Gasdermin D aggravates a mouse model of radiation-induced liver disease by promoting chemokine secretion and neutrophil recruitment

Abstract

Radiation-induced liver disease (RILD) severely impairs the outcome of patients receiving irradiation (IR); however, its underlying mechanism remains unknown. GSDMD drives the progression of pyroptosis, and can be induced by IR in the gut and bone marrow, but its role in RILD remains unknown. Here we show that GSDMD is significantly upregulated and positively correlated with RILD severity in a mouse model. Hepatocytes are identified as critical pyroptotic cells in RILD thorough scRNA-seq, immunofluorescence and fluorescence-activated cell sorting analysis. Functional and mechanistic analysis using Gsdmd knockout (Gsdmd^∆Hep) mice and cell models. Mechanistically, GSDMD is indispensable for triggering hepatocyte pyroptosis and initiating the activation of transcription factor STAT5A, which subsequently promoted CXCL1 expression to recruit neutrophil into liver to accelerate the severity of RILD. We also discovered that pharmacological targeting GSDMD and its downstream CXCL1 effectively alleviated RILD. Together, our study demonstrates that GSDMD as therapeutic targets to improve RILD.

Fig. 1. The upregulation of GSDMD-FL/N and pyroptotic phenotype in hepatocytes after irradiation.

A Immunoblots of GSDMD-FL/N in livers from Gsdmd^FL/FL mice after short-term (ST-RILD; 30 Gy, 7 days) or long-term (LT-RILD; 6 Gy × 3, 10 weeks) irradiation (n = 3, Ctrl; n = 5, ST/LT). B Gsdmd, Il1b and other pyroptosis-related genes expression in each immune cluster of Ctrl and IR groups. Y axis indicates log-normalized expression. C Representative low-magnification (Scale bars, 100×, 200 μm) and blow-up magnification (Scale bars, 400×, 50 μm) images of GSDMD-FL staining in Ctrl, ST-RILD and LT-RILD mice (five fields were observed per liver, n = 5/group). ****p = 5E-10, Ctrl vs ST; ****p = 3.499E-05, Ctrl vs LT. D Immunofluorescence visualization of GSDMD-FL (green), DAPI (blue), HNF4a/CD11b/CD3/CK19/ɑ-SMA (red) in liver sections from Ctrl and IR-treated mice (ST- and LT-RILD, n = 5/group). Up, representative images (Scale bars, 200×, 100 μm); Down, statistical analysis of the change of the cells co-stained with GSDMD-FL (five fields were observed per liver, n= 5roup). HNF4a: ****p = 2.063E-09, Ctrl vs ST; ****p = 7.967E-07, Ctrl vs LT; CD3: ****p = 3.230E-07, Ctrl vs ST; CK19: **p = 0.008, Ctrl vs ST; ****p = 2.740E-07, Ctrl vs LT; E GSDMD-N protein expressions from hepatocyte, monocyte-granulocyte, HSC, T cell and sinusoidal endothelial cell (SEC) were detected by immunoblot analysis. F Immunoblot analysis of GSDMD-FL and GSDMD-N expression from Gsdmd^FL/FL primary mouse hepatocytes exposed to 8 Gy irradiation at different times. G–I Gsdmd^FL/FL and Gsdmd^∆Hep hepatocytes 24 h after 8 Gy irradiation, cell morphology was observed under microscopy (black indicates bubble-like morphology) (G), and cell viability was assessed by LDH assay (H) and propidium iodide (PI) staining (I) (Each sample was from 3 biologically independent replicates for E-I). **p = 0.007, Gsdmd^FL/FL-Ctrl vs IR; *p = 0.02, Gsdmd^FL/FL-IR vs Gsdmd^∆Hep-IR; ****p = 1.177E-06, Gsdmd^FL/FL-Ctrl vs IR; ****p = 1.456E-06, Gsdmd^FL/FL-IR vs Gsdmd^∆Hep-IR; P values were calculated using one-way ANOVA with Tukey’s multiple comparison test (C, D, H) and two-way ANOVA following Fisher’s LSD test for multiple comparisions (I). Data are mean ± SEM. No data were excluded from the analyses. Source data are provided as a Source Data file.

Fig. 2. GSDMD deficiency ameliorates acute liver injury and chronic fibrosis in RILD.

