We applied a univariate Cox regression model to analyze the influence of TCA cycle and PCK2 expression levels on the prognosis of patients with non-WNT/non-SHH medulloblastoma. To estimate the relative risk of death among groups, hazard ratios (HRs) with 95% confidence intervals (CIs) were computed. Analyses specific to subtypes were conducted to assess the impact of the TCA cycle in various subtypes. The expression levels of the PCK2 were incorporated into the calculation of death risk HRs. Furthermore, to enhance interpretability, the GSVA scores were standardized using min–max scaling and then multiplied by 10 for HR prediction. Clinical data of non-WNT/non-SHH medulloblastoma in the GSE85217 (Group3 = 113, Group4 = 264) were obtained for overall survival (OS) Kaplan–Meier curves. For Kaplan–Meier survival analysis, samples were dichotomized based on the median expression value of the gene of interest. This approach was used consistently for all survival analyses to avoid bias associated with selecting cutoffs that minimize p-values. All survival models were constructed using the R package “survival (v.3.5-5)”. See supplementary methods for additional details for all methods. Results Transcriptomic continuum in non-WNT/non-SHH medulloblastoma Due to the intertumoral and intratumoral heterogeneity of non-WNT/non-SHH medulloblastoma, we introduced a continuum score. The continuum score was calculated by applying a logistic transformation 1/(1 + exp(− x)) to the Group3 and group4 MB metagenes. The two metagenes that described Group3 and Group4 samples were notably gradated and overlapping in an anticorrelative manner (Fig. S1a). Group3 and Group4 are not distinct transcriptional entities but rather exist as a continuum between two transcriptional polarities that we refer to here as G3 and G4. Regardless, the non-WNT/non-SHH medulloblastoma subtypes may be broadly ordered upon the G3/G4 continuum in partially overlapping domains from most group 3-like to most group 4-like (Group3γ, Group3α, Group3β, Group4γ, Group4α, Group4β, respectively) (Fig. S1b). For convenient comparison, we subdivided the expression continuum (G3/G4 score) into 3 purely notional quantiles, following the reference: G3 (0 − 0.4, n = 103/453 [23%]), G4G3 (0.4–0.6, n = 81/453 [18%]), G4 (0.6–1.0, n = 269/453 [59%]) [21]. Principal component analysis (PCA) revealed the presence of three distinct transcriptomic clusters in non-WNT/non-SHH medulloblastoma (Fig. S1c). Survival analysis revealed that the overall survival (OS) of the three groups was significantly different. G3 had the worst prognosis in terms of OS, while G4 had the best prognosis (Fig. S1d). Interestingly, the scale plot shows that the G4 group has more group4 samples, the G3 group has more group3 samples, and the G4G3 group has an unequal amount of group4 and group3 samples (Fig. S1e). The demographics and prognostic data of those patients are displayed in Supplementary Table 1. The distinct metabolic patterns and TIME of three groups associated with tumor continuum We examined the metabolic reprogramming mechanisms in non-WNT/non-SHH medulloblastoma by selecting 2022 human genes linked to 82 metabolic pathways from the KEGG database (Table S2). We used GSVA to estimate enrichment scores for 82 metabolic pathways in each sample from the GSE85217 dataset, aiming to uncover the metabolic heterogeneity of non-WNT/non-SHH medulloblastoma. In all 453 non-WNT/non-SHH medulloblastoma samples, we performed Pearson correlation assessment between the continuum score and the metabolic pathways GSVA score. G3 (p < 0.05, number of metabolic pathways = 4) exhibited upregulation in the Citrate cycle (TCA cycle). G4(p < 0.05, number of metabolic pathways = 7), was characterized by remarkable upregulation of Inositol phosphate metabolism, and Sphingolipid metabolism (Fig. 1a). To investigate the characteristics of the G3 and G4 groups, volcano plots were constructed to compare gene expression between these samples (Fig. 1b). The TCA cycle score is significantly elevated in G3, and suggesting a poor prognosis in non-WNT/non-SHH medulloblastoma (Fig. 1c). This is supported by KEGG and GSEA analyses of differentially expressed gene groups. The