Transcriptomic insights into GSDME-driven pyroptosis in head and neck cancer

Abstract

This study investigates the expression of pyroptosis-related genes (PRGs) in head and neck squamous cell carcinoma (HNSCC) and their potential role in the tumor immune microenvironment. Additionally, we conducted an in-depth transcriptomic analysis of HNSCC with high GSDME expression. This study utilized the curated TCGA-HNSCC and GSE65858 datasets to analyze differentially expressed genes (DEGs) associated with pyroptosis in HNSCC. Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses were performed on these DEGs. The infiltration of 22 immune cell types was assessed using the CIBERSORTx algorithm. GSDME-overexpressing cells were sequenced using the Illumina NovaSeq 6000 platform, and immunohistochemistry (IHC) was conducted for validation. A total of 501 TCGA-HNSCC patient data and 270 tumor samples from the GSE65858 dataset were collected. Seventeen pyroptosis-related genes were found to have mutations in HNSCC samples, with TP53, CASP8, and NLRP3 exhibiting the highest mutation frequencies. GO enrichment analysis revealed that the DEGs were enriched in cellular components such as myofibrils and collagen-containing extracellular matrix, as well as in molecular functions including actin binding and receptor ligand activity. KEGG pathway enrichment analysis showed that DEGs were enriched in the focal adhesion biological pathway. GSEA analysis indicated a negative correlation between DEGs and the cytochrome P450 pathway. Transcriptomic analysis of high GSDME expression revealed a significant decrease in CK13 (KRT13) and CK19 (KRT19) in the GSDME-overexpressing group. Downregulated genes were significantly enriched in cell adhesion molecules and the arachidonic acid pathway. Immunohistochemistry confirmed a statistically significant reduction in CK13 and CK19 protein expression in the high GSDME expression group. Transcriptomic analysis of GSDME overexpression showed significant upregulation or downregulation of transcription factor families, including E2F, ETS, HMG, MYB, BZIP, and C2H2. Specifically, ELF3 from the ETS transcription factor family and PBX1 from the Homeobox transcription factor family were downregulated. GSDME is potentially associated with the ETS transcription factor ELF3 and the Homeobox transcription factor PBX1. The reduced expression of CK13 and CK19 in the GSDME-overexpressing group may serve as a potential mechanism by which GSDME inhibits HNSCC growth.

Introduction

Head and neck squamous cell carcinoma (HNSCC) is a globally widespread malignant tumour that commonly occurs in the oral cavity, sinonasal cavity, pharynx, and larynx^1,2. It has a low 5-year survival rate and high rates of recurrence and metastasis³. It is estimated that in 2022, there were over 54,000 new cases of HNSCC in the United States, with approximately 117,000 new cases in China in the same year^2,4.

Pyroptosis is a form of programmed cell death mediated by the gasdermin protein family, characterized by cell swelling and the formation of numerous bubble-like protrusions on the plasma membrane, ultimately leading to membrane rupture and the release of intracellular contents accompanied by a pronounced inflammatory response^5–8. Pyroptosis can be triggered by a variety of extracellular and intracellular stimuli, including pathogens such as bacteria and viruses, toxins, and multiple anticancer agents^5–8. During tumor initiation and progression, pyroptosis exerts a dual role⁹. On one hand, pyroptosis lyses tumor cells, exposes tumor-associated antigens, and releases immunogenic molecules, thereby effectively activating immune surveillance and inhibiting tumor initiation, progression, and metastasis^9–15. For instance, the induction of pyroptosis has been shown to significantly suppress tumor progression in several malignancies, including ovarian cancer, breast cancer, and head and neck squamous cell carcinoma (HNSCC)^11,12,14,16. On the other hand, when pyroptotic responses are persistent or dysregulated, CAF-derived NLRP3/IL-1β signaling reshapes the tumor microenvironment. This remodeled niche not only fosters immune suppression but also promotes angiogenesis and tumor cell extravasation by upregulating endothelial adhesion molecules and matrix metalloproteinases, thereby facilitating metastasis^10,17–22. Thus, the role of pyroptosis in cancer is not paradoxical but context-dependent, manifesting as either tumor-suppressive or tumor-promoting effects depending on the microenvironment and regulatory status.

