Expression profile of GSDMB in oral squamous cell carcinoma and its impact on tumor immune microenvironment and prognosis

Abstract

Background

Oral squamous cell carcinoma (OSCC) is one of the most common malignant tumors of the head and neck region worldwide, with its development influenced by a multitude of factors such as genetic mutations and the immune microenvironment. Despite the availability of current treatment options, issues like drug resistance and recurrence persist. The Gasdermin B (GSDMB) gene encodes the protein Gasdermin B which exhibits complex biological functions in various cancers, notably playing a significant role in cell death and immune responses. However, the role of GSDMB in OSCC remains unclear. Therefore, this study aims to investigate the expression profile of GSDMB in OSCC and its potential biological functions.

Methods

This study employs a bioinformatics approach to analyze the expression landscape of GSDMB in OSCC based on multi-omics data. Initially, we analyzed the expression of GSDMB across pan-cancer types and categorized OSCC samples into high and low-expression groups according to the median expression value. Subsequently, differences between these two groups concerning tumor mutation burden (TMB), immune microenvironment, in silico drug sensitivity analysis, and prognosis were explored. Additionally, single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics data (stRNA-seq) were utilized to analyze the expression characteristics of GSDMB across different cell types. To further support the computational prediction, in vitro 5-fluorouracil (5-FU) sensitivity assays were performed in SCC25 cells with or without GSDMB knockdown.

Results

The expression of GSDMB was significantly higher in OSCC tumor tissues compared to normal tissues, with significant differences observed across tumor grades, age groups, and N stages. The TMB was significantly higher in the high-expression group than in the low-expression group, showing a positive correlation between GSDMB expression levels and TMB. Survival analysis revealed that patients in the high-TMB group had significantly lower survival rates compared to those in the low-TMB group, while high GSDMB expression correlated with poorer prognosis. Immune landscape analysis indicated higher infiltration levels of CD8+ T cells and natural killer cells in the high-expression group, along with enhanced immune functions such as antigen presentation, cytotoxic activity, and immune checkpoint activity. In silico drug sensitivity analysis suggested that patients with high GSDMB expression may be more responsive to multiple anticancer agents. Consistently, in SCC25 cells, GSDMB knockdown significantly increased the IC50 of 5-FU, indicating reduced chemosensitivity upon GSDMB depletion. Furthermore, scRNA-seq analysis uncovered differential expression of GSDMB among distinct immune cell populations, suggesting its potential significance in the tumor immune microenvironment.

Conclusion

GSDMB may play a crucial role in the development of OSCC and is closely associated with TMB, the immune microenvironment, and drug response prediction by in silico drug sensitivity analysis. Its expression not only influences the tumor immune response but also has the potential to serve as a biomarker and therapeutic target for OSCC. This study provides novel insights into the functional role of GSDMB in OSCC and its clinical application as a prognostic marker.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12903-026-07781-1.

Introduction

Oral squamous cell carcinoma (OSCC) is one of the most prevalent malignant tumors of the head and neck region globally, with a higher incidence particularly in developing countries. The pathogenesis of OSCC is complex, involving multiple carcinogenic factors such as long-term tobacco and alcohol use, human papillomavirus (HPV) infection, genetic susceptibility, and environmental influences [1–3]. The development of OSCC typically proceeds from normal epithelial cells through a series of genetic and epigenetic alterations to precancerous lesions and ultimately to cancer. Common driver mutations include alterations in genes such as p53, HRAS, and K-ras, which influence processes like cell cycle regulation and DNA repair, thereby promoting tumor initiation and progression [4, 5]. Despite advances in molecular biology, the heterogeneity of OSCC at both genetic and cellular levels remains a major obstacle to effective risk stratification and individualized treatment.

Current treatments for OSCC primarily rely on surgical resection, radiation therapy, and chemotherapy; however, due to issues of therapeutic resistance and tumor recurrence, patient survival rates remain low. Therefore, identifying reliable prognostic markers is of critical importance. Recent high-throughput genomic and transcriptomic studies have uncovered potential markers that are closely related to the aggressiveness and prognosis of OSCC. Moreover, the application of single-cell analysis techniques has further revealed the role of the tumor microenvironment and immune cell infiltration, offering new directions for early diagnosis and targeted therapy of OSCC [6–8]. In particular, emerging evidence suggests that tumor immune heterogeneity and mutation burden play pivotal roles in shaping therapeutic response and clinical outcome in OSCC.

Gasdermin B (GSDMB) encodes the protein Gasdermin B, which participates in regulating cell death, specifically pyroptosis and apoptosis. It promotes the release of cellular contents by forming membrane pores, thereby inducing cell death. In various cancers, GSDMB plays complex roles. Studies have shown that in non-small cell lung cancer, breast cancer, hepatocellular carcinoma, gastric cancer, colorectal cancer, and ovarian cancer, the expression of GSDMB is closely associated with tumor cell survival, metastasis, and therapeutic response [9, 10]. Specifically, GSDMB may enhance the efficacy of chemotherapy by mediating pyroptosis, regulate tumor immune responses, or participate in mechanisms of immune evasion [11, 12]. Consequently, GSDMB not only serves as a potential biomarker for various cancers but also represents a promising target for cancer therapy. However, despite these advances, the expression characteristics, prognostic significance, immune-related functions, and therapeutic relevance of GSDMB in OSCC have not been systematically investigated.

This study aims to integrate multi-omics data of OSCC and employ a bioinformatics approach to characterize the expression landscape of GSDMB across multiple cancer types. Based on the median expression level of GSDMB, OSCC samples were stratified into high- and low-expression groups. Differences between these groups were then analyzed in terms of tumor mutation burden (TMB), immune microenvironment, drug sensitivity, and patient prognosis. Additionally, scRNA-seq and stRNA-seq data from OSCC tissues were utilized to examine GSDMB expression across distinct cell types. Finally, bioinformatics findings were validated using immunohistochemistry in clinical specimens and functional assays in OSCC cell lines, providing experimental evidence to support the clinical and biological significance of GSDMB in OSCC.