TCA cycle was notably enriched in G3, whereas gene sets associated with substance metabolic processes, including inositol phosphate and sphingolipid metabolism, were enriched in G4 (Fig. 1d–f). The association of metabolic features with continuum score. a Heatmap showing level of metabolic pathway enrichment for 453 non-WNT/non-SHH MB individuals ordered by continuum score. b Volcano plots showing genes that are differentially expressed in G4 or G3 samples. c Difference analysis of TCA cycle score between continuum subgroup, and Kaplan–Meier plot showing significant differences (Log-Rank test for trend) in non-WNT/non-SHH MB overall survival by TCA cycle score. d Functional enrichment analyses of genes upregulated in G3. e Functional enrichment analyses of genes upregulated in G4. f GSEA enrichment analysis of feature genes in G3 and G4 group. Green indicate genes upregulated in G4; red indicate genes upregulated in G3 Then, we evaluated the differences in immune infiltration among G4, G4G3, and G3. As shown in Fig. S2a, G4G3 exhibited the highest ImmuneScore, StromalScore, and ESTIMATE score, alongside the lowest tumor purity, indicating a superior immune response compared to the other groups. A comprehensive analysis revealed distinct patterns of immune cell infiltration and immune checkpoint gene expression among G4, G4G3, and G3 (Fig. S2b–d). In our mind, this may be due to inconsistent analysis results since MB is a “cold” tumor and has high intratumor heterogeneity. Single-cell transcriptome sequencing reveals the TIME in G3 and G4 patients Based on the above results, analysis of the bulk transcriptome of non-WNT/non-SHH medulloblastoma showed differences in metabolism patterns. Next, we used 20 single-cell sequencing datasets (GSE155446) for the next step analysis (Fig. 2a). First, we determined the chromosomal copy number variants (CNVs) in each sample based on matched transcriptomic data to understand the malignancy cell subpopulation (Fig. 2b, Table S2). This image shows the results revealed subgroup according to the original classification based on NaNo-string method (Fig. 2c). Then, a total of 9 clusters were identified through Uniform Manifold Approximation and Projection (UMAP). Firstly, the tumor cell population with CNVs was determined to be 0–1 according to the metagene score, and the tumor cells with a score of less than 0.5 were located as malignant Group 3 cells (mG3) and the tumor cells with a score greater than 0.5 were located as malignant Group 4 cells (mG4) (Fig. 2d). Subsequently, according to the definition of marker genes as B cells (CD79A, CD22, CD79B), T cells (CD8B, CD3D, CD3E), OPC and oligodendrocyte (PDGFRA, SOX2, OLIG2), TAMs (TMEM119, CD74, A2M), Neutrophil (RETN, CDA, S100A12), Pericytes (PDGFRB), Endothelial cells (IGFBP7) (Fig. 2e). The heterogeneity of immune cells across different MB subtypes based on tumor continuum. a An analysis of 20 samples in non-WNT/non-SHH MB using the UMAP plot. b UMAP plot shows the distribution of malignant cells and un-malignant cells differentiated by CNVs. c UMAP plot shows the distribution of group3 and group4 of 20 non-WNT/non-SHH medulloblastoma samples. d Cellular populations identified. The UMAP projection shows the formation of 9 main clusters with label names. Each dot corresponds to a single cell, colored according to cell type. e Bubble plot showing expression of marker gene of immune cells. f Percentages of the cell types among three groups. g Circle plot shows the difference in the number of interactions per cell cluster at non-WNT/non-SHH MB. h Circle plot shows the difference in the intensity of interactions per cell cluster at non-WNT/non-SHH MB. i Immunofluorescence staining verified the difference in macrophages in group3 and group4 MB paraffin-embedded specimens samples. m stands for malignant cells The samples were categorized into G4, G4G3, and G3 groups based on the mG3 and mG4 score, resulting in 12 G4, 3 G4G3, and 5 G3 samples. Stacking plots indicated that patients with elevated mG4 scores might exhibit increased B cell infiltration levels. Patients with G4G3 may have higher proportions of TAMs (Fig. 2f). We then examined the internal factors contributing to