PGLS as an immune and prognostic biomarker: from pan-cancer analysis to validation

Abstract

6-phosphogluconolactonase (PGLS) is a metabolic enzyme of the pentose phosphate pathway that has been linked to tumorigenesis in several cancers. However, its role in pan-cancer remains unclear. This study reveals the expression profile and prognostic analysis of PGLS in different cancers by pan-cancer analysis. We utilized mRNA-seq data from TCGA and GTEx databases to analyze PGLS expression in different tumors. PGLS expression was significantly higher in almost all types of human cancer tissues compared to corresponding normal tissues. High PGLS expression was linked to poor prognosis. PGLS expression was significantly associated with immune regulatory genes, immune cell infiltration, tumor heterogeneity, tumor stemness, and involved in the regulation of anticancer drug sensitivity. Knockdown of PGLS resulted in slowed growth and diminished migratory and invasive capacity of Huh7 and A498 cells. Additionally, PGLS knockdown led to an increase in M1 macrophages, CD8⁺ T cells, and CD4⁺ T cells, while reducing the proportion of M2 macrophages and Tregs in tumor tissue. PGLS could serve as a potential therapeutic target and a reliable predictor of both prognostic factors and treatment sensitivity across various types of tumors. Further studies are warranted to explore its clinical application in cancer treatment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-03108-8.

Introduction

Cancer continues to pose a significant global health threat, accounting for approximately one sixth of all deaths annually worldwide [1, 2]. Despite advancements in clinical treatment and basic medical research, the possibility of achieving a cure for cancer remains elusive. However, recent developments in targeted combination immunotherapy regimens have shown promise in improving survival rates for certain cancer patients [3–6]. As the need for more personalized therapeutic strategies grows, leveraging big data to identify novel and effective targets has become a critical avenue of research. One such approach is pan-cancer analysis, which has gained traction as a powerful tool for uncovering broader patterns in cancer biology and treatment response.Through pan-cancer analysis, it is anticipated that key intersections in cancer-targeted immunotherapy combinations will be identified, offering novel insights for the advancement of cancer treatment.

A key feature of cancer cells is their metabolic transformation, and increasing evidence highlights the crucial role of metabolism-related enzymes and metabolites in tumorigenesis and the shaping of the tumor microenvironment (TME) [7–11]. Targeting cancer cell metabolism has emerged as a promising therapeutic strategy. The pentose phosphate pathway (PPP) plays a central role in cancer cell metabolism, involved in nucleic acid and fatty acid synthesis, as well as NADPH production [12]. PPP is also essential in maintaining redox homeostasis, enabling cancer cells to survive in the challenging TME [13]. Moreover, alterations in PPP have been linked to resistance to chemotherapy and targeted therapies in various cancers, including breast [14], colorectal [15–17], and renal clear cell carcinoma [18].

Among the key enzymes in the PPP, 6-phosphogluconolactonase (PGLS) has attracted attention due to its potential involvement in cancer progression. While studies have shown that PGLS is highly expressed in breast cancer [19], promotes hepatocellular carcinoma (HCC) progression [20], and correlates with poor prognosis in gastric [21] and ovarian cancers [22]. However, despite the growing body of evidence pointing to PGLS’s potential involvement in cancer treatment, there remains a significant gap in our understanding of how PGLS influences cancer progression and treatment response. Few studies have systematically explored the therapeutic implications of targeting PGLS, and its precise role in modulating treatment efficacy remains largely unexplored.Given the pivotal function of the PPP in tumor metabolism, further investigation into the role of PGLS in cancer progression, therapeutic resistance, and its potential as a novel target for cancer treatment is warranted. This study aims to address this gap by exploring the contribution of PGLS to tumorigenesis and its potential as a therapeutic target.

In this study, we utilized The Cancer Genome Atlas (TCGA) database to conduct pan-cancer analysis on PGLS expression in different cancers and explored its correlation with patient prognosis, immune regulation, tumor heterogeneity, tumor stemness, and mRNA methylation. Additionally, we conducted in vitro and in vivo experiments related to proliferation. Our findings indicate that PGLS expression is upregulated in most cancers and may serve as both a predictor of poor prognosis and a potential therapeutic target. In vivo, reducing the expression of PGLS in tumor cells will recombine the immune microenvironment to support anti-tumor immunotherapy.

Materials and methods

Data sources and preprocessing

The mRNA-seq data for cancer cases were obtained from The Cancer Genome Atlas (TCGA) database. Corresponding mRNA-seq data for normal tissues and patient clinical information were also obtained from TCGA. Due to the inadequate availability of normal tissue mRNA sequencing data in TCGA, we acquired additional normal tissue data from the Genotype-Tissue Expression project (GTEx), which is available at https://commonfund.nih.gov/GTEx. RNA-Seq preprocessing involved DESeq2-based variance stabilizing transformation for raw counts and log2(x + 1)-transformed FPKM/TPM with subsequent quantile normalization. Clinical data were harmonized through standardized monthly survival metrics (OS/PFS), a unified TNM staging framework aligned with AJCC 8th edition criteria, and rigorous missing value handling via multiple imputation for variables with ≤ 30% missingness. Data integrity was validated against the UCSC Xena reference dataset.