Methods

Dataset acquisition

RNA-seq data for the TCGA-HNSC cohort, including samples from 330 OSCC patients and 32 normal tissues, were obtained from The Cancer Genome Atlas Program (TCGA) database (https://www.cancer.gov/ccg/research/genome-sequencing/tcga). Corresponding clinical and survival information for these samples was collected from the UCSC Xena database (https://xena.ucsc.edu/). Additionally, single-cell RNA-seq (scRNA-seq) data from five OSCC tissue samples were retrieved from the GSE172577 dataset in the Gene Expression Omnibus (GEO) database, while stRNA-seq data from four OSCC samples were sourced from the GSE220978 dataset [13].

Pan-cancer analysis

Expression data across various cancers for the target genes were collected from the TIMER2.0 database (http://timer.cistrome.org/) [14]. Univariate Cox analysis was performed based on prognostic and expression data to identify which cancers showed significant associations with the target gene’s prognosis (p < 0.05).

Differential expression analysis

All OSCC samples were divided into high and low-expression groups based on GSDMB expression levels. Differential expression analysis between the two groups was conducted using the “limma” package [15], resulting in the identification of 310 differentially expressed genes (DEGs) with criteria |logFC|>0.5 and adj.p < 0.05.

TMB analysis

The TMB load of OSCC samples was assessed using the “maftools” package. The calcTMB(·) function evaluated TMB for each sample, and plotTMBDistribution(·) visualized the results. Patients were categorized into high and low TMB groups based on median TMB scores, and Kaplan-Meier (KM) survival curves compared survival outcomes between the groups.

Enrichment analysis (GO, KEGG, and Metascape)

Gene set enrichment analyses for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were conducted using the “clusterProfiler” package in R software, followed by visualization of significant pathways using “ggplot2”. Moreover, the Metascape website (http://metascape.org/gp/index.html) was utilized for additional enrichment analysis [16], enabling the visualization of significant pathways through bar charts and network diagrams.

Immune infiltration and immunotherapy analysis

Single-sample Gene Set Enrichment Analysis (ssGSEA) was employed to analyze immune function scores and the abundance of immune cell infiltration by evaluating the enrichment of immune gene sets within individual samples [17]. Additionally, the Immune Phenotype Score (IPS), calculated using a random forest algorithm based on the expression of representative genes or gene sets related to MHC molecules, immune modulators, effector cells, and inhibitory cells, was used to predict patient responses to anti-CTLA-4 and anti-PD-1 antibodies. IPS data were downloaded from The Cancer Imaging Archive (TCIA) database (https://www.cancerimagingarchive.net/).

Drug sensitivity prediction and molecular docking analysis

To predict the sensitivity of high and low-expression group OSCC patients to chemotherapeutic drugs, models were trained using the trainModel(·) and predictSensitivity(·) functions in the “oncoPredict” package, estimating the half-maximal inhibitory concentration (IC50) [18]. Differences in drug IC50 values between high-risk and low-risk groups were tested for statistical significance (p < 0.05). To identify potential therapeutic compounds for OSCC, small molecule compounds effective against OSCC were inferred from the DSigDB database (http://dsigdb.tanlab.org/DSigDBv1.0/) by combining prognostic gene expression profiles. After performing drug enrichment analysis using the “clusterProfiler” package, the top four small molecules (adj.p < 0.05) were selected as candidate drugs. Three-dimensional structure information for these compounds was retrieved from the PubChem database (https://www.ncbi.nlm.nih.gov/pccompound), and corresponding protein structures were obtained from the Protein Data Bank (PDB) database (http://www.rcsb.org/). Molecular docking simulations were conducted using the CB-Dock2 platform (https://cadd.labshare.cn/cb-dock2/index.php) [19, 20], assessing the efficacy and specificity of each candidate drug. The analysis was performed to predict the potential binding interactions between candidate drugs and the GSDMB protein at the structural level. The three-dimensional structure of GSDMB was retrieved from public databases, and docking simulations were conducted to estimate binding modes and binding affinities. This analysis was intended solely as an in silico prediction of drug-protein interactions rather than experimental validation of GSDMB expression regulation.

scRNA-seq baseline analysis and cell type annotation

scRNA-seq data from six OSCC samples were processed and analyzed using the Seurat package [20]. Data loading and object creation were performed using Read10X(·) and CreateSeuratObject functions, respectively. Quality control measures included filtering cells based on feature and count thresholds, normalization using LogNormalize, identification of variable features, Principal Component Analysis (PCA), batch effect correction with Harmony [21], clustering, and cell type annotation using SingleR [22]. Results were visualized via t-Distributed Stochastic Neighbor Embedding (t-SNE) plots to confirm cell type distribution.

Cell communication and pseudotime analysis

Cell communication and pseudotime analyses were performed on OSCC scRNA-seq data to uncover interactions among different immune cell populations and their temporal changes. CellChat was used to infer intercellular communications [23], and monocle was applied for pseudotime trajectory reconstruction [24]. Network and trajectory visualization methods revealed the dynamic changes and interactions within the immune cell population over time.

stRNA-seq baseline analysis

Spatial transcriptomics RNA-seq data from OSCC samples were analyzed using the “Seurat” package. Data preprocessing, quality control, normalization, PCA, and clustering were carried out similarly to scRNA-seq analysis. Spatial expression patterns of GSDMB were also explored.

IHC validation and in vitro functional assays

Tissue specimen collection

From May to October 2024, tissue specimens were collected from patients with OSCC who underwent surgical treatment at the Fourth Hospital of Hebei Medical University. Inclusion criteria required a histopathological diagnosis of primary OSCC, while patients who had received neoadjuvant chemotherapy, targeted therapy, or radiotherapy prior to surgery were excluded. During surgery, tumor tissues and paired adjacent normal tissues (≥ 3 cm from the tumor margin) were obtained. A total of 20 OSCC tumor and 20 matched normal tissue samples were collected. All specimens were fixed in formalin and embedded in paraffin for subsequent analysis. Written informed consent was obtained from each patient before participation. This study was approved by the Ethics Committee of the Fourth Hospital of Hebei Medical University (Approval No. 2024KY008).

IHC

Paraffin-embedded tissue sections were baked at 65 °C for 2 h, deparaffinized with xylene, and rehydrated through graded ethanol. Antigen retrieval was performed in 0.01 M sodium citrate buffer (pH 6.0) using high-pressure heating. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide and a commercial blocking solution (ZSGB-BIO, China). Sections were incubated overnight at 4 °C with rabbit anti-human GSDMB antibody (1:50, 12885-1-AP, Proteintech, China), followed by HRP-conjugated goat anti-rabbit IgG secondary antibody (ZSGB-BIO, China) at 37 °C for 30 min. DAB was used for visualization, and sections were counterstained with hematoxylin, dehydrated, and mounted.