TIME variations in non-WNT/non-SHH medulloblastoma. CellChat was utilized to analyze cell interactions in non-WNT/non-SHH medulloblastoma. The findings indicated no significant difference in interactions between mG3 and mG4 cells (Fig. 2g). And, the strength of signaling pathways in mG4 to TAMs was significantly higher than in mG3 cells (Fig. 2h). Immunofluorescence experiments confirmed that the number of TAMs showed no difference between group 3 and group 4 samples (Fig. 2i). These findings align with our previous analysis results derived from bulk transcriptome sequencing. It highlights the intricate heterogeneity and continuity in non-WNT/non-SHH medulloblastoma (Fig. S3a, b). Identification of PCK2 as a G3-MB-enriched gene and associated with tumor progression Similar patterns were observed in external independent cohorts using data from GSE85217 and E-MATB-10767 database (Fig. 3a). In the non-WNT/non-SHH medulloblastoma, TCA cycle (Fig. S3c–d), particularly PCK2 exhibited the highest expression and activity levels in G3 group cells (Figs. 3b, S3e), which was consistent with the PCK2 expression in Group3 MB (Fig. 3c). Survival analysis indicated that non-WNT/non-SHH medulloblastoma patients with high PCK2 expression had shorter OS (Fig. 3d, p= 0.02). Group 3 MB, accompanied by MYC amplification or overexpression, often has a very poor prognosis, which often results in a worse prognosis for Group 3 than Group 4 MB (Fig. 3e). As shown in Fig. 3f, we found that in this larger dataset, PCK2 was a significant correlation with MYC in non-WNT/non-SHH medulloblastoma. According to Elisa Caffarelli's research report, when the function of MYC is inhibited, the function of many genes will also be suppressed [23]. Our data indicate that MYC binds to the transcription start site of PCK2 and regulates the expression of PCK2 (Sup Fig. 4). This is consistent with the results obtained from public datasets (Fig. 3g) [23]. Because c‐MYC is a transcriptional factor, we further investigated whether c‐MYC directly binds to the PCK2 promoter. There are two binding sites for c‐MYC, located at 956–963 and 1871–1878 of the PCK2 promoter according to JASPAR, a program to predict transcription factor–binding sites in DNA sequences (Fig. S4a). These results were supported using ChIP-Seq public data (Fig. S4b). To better evaluate the role of c-MYC in regulating PCK2 expression, we inhibited c-MYC expression in D341 and D458 cells using c-MYC siRNAs (siMYC-1, and siMYC-2). c-MYC knockdown reduced PCK2 protein expression in D341 and D458 cells. Similarly, c-MYC inhibitor (10,058-F4) reduced PCK2 protein levels in D341 cells as the concentration of the drug increased (Fig. S4c). The results suggest that c-MYC may upregulate the expression of PCK2 at the transcriptional level by directly binding to its promoter. However, the specific results still need to be further validated through experiments. Identification of PCK2 as a G3-MB-enriched gene. a Correlation of TCA cycle-related genes with continuum score in GSE85217 and E-MATB-10767 datasets. b Difference analysis of PCK2 gene expression between continuum subgroup. c Difference analysis of PCK2 gene expression between MB subgroup and normal brain. d PCK2 expression in non-WNT/non-SHH MBs was classified as PCK2 high (red, n = 188, below median) or PCK2 low (green, n = 189, above median). Kaplan–Meier survival analysis was performed between the two groups using datasets from Cavalli et al. e Kaplan–Meier plot showing significant differences in group3 and group4 MB overall survival. f Correlation of the continuum score with PCK2 and MYC genes expression. g Inhibiting MYC, the expression of PCK2 also decreased. Bar chart of MYC and PCK2. NC, normal control. DOX, inhibit MYC function, by the MYC dominant negative mutant OMOMYC. *p ≤ 0.05, **p ≤ 0.01. h Representative immunostaining images for PCK2 across the non-WNT/non-SHH MB subgroups (group3, n = 3; group4, n = 3; and Normal brain, n = 3). Red arrows indicate myeloid cells, and Purkinje cells in the normal brain and cerebellum. i Difference analysis of PCK2 protein expression between MB subgroup and normal brain. Data are plotted as means ± SD. Using datasets from Samuel Rivero-Hinojosa. j