Statistical analysis

Statistical analyses were performed using R or GraphPad Prism 8. Descriptive statistics (means, SDs, ranges for continuous variables; frequencies, percentages for categorical variables) were calculated. Pearson’s correlation was used for normally distributed data, and Spearman’s for non-normally distributed data, with significance set at p < 0.05. Multiple comparisons were adjusted using the Benjamini-Hochberg procedure to control the false discovery rate (FDR). t-tests or ANOVA were used for group comparisons, with post-hoc analyses when necessary. A two-tailed p-value < 0.05 was considered significant.

Elimination of batch effect and confusion effect

Batch effect correction in TCGA/GTEx/TARGET datasets requires context-dependent implementation rather than universal application. Cross-project integration or multi-platform analyses necessitate rigorous correction using methods like Combat-seq for count-based RNA data or RefFreeEWAS for methylation arrays, while preserving biological signals through supervised approaches. Validation should combine technical metrics with biological sanity checks. Overcorrection risks can be mitigated by monitoring established biomarkers during correction workflows.

Use consensus non-negative matrix factorization across multi-omics platforms (RNA/methylation/CNV) to disentangle organ-specific vs. pan-cancer signatures, then apply inverse probability weighting in Cox models to balance subtype proportions. Validate findings through cross-cohort perturbation testing - systematically excluding each tumor subtype while monitoring hazard ratio stability (threshold: <15% variation). For mutation analyses, employ MutSigCV’s tumor-type-specific background model to distinguish driver events from tissue-of-origin artifacts. Crucially, maintain separate validation cohorts for each constituent tumor type to verify trans-cancer patterns.

Immunohistochemical staining

Representative immunohistochemical staining images of PGLS in cancer and corresponding normal tissues were obtained from the HPA database. Immunoreactivity was quantified based on staining intensity and the percentage of positively stained cells. Intensity was scored as 0 (none), 1 (weak), 2 (moderate), and 3 (strong), while the percentage of stained cells was scored as 0% (score 0), 1–25% (score 1), 26–50% (score 2), 51–75% (score 3), and > 75% (score 4). The final score, ranging from 0 to 12, was calculated by multiplying the intensity and percentage scores. Samples were classified into low or high expression groups based on a threshold score of 7.

Analysis of PGLS expression levels and prognosis in different cancers

To analyze the mRNA expression of PGLS in various cancers, we used Sangerbox (http://vip.sangerbox.com/), an online tool. Immunohistochemical images of PGLS in different cancers versus paired normal tissues were downloaded from the Human Protein Atlas (http://www.proteinatlas.org/). Cox regression analysis and Kaplan-Meier survival analysis were performed to compare the effect of different levels of PGLS expression on survival outcomes. Cancers with less than three samples in a single cancer species were excluded, and statistical differences between groups were analyzed using a t-test. A p-value of less than 0.05 was considered statistically significant. The single cell expression data of PGLS was sourced from the DISCO database (https://www.immunesinglecell.org/), and analyzed the functional states of PGLS across different cancersthrough the cancer single cell state atlas (http://biocc.hrbmu.edu.cn/CancerSEA/).

Tumour immunological correlation analysis

After excluding normal tissue samples, Pearson correlations were calculated between PGLS and immunomodulatory genes as well as immune checkpoints using Sangerbox. Additionally, Stromalscores, Immunescores, and ESTIMATEscores were computed for each patient in each tumor based on gene expression. Furthermore, the association between PGLS expression and immune cells was analyzed using TIMER and CIBERSORT.

Tumour heterogeneity analysis

Homologous recombination deficiency data for each tumor were retrieved from a previous study [23]. Using the online tool Sangerbox, we investigated the correlations between PGLS expression and tumor heterogeneity parameters, including tumor mutational burden (TMB), mutant-allele tumor heterogeneity (MATH), microsatellite instability (MSI), neoantigen (NEO), tumor purity, tumor ploidy, homologous recombination deficiency (HRD), and loss of heterozygosity (LOH).

Tumour stemness analysis

The tumour stemness analysis involves determining various characteristics of tumour stemness. In this study, we utilized the methodology from a previously published work [24]. We conducted a comprehensive assessment of tumour stemness features, including RNA expression-based stemness score (RNAss), DNA methylation-based stemness score (DNAss), differentially methylated probes-based stemness score (DMPss), Enhancer Elements/DNAmethylation-based stemness score (ENHss), epigenetically regulated RNA expression-based stemness score (EREG.EXPss), and epigenetically regulated DNA methylation-based stemness score (EREG.METHss), to investigate their correlation with the expression of PGLS gene in tumours.