All sections were independently evaluated in a double-blind manner by two experienced pathologists. IHC results were assessed using a semi-quantitative scoring system based on staining intensity and the percentage of positive cells. Staining intensity was scored as follows: 0 (no staining), 1 (light yellow), 2 (brownish-yellow), and 3 (dark brown). The proportion of positively stained cells was scored as: 0 (< 5%), 1 (5–25%), 2 (26–50%), 3 (51–75%), and 4 (> 75%). The final immunoreactivity score was calculated by multiplying the intensity score by the proportion score, yielding a total score ranging from 0 to 12. Samples were classified into two groups: low expression (score < 6) and high expression (score ≥ 6).

Cell culture and transfection

The normal human oral keratinocyte cell line HOK and the human OSCC cell line Tca8113 were obtained from Yaji Biotechnology Co., Ltd. (Shanghai, China). The human OSCC cell lines SCC4 and SCC25 were purchased from the American Type Culture Collection (ATCC, USA). Cells were routinely cultured in high-glucose Dulbecco’s Modified Eagle Medium (DMEM; Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Life-iLab, China) and 1% penicillin-streptomycin (P/S; Biosharp, China). Cultures were maintained at 37 °C in a humidified incubator with 5% CO₂ (Thermo Scientific, USA). For transfection, SCC25 cells in the logarithmic growth phase were seeded into 6-well plates. When the cell confluence reached approximately 60%, the medium was replaced with serum-free and antibiotic-free DMEM, and cells were incubated for an additional 4 h. Transfection was then performed using Lipofectamine™ 2000 reagent (Invitrogen, USA) according to the manufacturer’s instructions. After 8 h of transfection, the medium was replaced with complete DMEM containing 10% FBS, and the cells were further cultured. At 48 h post-transfection, cells were harvested for transfection efficiency evaluation by Western blotting. Small interfering RNAs (siRNAs) targeting GSDMB and the negative control (si-NC) were obtained from GenePharma Co., Ltd. (Shanghai, China). The siRNA sequences were as follows: si-GSDMB#1: 5’-GCCUUGUUGAUGCUGAUAGAUTT-3’, si-GSDMB#2: 5’-CCAGACAAGCCUCUCCUAATT-3’.

Western blot

Cells were lysed in RIPA buffer (Solarbio, China), and protein concentrations were measured using a BCA kit (Solarbio, China). Samples were denatured at 100 °C for 10 min, separated by 10% SDS-PAGE, and transferred to PVDF membranes (Millipore, Germany). After blocking with 5% non-fat milk, membranes were incubated with primary antibodies against GSDMB (1:1000, ab215729, Abcam) and β-Actin (1:10,000, R380624, Zenbio) at 4 °C overnight, followed by HRP-conjugated secondary antibody (1:10,000, 511203, Zenbio) for 1 h at room temperature. Signals were detected using ECL (MCE, USA) and visualized with the gel imaging system (Gene, UK).

Cell proliferation assay (CCK-8)

Cells were seeded into 96-well plates at a density of 1 × 10³ cells/well in 100 µL of complete medium. Cell proliferation was assessed at 0, 24, 48, and 72 h post-transfection. At each time point, 10 µL of CCK-8 reagent (Beyotime, China) was added to each well, followed by incubation in the dark for 4 h. The optical density (OD) was measured at 450 nm using a microplate reader (Biotek, USA) to evaluate cell proliferation.

Colony formation assay

Cells (2 × 10³/well) were seeded into 6-well plates and cultured in 2 mL complete medium, with medium replaced every 48 h. After 10–14 days, when colonies (> 50 cells) were visible under a microscope, cells were fixed with 4% paraformaldehyde for 20 min and stained with 0.5% crystal violet for 15 min. Plates were rinsed with PBS, air-dried, and imaged. Colonies were counted using ImageJ software (v1.54).

Wound healing assay

Cells were seeded into 6-well plates and cultured until reaching approximately 90% confluence. A linear scratch was made across the cell monolayer using a sterile 200 µL pipette tip. Detached cells were removed by gently washing twice with PBS, and the medium was replaced with serum-free DMEM. Images of the wound area were captured at 0 and 24 h using a microscope (Leica, Germany). The wound area was quantified using ImageJ software, and the migration rate was calculated using the formula: Migration rate (%) = (wound area at 0 h − wound area at 24 h) / wound area at 0 h × 100%.

Cell invasion assay (Transwell)

The upper chambers of Transwell inserts (LabSelect, China) were coated with Matrigel (Beyotime, China) to form a uniform gel layer. A total of 2 × 10⁵ cells were seeded in the upper chamber with 200 µL of serum-free medium. The lower chamber was filled with 600 µL of complete medium containing 10% FBS. After 24 h of incubation, non-invading cells were gently removed from the upper surface with a cotton swab. Inserts were washed with PBS, fixed with 4% paraformaldehyde for 20 min, and stained with 0.5% crystal violet for 15 min. Invaded cells on the lower surface were imaged under a microscope and quantified using ImageJ software.

Cytotoxicity assays

Cytotoxicity was assessed using the CytoTox 96^® Non-Radioactive Cytotoxicity Assay (Promega, USA) to measure lactate dehydrogenase (LDH) release, in accordance with the manufacturer’s instructions.

ELISA

The levels of IL-18 and IL-1β were quantified using ELISA kits (Beyotime, China) according to the manufacturer’s protocols.

Drug sensitivity assays

Cells (5 × 10³ per well) were seeded into 96-well plates. After attachment, cells were treated with 5-fluorouracil (5-FU; MCE, USA) at 0, 6.4, 32, 160, 800, and 4000 µM for 48 h. Cell viability was then measured using the CCK-8 assay. Dose–response curves were generated based on the relationship between vehicle-normalized absorbance and log10-transformed drug concentrations. The IC50 was defined as the concentration that reduced viability by 50% relative to untreated cells.