Among the known mediators of pyroptosis, members of the gasdermin family (e.g., GSDMA, GSDMC, GSDMD, and GSDME) are critically involved in regulating tumor cell fate^{14,21,23–29}. Among them, GSDME plays a central role in pyroptosis: upon cleavage by caspase-3, its N-terminal domain is released, which drives pyroptotic cell death and exerts potent tumor-suppressive effects^30,31. Recent studies have demonstrated that GSDME not only directly inhibits tumor cell growth, invasion, and metastasis but also promotes the release of inflammatory mediators and immunogenic factors, thereby enhancing the infiltration and functionality of immune cells, including CD8⁺T cells and macrophages, ultimately inducing a robust antitumor immune response^25,27,28,32. For example, in melanoma models, GSDME-dependent pyroptosis enhances macrophage phagocytic and cytotoxic activity^25,26; in breast cancer, activation of GSDME is closely associated with CD8⁺T cell–mediated antitumor immunity, thereby amplifying therapeutic efficacy³².

Pyroptosis, as a form of programmed cell death, shows unique potential in cancer therapy due to its ability to regulate inflammation and the immune microenvironment. The correlation between pyroptotic factors and immune cells in head and neck squamous cell carcinoma requires further investigation. Few studies have systematically investigated the targets and signaling pathways through which GSDME overexpression inhibits HNSCC growth and invasion, as well as its impact on the tumor immune microenvironment. In this study, we classified 501 HNSCC patients based on pyroptosis-related genes, analyzing pyroptosis-related differentially expressed genes, enrichment analysis, and immune infiltration profiles. Based on this, we conducted related studies and validation on the differential expression of oncogenes and signaling pathways in HNSCC with high GSDME expression.

Materials and methods

Data sources

We obtained gene expression profile data for head and neck squamous cell carcinoma (HNSCC) from the TCGA database (https://portal.gdc.cancer.gov/projects/TCGA-HNSC) using the TCGAbiolinks package, and retained data from 501 patients for analysis (Table 1). We used the GEO query package³³ in R software (version 4.2.0, http://r-project.org/) to download the HNSCC-related dataset GSE65858¹⁰ from the GEO database, which includes 270 tumor samples (Table 1).

Table 1.

Baseline data table related to pyroptosis.

Characteristic	TCGA			GSE65858
	Low score	High score	p	Low score	High score	p
n	251	250		135	135
Gender, n (%)
Female	66 (13.2%)	67 (13.4%)	0.979	26 (9.6%)	21 (7.8%)	0.521
Male	185 (36.9%)	183 (36.5%)		109 (40.4%)	114 (42.2%)
Stage, n (%)
'–	42 (8.4%)	26 (5.2%)	0.099			0.755
Stage I	13 (2.6%)	12 (2.4%)		7 (2.6%)	11 (4.1%)
Stage II	37 (7.4%)	32 (6.4%)		22 (8.1%)	15 (5.6%)
Stage III	40 (8%)	39 (7.8%)		19 (7%)	18 (6.7%)
Stage IVA	116 (23.2%)	132 (26.3%)		77 (28.5%)	78 (28.9%)
Stage IVB	2 (0.4%)	9 (1.8%)		7 (2.6%)	9 (3.3%)
Stage IVC	1 (0.2%)	0 (0%)		3 (1.1%)	4 (1.5%)
Age, median (IQR)	60.62 (53.92, 68.53)	62.03 (53.34, 69.66)	0.231	59.28 (54.86, 69.78)	57.08 (51.24, 66.47)	0.009

Differential expression of pyroptosis-related genes (PRGs)

A total of 51 pyroptosis-related genes (PRGs) were included in this study, sourced from the GeneCards³⁴ (https://www.genecards.org/), the Molecular Signature Database V7.4 (MsigDB)³⁵, and previous literature, with duplicate genes removed to form the final gene set^6,8,36,37. The Mann–Whitney U test was used to compare the expression differences of pyroptosis-related genes between tumor and adjacent tissues in the TCGA-HNSCC dataset, with visualization performed using the R package ggplot2³⁸. Dimensionality reduction analysis was then performed, followed by the construction of a PCA plot for the pyroptosis-related genes.

