GSDMD Promotes Epithelial-Mesenchymal Transition and Lymph Node Metastasis in Cervical Cancer in vitro and in vivo

Abstract

Background

Lymph node metastasis (LNM) is a critical determinant of prognosis in cervical cancer (CC). Tumor-associated macrophages (TAMs) and inflammatory signaling within the tumor microenvironment have been implicated in metastatic progression; however, the molecular mechanisms linking inflammation to lymphatic dissemination remain poorly understood.

Methods

We integrated single-cell RNA sequencing (scRNA-seq) of paired primary tumors and metastatic lymph nodes tissues from a single representative patient with clinical histological validation. Functional roles of gasdermin D (GSDMD) were evaluated using in vitro assays, including proliferation, migration, invasion, ELISA, and RNA sequencing, as well as in vivo xenograft and lymph node metastasis models.

Results

Single-cell transcriptomic profiling revealed significant activation of pyroptosis-related pathways in metastatic lymph nodes, accompanied by enhanced tumor–stromal interactions. Clinical tissue validation confirmed that GSDMD protein was significantly enriched in metastatic lesions compared to primary tumors. In vitro functional experiments demonstrated that GSDMD promotes epithelial–mesenchymal transition (EMT), enhances migration and invasion, and upregulates VEGFA and VEGFC expression. Mechanistically, GSDMD facilitated the release of pro-inflammatory cytokines such as IL-1β, contributing to a pro-metastatic microenvironment. Notably, GSDMD knockdown suppressed lymph node metastasis in vivo despite promoting primary tumor growth, indicating a context-dependent role in tumor progression.

Conclusion

Our findings reveal that GSDMD contributes to the pro-metastatic phenotype of cervical cancer. GSDMD dysregulation correlates with activated EMT and lymphangiogenic signaling, potentially facilitating tumor lymphatic dissemination. This study highlights GSDMD as a promising therapeutic candidate for restraining cervical cancer metastatic progression.

Introduction

Cervical cancer (CC) is uniquely characterized by its well-defined etiology and preventability. Although the incidence of CC can be effectively reduced through human papillomavirus (HPV) vaccination, as well as the screening and treatment of precancerous lesions, CC still ranks fourth in both cancer incidence and mortality among women globally.^1,2 Patients with early-stage CC generally have a favorable prognosis, with a 5-year survival rate exceeding 90%. However, the prognosis for advanced-stage CC is remarkably poor; particularly for patients with lymph node metastasis (LNM) or distant organ metastasis, the 5-year survival rate drops below 50%.^3,4 LNM serves not only as an independent prognostic factor for CC but also as a critical determinant in clinical treatment decisions.⁵ Lymph node metastasis in CC is a complex process involving tumor cell proliferation and apoptosis, epithelial-mesenchymal transition (EMT), lymphangiogenesis, non-coding RNAs, and the tumor microenvironment (TME).⁶

Pyroptosis is a unique form of programmed cell death, which is essentially an innate immune response primarily regulated by the gasdermin protein family. Unlike classic apoptosis, which is generally immunologically silent and characterized by membrane blebbing without cellular rupture, pyroptosis is distinguished by rapid plasma membrane pore formation, continuous cell swelling, and the massive lytic release of highly pro-inflammatory intracellular contents.⁷ The gasdermin family includes GSDMA, GSDMB, GSDMC, GSDMD, GSDME, and DFNA5, among which GSDMD has been demonstrated to be the key effector molecule mediating pyroptosis.⁸ GSDMD-mediated pyroptosis exhibits complex biological characteristics, presenting significant heterogeneity in its expression, mechanisms, and functional outcomes across different tissues, cell types, and pathological microenvironments.⁹ This heterogeneity is particularly prominent in inflammation-related diseases and tumors.

GSDMD, a critical regulator of pyroptosis, is ubiquitously expressed in both non-tumor and malignant cells. In non-neoplastic conditions, GSDMD aggravates inflammation and accelerates disease progression primarily through promoting the release of pro-inflammatory cytokines.¹⁰ In the context of cancer, however, GSDMD exerts bidirectional regulatory effects. Activation of the GSDMD signaling pathway can potently inhibit tumor growth, yet pyroptosis may paradoxically enhance tumor cell migration and metastasis.¹¹ Such dual functionality largely depends on the dynamic interactions among the tumor microenvironment, immune cells, and tumor cells. Previous studies have established that the release of pro-inflammatory cytokines, particularly IL-1β, acts as a critical paracrine bridge to stimulate localized lymphangiogenesis and facilitate tumor cell intravasation into lymphatic vessels.¹² In the context of the highly inflamed cervical cancer TME, we hypothesize that persistent GSDMD activation serves as the intrinsic driver for the release of these pro-metastatic cytokines.

While recent comprehensive reviews have highlighted the emerging biological relevance and clinical potential of pyroptosis in cervical cancer,¹³ the specific underlying mechanisms by which GSDMD drives lymphatic dissemination and TME remodeling remain poorly understood. Therefore, this study aims to perform an exploratory single-cell transcriptomic analysis of the lymph node metastasis microenvironment in CC to generate initial hypotheses, identify differential signaling factors, and further combine in vivo and in vitro experiments to elucidate the role of GSDMD in the process of lymphatic metastasis of CC.