Statistical tests

Statistical analyses were conducted using R software (v4.2.2). Wilcoxon tests were used for comparing differences between groups in differential expression analysis. Kruskal-Wallis rank sum tests were applied when comparing multiple clinical factors. Log-rank tests evaluated differences in survival outcomes. Spearman correlation coefficients measured the relationship between immune cell infiltration and GSDMB expression. In the experimental section, the chi-square test was employed to analyze IHC results. Data from the CCK-8 assay were analyzed using two-way repeated measures ANOVA. One-way ANOVA followed by Tukey’s HSD post-hoc test was used to assess differences among groups in colony formation, wound healing, and invasion assays. A p-value < 0.05 indicated statistical significance unless otherwise specified.

Results

Expression landscape of GSDMB in OSCC

Figure 1 provides a technical roadmap for this study. We initially explored the expression landscape of GSDMB in OSCC samples across different groupings. In terms of tumor grade, a significant difference in GSDMB expression was observed (Fig. 2A). Compared with Grade 1 (G1), GSDMB expression was significantly elevated in Grades 2 and 3 (G2, G3), while no statistically significant change was noted in Grade 4 (G4). Analysis based on patient age revealed that GSDMB expression was significantly higher in patients aged ≥ 65 years compared to those < 65 years (Fig. 2B). Furthermore, in N-stage analysis, GSDMB expression was higher in N1, N2, and N3 stages compared to the N0 stage, with a particularly significant increase in the N3 stage (Fig. 2C). Box plot analysis indicated that GSDMB expression in OSCC tumor tissues was significantly higher than in normal tissues, with statistical significance (Fig. 2D, p < 0.001). This suggests that GSDMB may play a potential role in the occurrence and development of OSCC. Moreover, GSDMB has diagnostic implications for OSCC. The ROC curve analysis showed that GSDMB has a high diagnostic performance, with an Area Under the Curve (AUC) value of 0.798 (95% CI: 0.718–0.877), indicating that it can distinguish between OSCC and normal tissues with relatively high sensitivity and specificity (Fig. 2E). Furthermore, in the supplementary materials, Figure S1 of this article presents the KM survival curves after dividing OSCC patients into high and low expression groups based on GSDMB expression levels, and then stratifying them according to other clinical stages (Stage, T, and N stages).

Fig. 1.

Schematic representation of the technical workflow employed in this study

Fig. 2.

Expression landscape of GSDMB across different groups (330 OSCC tissues, 32 normal tissues). A-C Differential expression of GSDMB among Grade categories, age groups, and N stages. D Boxplot showing the expression difference of GSDMB between OSCC cases and controls. E Receiver Operating Characteristic (ROC) curve for GSDMB. F Forest plot from univariate Cox regression analysis based on GSDMB expression levels and prognostic outcomes across various cancers. G Boxplot illustrating differences in GSDMB expression between control and disease groups across pan-cancer samples. “wilcox.test” is used to calculate the significance p-value (* p<0.05, ** p<0.01, *** p<0.001)

This paper further investigated the expression of GSDMB and its prognostic significance in pan-cancer analysis. Using univariate Cox regression analysis to examine the association between GSDMB expression levels and prognosis outcomes across various cancer types, the generated forest plot demonstrated that GSDMB had significant prognostic relevance in multiple cancer types (Fig. 2F). Additionally, GSDMB showed significant differences between tumor and normal groups across several cancer types (Fig. 2G). These findings suggest that GSDMB could be a key molecule in multiple cancers and possesses potential clinical value.

TMB landscape and expression differences between groups

We stratified all OSCC samples into high and low expression groups based on GSDMB expression levels. Through gene mutation analysis of the two groups, it was found that both exhibited a high frequency of mutations. In the high-expression group, 93.67% of the samples (148/158) had genetic mutations, with TP53, TTN, and FAT1 being the genes with the highest mutation frequencies (Fig. 3A). In the low-expression group, 96.25% of the samples (154/160) underwent genetic mutations, where TP53, TTN, and CDKN2A showed the highest mutation frequencies (Fig. 3B). This suggests that GSDMB expression may be associated with tumor mutation patterns. The TMB in the high-expression group was significantly higher than that in the low-expression group (Fig. 3C, p = 0.0027). Further analysis revealed a significant positive correlation between GSDMB expression and TMB (Fig. 3D, R = 0.22, p = 5.9e-05), indicating that GSDMB may play an important role in tumor mutational burden. Kaplan-Meier survival curve analysis showed that patients in the high TMB group had significantly lower overall survival rates compared to those in the low TMB group (Fig. 3E, p = 0.012). Patients were further categorized into four groups based on TMB and GSDMB expression levels (high TMB-high expression, high TMB-low expression, low TMB-high expression, low TMB-low expression). Combined analysis indicated that grouping based on TMB and GSDMB expression levels had significant prognostic discrimination power (Fig. 3F, p = 0.031). Notably, the high TMB-high expression group had the lowest survival rate, whereas the low TMB-low expression group had the highest.

Fig. 3.

Landscape of Tumor Mutational Burden (TMB) in high and low GSDMB expression groups (318 OSCC tissues). A-B Waterfall plots depicting mutations in high and low GSDMB expression groups. C Boxplot comparing TMB differences between the two groups. D Scatter plot of the correlation between TMB and GSDMB expression levels. E Kaplan-Meier survival curves for high and low TMB groups. F KM survival curves for four sample groups classified by combined TMB and GSDMB expression levels

Differential expression analysis between the two groups identified multiple significantly DEGs (Fig. 4A). A heatmap was generated to illustrate the clustering patterns of DEGs in the high-expression and low-expression groups, revealing clear stratification of DEG expression levels between the two groups (Fig. 4B). GO and KEGG enrichment analyses, visualized using circular plots combined with network diagrams (Fig. 4C) and bar plots (Fig. 4D), highlighted significantly enriched pathways such as inflammatory response (GO:0006954), antigen processing and presentation (GO:0019882), and adaptive immune response (GO:0002250). Moreover, KEGG pathways including ECM-receptor interaction (hsa04512), PI3K-Akt signaling pathway (hsa04151), and focal adhesion (hsa04510) were also significantly enriched, suggesting their critical roles in the biological differences between the two groups. Gene Set Enrichment Analysis (GSEA) further revealed significant differences in key pathways, including ECM-receptor interaction (Fig. 4E), focal adhesion (Fig. 4F), Mitogen-Activated Protein Kinase (MAPK) signaling pathway (Fig. 4G), natural killer cell-mediated cytotoxicity (Fig. 4H), primary immunodeficiency (Fig. 4I), T-cell receptor signaling pathway (Fig. 4J), Transforming Growth Factor Beta (TGF-β signaling pathway (Fig. 4K), and Toll-like receptor signaling pathway (Fig. 4L). Collectively, these results indicate substantial differences in the molecular mechanisms related to immune responses, cell signaling, and adhesion between the two groups. Additional pathways are detailed in the supplementary material’s “GSEA” folder.