Subtype analysis of pyroptosis-related genes

We performed subtype analysis of the TCGA dataset based on pyroptosis-related genes (PRGs), classifying the samples into two molecular subtypes: 251 samples were assigned to subtype A and 250 to subtype B. Consensus clustering was performed using the R package ConsensusClusterPlus (v1.62.0) to classify molecular subtypes of head and neck squamous cell carcinoma (HNSCC) based on PRG expression profiles. The algorithm was applied to the PRG expression matrix to explore an optimal cluster number (K) ranging from 2 to 6. For each K value, 1000 resampling iterations were performed (each time randomly selecting 80% of samples and 80% of genes) to construct the consensus matrix. The parameters were set as follows: clusterAlg = “hc” (hierarchical clustering), distance = “spearman”, and seed = 123. Based on a comprehensive assessment of the cumulative distribution function (CDF) and delta-area curves, proportion of ambiguous clustering (PAC), block structure of the consensus heatmap, and average silhouette width, K = 2 was determined as the optimal number of clusters, resulting in subtypes A and B. In addition, differential expression analysis between these subgroups was conducted using the R package DESeq2^13,39. Genes with log2 (Fold change) > 1.5 and P < 0.05 were defined as up-regulated differentially expressed genes (up_regulated_genes), whereas those with log2 (Fold change) <  − 1.5 and P < 0.05 were defined as down-regulated differentially expressed genes (down_regulated_genes).

Enrichment analysis (GO/KEGG/GSEA/GSVA)

Gene Ontology (GO)⁶ annotation and the Kyoto Encyclopedia of Genes and Genomes (KEGG)^40–42 were performed on differentially expressed genes using the clusterProfiler package⁴³, with an FDR < 0.05 considered statistically significant.

Gene set enrichment analysis (GSEA)⁴⁴ was conducted for further enrichment analysis. A false discovery rate (FDR) < 0.25 and p.adjust < 0.05 were considered significantly enriched. We employed the R package (GSVA)⁴⁵ (Gene Set Variation Analysis) and calculated the pathway scores for each sample using the gene set enrichment analysis (GSEA)⁴⁴ method based on their gene expression matrices. Differentially enriched functions or pathways were then identified using the limma package^46,47 (GSEA analysis: FDR < 0.25 and p.adjust < 0.05; GSVA analysis: Differential screening based on the limma package, with a statistical threshold of adj.P.Val < 0.05;).

Immune infiltration analysis

Based on the curated TCGA-HNSC dataset (501 cases) and GSE65858 dataset (270 cases), we applied univariate Cox regression analysis and Lasso regression analysis to further identify prognostic-related genes and construct a prognostic model accordingly. After constructing the Lasso-Cox model using the merged dataset, we stratified patient samples in the training and validation cohorts into high- and low-risk groups based on the assigned risk scores. The expression matrix data from the training and validation cohorts were uploaded separately to CIBERSORTx⁴⁸. In conjunction with the LM22 signature gene matrix, CIBERSORTx filtered out samples with p < 0.05, generating the immune cell infiltration matrix. Bar plots were generated using the ggplot2 package in R to visualize the distribution of 22 immune cell infiltrations across individual samples.

Cell culture

The human oral keratinocyte (HOK) cells were purchased from Tongpai Biotechnology (Shanghai, China; ZQY012; http://www.shtpbio.com/aboutus.html), and the human tongue squamous cell line (CAL-27) was obtained from hycyte (Jiangsu, China; TCH-C148; https://www.hycyte.com/sys-pd/195.html). Cells were cultured in DMEM (Gibco, USA) supplemented with 10% fetal bovine serum (FBS) (Gibco, USA). All cells were cultured in a humidified incubator at 37 °C and 5% CO₂. The lentivirus for GSDME overexpression was purchased from GeneChem Co., Ltd. (Shanghai, China). The GSDME overexpression group (LV-GSDME) and the control group (LV-NC) were transfected into CAL-27 cells, and stable transfected cell lines were selected using puromycin.

Quantitative realtime‑PCR validation

The expression of NOD1 and NOD2 was validated by analyzing HOK and CAL27 cells. RNA was extracted from CAL27, LV-GSDME, and LV-NC stable transfectants to validate the expression of GSDME. Total RNA was extracted using Trizol reagent (Thermo Fisher Scientific, USA), and reverse transcription was performed using the M5 Sprint qPCR RT Kit (Mei5 Biotechnology, China) with gDNA remover. qPCR quantification was performed using the MF797 (Mei5 Biotechnology, China) system, and the expression was normalized to that of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Relative expression levels were determined using the 2 − ΔΔCt method (Table 2).

Table 2.

Sequences of gene-specific primers used for real-time RT-qPCR.