Materials and Methods

Clinical Sample Collection, Single-Cell Sequencing, and Bioinformatic Analyses

Paired primary tumor (CX) and lymph node metastasis (LN) samples were collected from a single representative patient (a 34-year-old female) with cervical squamous cell carcinoma at Shandong Cancer Hospital and Institute. Prior to sequencing, the patient’s clinical and pathological characteristics were evaluated (HPV status was not tested prior to the surgery). The presence of tumor cells within the collected metastatic lymph node was definitively confirmed by senior pathologists via standard H&E staining. This study was approved by the Ethics Committee of Shandong Cancer Hospital (No: SDTHEC202508053). For the LN sample, tissue was specifically retrieved from the macroscopically identified metastatic cancer area within the enlarged lymph node, rather than the entire whole lymph node, to accurately capture the metastasis-specific microenvironment. Fresh tissue samples were obtained immediately post-surgery, washed, and minced into smaller pieces before enzymatic dissociation to obtain single-cell suspensions. Detailed clinical features of the patients are provided in Supplementary Table S1.

Single-cell transcriptomic libraries were constructed utilizing the 10x Genomics Chromium platform, followed by high-throughput sequencing on the Illumina NovaSeq platform. The generated single-cell RNA sequencing (scRNA-seq) data were then analyzed using the Seurat R package (version 4.3.0). To guarantee data reliability, strict quality control was performed by filtering out low-quality cells (typically those with < 200 or > 6,000 detected genes, and > 10% mitochondrial gene expression). The filtered gene expression matrices were normalized, and the top 2,000 highly variable genes were extracted. Following data scaling, Principal Component Analysis (PCA) and Uniform Manifold Approximation and Projection (UMAP) were sequentially applied for diensionality reduction. Cell clusters were annotated into major lineages—including epithelial, myeloid, fibroblast, endothelial, and T/NK cells—based on canonical markers.

To characterize dynamic changes in the tumor microenvironment during metastasis, intercellular communication between epithelial cells and surrounding stromal/immune cells was evaluated using the CellPhoneDB. The number of interaction pairs and specific metastasis-associated ligand–receptor pathways were systematically compared between CX and LN lesions. Gene Set Enrichment Analysis (GSEA) with clusterProfiler (version 4.4.4) revealed significant enrichment of pyroptosis-related pathways in the LN group. Finally, expression levels of key pyroptosis effectors and downstream pro-inflammatory/lymphangiogenic factors were extracted and visualized using Seurat’s DotPlot function, showing both relative expression and the percentage of expressing cells.

To guide the selection of appropriate cell models for in vitro functional assays, the baseline RNA (nTPM) and protein expression profiles of GSDMD across eight human cervical cancer cell lines (including HeLa, SiHa, and Ca Ski) were retrieved and visualized using The Human Protein Atlas (HPA) database (https://www.proteinatlas.org/).

Plasmid Construction and Cell Transfection

To silence GSDMD, three short hairpin RNAs (shRNAs) targeting distinct regions of human GSDMD (Gene ID: 79792) and a negative control sequence (shNC) were synthesized and packaged into the GV112 lentiviral vector (hU6-MCS-CMV-Puromycin) by GeneChem Co., Ltd. (Shanghai, China). The specific targeting sequences were as follows: shNC (CON054): 5′-TTCTCCGAACGTGTCACGT-3′; sh1 (PSC100462-1): 5′-AGCCACGTGCTTGCAGGGTGA-3′; sh2 (PSC100463-1): 5′-CAGCACCTCAATGAATGTGTA-3′; and sh3 (PSC100464-1): 5′-GATGAGGTGCCTCCACAACTT-3′.

For GSDMD overexpression (oe-GSDMD), the full-length human GSDMD coding sequence (Transcript: NM_024736.7) was cloned into the GV341 lentiviral vector (Ubc-MCS-3FLAG-SV40-puromycin) by the same manufacturer, with an empty vector serving as the control (oe-NC). Lentiviral particles (titer ≥1.0×108≥1.0×108 TU/mL) were transduced into HeLa and SiHa cells using HitransG A infection reagents according to the manufacturer’s instructions. Following lentiviral infection, stable cell lines were selected using puromycin. Crucially, based on Western blot validation, the “sh3” sequence exhibited the most profound GSDMD knockdown efficiency and was thus exclusively utilized for all subsequent functional assays (designated as “shGSDMD”).

Immunohistochemical (IHC) Staining

IHC staining was performed following the protocol detailed in our previous study.¹⁴ Briefly, tissue sections from primary tumors and metastatic lymph nodes were incubated overnight at 4 °C with primary antibodies against GSDMD (1:200), E-cadherin (1:100), and Ki-67 (1:500). Following incubation with the corresponding secondary antibodies for 15 min at 37 °C, the protein signals were visualized using 3,3′-diaminobenzidine (DAB) and the nuclei were subsequently counterstained with hematoxylin. The immunoreactivity of the target proteins was quantitatively assessed by calculating the mean optical density (MOD), which is derived from the ratio of integrated optical density (IOD) to the valid stained area. Details of the antibodies used in this study are provided in Supplementary Table S2.

Cell Lines and Cell Culture

HeLa and SiHa cell lines were obtained from china center for type culture collection (Wuhan, China). To ensure our in vitro findings account for the clinical and biological heterogeneity of cervical cancer, we specifically selected these two lines: SiHa represents an HPV16-positive squamous cell carcinoma, while HeLa represents an HPV18-positive adenocarcinoma origin. Together, they encompass the two most prevalent high-risk HPV types driving cervical carcinogenesis. They were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS). All media was supplemented with 100 U/mL of penicillin and 100 ug/mL of streptomycin. These cells were cultured in a humidified 5% CO₂ at 37 °C.