Fig. 4.

Analysis of differences between the two groups and GSEA analysis results (330 OSCC tissues). A Volcano plot displaying DEGs. B Heatmap of DEGs illustrating expression levels across samples. C Circle plot combined with a network view of GO/KEGG pathway enrichment analysis. D Bar plot summarizing the results of GO/KEGG enrichment analysis. E-L Gene set enrichment analysis (GSEA) results for key KEGG pathways. Specific pathways include ECM-receptor interaction (E), focal adhesion (F), MAPK signaling (G), natural killer cell-mediated cytotoxicity (H), primary immunodeficiency (I), T-cell receptor signaling (J), TGF-beta signaling (K), and Toll-like receptor signaling (L). Key statistics (NES: Normalized Enrichment Score; p-value; FDR: False Discovery Rate) are provided in the plots

Immune landscape between groups

In this section, we evaluated the immune landscape of GSDMB high and low expression groups based on the “ssGSEA” (single-sample Gene Set Enrichment Analysis) algorithm. Significant differences were observed in multiple aspects of immune cell infiltration and immune function between the two groups (Figs. 5A-B). In the high-expression group, the abundance of infiltrating immune cells such as CD8 + T cells and natural killer (NK) cells was significantly increased. Additionally, immune functions including antigen presentation, cytotoxic activity, and immune checkpoint activity were significantly enhanced (p < 0.05).

Fig. 5.

Immune landscape in high and low GSDMB expression groups. A-B Boxplots displaying differences in immune cell infiltration and function between the two groups as evaluated by ssGSEA algorithm. C-F Violin plots showing differences in immune phenotype scores between the two groups obtained from the TCIA database. The labels "ips_ctla4_neg_pd1_neg", "ips_ctla4_neg_pd1_pos", "ips_ctla4_pos_pd1_neg", and "ips_ctla4_pos_pd1_pos" refer to non-response or response to anti-CTLA-4 and anti-PD-1 antibodies. G-Q Scatter plots of correlations between significantly differing immune cells/functions and GSDMB expression between the two groups. "wilcox.test" is used to calculate the significance p-value

Furthermore, using the Tumor Immune Estimation Resource (TIMER) or TCIA database to analyze the distribution of different IPS, it was found that the high-expression group had significantly higher scores in the “ips_ctla4_neg_pd1_pos” and “ips_ctla4_pos_pd1_pos” immune phenotypes compared to the low-expression group (Figs. 5C-F). This suggests that the high-expression group of GSDMB may be more sensitive to PD-1 monoclonal antibody treatment and combined CTLA-4/PD-1 dual antibody therapy.

Finally, an analysis of the correlation between GSDMB expression levels and immune cell infiltration as well as immune function revealed significant positive correlations. Specifically, GSDMB expression showed a strong positive correlation with CD8 + T cell infiltration (R = 0.29, p = 1.7 × 10e^− 7), immune checkpoint activity (R = 0.23, p = 3.9 × 10e^− 5), and cytotoxic activity (R = 0.28, p = 2.6 × 10e^− 6) (Fig. 5G-Q). These results indicate that GSDMB might contribute to tumor microenvironment regulation by promoting immune cell infiltration and enhancing immune function.

Drug sensitivity and molecular docking analysis results

In this study, based on drug sensitivity analysis, it was found that the group with high expression of GSDMB exhibited significantly enhanced sensitivity to a variety of anticancer drugs, such as 5-fluorouracil, oxaliplatin, prednisolone, and topotecan, among others, with notably lower IC50 values compared to the low-expression group. These drugs encompass both classical chemotherapeutic agents and targeted therapies, including LGX-971 and MIM1, all of which showed a stronger therapeutic response in the high-expression group (Fig. 6A-I). This suggests that GSDMB may be closely associated with tumor cell drug sensitivity, providing a foundation for its potential as a predictive biomarker for drug sensitivity.

Fig. 6.

Results of drug sensitivity analysis. A-I Boxplots of compounds with significant differences in IC50 values between high and low expression groups. "wilcox.test" is used to calculate the significance p-value

On the other hand, molecular docking analysis was employed in this study to evaluate the binding modes between GSDMB and four key small molecules: HC toxin, vorinostat, rifabutin, and scriptaid. The candidate drugs may potentially interact with GSDMB through stable binding conformations with favorable binding energies. It indicates a putative structural basis for drug-GSDMB interactions, but do not provide direct evidence for suppression of GSDMB expression or activity, which requires further experimental validation (Fig. 7A-D). Specifically, HC toxin formed hydrogen bonds with Lys42 and Trp230, vorinostat established multiple hydrogen bonds with Glu153 and Lys154, rifabutin interacted with Asp156 and Phe156, whereas scriptaid engaged in strong interactions with Trp230 and Lys151. Furthermore, the chemical structures of HC toxin, vorinostat, rifabutin, and scriptaid revealed their potential for binding to GSDMB, offering theoretical support for their candidacy as potential GSDMB inhibitors (Fig. 7E-H).

Fig. 7.

. Molecular docking results for key small molecules (HC toxin, vorinostat, rifabutin, and scriptaid). A-D Docking results of GSDMB with four compounds. E-H Structures of key small molecules

Basic analysis of scRNA-seq data

In this section, we performed a comprehensive analysis of scRNA-seq data derived from OSCC tissues using the Seurat package, including data preprocessing, normalization, dimensionality reduction, clustering, and cell type annotation. The initial dataset consisted of 64,048 cells and 25,527 genes. After quality control filtering to remove low-quality cells (Fig. 8A), a total of 42,617 cells were retained for subsequent analyses.

Fig. 8.