Diagnostic ROC curves of PCK2 in Group3 and Group4 MB according to subgroup To evaluate the potential clinical utility of PCK2 as a therapeutic target. IHC staining was performed on non-WNT/non-SHH medulloblastoma and normal brain tissues. Binding the protein expression of PCK2 was found to be significantly higher in Group 3 MB tissue than in normal brain tissues and other MB subgroups (Fig. 3h, i). According to the AUC value of the ROC curve, PCK2 expression demonstrated excellent discriminatory power for distinguishing Group 3 from Group 4 tumors (AUC = 0.981) (Fig. 3j). Deficiency of PCK2 impeded MYC-amplified MB cell proliferation and metastasis The DAOY (SHH cell line) and D283 (Group3 cell line, non-MYC amplified) showed reduced PCK2 protein expression compared to the MYC-amplified Group3 cell lines D458 and D341 (Fig. 4a). Similarly, PCK2 gene expression is higher in Group3 cell lines than SHH group cell lines (Fig. S5a). Group3 tumor tissues exhibited higher protein levels of PCK2 and MYC compared to SHH and Group 4 MB tissues, while normal brain tissues showed relatively low expression levels of these proteins (Fig. 4b). Then, we created two PCK2-deficient Group3 cell lines, D458 and D341, to investigate PCK2's role in MB (Fig. 4c). After PCK2 knockdown, the D341 cell line exhibited significantly reduced cell growth and clonogenic capacity compared to the control group (Fig. 4d, e). To further examine the impact of PCK2 on tumor cell growth in vivo, we established PCK2-deficient D341 and D458 orthotopically transplanted models. Bioluminescence imaging (BLI) of mice with orthotopic xenografts of shPCK2 D341 and D458 cells demonstrated slower tumor growth and a reduced incidence of spinal metastasis compared to control mice with wild-type (WT) cells (Figs. 4f–h, S5b, c). Furthermore, knockdown PCK2 in D341 and D458 cells may extend animal lifespan (Figs. 4i, S5d). The PCK2-deficient group exhibited a decrease in Ki67-positive proliferating cells (Figs. 4j, S5e). To evaluate the impact of PCK2 regulation on the proliferation and migration of D458 adherent cells, we performed CCK8, colony formation assay, and wound healing assays. When PCK2 knockdown was in the D458 cell line, this cell line exhibited a reduction in the proliferation ability by CCK8 assay (Fig. 4k), decreased the colony numbers in plate and cell migration in wound healing assays (Fig. 4l, m). Together, these observations suggest that PCK2 may contribute to the malignant progression of MYC-amplified medulloblastoma. Loss of PCK2 attenuates proliferation and metastasis of G3-MB cells. a Representative western blot for c-MYC and PCK2 in group-3 (D283, D458 and D341) and non-group-3 (DAOY) MB cell lines. GAPDH was used as loading control. b Expression of PCK2 protein analyzed by Western Blot in MB and normal cerebellum samples. NC-C mean normal cerebellum- cerebellum sample1/2/3. c Western Blot for PCK2 expression in D341 and D458 cell lines with control and shPCK2 groups. d CCK-8 assay showed that D341 cell viability was significantly reduced after PCK2 knockdown. e PCK2 knockdown suppressed D341 cell stemness. f Representative luminescence images showing mice bearing D341 cells with PCK2 WT or knockdown after implantation. g Compare the fluorescence signals in the cerebellum of BALB/c nude mice in the PCK2 WT and knockdown groups. h Pie charts showing spinal metastasis number of mice bearing D341 cells with PCK2 WT or knockdown after implantation. * p ≤ 0.05, **p ≤ 0.01. i Survival curves of two intracranial tumor-bearing mice. j Cerebellum in situ tumors were stained with H&E staining and IHC for PCK2 and KI67. k CCK-8 assay showed that D341 cell viability was significantly reduced after PCK2 KO. l Colony formation assay showed that number of colonies in PCK2 knockdown cells was significantly lower than that in the control group. m Scratch assay showed that PCK2 knockdown significantly slowed down the migration of D458 cell line Combined analysis of transcriptomic and metabolomic data To better understand PCK2 effects against MYC-amplified MB, we performed transcriptomic analyses of high-MYC cell line D341 with negative control (NC) and PCK2 knockdown groups. Our results showed