Wound-Healing Assay

Cells were seeded at a density of 4×10⁵ cells per well in 6-well plates. Upon reaching confluence, a scratch wound was created in the cell monolayer using a 1000-μL pipette tip. Following treatment under various conditions for 0 and 24 hours, the cells were examined under an inverted microscope. The scratch area in each group was quantified using ImageJ software.

Migration Assay

Migration assays were performed using 24-well Transwell chambers with 8-μm pore size (Corning Life Sciences, USA). Cells were resuspended in 100 μL of serum-free RPMI 1640 medium and seeded into the upper chamber. The lower chamber was filled with RPMI 1640 supplemented with 10% fetal bovine serum (FBS). Following the designated incubation period, cells were fixed with 4% paraformaldehyde and stained with 0.4% crystal violet solution. Non-migratory cells on the upper surface of the filter were removed, and the number of cells that had migrated to the lower surface of the membrane was counted under an inverted microscope.

Western Blot

Western blot analysis was performed as previously described.¹⁴ Briefly, stably transfected HeLa and SiHa cells (shNC/shGSDMD or oe-NC/oe-GSDMD) were harvested and lysed in RIPA buffer [50 mM Tris-HCl (pH 8.0), 1% NP-40, 150 mM NaCl, 0.1% sodium dodecyl sulfate (SDS), and 1 mM EDTA] (Beyotime, China). Protein concentrations were determined using a BCA protein assay kit (Beyotime, China) according to the manufacturer’s instructions. Equal amounts of total protein were resolved by SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, USA). The membranes were blocked with 5% non-fat dry milk (Bio-Rad, USA) and then incubated overnight at 4°C with primary antibodies against GSDMD (1:1000), E-cadherin (1:1000), β-catenin (1:1000), Snail (1:1000), Slug (1:1000), VEGFA (1:1000), VEGFC (1:1000), and α-Tubulin (1:1000). After thorough washing, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (1:5000) for 1 hour at room temperature. Protein bands were visualized using enhanced chemiluminescence (ECL) detection reagents (Proteintech, USA).

Enzyme-Linked Immunosorbent Assay (ELISA)

Concentrations of IL-1β and IL-18 released into the culture medium were quantified using specific human ELISA kits (Boster Bio, USA). Briefly, HeLa and SiHa cells (shNC and shGSDMD groups) were cultured for the indicated durations, after which the cell culture supernatants were collected and centrifuged to remove cellular debris. Subsequently, 100 μL of the clarified supernatant from each group was aliquoted for the respective ELISA assays. All procedures were performed strictly according to the manufacturers’ instructions. The optical density (OD) was measured, and cytokine concentrations were calculated based on standard curves.

Cell Proliferation Assay

Cell viability and proliferation were evaluated using the MTT assay. Briefly, stably transfected HeLa and SiHa cells were seeded into 96-well plates at a density of 3,000 cells per well in 100 μL of standard culture medium and incubated overnight to allow attachment. To assess proliferation kinetics, cell viability was measured at 24, 48, 72, and 96 hours after seeding. At each time point, 5 μL of MTT solution (5 mg/mL) was added to the appropriate wells, and the plates were incubated for 1 hours at 37 °C in the dark. The MTT-containing culture medium was then carefully removed, and 100 μL of dimethyl sulfoxide (DMSO) was added to each well to dissolve the formazan crystals formed. Absorbance was measured at 570 nm using a microplate reader (Bio-Rad Laboratories, USA).

Colony Formation Assay

To evaluate long-term proliferative capacity, stably transfected HeLa and SiHa cells were seeded into 6-well plates at a density of 600 cells per well. The cells were cultured in standard medium at 37 °C in a 5% CO₂ incubator for 10–14 days, with the medium replaced every 3–4 days. After macroscopic colonies had formed, the plates were washed with phosphate-buffered saline (PBS), fixed with 4% paraformaldehyde for 15 minutes, and stained with 0.5% crystal violet for 20 minutes at room temperature. Following extensive washing and air-drying, the plates were photographed.

Immunofluorescence (IF) Staining

For tissue samples, paraffin-embedded sections were deparaffinized, rehydrated, and subjected to heat-induced antigen retrieval in sodium citrate buffer. For in vitro cellular assays, cells grown on glass coverslips were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. After blocking with 5% BSA for 1 hour, both tissue sections and cell-covered coverslips were incubated overnight at 4 °C with primary antibodies against GSDMD and E-cadherin. The samples were then incubated with a mixture of corresponding Alexa Fluor-conjugated secondary antibodies for 1 hour at room temperature in the dark. Nuclei were counterstained with DAPI for 5 minutes. Finally, the samples were mounted using an anti-fade medium, and double-fluorescence signals were captured and analyzed using a laser scanning confocal microscope.

