Raw count data were retrieved from the GSE193677 and GSE214695 datasets available in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/). The GSE193677 dataset includes 872 UC and 461 Control samples, whereas the GSE214695 dataset consists of 6 UC and 6 Control samples. Ensembl IDs in both datasets were converted to gene symbols using the reference genome hg19. The count expression matrices were subsequently processed using the “edgeR” package (v 3.40.1) in R (v 4.2.2, R Foundation for Statistical Computing, Vienna, Austria) to compute average values and normalize expression levels for genes with multiple entries.

The “edgeR” package was employed for screening differentially expressed genes (DEGs), with the screening criteria set as $|$log₂FC$|$ $>$1 & adj. p value 0.05. Gene ID conversion was performed on the identified DEGs using “org.Hs.eg.db”, and functional enrichment analysis was conducted using the “clusterProfiler” package (v 4.4.4).

Single-cell analysis was carried out using the “Seurat” package (v 4.3.0.1). Cells were screened based on the criteria of 200 nFeature RNA 6000. Cell type annotation was subsequently conducted using the SingleR package (v 2.2.0).

Pyroptosis-related genes were obtained through a systematic literature review [12]. Differential expression analysis of these genes was performed using the GSE193677 dataset. Heatmaps were plotted using the “pheatmap” package. Gene set variation analysis (GSVA) was carried out using the “GSVA” package to calculate pyroptosis scores for each sample group.

The “WGCNA” package (v 1.71) was applied for conducting WGCNA on the identified key genes, and the “clusterProfiler” package was employed for performing Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses.

The STRING database (https://string-db.org/, v 11.5) was used for carrying out protein-protein interaction (PPI) analysis. The MCODE plug-in in Cytoscape (v 3.9.1) was employed to identify key genes within the PPI networks. Hub gene screening was performed using the following three machine learning approaches: LASSO regression using the “glmnet” package (v 4.1-4), random forest analysis using the “randomForest” package (v 4.7-1.1), and support vector machine-recursive feature elimination (SVM-RFE) using the “CARET” (v 6.0-92) and “e1071” (v 1.7-11) packages. Venn diagrams were generated with the “Venn” package. Single-sample gene set enrichment analysis (ssGSEA) of aquaporin 9 (AQP9) was performed using GSEA (v 4.3.2).

Immune cell infiltration analysis was performed on normalized expression data using the “CIBERSORT” package (v 1.04). Bar plots were generated to visualize the relative proportions of immune cells. Correlation analysis among immune cell types was conducted using the “corrplot” package (v 0.92), with scatter plots and boxplots created using the “ggplot2” and “ggpubr” (v 0.6) packages.

Male C57BL/6 mice (SPF, 6–8 weeks old) were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd., and housed in an 18–22 °C facility with ad libitum access to water and food. Following a one-week acclimatization period, twelve mice were randomly assigned to two groups: Control (n = 6) and UC (n = 6). Mice in the UC group were continuously administered with drinking water containing 2% (w/v) dextran sulfate sodium salt (DSS) (Sigma, Shanghai, China, 265152-M) for 7 days, while mice in the Control group were fed with normal drinking water. Predefined humane endpoints were established such that mice exhibiting signs of severe pain during the experiment would be euthanized prior to the study completion. However, none of the mice were euthanized, and all mice survived until the end of the experiment. After 7 days of treatment as described above, mice were euthanized for colon tissue collection to evaluate whether the model was successfully constructed. Euthanasia was performed using CO₂ inhalation at a controlled replacement rate of 30%–70% of the container volume/minute to ensure rapid and humane loss of consciousness. This experiment was approved by the Ethics Committee of Laboratory Animal Management and Welfare of the First Affiliated Hospital of Harbin Medical University (Approval No.: IACUC-2023053), and was conducted following the ARRIVE guidelines.