Gene	Forward primer (5′‑3′)	Reverse primer (5′‑3′)
NOD2	CACCGTCTGGAATAAGGGTACT	TTCATACTGGCTGACGAAACC
NOD1	ACTGAAAAGCAATCGGGAACTT	CACACACAATCTCCGCATCTT
GSDME	ACATGCAGGTCGAGGAGAAGT	TCAATGACACCGTAGGCAATG
GAPDH	Ctctgctcctcctgttcgac	Ttaaaagcagccctggtgac

Transcriptomic analysis of GSDME high expression

RNA was extracted from CAL-27, LV-GSDME, and LV-NC stable transfectants, and sequencing of the libraries was performed using the Illumina Novaseq 6000 platform. Gene counts were analyzed using PCA and plotted with R (v 3.2.0) to assess biological replicates of the samples. Differential analysis was performed using DESeq2, followed by GO functional enrichment, KEGG pathway enrichment, transcription factor and target gene prediction. Transcriptomic sequencing and analysis were performed by Shanghai OuYi Biotechnology Co., Ltd. (Shanghai, China).

Xenograft models

Female nude mice (nu/nu, 5–6 weeks old) were randomly assigned into three groups: LV-GSDME, LV-NC, and CAL-27. Subcutaneous injection of 2 × 10⁶ cells was performed, and after 4 weeks, the mice were euthanized for immunohistochemical analysis.

All experimental animals were housed in a specific pathogen-free (SPF) facility under standardized environmental conditions (temperature 22 ± 2 °C, humidity 50 ± 10%, 12-h light–dark cycle). All procedures strictly adhered to institutional and national laboratory animal management regulations (Regulations on the Management of Laboratory Animals in China). Nude mice were anesthetized with 2% isoflurane for 2–3 min and euthanized by cervical dislocation once a surgical depth of anesthesia was achieved. All animal experiments were conducted in accordance with the ARRIVE guidelines and the NIH Guide for the Care and Use of Laboratory Animals. All procedures were approved by the Animal Experiment Ethics Committee of Shanxi University of Chinese Medicine (Approval No. 2023KY-07013).

Immunohistochemical (IHC) staining

Tumor tissues from GSDME overexpression and control xenograft nude mice were fixed overnight in 4% formaldehyde and embedded in paraffin. Sections were incubated overnight with the primary antibody, followed by incubation with the secondary antibody and hematoxylin staining. The primary antibody detection was performed for CK13 (1:100; 66,684–1-Ig, proteintech), CK19 (1:100; 260,187–1-Ig, proteintech), and GSDME (1:400; 13,075–1-AP, proteintech), with images acquired using the slide scanning system NanoZoomer®S360.

Statistical analysis

All data calculations and statistical analyses were performed using R programming (https://www.r-project.org/, version 4.1.2). For comparison of two continuous variables, statistical significance of normally distributed variables was assessed using the independent Student’s t test, while the Mann–Whitney U test was used for non-normally distributed variables. Correlation analysis was performed using the Spearman method. All statistical P values were two-tailed, with P < 0.05 considered statistically significant. Comparisons between different groups were made using non-parametric Wilcoxon/Mann–Whitney tests. Qualitative variables were analyzed using the Pearson × 2 test or Fisher’s exact probability test. P < 0.05 was considered statistically significant.

Result

Differential expression and mutations of pyroptosis-related genes

The expression of pyroptosis-related genes between the two groups was compared, resulting in 17 genes with significant differential expression (Fig. 1A). Principal component analysis (PCA) revealed a clear distinction between the two groups (Fig. 1B). Seventeen pyroptosis-related genes were found to potentially harbor mutations in head and neck cancer samples (Fig. 1C). The three genes with the highest mutation frequencies were TP53, CASP8, and NLRP3, each of which may contain multiple mutation sites (Fig. 1D). We then analyzed the overall copy number variation in the genomes of HNSCC patients, illustrating both amplifications and deletions (Fig. 1E).

Fig. 1.

Differential expression and mutation of PRGs in HNSCC. (A) Expression Differences of Mutated Pyroptosis Genes in HNSCC. (B) PCA Analysis of the Distribution of Pyroptosis Genes in HNSCC Tumor and Adjacent Normal Tissues. (C) Panoramic Overview of Pyroptosis Genes in HNSCC. (D) Mutation Sites of TP53, CASP8, and NLRP3 Genes. (E) Copy Number Variation of Pyroptosis Genes. (*p < 0.05; **p < 0.01; ***p < 0.001).