that PCK2 and c-MYC expression was decreased (Fig. 5a). The D458 and D341 cell lines were identified as MYC-amplified G3 models. As a principal driver of MB, c-MYC total protein and Ser62 phosphorylation levels exhibited a marked decline following PCK2 knockdown (Fig. 5b). Ser62 phosphorylation is a critical post-translational modification that stabilizes MYC protein and enhances its transcriptional activity and oncogenic function [23]. These findings demonstrated that the PCK2 knockdown inhibited the growth of MYC-amplified MB cells and decreased c-MYC stability, highlighting the potential of PCK2 as a therapeutic target for the treatment of MYC-amplified MB. PCK2 maintains stemness and inhibits differentiation processes through metabolic reprogramming in G3-MB cells. a The statistical graph on the RNA expression of PCK2 and MYC. b Western blotting was used to detect the changes in c-MYC and ser62-MYC activities after knockdown and inhibition of PCK2 in MB cell lines. c GO enrichment analyses of DEGs, including GO Biological Process (BP), GO Molecular Function (MF), and GO Cellular Compartment (CC). d KEGG enrichment analysis of the differential metabolites. e Statistical histogram of the top 20 pathways with the most differential metabolites. f Transcriptomic and metabolomic differential analysis intersection Venn plot. g KEGG compound classification analysis of differential metabolites. h The statistical graph on the RNA expression of NEUROD1 and PROM1, and western blotting was used to detect the changes in SOX2 and NEUROD1 expression after knockdown of PCK2 in D341 cell line. *p ≤ 0.05, **p ≤ 0.01. Red and blue represent systems that are heavily affected Integrated pathway analysis of transcriptomic and metabolomics data. We detected by RNAseq that 1089 genes were significantly downregulated (log2 fold change < − 1, FDR < 0.05) and 233 genes were upregulated (log2 fold change > 1, FDR < 0.05), respectively (Table S3). Moreover, GO pathway analyses showed that these genes were enriched in the metabolic process, immune system process, and developmental process (Fig. 5c). The metabolite analysis showed that there were 137 differential metabolites in all 713 metabolites (Table S3). KEGG pathway analyses showed that these differential metabolites were enriched in the immune system, nucleotide metabolism, and carbohydrate metabolism (Fig. 5d). After combining analysis of transcriptomic and metabolomic data, we find that PCK2 may regulate cell fate and TIME by influencing nucleotide metabolism and purine metabolism most likely (Fig. 5e–g). While multiple metabolic pathways were perturbed, the immune-related pathways were of particular interest given the established role of the tumor microenvironment in medulloblastoma progression and the emerging link between PCK2 and immune modulation. Given the link of the PCK2 to the immune system, especially macrophages [16], we wanted to further explore the connections of PCK2 to macrophage polarization in MB. High expression of PCK2 in tumor cells does not modulate M1 macrophage polarization, but promotes M2 polarization in vitro (Fig. S5f, g). Phosphoenolpyruvate (PEP) is a key metabolite that has been shown to regulate macrophage polarization, with elevated PEP levels promoting an M2-like phenotype associated with tumor progression. About the developmental process, Yang et al. thinks that the differentiation of tumor cells is regulated by NeuroD1 expression, which is repressed by H3K27me3 in tumor cells, and EZH2 can facilitate the differentiation of MB tumor cells by modulate H3K27me3 [24, 25]. Because of this, we found that PCK2 knockdown can promote upregulation of NEUROD1 and inhibit the expression of CD133 (PROM1) and SOX2 (Figs. 5h, S5h). We also found that after knockdown of PCK2, EZH2 expression in tumor cells decreased (Fig. S5i). It means PCK2 knockdown can promote MB tumor cell differentiation and inhibit tumor cell stemness. Discussion Our study systematically delineates the metabolic reprogramming features of non-WNT/non-SHH medulloblastoma (MB), revealing TCA cycle as key pathways linked to poor overall survival (OS). Despite the classical emphasis on glucose metabolism in cancer, these findings challenge