Human neutrophils (Immocell, Xiamen, Fujian, China, IMP-H209) and human intestinal epithelial cells (YaJi Biological, Shanghai, China, YS3102C) were utilized in this study. Neutrophils were cultured in a specialized medium (Immocell, Xiamen, Fujian, China, IMP-H209-1), while intestinal epithelial cells were maintained in Opti-MEM medium supplemented with fetal bovine serum (ThermoFisher, Shanghai, China, A5256701), EGF, and GlutaMAX™ (ThermoFisher, Shanghai, China, 42360032). Cells were cultured under standard conditions (37 °C, 5% CO₂). Formation of NETs was induced by stimulating neutrophils with 100 nM/L phorbol 12-myristate 13-acetate (PMA) for 4 h. Cells were transfected with either the constructed AQP9-targeting siRNA or si-NC. Pyroptosis was induced by treating the cells with 10 µM nigericin for 24 h. Activation of the JAK2-STAT3 pathway was achieved by exposing the cells to 20 µM Colivelin for 12 h. All cell lines were validated by STR profiling and tested negative for mycoplasma.

Tissue samples were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned into 4 µm slices. Sections were stained with HE (Beyotime, Shanghai, China, C0105S) and examined microscopically.

1 $\times$ 10⁶ cells were inoculated into 24 mm sterile cell slides and placed in 6-well plates. the following adherence, cells were fixed with 4% paraformaldehyde and blocked using a blocking buffer. Cells were then incubated sequentially with primary and secondary antibodies, followed by nuclear staining with DAPI. Slides were mounted and visualized under a microscope. Primary antibodies used were myeloperoxidase (MPO, Cell Signaling Technology, Danvers, MA, USA, 14569) and citrullinated histone H3 (CitH3, Cell Signaling Technology, Danvers, MA, USA, 97272).

Total protein was extracted from colon tissues or cultured cells using a lysis buffer containing 1% protease inhibitor, and then quantified using the BCA kit (Beyotime, Shanghai, China, P0012S). Electrophoresis, membrane transfer, antibody incubation, and color development were subsequently performed. The antibodies used included cleaved N-terminal gasdermin D (GSDMD-N, Cell Signaling Technology, Danvers, MA, USA, 36425), GSDMD (Cell Signaling Technology, Danvers, MA, USA, 39754), p-STAT3 (Cell Signaling Technology, Danvers, MA, USA, 9145), p-JAK2 (Cell Signaling Technology, Danvers, MA, USA, 3776), JAK2 (Cell Signaling Technology, Danvers, MA, USA, 3230), STAT3 (Cell Signaling Technology, Danvers, MA, USA, 12640), AQP9 (Invitrogen, Shanghai, China, PA5-114872), ZO-1 (Invitrogen, Shanghai, China, 61-7300), and occludin (Cell Signaling Technology, Danvers, MA, USA, 91131). $\beta$-actin (Cell Signaling Technology, Danvers, MA, USA, 4970S) was set as the internal reference protein.

The expression levels of corresponding inflammatory factors were measured using the following ELISA kits: mouse interleukin (IL)-1$\beta$ (Beyotime, Shanghai, China, PI301), mouse IL-18 (Beyotime, Shanghai, China, PI553), human IL-1$\beta$ (Beyotime, Shanghai, China, PI305), human IL-18 (Beyotime, Shanghai, China, PI558), and MPO-DNA (COIBO BIO, Shanghai, China, CB21448-Hu).

Total RNA was extracted from colon tissues using the TRIzol reagent. RNA purity and integrity were assessed via agarose gel electrophoresis, and the A260/A280 ratio was measured by NanoDrop 2000 spectrophotometry. cDNA synthesis was performed using the PrimeScript RT kit (TaKaRa, Beijing, China, RR014A). qRT-PCR was conducted with PowerTrack™ SYBR Green Master Mix (Applied Biosystems, Shanghai, China, A46012) as per the manufacturer’s instructions. Primer sequences are listed in Table 1. The relative expression of mRNA was calculated using the 2^{-Δ⁢Δ⁢Ct} method.