RNA Sequencing (RNA-Seq) Analysis

To explore the downstream mechanisms of GSDMD, total RNA was extracted from HeLa-shNC and HeLa-shGSDMD cells (three biological replicates per group) using TRIzol reagent and sequenced on an Illumina platform. Clean reads were aligned to the human reference genome. Differential expression analysis between groups was performed strictly using the DESeq2 R package based on the unnormalized raw count data, with thresholds of |log₂ fold change| > 1 and a False Discovery Rate (FDR) adjusted P-value < 0.05 (calculated via the Benjamini-Hochberg procedure to correct for multiple testing). Additionally, gene expression levels were normalized as Fragments Per Kilobase Million (FPKM) specifically for the purpose of generating heatmaps and expression visualization. In addition, Gene Set Enrichment Analysis (GSEA) was performed on the fully ranked gene list against the KEGG database using the clusterProfiler package, which revealed significant enrichment of the “Cell Adhesion Molecules” pathway.

In vivo Tumor Xenograft and Lymph Node Metastasis Models

All animal experiments were approved by the Ethics Committee of Shandong Cancer Hospital and Institute (Approval No.: SDTHEC202508053). Female BALB/c nude mice (4–5 weeks old) were purchased from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and housed under specific pathogen-free (SPF) conditions.

To evaluate primary tumor growth, a subcutaneous xenograft model was established. Briefly, stably transfected cervical cancer cells (HeLa-shNC and HeLa-shGSDMD, 5 × 10⁶ cells in 100 μL PBS) were injected subcutaneously into the right flank of nude mice (n = 8 per group). Tumor dimensions were measured every few days using a digital caliper, and tumor volumes were calculated using the standard formula: volume = (length × width²) / 2. At the experimental endpoint, mice were euthanized, and subcutaneous tumors were carefully excised, photographed, and weighed. The harvested tumor tissues were then fixed in 4% paraformaldehyde, embedded in paraffin, and subjected to immunohistochemical (IHC) staining for Ki-67, GSDMD, and E-cadherin to assess proliferation and EMT status.

To further investigate the impact of GSDMD on lymphatic dissemination in vivo, a popliteal lymph node metastasis model was employed, as classically described in previous studies.¹⁵ Stably transfected cells (HeLa-shNC and HeLa-shGSDMD, 2 × 10⁶ cells in 50 μL PBS) were injected directly into the footpads of the hind limbs of nude mice (n = 5 per group). After an appropriate growth period, the mice were sacrificed, and the popliteal lymph nodes were surgically dissected and photographed in situ. The metastatic lymph nodes were then isolated, and their volumes and weights were accurately measured and recorded. Finally, the excised lymph nodes were processed for IHC staining (Ki-67, GSDMD, and E-cadherin) to validate the metastatic burden and investigate the underlying mesenchymal-to-epithelial transition (MET) mechanisms in vivo.

Statistical Analysis

All statistical analyses were performed using GraphPad Prism software (version 9.0) or R software (version 4.2). All in vitro experiments were performed in at least three independent biological replicates. Continuous variables are presented as the mean ± standard deviation (SD). Differences between two groups were analyzed using the two-tailed Student’s t-test, while comparisons among multiple groups were conducted using One-way Analysis of Variance (ANOVA). P-value < 0.05 was considered statistically significant.

Results

Exploratory Single-Cell Transcriptomic Profiling Indicates the Activation of GSDMD-Mediated Pyroptosis in Cervical Cancer Lymph Node Metastasis

To initially explore the heterogeneity of the TME during CC progression and generate mechanistic hypotheses, we analyzed the single-cell transcriptomic profiles of primary cervical cancer and matched lymph node metastases from a representative patient. Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction classified the cells from the single patient into ten distinct major cell lineages, including T/NK cells, B cells, Plasma cells, Cycling cells, Epithelial (tumor) cells, Myeloid cells, Neutrophils, Fibroblasts, Endothelial cells, and Dendritic cells (DC) (Figure 1A). The identities of these cell clusters were robustly validated by the high expression of canonical marker genes, such as KRT8/KRT19 for Epithelial cells, CD68/LYZ for Myeloid cells, COL1A1/ACTA2 for Fibroblasts, and VWF/PLVAP for Endothelial cells (Figure 1B).

Figure 1.

Exploratory single-cell transcriptomic profiling and clinical validation reveal the enrichment of GSDMD in cervical cancer lymph node metastasis. (A) UMAP plot visualizing the integrated clustering of all cells derived from both the primary cervical tumor (CX) and the matched lymph node metastasis (LN) of a single patient into ten distinct major cell lineages. Different colors distinguish the identified cell clusters. (B) Heatmap displaying the normalized mean expression of canonical marker genes used to annotate the identified cell clusters. (C) Intercellular communication analysis between Epithelial (tumor) cells and the tumor microenvironment (TME). Left: A bidirectional bar chart comparing the total number of inferred interaction pairs originating from epithelial cells to other cellular components. The red bars extending to the right represent the number of interactions identified in the lymph node metastases (LN), whereas the blue bars extending to the left (plotted on the negative axis for visual contrast) represent the baseline interactions detected in the primary tumor (CX). Right: Dot plot highlighting the significant ligand-receptor signaling pathways mediating the interactions between the tumor microenvironment and epithelial cells. (D) GSEA revealing the significant enrichment of biological processes related to “pyroptosis” and “positive regulation of pyroptosis” in the LN metastatic group. (E) Dot plot showing the expression patterns of key pyroptosis-related regulators and executioners (AIM2, GSDMD, NLRP3, and CASP1) in the LN and CX groups. (F) Dot plot evaluating the expression levels of crucial downstream functional factors, including pro-inflammatory cytokines and pro-angiogenic drivers. In (E) and (F), the dot size indicates the percentage of cells expressing the target gene, and the color gradient (from grey to blue) reflects the average expression level. (G) Validation of GSDMD protein expression in paired clinical samples. Representative immunohistochemistry (IHC, top) and immunofluorescence (IF, bottom) images demonstrate the accumulation of GSDMD in metastatic lymph nodes compared to the matched primary cervical tumor.