Gene mutation and drug sensitivity analysis

The present study conducted an analysis of gene mutations and drug sensitivity, specifically focusing on PGLS. The cBioPortal platform (http://www.cbioportal.org) was utilized to investigate the frequency of PGLS mutations across various cancer types, as well as to assess differences in copy number variations (CNVs). Additionally, CellMiner (NCI-60) was implemented to perform a thorough drug sensitivity analysis.

In vitro and in vivo proliferation assays

This study aimed to examine the effect of PGLS on tumor proliferation both in vitro and in vivo. After identifying cancers with high PGLS expression and poor prognosis, we determined that HCC and renal carcinoma were representative models. HCC was characterized by a high incidence rate, while renal carcinoma showed a steadily increasing incidence over the years. Both cancers are also associated with a poor prognosis. To achieve this goal, lentiviral transfection of shRNA was employed to knock down PGLS expression in cancer cells, followed by assessment of cell proliferation via CCK8 and plate cloning assays. Additionally, changes in migratory capacity were determined via scratch and Transwell assays. To further explore the impact of PGLS on tumour proliferation in vivo, we established a BALB/c nude (6–8 weeks, females) subcutaneous graft tumour model. The experimental protocol of this study was approved by the Experimental Animal Ethics Committee of Huazhong University of Science and Technology (IACUC Number: 2910). This study strictly adhered to guidelines approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology. The maximum permitted tumor burden was 2 cm in diameter, and no animals exceeded this limit. Anesthesia was induced with pentobarbital sodium, followed by euthanasia via cervical dislocation.

Results

Expression of PGLS in various tumours

In this study, we investigated the expression patterns of PGLS in various tumors using bioinformatics analysis of TCGA and GTEx databases. By comparing the expression levels between cancerous and paired normal tissues, we found that PGLS was significantly upregulated in 19 tumors including breast invasive carcinoma (BRCA), cholangiocarcinoma (CHOL), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) (Fig. 1A). Furthermore, by incorporating normal tissue data from GTEx, we identified 27 cancer species with high PGLS expression levels (Fig. 1B). Immunohistochemical staining images of PGLS in several cancers and paired normal tissues, as shown in Fig. 1C, further confirmed its upregulation at the protein level in cancerous tissues. Our findings suggest that PGLS is highly expressed in almost all types of cancer, indicating its potential as a diagnostic and therapeutic target for cancer treatment.

Fig. 1.

Expression of PGLS in varied tumours. (A) Expression of PGLS mRNA in the TCGA database between multiple carcinomas and normal paracancerous tissues. (B) Expression of PGLS mRNA in the TCGA and GTEx database between multiple carcinomas and normal paracancerous tissues. (C) Representative immunohistochemical staining images of PGLS in cancer and corresponding normal tissue from the HPA database. -, no statistical difference; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001

Correlation between PGLS expression and patient prognosis

To evaluate the prognostic significance of PGLS expression in various cancer types, we utilized Cox regression and Kaplan-Meier survival analysis to analysis TCGA and GTEx databases. Our findings revealed that heightened PGLS expression was significantly associated with unfavorable overall survival (OS) across six tumor types, including brain lower grade glioma (LGG), kidney renal clear cell carcinoma (KIRC), LIHC, skin cutaneous melanoma (SKCM), metastases from SKCM (SKCM-M) and adrenocortical carcinoma (ACC) (Figs. 2 A and S1). Interestingly, PGLS may serve as a protective factor in acute myeloid leukemia (LAML). Moreover, our results indicated that elevated PGLS levels were predictive of worse disease-free interval (DFI) in male testicular germ cell tumors (TGCT) (Figs. 2B and S2A). Additionally, PGLS expression was an equally significant predictor for progression-free interval (PFI) and disease-specific survival (DSS), wherein heightened levels were indicative of poorer PFI and DSS within four and five cancers, respectively (Figs. 2 C, D, and S2B, S2C). These findings were consistent with the previously reported worse OS prediction group.

Fig. 2.

Univariate COX regression analysis of PGLS expression levels and prognosis in different cancer types. (A) Impact of PGLS expression on overall survival (OS) in different cancers. (B) Impact of PGLS expression on disease free interval (DFI) in different cancers. (C) Impact of PGLS expression on progression free interval (PFI) in different cancers. (D) Impact of PGLS expression on disease specific survival (DSS) in different cancers

Association between PGLS expression and immune infiltration

Immunotherapy has emerged as a promising approach to cancer treatment in recent years, with its efficacy being governed by the immune infiltration of the cancer. In this study, we analyzed the association between PGLS and immunomodulatory genes, immune checkpoints, immune infiltration scores, and immune cells. By analyzing the correlation of 150 marker genes of the five classes of immune pathways (chemokine, receptor, MHC, immunoinhibitor, and immunostimulator), we found that PGLS was negatively correlated with most genes with antitumor activity, particularly in prostate adenocarcinoma (PRAD), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), thymoma (THYM), and TGCT (Fig. 3). Conversely, PGLS was positively associated with most immune resistance factors, such as CD276 and PVRL2 (Fig. 3), which have been reported to be detrimental to T cell-mediated anti-tumour immune activity [25, 26]. At the immune checkpoint, PGLS was found to be significantly positively correlated with VEGFB, CD276, and TGFB1 in almost all cancers (Figure S3). However, inconsistent results have been observed in some cancers, which may be related to the complex TME.