Enrichment analysis of pyroptosis genes

We first performed functional enrichment analysis of the differentially expressed pyroptosis genes (DEGs) through Gene Ontology (GO) enrichment analysis, focusing on biological processes (BP), cellular components (CC), and molecular functions (MF). Significant enrichment was observed in cellular components such as myofibrils and collagen-containing extracellular matrix, and in molecular functions such as actin binding and receptor-ligand activity (Fig. 2A). Next, KEGG pathway enrichment analysis was performed on the DEGs, which revealed significant enrichment in the focal adhesion biological pathway (Fig. 2B). GSEA analysis of pyroptosis genes showed a positive correlation with collagen formation, deacetylation of histones, and DNA methylation, and a negative correlation with the cytochrome P450 metabolic pathway (Fig. 2C–H).

Fig. 2.

Enrichment analysis. (A) GO enrichment analysis: The x-axis represents the gene ratio, the y-axis represents the GO terms, the node size indicates the number of genes enriched in each pathway. (B) KEGG Pathway Enrichment Analysis: The x-axis represents the gene ratio, the y-axis represents the pathway name, the node size indicates the number of genes enriched in each pathway. (C–E) GSEA Analysis of Pyroptosis Genes: Three significantly upregulated pathways are shown. (F–H) GSEA Analysis of Pyroptosis Genes: Three significantly downregulated pathways are shown (GO/KEGG: Significant enrichment is defined as p < 0.1 and q < 0.2; GSEA: Significant enrichment is defined as a false discovery rate (FDR) < 0.25 and p < 0.05.)

Immune infiltration analysis of pyroptosis genes

Analysis of the TCGA training cohort and GEO validation cohort revealed significant differences in the immune cell infiltration levels between high and low pyroptosis groups, including naive B cells, plasma cells, CD8 + T cells, activated memory CD4 + T cells, M1 macrophages, and activated dendritic cells (Fig. 3C, D). In the GEO validation cohort, the expression of M1 macrophages was higher in the high pyroptosis group compared to the low pyroptosis group, suggesting that pyroptosis genes can influence the activation pathways of macrophages, B cells, plasma cells, CD8 + T cells, and activated memory CD4 + T cells in HNSCC, thereby affecting the tumor immune microenvironment.

Fig. 3.

(A, B) The expression levels of NOD1 and NOD2 genes were measured using quantitative real-time polymerase chain reaction (qRT-PCR, ***p < 0.001). (C) Expression of various immune cells grouped by pyroptosis score in the TCGA training cohort. (D) Expression of various immune cells grouped by pyroptosis score in the GEO validation cohort.

Validation of sequencing data using real‑time PCR

PCR results showed that the expressions of NOD1, NOD2 in CAL-27 cells were higher than those in HOK cells, and the differences were statistically significant (Fig. 3A, B). GSDME overexpression validation showed that GSDME expression was significantly higher in the LV-GSDME group compared to both the LV-NC and CAL groups. The LV-NC group showed similar expression levels to the CAL group, with no statistically significant difference (Fig. 5A).

Fig. 5.

(A) The expression levels of GSDME were measured by quantitative real-time polymerase chain reaction (qRT-PCR). (B) Volcano plot showing the differential gene expression between the LV-GSDME and LV-NC groups. Gray dots represent genes with no significant difference, while red and blue represent genes with significant differential expression. (C) Radar plot of differential gene expression between the LV-GSDME and LV-NC groups. The first ring: upregulated genes (light red) and downregulated genes (light blue); the second ring: outer circle (average expression in the experimental group) and inner circle (average expression in the control group); the third ring: average expression levels of individual genes in both experimental and control groups. (D) KEGG pathway enrichment analysis of differential genes between the LV-GSDME and LV-NC groups. (E) GO enrichment analysis of differential genes between the LV-GSDME and LV-NC groups.

Transcriptome analysis of GSDME overexpression

Transcriptome analysis of GSDME overexpression showed significant upregulation or downregulation of the E2F, ETS, HMG, MYB, BZIP, and C2H2 transcription factor families (Fig. 4B). Among them, the ETS transcription factor ELF3 was downregulated, along with its target genes CCL17, CENPF, CLIC3, KRT13, KRT19 and USP1. The Homeobox transcription factor family, particularly PBX1, was downregulated, along with its target genes POU2F3, RGS9, TJP3, and UPK2 (Fig. 4A).

Fig. 4.

(A) Distribution of transcription factor family target genes among differentially expressed genes in GSDME overexpression. (B) Distribution of differentially expressed transcription factor families in GSDME overexpression.

Differential gene analysis of GSDME overexpression

Comparison between the LV-GSDME and LV-NC groups revealed significant downregulation of genes including CK13 (KRT13), CK19 (KRT19), CD99, AGR2, KRT15 and MMP13 in the GSDME overexpression group (Fig. 5B, C). GO and KEGG enrichment analysis showed that downregulated genes were significantly enriched in cell adhesion molecules and the arachidonic acid pathway (Fig. 5D, E).