CCK-8 assay was carried out according to the manufacturer’s manual (Beyotime, Shanghai, China, C0037). A total of 2000 cells were added to each well of the 96-well plate and incubated for 24 h. Subsequently, 10 µL of the CCK-8 solution was added to each well, and incubation was performed at 37 °C for 2 h. Absorbance at 450 nm was measured using a microplate reader.

Cell death was detected using a TUNEL assay kit (Beyotime, Shanghai, China, C1086). Briefly, cells were fixed with 4% paraformaldehyde for 30 min, and incubated with PBS containing 0.3% Triton X-100, at room temperature for 5 min. Subsequently, cells were incubated with 50 µL of TUNEL reaction mixture at 37 °C for 60 min. After mounting with an anti-fade sealing solution, samples were analyzed via fluorescence microscopy.

Intracellular ROS levels were measured using a DCFH-DA probe (Beyotime, Shanghai, China, S0033S). Briefly, 5 $\times$ 10³ cells were seeded into 96-well plates and incubated with 8 µM DCFH-DA at 37 °C for 15 min. The results were observed under a fluorescence microscope.

GraphPad Prism 9.0.0 (GraphPad Software, Boston, MA, USA) was utilized for carrying out statistical analysis and data visualization. Data were expressed as mean $\pm$ standard deviation. A t-test was applied for making comparisons between groups, with p 0.05 denoting a statistically significant difference. All animal experiments were each conducted with six biological replicates, and all cell experiments were each carried out with three biological replicates.

Differential expression analysis was performed on the GSE193677 dataset using the thresholds of $|$log₂FC$|$ $>$1 & adj. p value 0.05. A total of 516 DEGs were identified, comprising 84 downregulated and 432 upregulated genes (Fig. 1A,B). Functional enrichment analysis of these DEGs was subsequently conducted. In the GO-biological process (BP) category (Fig. 2A), DEGs were predominantly enriched in processes such as defense response to bacteria and humoral immune response. For the GO-cellular component (CC) category (Fig. 2B), enrichment of DEGs was observed in the immunoglobulin complex and the circulating immunoglobulin complex. In the GO-molecular function (MF) category (Fig. 2C), DEGs were mainly associated with antigen binding, immunoglobulin receptor binding, and cytokine activity. KEGG pathway analysis (Fig. 2D) revealed that these DEGs were principally enriched in the cytokine-cytokine receptor interaction pathway.

Fig. 1.

Screening of DEGs. (A) Volcano map showing the results of differential expression analysis from the GSE193677 dataset. (B) Heatmap showing the expression of DEGs across samples from different groups in the GSE193677 dataset. DEGs, differentially expressed genes.

Fig. 2.

Functional enrichment analysis of the identified DEGs. (A) GO-BP. (B) GO-CC. (C) GO-MF. (D) KEGG pathway enrichment analysis results. GO, gene ontology; BP, biological process; CC, cell component; MF, molecular function; KEGG, kyoto encyclopedia of genes and genomes.

Differential expression analysis was conducted on pyroptosis-related genes within the GSE193677 dataset, with the results presented in Fig. 3A. Pyroptosis scores for each sample were calculated via GSVA. The results demonstrated that the UC group exhibited elevated pyroptosis scores compared with the Control group (Fig. 3B).

Fig. 3.

Pyroptosis score. (A) Heatmap illustrating the expression of pyroptosis-related genes across samples from different groups in the GSE193677 dataset. (B) Bar chart showing the pyroptosis scores for each group. *p 0.05; **p 0.01; ***p 0.001.