Basic analysis results of scRNA-seq data. A Violin plots of six OSCC samples' scRNA-seq data post-merging and quality control, indicating the number of nonzero-expressed genes per cell, total UMI counts, and mitochondrial gene percentage. B Two-dimensional distribution of six samples with batch effects. C Distribution of different samples after removing batch effects using the "Harmony" package in R. D Heatmap predicting cell cluster identities using the "singleR" software package. E Two-dimensional distribution of different cell types. F Expression of GSDMB in different cell types. G Distribution of six OSCC samples across different cell types

PCA was conducted based on highly variable genes, and the top 12 principal components were selected for downstream analysis. t-SNE visualization demonstrated clear separation among distinct cell populations in the reduced dimensional space (Fig. 8B). To minimize technical variability, batch effects were corrected using the Harmony algorithm, resulting in improved integration and more coherent clustering of cells across samples (Fig. 8C).

Unsupervised clustering analysis classified the cells into 28 distinct clusters. Each cluster was annotated according to canonical marker genes, and cell identity was further validated using the SingleR algorithm. A total of nine major cell types were identified, including epithelial cells, keratinocytes, macrophages, T cells, monocytes, tissue stem cells, fibroblasts, B cells, and endothelial cells (Fig. 8D-E). Notably, GSDMB expression was predominantly enriched in the T cell population (Fig. 8F). In addition, substantial heterogeneity was observed in the relative proportions of different cell types across OSCC samples (Fig. 8G), highlighting the complexity of the tumor microenvironment.

Analysis of high and low expression cell populations and stRNA-seq data

During immune cell subset identification, four major immune cell populations-macrophages, monocytes, T cells, and B cells-were extracted from the scRNA-seq dataset for further analysis. To explore intercellular communication among these immune populations, the CellChat framework was applied to infer and visualize ligand-receptor interactions. Analysis of communication number and interaction strength revealed that signaling was most frequent among T cells, macrophages, and monocytes (Fig. 9A). Notably, the strongest communication intensity was observed between T cells and macrophages, suggesting their central roles in immune regulation within the OSCC microenvironment. Bubble plots further illustrated ligand–receptor interactions between T cells and other immune cell types (Fig. 9B). Specifically, T cells and B cells were major contributors to CCL signaling pathways (Fig. 9C), whereas macrophages showed prominent receptor activity in the TGFβ signaling pathway (Fig. 9D).

Fig. 9.

Results of cell communication and pseudotime trajectory analysis. A Number (left) and strength (right) of communications among different immune cells. B Pathway bubble charts for T-cell interactions as ligands (top) and receptors (bottom) with other cells. C-D Violin plots of CCL and TGFβ pathway genes expressed in different cell types. E-F Differentiation maps of cell types reduced by pseudotime. G Expression of GSDMB across different cell states and pseudotimes

To investigate immune cell differentiation dynamics, pseudotime trajectory analysis was performed (Fig. 9E-F). The results indicated that T cells and B cells were primarily distributed at earlier stages of differentiation, while macrophages and monocytes occupied intermediate to later stages. GSDMB expression exhibited dynamic and non-linear changes along the pseudotime trajectory (Fig. 9G), suggesting a potential association with immune cell state transitions and differentiation processes. In addition, stRNA-seq data from OSCC samples were analyzed to further characterize the spatial distribution of GSDMB expression. Cells were stratified into high- and low-expression groups based on the median GSDMB expression level (Fig. 10A). Comparison of cell-type composition between these groups revealed significant differences in cellular distribution (Fig. 10B). GSEA demonstrated that differential GSDMB expression was associated with multiple biological processes and signaling pathways related to tumorigenesis (Fig. 10C).

Fig. 10.

Classification of high and low expression cell clusters and GSDMB expression in stRNA-seq data. A All cells were divided into high and low expression clusters based on the median value of GSDMB. B Proportions of different cell types within high and low expression clusters. C Results of GSEA analysis for the two clusters. D Clustering results for four slices based on OSCC stRNA-seq data. E Cell type annotation results for four slices from OSCC stRNA-seq data. F Expression of GSDMB in different cell types

Spatial clustering analysis of four OSCC tissue sections revealed distinct GSDMB expression patterns across regions, reflecting spatial heterogeneity within the tumor microenvironment (Fig. 10D). Subsequent cell type annotation identified diverse cellular components within each section and further clarified the relationship between GSDMB expression and specific cell types (Fig. 10E). Collectively, these results indicate that GSDMB exhibits cell-type- and spatially dependent expression patterns, suggesting diverse functional roles in different cellular contexts within the OSCC tumor microenvironment (Fig. 10F).

GSDMB silencing attenuates the malignant properties and modulates 5-FU induced pyroptosis related responses in SCC25 cells

IHC analysis revealed that GSDMB protein was highly expressed in 14 of 20 OSCC tissue samples, compared with 7 of 20 normal oral mucosa samples (χ² = 4.912, p = 0.027; Fig. 11A). Among the cell lines analyzed, GSDMB expression was highest in SCC25 cells relative to HOK cells (Fig. 11B); therefore, SCC25 was selected for subsequent siRNA knockdown experiments. As shown in Fig. 11C, both si-GSDMB#1 and si-GSDMB#2 efficiently reduced GSDMB protein levels. Functionally, compared with the negative control (si-NC) group, SCC25 cells transfected with si-GSDMB#1 or si-GSDMB#2 exhibited decreased proliferation, colony formation, migration, and invasion (Fig. 11D–G).

Fig. 11.

GSDMB silencing attenuates the malignant properties and modulates 5-FU induced pyroptosis related responses in SCC25 cells. A IHC analysis of GSDMB expression in OSCC tissues and normal oral mucosa (n = 40). B–C Western blot analysis of GSDMB protein levels in HOK and OSCC cell lines and in SCC25 cells following siRNA transfection. D Cell viability assessed by the CCK-8 assay. E Representative images and quantification of colony formation. F Wound healing assay evaluating cell migration. G Transwell invasion assay. H Dose–response curves and corresponding IC50 values for 5-fluorouracil (5-FU). I Lactate dehydrogenase (LDH) release assay assessing 5-FU induced lytic cell death. J–K ELISA quantification of IL-1β and IL-18 levels in culture supernatants after 5-FU treatment. (* p < 0.05, ** p < 0.01, *** p < 0.001)

To further clarify the impact of GSDMB on chemosensitivity, we evaluated the response of SCC25 cells to 5-FU after GSDMB knockdown. Compared with the si-NC group [IC50 = 29.86 µM (27.12–32.60 µM)], the IC50 values of 5-FU were increased in si-GSDMB#1 and si-GSDMB#2 cells [51.55 µM (46.80–56.29 µM) and 41.37 µM (37.52–45.22 µM), respectively], indicating that GSDMB silencing reduced the sensitivity of OSCC cells to 5-FU (Fig. 11H). These findings suggest that OSCC patients with higher GSDMB expression may be more likely to benefit from 5-FU based chemotherapy, consistent with our bioinformatic analysis.