Fig. 3.

Correlation analysis between PGLS expression and immunomodulatory genes

Furthermore, we calculated the correlation between PGLS expression and immune infiltration scores using three different score models. We found that four cancers were positively correlated with ESTIMATEScore (Figures S4), ImmuneScore (Figures S5), and StromalScore (Figures S6), including LGG, LAML, sarcoma (SARC), and pheochromocytoma and paraganglioma (PCPG), while five other cancers were negatively correlated, including colon adenocarcinoma (COAD), PRAD, LUSC, thyroid carcinoma (THCA), and pancreatic adenocarcinoma (PAAD). We also calculated the correlation between PGLS and various immune cells using TIMER and CIBERSOR and found that PGLS expression was negatively correlated with CD8⁺ T cells with anti-tumour immune activity and positively correlated with CD4⁺ T cells and macrophages (Fig. 4). Additionally, most of these CD4⁺ T cells and macrophages were regulatory T cells (Tregs) and M2 macrophages that exert immunosuppressive functions (Figure S7). These findings suggest a potential role for PGLS in modulating the immune response to cancer and may have implications for the development of novel immunotherapeutic strategies.

Fig. 4.

Association of PGLS expression with immune cell infiltration score by TIMER

Correlation analysis between PGLS expression and tumour heterogeneity

Tumour heterogeneity is a critical factor contributing to the failure of anti-cancer therapies and patient mortality [27]. In this study, we conducted an analysis of the relationship between PGLS and eight heterogeneous tumour characteristics. Tumour mutational burden (TMB) was significantly correlated with immunotherapy sensitivity [28], with PGLS expression displaying a positive correlation with TMB in KIRC and ACC, and a negative correlation with TMB in COAD and CHOL (Fig. 5A). The Mutant Allele Tumour Heterogeneity (MATH) algorithm is a measure of tumour heterogeneity based on allelic mutations, and higher values correspond to increased tumour heterogeneity [29]. Our results showed that PGLS was positively correlated with MATH in SARC, LUSC, LIHC, and negatively correlated with MATH in LGG, BRCA, stomach adenocarcinoma (STAD), and bladder urothelial carcinoma (BLCA) (Fig. 5B). Microsatellite Instability (MSI) has been reported as more responsive to immune therapy [30], potentially because MSI-H cells are more easily recognized by immune cells. PGLS exhibited a positive correlation with MSI in six tumours and a negative correlation in two tumours (Fig. 5C). Tumour neoantigen (NEO) is a crucial target for T cell-mediated anti-tumour immunity [31]. Our findings indicate that PGLS was positively correlated with NEO in HNSC and ACC and negatively correlated in COAD and PAAD (Fig. 5D). Tumour purity and ploidy are also essential features of tumour heterogeneity. We observed a positive correlation between PGLS and tumour purity in 11 tumours and a negative correlation in four tumours (Fig. 5E). Moreover, PGLS expression positively correlated with tumour ploidy in SARC and LIHC and negatively correlated in kidney chromophobe (KICH) and DLBC (Fig. 5F). Homologous recombination deficiency (HRD) predicts the sensitivity of PARP inhibitors and platinum-based drugs [32]. PGLS expression showed a positive correlation with HRD in 12 cancers and a negative correlation with HRD in LAML, STAD, and CHOL (Fig. 5G). Loss of heterozygosity leads to repressor gene inactivation [33], and our results showed that PGLS expression was positively correlated with LOH in nine tumours while it was negatively correlated with LOH in BRCA and STAD (Fig. 5H). These findings suggest that PGLS has predictive potential for treatment modality sensitivity in some tumours. For instance, high PGLS expression may predict immunotherapy tolerance in colorectal cancer. Conversely, in ACC, high PGLS expression indicates the opposite outcome. In STAD, high PGLS expression results in insensitivity to both chemotherapy and targeted therapies.

Fig. 5.

Correlation of PGLS expression with tumour heterogeneity. (A) Tumor mutational burden (TMB). (B) Mutant-allele tumor heterogeneity (MATH). (C) Microsatellite instability (MSI). (D) Neoantigen (NEO). (E) Tumor purity. (F) Tumor ploidy. (G) Homologous recombination deficiency (HRD). (H) Loss of heterozygosity (LOH)