WGCNA was performed on the identified DEGs. Initial sample clustering led to the exclusion of four outlier samples (Fig. 4A), after which the remaining samples were re-clustered and a heatmap of clinical features was generated (Fig. 4B). Based on the scale-free topology criterion, a soft-thresholding power of 7 was selected to construct the weighted co-expression network (Fig. 4C–E). A topological overlap matrix (TOM) was then constructed (Fig. 4F), and inter-module correlations were analyzed (Fig. 4G). Through correlation analysis between clinical traits (pyroptosis score and UC status) and gene modules (Fig. 4H), two modules, blue (76 genes) and turquoise (252 genes), were identified with strong correlations (p 0.05, r $>$ 0.5). Gene significance (GS) and module membership (MM) analyses (Fig. 4I,J) confirmed that both modules were positively correlated with the pyroptosis score. A total of 149 key genes were subsequently identified using the criteria of MM $>$0.6 and GS $>$0.4, comprising 110 genes from the turquoise module and 39 from the blue module. Functional enrichment analysis of these key genes (Fig. 5A,B) revealed their predominant involvement in cytokine-cytokine receptor interaction, leukocyte migration, immune receptor activity, and positive regulation of cytokine production.

Fig. 4.

WGCNA. (A) Sample clustering. (B) Sample clustering and clinical feature heatmaps. (C) Scale-free fitting index and average connectivity analysis under different soft threshold powers. (D) Visual testing of scale-free topologies. (E) Heatmap showing the correlation between module characteristic genes and clinical status. (F) TOM diagram. (G) Heatmap showing the correlation between modules. (H) Heatmap showing the correlation between modules and clinical features. (I) Correlation analysis of the blue module with UC. (J) Correlation analysis of the turquoise module with UC. WGCNA, weighted gene co-expression network analysis; TOM, topological overlap matrix; UC, ulcerative colitis.

Fig. 5.

Functional enrichment analysis. (A) GO enrichment. (B) KEGG pathway enrichment.

PPI analysis was conducted on the 149 key genes identified through WGCNA (Fig. 6A). Using Cytoscape, a core network comprising 12 genes (all upregulated in the UC group) was generated (Fig. 6B). Three machine learning algorithms were applied to these 12 genes to further identify hub genes: LASSO regression (Fig. 6C,D), random forest (Fig. 6E,F), and SVM-RFE (Fig. 6G). The intersection of results from all three algorithms yielded five hub genes: AQP9, S100A8, S100A9, S100A12, and VNN2 (Fig. 6H).

Fig. 6.

Hub gene screening. (A) PPI network of 149 WGCNA-derived genes. (B) Key genes identified within the PPI network. (C) Calculation of regression coefficients. (D) Best forecasting models. (E) The effect of the decision tree number on error rate. (F) Results of the Gini coefficient method in a random forest classifier. (G) RMSE evaluates SVM regression models. (H) A Venn diagram illustrating the similarities and differences among the results of the three machine learning algorithms. PPI, protein-protein interaction; RMSE, root mean square error; SVM, support vector machine.

Immunoinfiltration analysis was conducted on the GSE193677 dataset (Fig. 7A), and inter-correlations among immune cell populations were assessed (Fig. 7B). A strong positive correlation was revealed between neutrophils and Gamma delta T cells. The relative abundance of individual immune cell types was compared between the UC and Control groups (Fig. 7C), with statistically significant differences observed in 17 of 22 immune cell types. Notably, AQP9 expression was positively correlated with neutrophils, macrophages M1, and plasma cells, while negatively correlated with resting mast cells, memory B cells, naïve CD4⁺ T cells, resting dendritic cells, and macrophages M2 (Fig. 7D). These results suggest that AQP9 may modulate immune cell infiltration, thereby contributing to the pathogenesis of UC.

Fig. 7.

Immunoinfiltration analysis. (A) Proportions of 22 immune cells in the samples within the GSE193677 dataset. (B) Correlations between immune cells. (C) Comparison of the differences in immune cells between the Control and UC groups. (D) Lollipop plot showing the correlation between immune cells and AQP9 expression (In the lollipop plot, the color of the circles represents the statistical significance of the correlation between genes and immune cells).

Correlation analysis revealed a significant association between AQP9 expression and pyroptosis scores (Fig. 8A). Consistently, AQP9 was upregulated in the UC group within the GSE193677 dataset (Fig. 8B). ssGSEA indicated that AQP9 may influence the JAK-STAT pathway (Fig. 8C). According to recent literature, the JAK-STAT pathway plays a pivotal role in the regulation of pyroptosis [13].

Fig. 8.