Immunohistochemical analysis revealed that GSDME expression was significantly elevated in the high-expression group, while CK13 and CK19 protein levels were significantly reduced in the same group, with statistical significance observed. No significant differences were observed between the NC and CAL groups (Fig. 6).

Fig. 6.

Immunohistochemical (IHC) images and analysis of GSDME, CK13 and CK19 expression in xenograft tumor tissue sections from nude mice Scale bars, 50 μm. (ns, no statistically significant difference, *p < 0.05; **p < 0.01; ***p < 0.001, ****p < 0.0001).

Discussion

The treatment of head and neck squamous cell carcinoma (HNSCC) is typically multimodal, involving surgery, radiotherapy, and chemotherapy⁴⁹. Owing to the anatomical complexity of the head and neck, radiotherapy for head and neck squamous cell carcinoma (HNSCC) frequently results in functional impairments—most notably dysphagia and speech impairment—leading to substantial declines in quality of life^49,50. Suicide rates among HNSCC survivors are second only to those for pancreatic cancer, underscoring the importance of quality-of-life interventions and psychological support alongside oncologic treatment^49,50. Treatment for HNSCC should shift from single-modality approaches to precision, integrative immuno-oncology strategies.

This study compared the differential expression of pyroptosis-related genes between HNSCC tumor tissues and adjacent normal tissues, revealing that the three genes with the highest mutation frequencies were TP53, CASP8, and NLRP3. TP53 is among the most frequently mutated driver genes across cancers; it regulates cell-cycle arrest, apoptosis, DNA repair and metabolic reprogramming, and is mutated in approximately 30% of breast cancers⁵¹. Prior studies report CASP8 mutations in 56% of oral squamous cell carcinoma samples^52,53; caspase-8 functions as a key node in apoptosis and necroptosis and serves as a “molecular switch” for pyroptosis, activating canonical and non-canonical inflammasome pathways, including those involving NLRP3^53,54. These alterations suggest that multiple cell-death programmes are rewired in HNSCC and may critically influence tumour initiation and progression.

The classical pyroptosis pathway primarily involves the activation of NLRP3 inflammasomes, the assembly of pro-caspase-1 with ASC, and subsequent activation of caspase-1, which promotes the production of pro-inflammatory cytokines such as IL-1β and IL-18. Additionally, caspase-1 can cleave gasdermin D (GSDMD), leading to the formation of pores in the cell membrane⁵⁴. In addition to the classical pyroptosis pathway, recent studies have identified atypical activation pathways for pyroptosis. In the human body, the activation of caspase-4 and caspase-5, triggered by intracellular lipopolysaccharide (LPS), activates NLRP3 inflammasomes, leading to IL-1β and IL-18 secretion and pyroptosis induction^18,55–59. Mutations in TP53, as well as the pyroptosis-inducing pathways mediated by CASP8 and NLRP3, may provide new therapeutic strategies for HNSCC treatment.

Differential gene analysis of pyroptosis-related genes revealed that NOD1 and NOD2 were upregulated in HNSCC, while NLRP2 expression was reduced. Studies have confirmed that NOD1 and NOD2 promote cell proliferation and enhance metastatic potential in metastatic cervical squamous cell carcinoma (CSCC)⁶⁰. Expression levels of NOD1 and NOD2 are positively correlated with immune-infiltration scores and immune cell–related pathways^61–65. Activation of the NOD pathway upregulates NOD-like receptor (NLR) and inflammasome components together with pro-IL-1β/IL-18, thereby priming conditions for GSDME-mediated pore formation, release of inflammatory mediators, and enhanced antigen presentation; this provides a mechanistic basis for the heterogeneous magnitude of immune amplification across tumours under comparable GSDME execution strength^61–65. NLRP2 is highly expressed in non-triple-negative breast cancer (TNBC) cell lines, but nearly absent in TNBC cell lines, with low NLRP2 levels being associated with poor clinical prognosis in TNBC patients⁶⁶. NLRP2 effectively inhibits the growth of in situ breast cancer, and the NLRP2 group exhibits fewer and smaller metastatic lesions⁶⁶. The findings of this study regarding NOD1, NOD2, and NLRP2 are consistent with studies on cervical cancer and triple-negative breast cancer (TNBC), further supporting the idea of targeting NOD1, NOD2, and NLRP2 for HNSCC treatment. PCR validation results for NOD1 and NOD2 further corroborated this hypothesis.