Analysis of the association between PGLS expression and tumour stemness

Tumor stem cells exhibit indefinite proliferative capacity, metastatic potential, and resistance to chemical toxicants [34]. The assessment of tumor stemness was performed using six indices that included RNA expression-based stemness score (RNAss), DNA methylation-based stemness score (DNAss), differentially methylated probes-based stemness score (DMPss), enhancer elements/DNA methylation-based stemness score (ENHss), epigenetically regulated RNA expression-based stemness score (EREG.EXPss), and epigenetically regulated DNA methylation-based stemness score (EREG.METHss), which were calculated based on mRNA expression levels and DNA methylation. Significant correlations were observed between PGLS expression and RNAss in 14 tumors, with five showing positive results (Fig. 6A). Meanwhile, PGLS expression was positively correlated with DNAss in LGG, and SARC and negatively correlated with it in the remaining 9 tumors (Fig. 6B). In DMPss, a positive correlation was found with PGLS expression in LGG, SARC, and PRAD, whereas a negative correlation was observed in 11 tumors (Fig. 6C). Additionally, a significant positive correlation was noted between PGLS expression and ENHss, EREG.EXPs, and EREG.METHss in LGG, and SARC (Fig. 6D, E and F), indicating that PGLS plays a diverse role in modulating tumor stemness in different microenvironments. In glioma and sarcoma, which currently lack effective curative options, PGLS appears to significantly promote their growth.

Fig. 6.

Association between PGLS expression and tumor stemness or RNA methylation. (A) RNA expression-based stemness score (RNAss). (B) DNA methylation-based stemness score (DNAss). (C) Differentially methylated probes-based stemness score (DMPss). (D) Enhancer Elements/DNAmethylation-based stemness score (ENHss). (E) Epigenetically regulated RNA expression-based stemness score (EREG.EXPss). (F) Epigenetically regulated DNA methylation-based stemness score (EREG.METHss). (G) RNA methylation-modified genes

Correlation analysis between PGLS expression and mRNA methylation

mRNA methylation is a crucial pathway that plays a vital role in regulating mRNA stability and gene expression, including the modification of m1A, m5C, m6A, and others [35]. This process is regulated by a complex interplay between the “writer”, “reader” and “eraser”. In this study, we investigated the association between PGLS expression and mRNA methylation-regulated genes, and found a significant positive correlation in 10 cancer types, including ovarian serous cystadenocarcinoma (OV), high-risk wilms tumor (WT), ACC, GBM, DLBC and COAD (Fig. 6G). Notably, we observed a significant positive correlation between PGLS and six regulatory genes (TRMT61A, ALKBH3, NSUN5, DNMT1, ALYREF, and KIAA1429) across almost all cancer types (Fig. 6G). These genes have been previously implicated in tumorigenesis, cancer development, and anti-tumor immune effects mediated by T cells to varying degrees [36–41]. Our findings suggest that PGLS may play a role in regulating mRNA methylation, leading to tumorigenesis, disease progression, and poor prognosis.

The mutational landscape of PGLS in various cancers

Using the cBioPortal database, we assessed the prevalence of PGLS mutations in various cancer types. Our analysis revealed that PGLS exhibits a mutation frequency exceeding 5% in both ovarian and lung cancers (Fig. 7A). Furthermore, within certain subtle pathological subtypes, such as uterine serous carcinoma/uterine papillary serous carcinoma, ovarian epithelial tumor, and pancreatic neuroendocrine tumor, the frequency of PGLS mutations is greater than 20% (Figure S8). Conversely, significant differences in copy number variations (CNV) of PGLS were observed across 16 tumor types (Fig. 7B). A total of 16 mutant loci were identified in PGLS among 16 cancers, including 15 missense, 2 truncating, 1 splice, and 1 fusion (Fig. 7C). Notably, H162R and I105F were the most commonly detected mutant loci.

Fig. 7.

The mutational landscape of PGLS in different cancer types. (A) PGLS mutation frequency based on the cBioPortal database. (B) Copy number variations (CNVs) differences of PGLS in different cancer types. (C) Mutation loci of PGLS in different cancer types

Drug sensitivity prediction

Subsequently, we conducted an analysis to evaluate the impact of PGLS expression on drug sensitivity by utilizing the CellMiner database. Supplementary Table 1 displays the complete list of drugs demonstrating a positive correlation which encompasses Perifosine, PX-316, BMS-777,607, TP-3654, and Foretinib, all of which are under clinical trials and have been approved by the FDA. The majority of these therapeutics target protein kinases (PK) and AKT, with further investigation required to elucidate the pathways involved in PGLS-mediated modulation of their efficacy.

Single cell landscape of PGLS

We conducted a further analysis on the expression distribution of PGLS at the single-cell level. The findings demonstrated that within the single-cell data set of breast cancer, PGLS exhibited greater distribution in cancer cells. Moreover, when compared to ductal carcinoma in situ (DCIS), which has lower malignancy, PGLS displayed a higher level of distribution in triple negative breast cancer (TNBC), which is characterized by higher malignancy and poorer prognosis (Figure S9A). Furthermore, PGLS was found to be enriched in functional or proliferative cell types including CD14⁺ myeloid cells and fibroblasts, while its distribution was relatively lower in immune cells with anti-tumor effects such as CD8⁺ T cells (Figures S9B and S9C). This pattern was also observed in pancreatic cancer and ovarian cancer, as evidenced through analysis of single-cell data sets specific to these cancer types (Figures S10 and S11). We proceeded to examine the functional status of PGLS across different cancers based on single-cell data, discovering a significant positive correlation between PGLS and at least three malignant phenotypes in most cancers (Figure S12A). Notably, in acute myeloid leukemia (AML), high expression of PGLS exhibited a strong positive correlation with nearly all malignant functional phenotypes of tumors (Figure S12B). The differential expression of PGLS at the single-cell level contributes to the formation of a hostile tumor microenvironment, which hinders the survival and function of anti-tumor immune cells. Nevertheless, this observation underscores the therapeutic potential of targeting PGLS, serving as an effective modulator in conjunction with immunotherapy.