Correlation between AQP9 and pyroptosis. (A) Correlation between AQP9 expression and pyroptosis scores. (B) Boxplot showing the expression of AQP9 in both the UC and Control groups. (C) Association between AQP9 expression and JAK-STAT pathway activity assessed by ssGSEA. ssGSEA, single-sample gene set enrichment analysis.

We interrogated the single-cell RNA sequencing dataset GSE214695 to pinpoint the specific cell type in which AQP9 exerts its influence on the pathogenesis of UC. Upon clustering, a total of 24 distinct cell clusters were identified (Fig. 9A). These clusters were subsequently annotated into nine cell types: T cells, tissue stem cells, neutrophils, epithelial cells, macrophages, B cells, common myeloid progenitors, endothelial cells, and neurons (Fig. 9B). Differential expression analysis of AQP9 across these cell types revealed significant upregulation of AQP9 in epithelial cells and neutrophils, with no significant differential expression observed in B cells, T cells, endothelial, or macrophages and no expression in the remaining cell types (Fig. 9C). Therefore, AQP9 is predominantly expressed in neutrophils and, notably, its expression is elevated in neutrophils from UC patients compared to those from individuals in the Control group (Fig. 9C).

Fig. 9.

AQP9 is predominantly expressed in neutrophils. (A) t-distributed stochastic neighbor embedding (t-SNE) plot showing cell clustering. (B) t-SNE plot illustrating cell annotation situation. (C) Bar chart displaying the differential expression of AQP9 in each cell across groups. CMP, common myeloid progenitors. ^nsp $>$ 0.05, ****p 0.0001.

To validate findings from the bioinformatics analysis described above, a mouse model of UC was established. Mice in the Control group were maintained on a standard diet and water intake, and they exhibited normal activity and stable body weight. In contrast, mice in the UC group displayed reduced activity, curled postures, and significant body weight loss after 4–5 days of modeling (Fig. 10A). Additionally, the spleen index was elevated in the UC group compared to the Control group (Fig. 10B). Macroscopically, colon tissues from the Control group showed no signs of hemorrhage or ulceration, whereas those from the UC group exhibited marked intestinal congestion, hemorrhagic changes, and substantial colon shortening (Fig. 10C). Histological examination revealed that colonic architecture was preserved in the Control group, with intact glandular structures, well-aligned epithelial cells, normal crypt morphology, and preserved lamina propria and muscular layers. In contrast, the UC group demonstrated extensive inflammatory cell infiltration, disrupted mucosal architecture, crypt loss, ulceration, and epithelial necrosis (Fig. 10D). Consistent with these pathological changes, the expression of pro-inflammatory cytokines IL-18 and IL-1$\beta$ was significantly upregulated in the colon tissues of mice with UC (Fig. 10E), while that of ZO-1 and occludin were markedly downregulated (Fig. 10F). Subsequent WB and qRT-PCR analyses confirmed a significant upregulation of AQP9 expression in the UC group (Fig. 10F–H), consistent with the findings from bioinformatics analysis. Ly6G is a specific surface marker for neutrophils. In the merged immunofluorescence image, we observed significant overlap between the Ly6G and AQP9 signals, which is displayed as a yellow signal. This indicates that AQP9 is expressed in neutrophils (Fig. 10H). Moreover, the levels of NET-associated markers MPO and CitH3 were found to be elevated in the UC group compared to the Control group. Immunofluorescence analysis showed their co-localization within the web-like structures of NETs, which is a hallmark of NETs (Fig. 10I).

Fig. 10.