Given that GSDMB has been implicated in the regulation of pyroptosis, we next investigated whether GSDMB mediated chemosensitivity was associated with pyroptotic cell death under 5-FU treatment. LDH release assays showed that GSDMB knockdown attenuated 5-FU induced lytic cell death in SCC25 cells (Fig. 11I). Consistent with this, the secretion of the canonical pyroptosis related cytokines IL-1β and IL-18 was decreased in si-GSDMB#1 and si-GSDMB#2 cells compared with si-NC after 5-FU exposure (Fig. 11J–K). Taken together, these data support a role for GSDMB in promoting malignant phenotypes in OSCC cells and further suggest that GSDMB may enhance the cytotoxic efficacy of 5-FU by facilitating pyroptosis associated lytic cell death and inflammatory cytokine release.

Discussion

This study systematically analyzed the expression characteristics of GSDMB and its potential biological functions and clinical significance in cancer. Our results revealed that GSDMB plays a significant role in tumorigenesis, immune regulation, and patient prognosis, providing theoretical support for its application in cancer diagnosis, therapy, and prognostic evaluation. By integrating bulk transcriptomics, tumor mutation burden analysis, immune profiling, scRNA-seq, spatial transcriptomics, and experimental validation, this study offers a comprehensive and multi-dimensional understanding of GSDMB in OSCC.

Firstly, our data indicated a significant correlation between GSDMB expression levels and clinical stage, age, and lymph node metastasis. The high expression of GSDMB in OSCC tissues suggests its involvement in tumor development and progression. These associations imply that GSDMB expression may reflect the biological aggressiveness of OSCC. ROC curve analysis demonstrated that GSDMB has high sensitivity and specificity (AUC = 0.798) in distinguishing between tumor and normal tissues, indicating its potential as a diagnostic biomarker for cancer. Taken together, these findings support the feasibility of GSDMB as a clinically relevant indicator for OSCC diagnosis and disease stratification.

Secondly, we found a significant positive correlation between GSDMB expression and TMB, suggesting that GSDMB may be linked to genomic instability or mutation accumulation in OSCC. Although a direct causal relationship cannot be established based on the current data, this association indicates that GSDMB may participate in biological processes related to tumor evolution and immune recognition. Survival analysis showed that patients with high TMB had poorer prognosis, while GSDMB combined with TMB stratification could significantly distinguish patient prognosis risk. This finding underscores the potential clinical value of GSDMB as an integrated biomarker for guiding precision treatment strategies.

Differential expression and functional enrichment analyses between the two groups revealed that high expression of GSDMB was significantly associated with various immune-related biological processes, including “antigen processing and presentation,” “inflammatory response,” and “T cell activation”. These pathways are essential for shaping tumor–immune interactions and are consistent with the observed immune infiltration patterns in high-GSDMB tumors. Elevated expression of Focal Adhesion Kinase (FAK) in OSCC was significantly correlated with larger tumor size, cervical lymph node metastasis, local recurrence, and poorer prognosis, suggesting its critical role in cancer invasion and metastasis [25]. The MAPK signaling pathway was activated in over 50% of OSCC cases and was associated with high malignancy and invasiveness, highlighting it as a key target for improving therapeutic strategies [26]. Additionally, recent studies reported that in head and neck squamous cell carcinoma (HNSCC) models, CHMP2A mediates tumor resistance to NK cells by secreting extracellular vesicles expressing MICA/B and TRAIL [27]. In OSCC, the T-cell receptor (TCR) signaling pathway demonstrated its potential as a biomarker for personalized therapy, as high TCR repertoire diversity was associated with better immunotherapy responses through activation of specific T-cell responses and the formation of tertiary lymphoid structure-like phenotypes [28]. TGF-β signaling, particularly through reduced or lost Smad4 expression, appears to promote OSCC progression by indicating abnormalities in this pathway [29]. Moreover, Toll-like receptor 2 signaling in OSCC was found to drive cancer progression and confer resistance to cisplatin-induced apoptosis by activating the miR-146a pathway and downregulating CARD10 expression [30]. The enrichment of these pathways in the high-GSDMB group suggests that GSDMB may be functionally linked to multiple oncogenic and immune-related signaling networks in OSCC.

Thirdly, this study first uncovered the crucial role of GSDMB in the tumor immune microenvironment. Through immune infiltration analysis, immunophenotype scoring, and correlation analysis, we found that GSDMB might regulate the tumor immune microenvironment via multiple mechanisms, offering potential value as an immunotherapy target. High GSDMB expression was significantly associated with the infiltration abundance of anti-tumor immune cells such as CD8 + T cells [31] and NK cells [32], suggesting it could enhance anti-tumor immunity by strengthening tumor immune surveillance. This phenomenon may reflect a compensatory immune activation in response to increased tumor immunogenicity in high-GSDMB tumors. TCIA analysis suggested that patients in the high GSDMB expression group might be more sensitive to PD-1 inhibitors or combined CTLA-4/PD-1 immunotherapy, providing theoretical support for future precision immunotherapy based on GSDMB expression levels. Furthermore, the significant correlation between GSDMB and immune checkpoint activity indicates it might be a key molecule affecting the efficacy of immunotherapy.

Fourthly, high GSDMB expression potentially modulates the tumor cell response to drugs, enhancing the effectiveness of chemotherapeutic and targeted therapies, and exhibits good binding affinity with several potential small molecules, offering a new perspective for precision medicine. Increased sensitivity to traditional chemotherapeutics like oxaliplatin [33] and 5-fluorouracil [6] in the high-expression group implies its potential value in conventional chemotherapy regimens. Molecular docking results showed that HC toxin, vorinostat [34], rifabutin, and scriptaid [35] can stably bind to the active sites of GSDMB, interfering with its function, possibly by preventing interactions with other molecules or downstream proteins to regulate immune responses and cellular biology in the tumor microenvironment. Notably, vorinostat and scriptaid, as histone deacetylase inhibitors, may exert synergistic effects by simultaneously modulating epigenetic regulation and GSDMB-related pathways, thereby enhancing chemosensitivity and immune activation.