 PGLS promotes cancer cell progression both in vitro and in vivo

We selected two types of cancer, liver and kidney cancers, which had high PGLS expression with poor prognosis, for in vitro and in vivo experiments to investigate the role of PGLS in their growth. After knockdown of PGLS, we observed significantly reduced viability and slower cell growth of Huh7 and A498 cells (Fig. 8A and C). Additionally, there was a decrease in colony-forming capacity, with knockdown of PGLS leading to a reduction in the number of colonies formed by Huh7 and A498 by 59.7% and 47.8%, respectively (Fig. 8B). Scratch results showed that knockdown of PGLS decreased the 24-hour wound healing capacity of Huh7 and A498 by 24.0% and 17.1%, respectively (Fig. 8D). Furthermore, Transwell assay demonstrated that the migration ability of Huh7 cells in the shPGLS group was only 33.7% of that in the shControl group, and the invasion ability was only 27.6%. Similarly, migration and invasion after A498 knockdown of PGLS were only 49.2% and 50.0% of that in the control group (Fig. 8E). In vivo tumorigenesis assays revealed that cancer cells with PGLS knockdown formed smaller tumors and had significantly slower tumor growth rates (Fig. 8F and G). The above results suggest that knockdown of PGLS reduces the proliferation and migration ability of cancer cells in vitro. In vivo, PGLS has a tumour growth-promoting effect.

Fig. 8.

Functional validation of PGLS knockdown on tumor growth through in vitro and in vivo models. (A) Cell viability analysis of Huh7 and A498 cells treated with PGLS-specific shRNA for 48 h, measured by CCK-8 assay. (B) Colony formation assay of Huh7 and A498 cells cultured for 14 days. (C) Growth curve quantification using CCK8 assay showing doubling time changes (24–96 h post-transfection). (D) Scratch assay to detect changes in cell migration capacity by knockdown of PGLS. Bar: 200 μm. (E) Transwell assay for the effect of knockdown of PGLS on cell migration and invasion capacity. Bar: 100 μm. (F) Gross images of harvested tumours taken 15 days after the establishment of the BALB/c nude mouse subcutaneous graft tumour model, 1 × 10^6 cells/mouse, n = 6. Bar: 2 cm. (G) Tumour growth curve. Tumor volume progression measured every 3 days with digital vernier caliper using the formula: Volume = (length × width^2)/2

 Knocking down PGLS in cancer cell reshaped tumor microenvironment

Finally, we have established an in situ model of hepatocellular carcinoma (HCC) using Hep1-6 cells (Fig. 9A). It was observed that the knockdown of PGLS resulted in a remarkable decrease in the liver weight relative to the overall body weight of the mice (Fig. 9B). Furthermore, the survival rate of mice bearing tumors was significantly extended (Fig. 9C). Subsequently, we examined immune cells obtained from the tumor tissues using flow cytometry. The data revealed that the knockdown of PGLS effectively suppressed the accumulation of M2 macrophages (Fig. 9D) while promoting the increase of M1 macrophages (Fig. 9E), CD8⁺ T cells (Fig. 9F), and CD4⁺ T cells (Fig. 9G) within the tumor microenvironment. Additionally, the proportion of Treg cells, a subset of immunosuppressive CD4⁺ T cells, showed a significant decline after PGLS knockdown (Fig. 9H). Furthermore, the expression of Foxp3, a pivotal functional molecule for Treg cells, was found to be significantly reduced (Fig. 9I). These findings collectively suggest that PGLS knockdown exhibits positive effects on both survival extension and amelioration of the tumor immune microenvironment in HCC-bearing mice. Importantly, these results emphasize the regulatory role of PGLS in immune function and its potential as a viable therapeutic target for cancer treatment.

Fig. 9.