AQP9 is highly expressed in the colon tissues of mice with UC. (A) Daily body weight changes of mice in both the Control and UC groups (n = 6). (B) Spleen index was measured in mice from both the Control and UC groups on day 7 post-modeling (n = 6). (C) Colon length was measured in mice from both the Control and UC groups on day 7 post-modeling (n = 6). (D) Histopathological changes in mouse colon tissue were examined in both the Control and UC groups on day 7 post-modeling (n = 6, scale bar = 100 µm). (E) Measurement of the levels of IL-18 and IL-1$\beta$ in mouse colon tissues by ELISA in both the Control and UC groups on day 7 post-modeling (n = 6). (F) Detection of the expression of AQP9 and pyroptosis-related proteins in mouse colon tissues by WB in both the Control and UC groups on day 7 post-modeling (n = 6). (G) Detection of the expression of AQP9 in mouse colon tissue by qRT-PCR in both the Control and UC groups on day 7 post-modeling (n = 6). (H) Assessment of the expression of AQP9 and Ly6G in mouse colon tissue by immunofluorescence in both the Control and UC groups on day 7 post-modeling (n = 6, scale bar = 100 µm). (I) Assessment of the expression of MPO and CitH3 in mouse colon tissue by immunofluorescence in both the Control and UC groups on day 7 post-modeling (n = 6, scale bar = 100 µm). **p 0.01. IL, interleukin; ELISA, enzyme linked immunosorbent assay; WB, western blotting; Ly6G, lymphocyte antigen 6 complex locus G6D; MPO, myeloperoxidase; CitH3, citrullinated histone H3.

Neutrophils were treated with PMA in vitro to induce the formation of NETs. PMA stimulation was revealed to promote the expression of NETs markers peptidylarginine deiminase 4 (PAD4), CitH3, and MPO (Fig. 11A,C), the expression of pyroptosis marker GSDMD-N (Fig. 11A), as well as the expression of p-JAK2 and p-STAT3 (Fig. 11B). Additionally, PMA treatment upregulated AQP9 expression (Fig. 11B), elevated ROS levels (Fig. 11D), and increased IL-1$\beta$, Lactate Dehydrogenase (LDH), and MPO-DNA complex levels (Supplementary Fig. 1). To determine the regulatory role of AQP9, AQP9 expression was silenced in the PMA-treated neutrophils. Knockdown of AQP9 significantly reduced the expression of PAD4, MPO, CitH3 (Fig. 11A,C), GSDMD-N (Fig. 11A), p-JAK2 and p-STAT3 (Fig. 11B), ROS levels (Fig. 11D), as well as levels of IL-1$\beta$, LDH, and MPO-DNA complex (Supplementary Fig. 1). Co-culture experiments with neutrophils and intestinal epithelial cells revealed that PMA treatment reduced epithelial cell viability (Fig. 11E), increased cell death (Fig. 11F), and suppressed the expression of ZO-1 and occludin (Fig. 11G). Conversely, AQP9 knockdown restored epithelial cell viability, decreased cell death, and promoted the expression of ZO-1 and occludin (Fig. 11E–G).

Fig. 11.

AQP9 knockdown inhibits PMA-induced formation of NETs to alleviate intestinal epithelial cell injury. (A) Detection of intracellular expression of PAD4, GSDMD-N, and GSDMD in different groups by WB (n = 3). (B) Detection of intracellular expression of AQP9 and JAK2-STAT3 pathway-related proteins in different groups by WB (n = 3). (C) Assessment of intracellular expression of MPO and CitH3 in different groups by immunofluorescence (n = 3, scale bar = 50 µm). (D) Determination of intracellular ROS levels in different groups using a fluorescent probe (n = 3, scale bar = 100 µm). (E) Assessment of intestinal epithelial cell viability in each group after co-culture with the aforementioned neutrophils using the CCK-8 assay (n = 3). (F) Evaluation of intestinal epithelial cell death in each group by TUNEL assay (n = 3, scale bar = 50 µm). (G) Determination of the expression of ZO-1 and occludin in intestinal epithelial cells from each group by WB (n = 3). ^nsp $>$ 0.05; *p 0.05; **p 0.01. NETs, neutrophil extracellular traps; PAD4, peptidylarginine deiminase 4 ; GSDMD-N, cleaved N-terminal gasdermin D; ROS, reactive oxygen species; CCK-8, cell counting kit-8; TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling; ZO-1, zonula occludens 1.