Given the critical role of the TME in tumor metastasis, we evaluated the intercellular communication networks between Epithelial cells and other cellular components. The number of interaction pairs revealed a significantly intensified crosstalk in the LN metastatic lesions compared to the primary CX. Specifically, interactions between Epithelial cells and Endothelial cells, Fibroblasts, and Myeloid cells were remarkably enriched in the LN microenvironment (Figure 1C). Ligand-receptor pair analysis further demonstrated that signaling pathways closely associated with EMT and inflammation, including TGF-β, WNT, NOTCH and TNF, were predominantly upregulated in the LN group (Figure 1C). Detailed results of the bioinformatics analysis are provided in Supplementary Tables S3 and S4.

To further pinpoint the underlying molecular mechanisms driving LN metastasis, GSEA was performed. The results revealed that biological processes associated with “Pyroptosis” (NES = 1.59, P = 0.024) and “Positive Regulation of Pyroptosis” (NES = 1.52, P = 0.037) were significantly enriched in the LN metastatic lesions (Figure 1D). Subsequently, we assessed the expression profiles of key pyroptosis-related regulators. The dot plot demonstrated a striking upregulation of GSDMD, the critical executioner of pyroptosis, in the LN group compared to the primary CX, along with an elevated expression of AIM2 (Figure 1E). Consistent with the activation of GSDMD-associated signaling, the expression of its downstream pro-inflammatory cytokine IL1B was specifically increased in the LN group (Figure 1F). To directly validate these in silico transcriptomic predictions at the protein level, we evaluated the spatial expression of GSDMD in clinical tissues. We performed IHC and IF staining on paired primary cervical tumor tissues and their corresponding metastatic lymph nodes from the same patient cohort. Consistent with our single-cell data, both IHC and IF assays demonstrated a significant enrichment of GSDMD protein specifically within the metastatic lymph nodes, whereas the primary tumors displayed relatively weaker staining (Figure 1G). Collectively, these transcriptomic and histological findings strongly indicate that elevated GSDMD expression is closely associated with lymphatic dissemination in cervical cancer.

GSDMD Knockdown Dampens Pyroptotic Cytokine Release, Accelerates in Vitro proliferation, but Induces Mesenchymal-to-Epithelial Transition (MET)

To delineate the biological functions of GSDMD in cervical cancer in vitro, we established stable GSDMD knockdown models in HeLa and SiHa cell lines using three distinct shRNAs. The knockdown efficiency was evaluated by Western blotting. While sh1 and sh2 exhibited only minimal to moderate effects, sh3 demonstrated a profound reduction in GSDMD protein levels compared to the negative control (shNC) (Figure 2A). Consequently, the sh3 construct was exclusively utilized for all subsequent functional experiments (hereafter referred to as shGSDMD). ELISA revealed that the concentrations of released IL-1β (Figure 2B) and IL-18 (Figure 2C) in the culture supernatants were significantly decreased following GSDMD depletion in both HeLa and SiHa cells, confirming the functional blockade of the pyroptotic pathway. MTT assays demonstrated that GSDMD knockdown enhanced the cell viability and proliferation rates of both HeLa and SiHa cells over a 96-hour period (Figure 2D). Consistent with this, colony formation assays revealed that shGSDMD cells formed more and larger colonies compared to their shNC counterparts (Figure 2E).

Figure 2.

GSDMD knockdown dampens pyroptotic cytokine release and accelerates in vitro proliferation but triggers an epithelial state transition. (A) Verification of GSDMD knockdown efficiency. Western blot analysis showing the profound reduction of GSDMD protein levels in HeLa and SiHa cells stably transfected with three distinct short hairpin RNAs (sh1, sh2, and sh3) compared to the negative control (shNC). (B and C) GSDMD depletion inhibits the release of classic pyroptosis-associated pro-inflammatory factors (IL-1β and IL-18). (D) MTT assays evaluating the cell viability and proliferationof the designated HeLa and SiHa cell groups over 96 hours. (E) Restraining GSDMD-mediated cell death promotes long-term colony formation capability. (F) Hierarchical clustering heatmap generated from the bulk RNA sequencing of HeLa-shNC and HeLa-shGSDMD cells, illustrating the distinct transcriptomic profiles of top invasion-related genes and the remarkable upregulation of CDH1upon GSDMD knockdown. (G) GSEA plot highlights the dynamic shift in the “Cell Adhesion Molecules” gene set and the specific core enrichment of the CDH1 transcript driven by GSDMD depletion.

Despite the enhanced proliferation, our clinical data (Figure 1G) indicated a pro-metastatic role for GSDMD. To reconcile this, we performed RNA-sequencing on HeLa-shNC and HeLa-shGSDMD cells to explore downstream molecular alterations. Heatmap clustering of top invasion-related genes revealed distinct transcriptomic profiles between the two groups (Figure 2F). Notably, CDH1 was remarkably upregulated upon GSDMD knockdown. Furthermore, GSEA evaluating cell adhesion molecules highlighted the significant enrichment and prominent positional shift of CDH1 towards the shGSDMD phenotype (Figure 2G). Detailed sequencing results are provided in Supplementary Table S5. Collectively, these findings indicate that while GSDMD depletion accelerates in vitro proliferation by avoiding pyroptotic death, it concurrently induces a MET by upregulating CDH1, thereby potentially compromising the invasive and metastatic capabilities of cervical cancer cells.