PGLS knockdown improves survival and remodels the tumor immune microenvironment in an orthotopic HCC model. (A) Orthotopic HCC model was established by intrahepatic injection of 5 × 10^5 Hep1-6 cells/mouse. Representative liver images collected on day 15 post-implantation. (B) Quantification of liver-to-body weight ratio in the orthotopic HCC model (n = 5). (C) Kaplan-Meier survival curve of tumor bearing mice (n = 7). (D) Detection of tumor tissue M2 macrophages (CD45⁺CD11b⁺F4/80⁺CD206⁺) by flow cytometry (n = 5). (E) Detection of tumor tissue M1 macrophages (CD45⁺CD11b⁺F4/80⁺CD86⁺) by flow cytometry (n = 5). (F) Detection of tumor tissue CD8⁺ T cells (CD45⁺CD3⁺CD8⁺) by flow cytometry (n = 5). (G) Detection of tumor tissue CD4⁺ T cells (CD45⁺CD3⁺CD4⁺) by flow cytometry (n = 5). (H) Detection of tumor tissue Tregs (CD45⁺CD3⁺CD4⁺CD25⁺) by flow cytometry (n = 5). (I) Flow cytometry detection of Foxp3 content of CD4⁺CD25⁺ cells in tumor tissues, with average fluorescence in the upper left corner (n = 5)

 Protein interaction networks and functional analysis of PGLS

In order to further analyze the potential functions and mechanisms of action of PGLS, we used STRING and GeneMANIA to analyze the protein interaction networks of PGLS (Figures S13A and S13B). The results showed that in these two databases, PGLS mainly interacts with other enzymes in the PPP pathway, including G6PD, RPIA, H6PD, PGD. Further GO analysis of the protein interaction networks of PGLS revealed that these genes are involved in multiple core processes in the PPP pathway, including capacity generation, redox balance, carbon conversion, and pentose biosynthetic process (Figure S13C). This further highlights the critical role of PGLS in the PPP pathway and cellular metabolism.

To further identify key molecules regulating the PGLS signaling pathway, we stratified the TCGA-LIHC cohort into PGLS high- and low-expression groups based on median expression levels. Differential gene analysis revealed 67 upregulated genes in the PGLS high-expression group (Figure S14A). GO enrichment analysis demonstrated these genes participate in multiple biological processes, molecular functions, and cellular components (Figure S14B), highlighting the functional diversity of PGLS. Among these genes, ETV4, CD24, and EPCAM were prioritized for investigation. Using the HCC-related single-cell database PRHCdb, we observed significant co-expression patterns of PGLS with these three genes in cancer cells (Figure S14C). JASPAR database analysis further identified evolutionarily conserved ETV4 binding sites in the promoter regions of both human and murine PGLS genes (Figure S14D), suggesting ETV4 may transcriptionally regulate PGLS expression. However, ChIP and luciferase reporter assays are required for validation. CD24, an immune checkpoint molecule, and EPCAM, implicated in proliferation and migration, showed significant downregulation upon PGLS knockdown in Huh7 cells (Figure S14E), indicating their potential role as PGLS downstream effectors. Nevertheless, additional experimental evidence is needed to establish PGLS-mediated regulation of the tumor immune microenvironment and progression via CD24 and EPCAM signaling.

Discussion

PGLS is a crucial enzyme involved in the metabolism of PPP and is fundamental for the growth and survival of normal cells [42]. However, research on its application in tumors is limited. In the current investigation, pan-cancer analysis revealed that PGLS expression was significantly upregulated in almost all cancerous tissues compared to their normal paracancerous counterparts. Furthermore, high levels of PGLS were associated with unfavorable prognosis in several cancer types, indicating its potential as a therapeutic target. Nonetheless, the precise role of PGLS in different cancers remains elusive. We conducted in vitro experiments and preliminarily confirmed the importance of PGLS for the proliferative and migratory capacity of Huh7 and A498 cell lines. Additionally, in vivo studies demonstrated that PGLS had a pronounced effect on promoting tumor growth. Nevertheless, the molecular mechanism underlying this activity requires further elucidation.

The tumour microenvironment is a significant limiting factor in the efficacy of T cell-mediated anti-tumour immunotherapy [43, 44]. Enhancing immunotherapy efficacy primarily involves increasing the infiltration of cytotoxic T lymphocytes into the tumour while mitigating factors in the tumour microenvironment that inhibit their function. A multitude of cytokines, chemokines and immune checkpoints regulate this process. This study examines the correlation between PGLS expression in different cancers and the expression of immunomodulatory genes as well as immune cells. Notably, PGLS was significantly associated with a large number of immunomodulatory genes, including CD276, PVRL2, VEGFB and TGFB1, of which are known to impede immunotherapy [25, 26, 45, 46]. Furthermore, high PGLS expression correlated with increased infiltration of Tregs and M2 macrophages. The presence of these immunosuppressive genes and cells severely limits the effectiveness of immunotherapy. While it is unclear how PGLS regulates immunomodulatory genes and immune cell infiltration, it presents an attractive entry point for synergistic immunotherapy.

Tumor heterogeneity reflects the diverse growth characteristics of tumor cells and can be utilized to prognose treatment sensitivity [27]. In this study, we computed the association between PGLS and eight distinct features of tumor heterogeneity, encompassing TMB, MATH, MSI, etc. We discovered that PGLS was significantly linked with various tumor heterogeneity features in several tumors. For instance, in colorectal cancer, upregulated PGLS expression often coincided with reduced levels of TMB, MSI, and NEO, thus hindering T-cell-mediated anti-tumor immunity against the malignant cells, ultimately compromising the efficacy of immunotherapy. Conversely, STAD showing high PGLS expression is predicted to exhibit lesser sensitivity to PARP inhibitors and platinum drugs owing to higher HRD levels. Thus, PGLS may play a crucial role in regulating therapeutic sensitivity via tumor heterogeneity control, making it a useful molecular marker for predicting therapeutic efficacy.