To elucidate the molecular mechanisms of pyroptosis in HNSCC, we performed functional enrichment analysis of differentially expressed pyroptosis-related genes in HNSCC using GO and KEGG. Pyroptosis-related genes were enriched in the collagen-containing extracellular matrix cell component and the molecular functions of actin binding and receptor ligand activity. KEGG pathway enrichment analysis revealed a significant enrichment of pyroptosis-related genes in the focal adhesion biological pathway. The collagen-containing extracellular matrix (ECM) is an important structural component of the ECM, which is a major constituent of tumors, regulating cell differentiation, migration, and proliferation, and is associated with immune suppression^67–69. Studies have shown that actin-binding proteins (ABPs) such as SATB1 (special AT-rich binding rotein 1), WASP (Wiskott-Aldrich syndrome protein), nesprin, and villin participate in the early stages of carcinogenesis by regulating oncogene expression. The migration and invasion of cancer cells are driven by the formation of actin-rich protrusions⁷⁰. Differentially expressed genes in HNSCC were primarily involved in drug metabolism via the cytochrome P450 pathway⁷¹. Cytochrome P450 reductase (POR) induces ferroptosis in various lineages and cell states and responds to different ferroptosis-inducing mechanisms⁷². Our work suggests that the cytochrome P450 pathway is negatively correlated with pyroptosis and represents a potential drug target for developing pyroptosis-related therapies for HNSCC.

Increasing evidence suggests that the prognosis of cancer patients is related to the level of immune infiltration in tumors⁷³. Our experimental results showed significant differences in the levels of immune cell infiltration, including naive B cells, plasma cells, CD8 + T cells, activated memory CD4 + T cells, M1 macrophages, and activated dendritic cells, between the high and low pyroptosis groups in both the TCGA training and GEO validation datasets. In the GEO validation set, the expression of M1 macrophages was higher in the high pyroptosis group than in the low pyroptosis group, suggesting that pyroptosis-related genes can influence the activation pathways of macrophages, thereby affecting the tumor immune microenvironment in HNSCC. This indicates that in head and neck squamous cell carcinoma, the features of pyroptosis-related genes significantly affect the immune status of the body.

Additionally, immune-invasive head and neck squamous cell carcinoma⁷⁴ is characterized by a high recurrence rate, metastasis, poor prognosis, and an abnormal tumor microenvironment^50,74-77. The tumor microenvironment plays a critical role in HNSCC treatment and tumor progression⁵⁰. Studies have confirmed that GSDME is associated with tumor immunity, inducing pyroptosis in tumor cells by affecting Gastric cancer, triple-negative breast cancer, and colorectal cancer^32,78,79. In cervical and breast cancer, GSDME suppresses tumor growth and modulates the tumor microenvironment²⁴. Its expression enhances the phagocytic capacity of tumor-associated macrophages toward tumor cells and increases both the quantity and functionality of tumor-infiltrating natural killer (NK) cells and CD8 + T lymphocytes²⁷.

GSDME is a terminal effector of the pyroptosis pathway; caspase-3-mediated cleavage directly permeabilizes the plasma membrane, thereby converting apoptosis into highly immunogenic pyroptosis^24,25. GSDME exerts immunomodulatory activity. GSDME-dependent pyroptosis is accompanied by the release of IL-1β, IL-18 and damage-associated molecular patterns (DAMPs), which potently recruit and activate CD8⁺T cells and NK cells, remodel the tumour immune microenvironment, and amplify antitumour immunity^24,25. Notably, pyroptosis in fewer than 15% of tumour cells is sufficient to eradicate entire tumours in immunocompetent hosts, whereas this effect is abolished in immunodeficient or T-cell-depleted mice^24,25. In head and neck squamous cell carcinoma (HNSCC), where response rates to immune-checkpoint inhibitors remain below 20%^49,50, inducing GSDME-mediated pyroptosis may offer a strategy to convert “cold” tumours into “hot” ones.

Transcriptomic analysis of GSDME overexpression revealed decreased expression of CK13 (KRT13), CK19 (KRT19), CD99, AGR2, KRT15, and MMP13 in the overexpression group. Overexpression of KRT13 increases proliferation, migration, and invasion in breast cancer, promoting tumorigenesis and metastasis⁸⁰. Additionally, KRT19 promotes liver cancer development⁸¹. Immunohistochemistry (IHC) results further confirmed the decreased expression of CK13 (KRT13) and CK19 (KRT19) in the GSDME overexpression group. The IHC results for KRT13 and KRT19 further support this hypothesis.