GSDMD Promotes Cervical Cancer Cell Migration and Invasion in vitro by Driving EMT and Upregulating VEGFA/C

Driven by the transcriptomic finding that GSDMD depletion upregulates the epithelial marker CDH1, we investigated the functional impact of GSDMD on the metastatic potential of cervical cancer cells. Transwell invasion assays demonstrated that shGSDMD remarkably impaired the invasive capacity of both HeLa and SiHa cells compared to shNC (Figure 3A). Consistently, wound-healing assays revealed that the migratory ability of shGSDMD cells was significantly delayed, with a substantially lower scratch healing rate at 24 hours (Figure 3B).

Figure 3.

GSDMD knockdown suppresses cervical cancer cell migration and invasion by enforcing MET and repressing VEGFA/C expression. (A) Transwell assay demonstrating the impaired invasive capacity of HeLa and SiHa cells following GSDMD depletion. (B) Wound-healing assay assessing the migration kinetics of the indicated cell lines. (C) Western blot profiling of EMT markers and lymphangiogenic drivers. (D and E) Subcellular visualization affirming the reciprocal relationship between GSDMD and E-cadherin expression.

To elucidate the molecular mechanism underlying this restricted motility, we evaluated the expression of metastasis-associated proteins using Western blotting. Strikingly, GSDMD knockdown led to a upregulation of E-cadherin and downregulation of Snail in both cell lines (Figure 3C). Furthermore, the expression of VEGFA and VEGFC was significantly suppressed upon GSDMD knockdown (Figure 3C).

Additionally, double-labeling IF staining revealed that GSDMD knockdown triggered membrane/cytoplasmic accumulation of E-cadherin in both HeLa (Figure 3D) and SiHa (Figure 3E) cells. It is noteworthy that the baseline GSDMD immunofluorescence in HeLa-shNC cells appears relatively faint compared to SiHa-shNC cells. This visual disparity aligns with the intrinsically lower basal GSDMD expression in the HeLa cell line. While Western blotting readily detects these low-abundance proteins by analyzing concentrated whole-cell lysates under prolonged exposure (Figure 2A), standard in situ immunofluorescence under uniform confocal laser settings renders the lower baseline in HeLa cells less visually prominent. Taken together, these data suggest that GSDMD promotes cervical cancer metastasis by sustaining the EMT process and VEGFA/C expression.

GSDMD Overexpression Promotes EMT and Upregulates Lymphangiogenic Factors to Enhance Invasion

To further substantiate the pro-metastatic role of GSDMD, we performed gain-of-function experiments. First, we evaluated the baseline expression levels of GSDMD across various cervical cancer cell lines using HPA database (Figure 4A). The analysis revealed that HeLa cells exhibit relatively low endogenous GSDMD expression at both RNA and protein levels compared to other cell lines. Consequently, HeLa cells were selected to establish a stable GSDMD-overexpressing model. Transwell assays demonstrated that overexpression of GSDMD enhanced the invasive capability of HeLa cells, resulting in a higher number of invaded cells compared to the oe-NC group (Figure 4B). Similarly, wound-healing assays revealed that oe-GSDMD cells exhibited increased scratch healing rate at 24 hours (Figure 4C).

Figure 4.

GSDMD overexpression drives EMT and upregulates lymphangiogenic factors to fuel cell invasion. (A) The transcriptomic and translational expression profile of GSDMD across eight distinct human cervical cancer cell lines was retrieved from The Human Protein Atlas (HPA) database. (B and C) Functional validation affirming the pro-metastatic properties of GSDMD. (D) GSDMD overexpression represses the epithelial marker E-cadherin, activates an entire battery of mesenchymal core transcription factors (Snail, Slug, and β-catenin), and stimulates the production of VEGFA and VEGFC. (E) Double-labeling IF substantiating the loss of epithelial adhesion.

To delineate the molecular alterations orchestrating this invasive phenotype, we assessed the EMT signaling cascade and lymphangiogenic factors via Western blotting (Figure 4D). GSDMD overexpression repressed the expression of E-cadherin. Concomitantly, the core mesenchymal transcription factors, including Snail, Slug, and β-catenin were significantly upregulated in the oe-GSDMD group. Furthermore, we observed a remarkable induction of VEGFA and VEGFC, the primary drivers of angiogenesis and lymph node metastasis, upon GSDMD overexpression. Finally, we corroborated this EMT progression at the subcellular level using IF (Figure 4E). Consistent with the Western blot data, oe-GSDMD cells showed intense green GSDMD fluorescence with a concurrent loss of E-cadherin expression. Collectively, these results demonstrate that GSDMD acts as a driver of CC metastasis by triggering EMT and activating VEGFA/C-mediated signaling.

In vivo models Reveal that GSDMD Knockdown Accelerates Primary Tumor Growth but Profoundly Impairs Lymph Node Metastasis

To translate our in vitro findings in vivo, we used a subcutaneous xenograft model. Consistent with our proliferation data, mice bearing shGSDMD tumors exhibited an accelerated primary tumor growth trend, with larger tumor volumes and weights compared to the shNC group (Figure 5A). IHC confirmed GSDMD knockdown, increased Ki-67 expression, and upregulated E-cadherin, indicating that GSDMD depletion induces MET in vivo microenvironment (Figure 5B).