the dogma by highlighting alternative metabolic dependencies in aggressive MB subtypes [4, 26]. Notably, the prognostic relevance of TCA cycle aligns with the observations of Northcott et al. in MB, they uncover an IDH1/c-MYC dependent vulnerability that regulates DLAT levels and can be targeted to kill Group 3 MB by cuproptosis [27]. Not only that, we also found that most immune cells in Group 3 and Group 4 showed no significant differences. Nevertheless, G4 cells characterized by low PCK2 expression demonstrated a markedly increased interaction with macrophages. This phenomenon may be attributed to the competition for the metabolic capacity of macrophages by cells exhibiting high PCK2 expression, resulting in alterations in metabolites that facilitate immune evasion [28]. Central to our findings is PCK2, a gluconeogenic enzyme repurposed by MYC-amplified MB to fuel anabolic pathways [29]. Its knockdown attenuates glycolysis and malignant phenotypes, consistent with PCK2’s role in redirecting mitochondrial metabolites toward biosynthetic processes. Intriguingly, PCK2 depletion also disrupts M2 macrophage polarization, likely through altering lactate or succinate levels-metabolites known to influence myeloid cell differentiation [30]. This dual metabolic-immunologic function positions PCK2 as a therapeutic vulnerability, particularly given MYC’s established role in therapy resistance [29]. While prior studies focused on PCK1 in cancer metabolism, our work highlights PCK2’s paralogue -specific relevance in MB, potentially reflecting tissue-specific metabolic dependencies [16]. Phosphoenolpyruvate carboxykinase (PEPCK) exists in two paralogues: PCK1, which functions in the cytoplasm, and PCK2, which is active in the mitochondria [16]. Unlike PCK2, PCK1 promotes tumor cell proliferation in hepatocellular carcinoma (HCC) by modulating the TCA cycle and fatty acid synthesis [31]. Similarly, PCK1 exhibits comparable functions in non-small cell lung cancer (NSCLC) and esophagealcarcinoma [32]. Beyond its role in metabolic regulation, PCK1 is closely associated with the tumor immune microenvironment, where its expression level influences the function and metabolic state of immune cells [32]. In medulloblastoma, however, PCK1 expression is nearly undetectable, with PCK2 being the predominant paralogue (Fig. S5j). Notably, there are significant differences in the expression levels and mechanisms of action of PCK2 across various tumor types [16]. For instance, PCK2 expression is high in breast cancer, prostatic cancer, and lung cancer, and is associated with tumor aggressiveness and poor prognosis [16]. Conversely, in HCC, PCK2 expression is low, yet this low expression correlates with poor prognosis [32]. These disparities may be related to the metabolic demands and immune microenvironments of different tumors [33]. MYC, a crucial transcription factor, influences cell metabolism and proliferation by regulating the expression of multiple genes [34]. Our results are consistent with those in the literature, medulloblastoma tumors with MYC amplification or overexpression have a poor prognosis, often characterized by rapid progression and shortened overall survival [35]. This suggests that PCK2 may support MYC-mediated metabolic reprogramming, thereby promoting the rapid proliferation and survival of tumor cells (Fig. S5k). Beyond its role in metabolic regulation, PCK2 may also affect the stability of MYC. Studies in melanoma have shown that PCK2 knockdown can reduce MYC protein levels and its Ser62 phosphorylation, thereby inhibiting tumor cell growth [28]. This indicates that PCK2 may influence the stability of MYC through metabolic pathways, thereby regulating the proliferative capacity of melanoma cells [28, 29]. Moreover, PCK2 and MYC have a synergistic effect on the tumor microenvironment, influencing the infiltration and function of immune cells [16]. For example, high PCK2 expression is associated with an immunosuppressive tumor microenvironment, characterized by immune-excluded tumors that may respond to immune checkpoint inhibitor therapy [36]. Similarly, high MYC expression is also associated with an immunosuppressive microenvironment, promoting tumor immune