Tumour stem cells possess an infinite capacity for proliferation, facilitate tumour growth and metastasis, and contribute to treatment resistance [34]. The initiation of tumour stem cell formation is reliant on oncogenic metabolic alterations, with the pentose phosphate pathway (PPP) playing a crucial role [47]. Our current investigation has identified that PGLS expression in a variety of tumours is significantly associated with several tumour stemness characteristics. Particularly in gliomas and sarcomas, PGLS expression demonstrated a significant and positive correlation with all stemness traits, suggesting that PGLS may promote the expression of tumour stemness genes and consequently augment tumour proliferation, metastasis, and recurrence potential.

Animal models are essential tools for validating PGLS function in vivo. When analyzing the tumor growth-promoting effects of PGLS, we employed subcutaneous xenograft models. This approach offers advantages such as ease of tumor volume monitoring and technical simplicity, but suffers from a distorted tumor microenvironment, limiting its application to evaluating PGLS effects on in vivo proliferation. For assessing the impact of PGLS knockdown on the tumor immune microenvironment, we utilized orthotopic liver cancer models. These models preserve liver-specific niches and better simulate the tumor microenvironment, though their technical complexity and challenges in real-time tumor quantification remain constraints. The subcutaneous model provides cost-effectiveness, while the orthotopic model more closely mirrors clinical complexity. Their complementary use balances experimental efficiency and physiological relevance.

This study identified distinct PGLS expression patterns across various cancers, with differential impacts on prognosis. These patterns correlated differently with immune checkpoint markers, immune molecules, and immune cell subsets, likely due to tumor heterogeneity. Tumors exhibit genetic, epigenetic, and expression diversity, leading to varying treatment responses. Additionally, the tumor microenvironment, including immune infiltration and checkpoint presence, varies by cancer type and tumor region, causing differences in results. Further in vitro and in vivo research is needed to explore PGLS’s role in cancer progression and its interaction with immune pathways, which could help develop more targeted therapies.

Conclusion

Overall, in our investigation, we have observed a notable upregulation in the expression of the PGLS gene in a majority of tumor samples. Additionally, we have identified a correlation between elevated levels of PGLS expression and the progression of tumors and unfavorable prognoses. These results signify that PGLS holds promise as both a prospective therapeutic target and a dependable predictor for prognostic factors and treatment response across diverse tumor types. Experimental downregulation of PGLS has been demonstrated to significantly impede the proliferation of tumor cells both in vitro and in vivo. Moreover, it leads to a prolonged survival period in mice bearing tumors and effectively enhances the immune microenvironment.

Limitations of the study

This study has several limitations that should be acknowledged. First, in our bioinformatics analysis, we solely relied on datasets from the TCGA database, without incorporating external datasets to validate or broaden the scope of our findings. This may limit the generalizability of our conclusions, as the inclusion of additional datasets could provide more robust evidence and help confirm the reliability of our results.Additionally, this article does not provide a comprehensive or in-depth analysis of the single-cell landscape of PGLS, an area that could significantly enhance our understanding of its cellular heterogeneity and the underlying mechanisms driving its pathology. A more detailed exploration of the single-cell transcriptomic profiles would offer valuable insights into the specific cell types and pathways involved.Moreover, while this study identifies certain biological effects and immune regulatory functions associated with PGLS, it does not fully investigate the specific molecular mechanisms behind these effects. Understanding the molecular pathways and interactions that govern the biological and immune responses in PGLS is crucial for unraveling the disease’s complexity and could offer new therapeutic targets.Addressing these limitations through future research will be important to gain a deeper understanding of PGLS and its broader implications in both basic and clinical settings.

Author contributions

S.L. analyzed the data and performed the cell and animal experiments. S.L. and W.W. wrote the manuscript. S.H. and W.W. supervised the research and revised the manuscript. All authors reviewed the manuscript.

Funding

This research was supported by the National Natural Science Foundation of China (Nos. 81874231, 81903173) and the Hubei Provincial Natural Science Foundation (No. 2025AFB040).

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

This retrospective study utilized de-identified data from public databases, containing no personally identifiable information; thus, ethics committee approval was not required. The research complied with the Helsinki Declaration and Ethical Guidelines for Public Health Research regarding public data use. The experimental protocol was approved by the Animal Ethics Committee of Huazhong University of Science and Technology (IACUC No. 2910), and all animal experiments strictly adhered to its guidelines.

Consent to publish

All authors consented to the publication.

 Human and animal rights

This study involved no direct human participation or recruitment; therefore, informed consent was not required. All data were obtained from authorized public databases and used in compliance with their access agreements.

Competing interests

All authors have no conflicts of interest with regard to this study.