Figure 5.

In vivo, GSDMD knockdown accelerates primary subcutaneous tumor growth but drastically impairs lymph node metastasis. (A) Subcutaneous tumor xenograft model evaluating the impact of GSDMD on primary tumor growth (experimental duration: 4 weeks). The scatter plots below quantify the resulting tumor volumes and weights. (B) IHC images of the subcutaneous tumors validates the knockdown of GSDMD, the elevated Ki-67 proliferation index, and the induction of the epithelial marker E-cadherin in vivo. (C) shGSDMD mice exhibited a pronounced suppression of metastatic lesion formation in situ anatomical images of the popliteal lymph nodes (experimental duration: 8 weeks). (D–E) Quantitative comparisons showing that both the volume and the weight of the metastatic lymph nodes were reduced following GSDMD knockdown compared to the shNC control group. (F) Representative IHC staining of the excised metastatic lymph node tissues.

To determine if this forced MET halts lymphatic spread, we employed an in vivo lymph node metastasis model. Strikingly, despite accelerating primary tumor growth, GSDMD knockdown suppressed lymph node metastasis (Figure 5C). Both the volume (Figure 5D) and weight (Figure 5E) of metastatic lymph nodes were significantly reduced in the shGSDMD group. Immunohistochemical staining of lymph node metastatic lesions showed effective GSDMD knockdown in the shGSDMD group compared with shNC controls. Ki-67 positivity remained readily detectable in both groups, suggesting that proliferative activity was largely unaffected. In contrast, E-cadherin expression, which was nearly absent in control metastases, was consistently maintained in shGSDMD metastases (Figure 5F). Collectively, these in vivo data unveil a dual role for GSDMD in cervical cancer. Although GSDMD knockdown accelerates primary tumor growth, it significantly suppresses lymph node metastasis by inducing MET.

Discussion

LNM is the primary route of dissemination and a critical determinant of poor prognosis in CC. Despite advances in identifying metastatic biomarkers, the underlying mechanisms driving lymphatic spread remain incompletely understood. In this study, we uncovered a dual role of GSDMD in CC progression by integrating exploratory single-cell transcriptomic profiling with in vitro and in vivo experimental models. We demonstrated that while GSDMD depletion accelerates primary tumor growth, its elevated expression acts as a promoter of EMT and VEGFA/C-dependent lymphangiogenesis, thereby contributing to lymph node metastasis.

GSDMD is conventionally regarded as the primary effector of pyroptosis, an intensely inflammatory form of regulated cell death.¹⁶ In the context of neoplastic diseases, GSDMD exhibits a bidirectional regulatory function. On one hand, the activation of the GSDMD signaling pathway can effectively suppress primary tumor growth; on the other hand, pyroptosis may paradoxically promote tumor cell migration and metastasis.¹⁷ Our clinical tissue validation revealed that elevated GSDMD protein expression is enriched in metastatic lymph nodes. This paradoxical phenomenon suggests that during the metastatic cascade, the elevated expression of GSDMD provides evolutionary advantages to tumor cells that far outweigh its direct cytotoxic effects.

EMT is a necessary precondition for metastasis, during which epithelial cells dismantle their intercellular junctions and gain the ability to migrate.⁶ Our RNA-seq and rescue functional assays demonstrated that GSDMD depletion upregulated E-cadherin (CDH1) and downregulated mesenchymal transcription factors (Snail, Slug), forcing the cells into a strict epithelial state. Previous studies have reported that in pancreatic cancer, FABP4-mediated saturated fatty acids induces macrophage pyroptosis in a caspase-1/GSDMD-dependent manner.¹⁸ This specific activation of the NLRP3/IL-1β axis subsequently upregulates EMT signaling, promoting tumor cell migration, invasion, and metastasis. Consistent with this paradigm, our in vitro results showed that GSDMD knockdown markedly decreased IL-1β secretion. However, the precise intermediate mechanism directly linking GSDMD to the EMT transcriptional cascade remains unproven in our current study. We propose several distinct but non-mutually exclusive possibilities: First, similar to findings in other cancers, GSDMD might promote EMT indirectly via an autocrine or paracrine inflammatory loop sustained by secreted IL-1β. Second, the upregulated VEGFA/C axis itself could be the upstream driver that reshapes the cellular cytoskeletal and mesenchymal features. Alternatively, EMT could be a direct consequence of intracellular ion flux caused by caspase-dependent GSDMD pore activity. Finally, it is also highly plausible that full-length GSDMD exerts a pyroptosis-independent, direct scaffolding or transcriptional regulatory function to suppress E-cadherin expression. Dissecting these precise causal mechanisms—potentially using recombinant cytokines or cleavage-resistant GSDMD mutants—represents a critical direction for our future investigations.