evasion by regulating the expression of immune checkpoint genes such as PD-L1 [37, 38]. Both PCK2 and MYC play important roles in tumor metabolic reprogramming, with PCK2 promoting gluconeogenesis and the TCA cycle to provide energy and metabolic intermediates for tumor cells, while MYC supports rapid tumor cell proliferation by regulating the expression of multiple metabolic genes [16]. Our results suggest that in MYC-amplified medulloblastoma, the synergistic action of PCK2 and MYC may further enhance metabolic reprogramming and promote tumor progression. MB are thought to be relatively immunologically “cold” due to paucity of mutations. Based on CIBERSORT analysis of the integrated transcriptomic dataset and the results of immune cell deconvolution in Sup Fig. 2, results show significantly greater CD8 T cells in MB Group3 compared to MB Group4, and greater NK cells in MBGroup4 compared to other subgroups [39]. This is the same as the results we obtained in Sup Fig. 2b, but the conclusions we drew from single-cell data analysis are not consistent, which may be because bioinformatics analysis cannot accurately reflect reality, so we still need further experimental validation. In conclusion, this study redefines non-WNT/non-SHH MB through the lens of metabolic-immune interplay, proposing PCK2 as a linchpin molecule coordinating tumor glycolysis and immunosuppression. The continuum score challenges conventional subtyping frameworks, advocating for gradient-based risk assessment in heterogeneous malignancies [40]. These insights not only advance MB biology but also inform therapeutic development for other MYC-driven cancers where metabolic plasticity underlies therapeutic resistance. Regrettably, the study only used human cell lines, preventing observation of macrophage changes in live mouse models. Whether the PCK2 inhibitor 3-MPA can cross the blood–brain barrier and exert effects in vivo still requires further research. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Acknowledgements Thanks to Professor Jiajia Wang (Department of Pediatric Neurosurgery, Xin Hua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine), Professor Wei Yang (Department of Neurosurgery, Beijing Children's Hospital, Capital Medical University, National Center for Children's Health), and Professor Aidong Zhou (Department of Cell Biology, School of Basic Medical Science, Southern Medical University) for their guidance. Author contributions The article was contributed to by all authors and the submitted version was approved by them. Ye Song, Hao Long, Xian Zhang, and Peng Li were responsible for revising the paper. Funding This work was supported by the National Nature Science Fund of China (No. 82272879), the Natural Science Fund of Guangdong Province, China (2023A1515010633, 2023A1515012382, 2024A1515012949), Guangzhou Science and Technology Bureau Key Research and Development Plan (2024B03J0352), Jiangxi Provincial Key Discipline Leader Cultivation Plan—Youth Talent Project (20225BCJ23001). Data availability The data are presented in the supplementary and can be available by contacting the authors. (Table S2 collects all the pathway information used in the paper. Table S3 shows the results of cell line RNA sequencing and metabolites sequencing). Declarations Ethics approval and consent to participate Ethics approval and consent to participate, the animal studies were performed according to the guidelines and approval of the Ethical Committee of Nanfang Hospital, Southern Medical University. (Approval No.: IACUC-LAC-20240308-002). Consent for publication All authors consent to the publication of this work. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Shuaishuai Xue, Zhengming Zhan, Yonghua Cai and Wei Chen have contributed equally to this work. Contributor Information Peng Li, Email: zhimeng_li@163.com. Hao Long, Email: longh812@163.com. Ye Song, Email: songye0724@126.com. References Associated Data Data Availability Statement The data are presented in the supplementary and can be available by contacting the authors. (Table S2 collects all the pathway information used in the paper. Table S3 shows the results of cell line RNA sequencing and metabolites sequencing).