To further elucidate the interplay between AQP9, pyroptosis, and the formation of NETs, neutrophils from the PMA+si-AQP9 group were treated with the pyroptosis agonist Nigericin. Nigericin treatment did not alter AQP9 expression (Fig. 12A), but significantly enhanced GSDMD-N expression (Fig. 12B), upregulated the expression of NETs markers PAD4, MPO, and CitH3 (Fig. 12B,C), elevated ROS levels (Fig. 12D), and increased the levels of IL-1$\beta$, LDH, and MPO-DNA complex (Supplementary Fig. 2). Subsequent co-culture with intestinal epithelial cells demonstrated that, compared to the PMA+si-AQP9+EC group, the PMA+si-AQP9+Nigericin+EC group exhibited reduced epithelial cell viability (Fig. 12E), increased cell death (Fig. 12F), and decreased expression of ZO-1 and occludin (Fig. 12G).

Fig. 12.

AQP9 knockdown inhibits pyroptosis-mediated formation of NETs to alleviate intestinal epithelial cell injury. (A) Detection of intracellular AQP9 expression in different groups by WB (n = 3). (B) Detection of intracellular expression of PAD4, GSDMD-N, and GSDMD in different groups by WB (n = 3). (C) Assessment of intracellular expression of MPO and CitH3 in different groups by immunofluorescence (n = 3, scale bar = 50 µm). (D) Measurement of intracellular levels of ROS in different groups using a fluorescent probe (n = 3, scale bar = 100 µm). (E) Assessment of intestinal epithelial cell viability in each group after co-culture with the aforementioned neutrophils using the CCK-8 assay (n = 3). (F) Evaluation of intestinal epithelial cell death in each group by TUNEL assay (n = 3, scale bar = 50 µm). (G) Determination of the expression of ZO-1 and occludin in intestinal epithelial cells from each group by WB (n = 3). **p 0.01.

Based on the results from our previous experiments showing that AQP9 knockdown suppresses the JAK2-STAT3 pathway (Fig. 11B), we hypothesized that AQP9 may regulate pyroptosis via modulation of this pathway. To make further validation, neutrophils from the PMA+si-AQP9 group were treated with the JAK2-STAT3 pathway agonist Colivelin. It was revealed that treatment with Colivelin did not affect AQP9 expression (Fig. 13A) but markedly enhanced the expression of JAK2, STAT3, and GSDMD-N (Fig. 13A,B). Additionally, the levels of NET-associated proteins PAD4, MPO, and CitH3 were upregulated (Fig. 13B,C), ROS levels were elevated (Fig. 13D), and levels of IL-1$\beta$, LDH, and MPO-DNA complex were increased (Supplementary Fig. 3). Co-culture with epithelial cells demonstrated that Colivelin-treated neutrophils significantly inhibited epithelial cell viability (Fig. 13E), increased cell death (Fig. 13F), and downregulated ZO-1 and occludin expression (Fig. 13G), thereby reversing the protective effects conferred by AQP9 silencing.

Fig. 13.

AQP9 knockdown inhibits the JAK2-STAT3 pathway to regulate pyroptosis-mediated formation of NETs to alleviate intestinal epithelial cell injury. (A) Detection of intracellular expression of AQP9 and JAK2-STAT3 pathway-related proteins in different groups by WB (n = 3). (B) Detection of intracellular expression of PAD4, GSDMD-N, and GSDMD in different groups by WB (n = 3). (C) Assessment of intracellular expression of MPO and CitH3 in different groups by immunofluorescence (n = 3, scale bar = 50 µm). (D) Measurement of intracellular ROS levels in different groups using a fluorescent probe (n = 3, scale bar = 100 µm). (E) Assessment of intestinal epithelial cell viability in each group after co-culture with the aforementioned neutrophils using the CCK-8 assay (n = 3). (F) Evaluation of intestinal epithelial cell death in each group by TUNEL assay (n = 3, scale bar = 50 µm). (G) Determination of the expression of ZO-1 and occludin in intestinal epithelial cells from each group by WB (n = 3). **p 0.01.