Lymph node metastasis in cervical cancer is a complex cascade involving multiple biological events. Tumor cells must not only gain invasive and transmigratory abilities via EMT but also stimulate angiogenesis and lymphangiogenesis within the tumor microenvironment. These processes enable the cells to breach tissue barriers and establish the structural basis for distant spread.¹⁹ Our study provides molecular evidence that GSDMD functions both as an intracellular remodeler of the EMT transcriptional network and as a potent upregulator of the protein expression of VEGFA (pro-angiogenic) and VEGFC (pro-lymphangiogenic) in tumor cells. Many studies has conclusively demonstrated that VEGFC secreted by tumor cells, through activating the VEGFC/VEGFR-3 signaling axis in the microenvironment, constitutes the most critical mechanism driving tumor-associated lymphangiogenesis and lymphatic metastasis. Meanwhile, trace amounts of inflammatory mediators (such as IL-1β), facilitated by active GSDMD signaling, have also been proven to potently stimulate the paracrine expression of VEGF family members in the surrounding stroma and drive vascular network remodeling.²⁰ Synthesizing our results, we hypothesize that cervical cancer cells with high GSDMD expression, while acquiring an EMT-mediated highly invasive phenotype, are highly likely to actively participate in the remodeling of the lymphatic microenvironment through the paracrine secretion of key pro-lymphangiogenic factors such as VEGFA and VEGFC.

In addition to intrinsic cell motility, effective dissemination through the lymphatic system also requires dynamic interactions with TME.^21–23 Consistent with this notion, our CellChat intercellular communication analysis revealed a markedly intensified crosstalk between tumor cells and the surrounding endothelial cells as well as fibroblasts in lymph node metastatic lesions. This amplified interactome was specifically enriched in the TNF and TGF-β signaling pathways. Immune cells or stromal cells in the tumor microenvironment secrete TNF, which induces GSDMD-mediated pyroptosis/inflammatory release in tumor cells.^24,25 The released inflammatory factors may in turn synergize with TGF-β signaling to promote EMT in tumor cells, thereby enhancing their invasive and migratory capacities.²⁶

While our study provides a mechanistic rationale for targeting GSDMD in metastatic CC, several critical limitations remain. First, our exploratory scRNA-seq was restricted to paired tissues from a single patient, and our spatial validation lacked a large, independent clinical IHC cohort. Additionally, our histological validation primarily contrasted primary tumors with matched metastatic lesions, lacking metastasis-negative (normal) lymph nodes from the same patients as anatomical controls. Given the profound clinical heterogeneity of cervical cancer, our data currently serves as a hypothesis-generating starting point. Second, our in vivo experiments utilized immunodeficient nude mice. Since inflammatory signaling interacts deeply with adaptive immunity, future studies must employ immunocompetent models. Third, our study primarily focused on total GSDMD expression and downstream macroscopic phenotypes without mapping the canonical biochemical cascade. The lack of direct evidence regarding upstream cleavage (Caspase-1/4/5), the accumulation of the active GSDMD-N terminal domain, and real-time membrane pore formation leaves a mechanistic gap. Therefore, identifying the specific mechanisms by which GSDMD drives EMT and lymphangiogenesis represents a critical direction for our next phase of research. Finally, from a translational perspective, we must emphasize that this is predominantly a preclinical experimental study. Although lymph node metastasis is a key issue in daily oncological management, GSDMD is not yet ready to guide treatment decisions. Extensive independent validation in large, multi-center cohorts is strictly required to determine whether these experimental findings can eventually translate into real patient care.

In conclusion, through an exploratory single-cell transcriptomic analysis of paired cervical cancer tissues, our study suggests that GSDMD acts not simply as an executor of cell death, but also contributes to tumor invasiveness. By promoting EMT and stimulating VEGFA/C-mediated microenvironmental remodeling, GSDMD is actively involved in lymphatic dissemination, making it a potential experimental target. While targeting the tumor-promoting and pro-metastatic functions of GSDMD may represent a preliminary, hypothesis-generating therapeutic vulnerability, we fully acknowledge that these findings are currently restricted to preclinical experimental models. Stronger independent validation in larger patient cohorts and clinically relevant models is strictly necessary before establishing GSDMD as a definitive biomarker or a viable therapeutic target in routine clinical practice.

Funding Statement

This study received funding from the Medical Research Project of China Medical and Health Development Foundation (chmdf2025-xrky11-16): Molecular Mechanism Study of GSDMD Mediated Cell Pyroptosis in Cervical Cancer Lymphatic Metastasis, during the conduct of the study.

Data Sharing Statement

Publicly available datasets analyzed in this study can be found in online databases, including the TCGA database and the GEO database. The original data supporting the conclusions of this article will be made available without reservation by the first author upon reasonable request.

Ethics Approval and Informed Consent

The study was approved by the Institutional Review Board of Shandong cancer hospital (No. SDTHEC202508053). Informed consent was obtained from all participants, and the research was conducted in accordance with the Declaration of Helsinki.

Author Contributions

Fufeng Gao: Conceptualization, Study Design, Execution, Acquisition of Data, Analysis and Interpretation, Visualization, Writing-Original Draft, Writing-Review & Editing. Hang Yu: Conceptualization, Acquisition of Data, Visualization, Supervision, Writing-Review & Editing. Hui Guo: Study design, Analysis and Interpretation, Writing-Review & Editing. Jiangtao Fan: Conceptualization, Study Design, Analysis and Interpretation, Visualization, Supervision, Funding acquisition, Writing-Review & Editing. All authors took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

Dr Fufeng Gao reports Support for the manuscript from Medical Research Project of China Medical and Health Development Foundation: Molecular Mechanism Study of GSDMD Mediated Cell Pyroptosis in Cervical Cancer Lymphatic Metastasis, during the conduct of the study.

Dr Hui Guo reports Support for the manuscript from Medical Research Project of China Medical and Health Development Foundation, during the conduct of the study.

The authors declare no other conflicts of interest in this work.