Fifthly, we conducted an in-depth analysis of scRNA-seq data from OSCC samples to reveal cell composition, batch effects, and gene expression patterns within the tumor microenvironment. We explored the significant heterogeneity in cell composition among different OSCC tissues, which may relate to differences in immune infiltration or stromal components within the tumor microenvironment. Through cell communication and pseudotime trajectory analysis, we delved into the interactions between immune cells and the developmental process in the OSCC tumor microenvironment. Communication analysis revealed the quantity and strength of communications between different immune cells, especially multiple important interaction pathways involving T cells as ligands and receptors with other cells, which may play a critical role in tumor immune escape mechanisms. Additionally, the expression analysis of genes in CCL and TGFβ pathways across various cell types indicated their potential roles in immune modulation and tumor metastasis. Nils Ludwig et al. discovered that novel TGF-β inhibitors could improve the progression of OSCC and enhance antitumor immune responses to anti-PD-L1 immunotherapy [36]. Pseudotime analysis further illustrated the temporal changes of different cell types during cell differentiation, providing a new perspective on the dynamic interactions between tumor and immune cells. The dynamic expression of GSDMB along pseudotime trajectories highlights its potential involvement in immune cell differentiation and functional state transitions.

Finally, we explored the role of GSDMB in the OSCC tumor microenvironment by dividing cells into high and low expression groups and analyzing GSDMB expression in stRNA-seq data. We observed differences in the proportion of high and low expression groups among different cell types, indicating that GSDMB expression varies markedly across tumor cells, immune cells, and other cell populations, suggesting potential differences in function across cell types. To better understand the biological functions of GSDMB, Fig. 1D shows the results of GSEA between high and low expression groups, revealing differences in multiple biological pathways, particularly those related to immune response and cell migration, hinting at its possible role in regulating these pathways during OSCC progression. Dong et al.‘s study first revealed the negative impact of lysine acetylation (especially downregulated acetylation of RPS6 and RPS3) on the ribosome biogenesis pathway in OSCC, which might contribute to cancer initiation and development [37]. Spatial transcriptomics further revealed that GSDMB expression is spatially heterogeneous, reinforcing the importance of spatial context when interpreting its biological function in the tumor microenvironment.

Despite the significant insights into the role of GSDMB in tumorigenesis, immune microenvironment, and patient prognosis, our study has certain limitations. First, although systematic analysis was based on multi-omics data, the generalizability of the results might be affected by the heterogeneity of tumor samples and diversity of data sources. Second, while we explored the relationship between GSDMB and drug sensitivity, in vivo or clinical trials are still needed to validate its effectiveness as a predictive biomarker. Future research should focus on verifying GSDMB expression pattern and predictive value for treatment response using large-scale clinical cohorts. Further investigation into GSDMB role in tumor immune escape mechanisms, particularly in immune checkpoint inhibitor therapies, is warranted. Additionally, our molecular docking analysis provides computational evidence suggesting potential physical interactions between candidate drugs and GSDMB, offering a structural rationale for future drug development. However, it should be emphasized that these findings are based exclusively on in silico predictions, and additional in vitro and in vivo experiments are required to determine whether these compounds can directly modulate GSDMB expression or function in OSCC. The integration of spatial transcriptomics with functional experiments will be particularly important for elucidating cell-type-specific roles of GSDMB in OSCC.

Conclusion

In summary, this study provides an in-depth analysis of the expression characteristics and biological functions of GSDMB in cancer, revealing its potential roles in tumorigenesis, immune regulation, and patient prognosis. The findings suggest that GSDMB serves as a promising biomarker for cancer diagnosis, effectively distinguishing between tumor and normal tissues, with close associations to TMB and the tumor immune microenvironment. Further analyses indicate that GSDMB might offer new targets for cancer immunotherapy by modulating immune cell infiltration, immunophenotypes, and responses to immunotherapy. Moreover, the results of molecular docking analysis provide theoretical support for the development of small molecule inhibitors targeting GSDMB.

Abbreviations

OSCC

Oral Squamous Cell Carcinoma

GSDMB

Gasdermin B

TMB

Tumor Mutation Burden

TCGA

The Cancer Genome Atlas

UCSC

University of California, Santa Cruz

GEO

Gene Expression Omnibus

scRNA-seq

Single-Cell RNA Sequencing

stRNA-seq

Spatial Transcriptomics RNA Sequencing

DEG(s)

Differentially Expressed Gene(s)

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

ssGSEA

Single Sample Gene Set Enrichment Analysis

IPS

Immune Phenotype Score

IC50

Half Maximal Inhibitory Concentration

GSEA

Gene Set Enrichment Analysis

CB-Dock2

Cavity-Based Docking 2

PDB

Protein Data Bank

ROC

Receiver Operating Characteristic

AUC

Area Under the Curve

TCIA

The Cancer Imaging Archive

FAK

Focal Adhesion Kinase

MAPK

Mitogen-Activated Protein Kinase

TCR

T Cell Receptor

TGF-β

Transforming Growth Factor Beta

NK cells

Natural Killer Cells

CTLA-4

Cytotoxic T-Lymphocyte-Associated Antigen 4

PD-1

Programmed Cell Death Protein 1

PCA

Principal Component Analysis

t-SNE

t-Distributed Stochastic Neighbor Embedding

NES

Normalized Enrichment Score

FDR

False Discovery Rate

Authors’ contributions

Zifeng Cui: Conceptualization, Methodology, Writing—Original Draft.Yan Gao: Methodology, Writing—Original Draft, Writing—Review & Editing.Kaicheng Yang: Conceptualization, Writing—Review & Editing.Zhiming Dong: Writing—Review & Editing, Funding acquisition.All authors have read and approved the final manuscript.

Funding

This work was supported by Hebei Natural Science Foundation (H2024206136).

Data availability

The data were obtained from the Cancer Genome Atlas Program (TCGA) database, the UCSC Xena database and the GEO database.

Declarations

Ethics approval and consent to participate

Written informed consent was obtained from each patient before participation. This study was approved by the Ethics Committee of the Fourth Hospital of Hebei Medical University (Approval No. 2024KY008). All procedures were conducted in accordance with the principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.