Transcriptomic analysis of GSDME overexpression revealed significant upregulation or downregulation of several transcription factor families, including E2F, ETS, HMG, MYB, BZIP, and C2H2. In the GSDME overexpression group, the ETS transcription factor ELF3 was downregulated, along with its target genes CCL17, CENPF, CLIC3, KRT13, KRT19, and USP1. The Homeobox transcription factor family, including PBX1, was downregulated, along with its target genes POU2F3, RGS9, TJP3, and UPK2. PBX1 is involved in various cancer-related processes, including anti-apoptosis, tumor-associated angiogenesis, epithelial-mesenchymal transition (EMT), metastasis, immune evasion, genomic instability, and cellular metabolic dysregulation. It mediates these processes by recruiting numerous downstream targets⁸². ELF3 is highly expressed in epithelial ovarian cancer (EOC) under hypoxic conditions, acting as a transcription factor for IGF1⁸³. ELF3-mediated increases in IGF1 and VEGF secretion promote endothelial cell proliferation, migration, and angiogenesis in EOC⁸³. The downregulation of ELF3 and PBX1, along with changes in their target genes, in the GSDME overexpression group suggests that GSDME, as a tumor suppressor gene, has significant research implications in relation to the ETS transcription factor ELF3 and the Homeobox transcription factor PBX1.

This study further validates the role and relevance of these pyroptosis-related factors in HNSCC, particularly CASP8, NOD1, NOD2, and GSDME, all of which may serve as potential therapeutic targets for advanced HNSCC. The potential association between GSDME and the ETS transcription factor ELF3, as well as the Homeobox transcription factor PBX1, provides new avenues for cancer therapy. However, the current study has certain limitations, and further molecular experiments are needed to validate the findings. As research on GSDME and its role in tumor immunity deepens, it is expected to be developed as a new anti-tumor therapeutic strategy, particularly for immune therapy in tumors like head and neck squamous cell carcinoma (HNSCC), offering new insights and potential targets.

Abbreviations

HNSCC

Head and neck squamous cell carcinoma

TGCA

The cancer genome atlas

GEO

Gene expression omnibus

PRG

Pyroptosis-related gene

LASSO

Least absolute shrinkage and selection operator

TME

Tumour microenvironment

GSDM

Gasdermin

RNA-seq

RNA sequencing

FDR

False discovery rate

GO

Gene ontology

KEGG

Kyoto encyclopedia of genes and genomes

ssGSEA

Single-sample gene-set enrichment analysis

Author contributions

Fengxiang Hao conducted most of the experiments, analyzed the results, and wrote the manuscript. Jiayi Hou and Danlei Qin performed some experiments. Xia Li and Ran Li provided technical support for the experiments. Bin Zhao was responsible for conceptualizing, designing, and supervising the study.

Funding

This work was supported by the National Natural Science Foundation Youth Science Fund project of china (No. 82202622) and the overseas students selected Foundation of Shanxi Province science (No. 20220020). This work was financially supported by the Shanxi Key R&D Planning Program (202202130501009).

Data availability

This study analyzed publicly available data sets that were obtained at GEO (https://www.ncbi.nlm.nih.gov/gds/?term=GSE65858), and TCGA repository (https://portal.gdc.cancer.gov/projects/TCGA-HNSC). The data used to support this study’s results are in the Published text. The original transcriptomic sequencing data from all experiments have also been uploaded to the GSA database as required (link: https://ngdc.cncb.ac.cn/gsa-human/, Project ID:HRA011698). If anyone wants to request data from this study, please contact sxmu0688@126.com.

Competing interests

The authors declare no competing interests.

Ethical approval

All experimental animals were housed in a specific pathogen-free (SPF) facility under standardized environmental conditions (temperature 22 ± 2 °C, humidity 50 ± 10%, 12-h light–dark cycle). All procedures strictly adhered to institutional and national laboratory animal management regulations (Regulations on the Management of Laboratory Animals in China). Anesthesia induction and maintenance were performed using 2% isoflurane, and euthanasia was carried out via cervical dislocation once deep anesthesia was achieved. All animal experiments were conducted in accordance with the ARRIVE guidelines and the NIH Guide for the Care and Use of Laboratory Animals. All procedures were approved by the Animal Experiment Ethics Committee of Shanxi University of Chinese Medicine (Approval No. 2023KY-07013).