A Representative high magnification images of Hematoxylin Eosin staining (HE) staining in Gsdmd^FL/FL and Gsdmd^∆Hep mice (n = 5/group) treated with or without one single 30 Gy hepatic radiation. (Scale bars, 200×, 200 μm, 400×, 50 μm). B-D Representative liver macroscopic image (B), ɑ-SMA staining (C) and sirius-red staining (D) in Gsdmd^FL/FL or Gsdmd^∆Hep mice in LT-RILD. Scale bars, 400×, 50 μm; 200×, 100 μm. Statistical analysis of ɑ-SMA and collagen area were present in the right, 5 fields were observed per mice (n = 5/group). Gsdmd^FL/FL-IR vs Gsdmd^∆Hep-IR: ɑ-SMA, ***p = 0.0001; collagen area, ***p = 0.0004. E Immunoblot analysis of liver protein expression of COL1A1, Fibronectin and ɑ-SMA from Gsdmd^FL/FL and Gsdmd^∆Hep mice (n = 1, no IR; n = 3, Gsdmd^FL/FL-IR; n = 5, Gsdmd^∆Hep-IR). F Key hepatic profibrogenic genes of Col1a1, Tgfb1 and Timp1 in Gsdmd^FL/FL and Gsdmd^∆Hep mice treated with IR (n = 3/group, Ctrl; n = 5/group, LT). Gsdmd^FL/FL-IR vs Gsdmd^∆Hep-IR: Col1a1, p = 0.0008; Tgfb1, p = 0.0003; Timp1, p = 0.0003. G Serum levels of liver function markers ALT, AST and ALB were determined in Ctrl-, ST- and LT-RILD groups (n = 3/group, Ctrl; n = 5/group, ST/LT). ST-Gsdmd^FL/FL vs Gsdmd^∆Hep: ALT, p = 2.087E-06; AST, p = 6.926E-05; ALB, p = 0.002. LT-Gsdmd^FL/FL vs Gsdmd^∆Hep: ALT, p = 1.857E-05; AST, p = 4.838E-05; ALB, p = 1.273E-05. H Supernatant levels of liver function markers ALT, AST and ALB were determined in hepatocytes with and without IR treatment (Each sample was from 3 biologically independent replicates). ALT, p = 1.156E-06; AST, p = 2.89E-07; ALB, p = 0.002. Data are expressed as mean ± SEM. p values were calculated using two-way ANOVA with Fisher’s LSD multiple comparisions test (C, D, F–H). No data were excluded from the analyses. Source data are provided as a Source Data file.

Fig. 3. GSDMD regulates the recruitment of neutrophils to the irradiated liver.

A, B Workflow and Proportion plot of immune cells. A Livers from Gsdmd^FL/FL -Ctrl, Gsdmd^FL/FL-ST and Gsdmd^∆Hep-ST group were used. 7 days after the last 30 Gy irradiation, the livers were digested and 10 × Genomics scRNA-seq was used to profile the cells; B UMAP plot of liver single CD45⁺ cells from three group mice, colored by the group. C Colors indicate the cell type annotated using Single R and manual annotation and validated by cell-type-specific gene expression. D Flow Cytometric analysis of liver-infiltrated neutrophils (FVD^-CD45⁺CD11b⁺Ly6G⁺) of Gsdmd^FL/FL and Gsdmd^∆Hep mice. Gsdmd^FL/FL vs Gsdmd^∆Hep: ST, p = 3.1E-09; LT, p = 0.002. E Neutrophil infiltrationwas analyzed by immunohistology staining (left, scale bars, 200×, 50 μm) with antibodies against MPO, and following morphometric quantification (right). D, E n = 3/group, Ctrl; n = 5/group, ST/LT. Gsdmd^FL/FL vs Gsdmd^∆Hep: ST, p = 1.218E-07; LT, p = 2.86E-07. F, G Detection of the neutrophil infiltration in liver of Gsdmd^∆Hep mice treated with or without AAV9-GSDMD-FL adeno-associated virus. F Representative plots (left) and statical analysis (right) of the percentage of CD11b⁺Ly6G⁺ cells with the flow cytometry. **p = 0.005. G MPO staining images (left) and quantification (right). ****P < 0.0001. F, G n = 4/group. H Quantification of neutrophil migration in response to Gsdmd^FL/FL and Gsdmd^∆Hep-hepatocytes pretreated with irradiation or not. Each sample was from 3 biologically independent replicates. ***P = 0.0003. I The depletion effect of neutrophils in liver were tested by flow cytometry in ST-RILD mice. ****P < 0.0001. J Representative liver HE images from ST-RILD mice (Scale bars, 200×, 50 μm). K Serum levels of ALT, AST, and ALB.****P < 0.0001. n = 5 mice/group for I–K. p values were using two-way ANOVA with Fisher’s LSD multiple comparisions test (D, E, H) and unpaired two-sided Student’s t test (F, G and I). No data were excluded from the analyses. Data are expressed as mean ± SEM. Source data are provided as a Source Data file.

Fig. 4. Selective CXCL1 neutralization alleviates the severity of RILD through hampering neutrophil recruitment.

A Volcano plot depiction of neutrophil chemokine genes between Gsdmd^FL/FL and Gsdmd^∆Hep mice (n = 5/group). B Detection of the hepatic mRNA level of Cxcl1 in Gsdmd^FL/FL and Gsdmd^∆Hep mice (n = 5/group) accepted or were free of 30 Gy or 6 Gy×5 IR. ***p = 0.001, ST: Gsdmd^FL/FL-Ctrl vs Gsdmd^∆Hep; *p = 0.01, LT: Gsdmd^FL/FL-Ctrl vs Gsdmd^∆Hep. C Representative immunohistochemistry images of CXCL1 in liver tissue of Ctrl and IR group from Gsdmd^FL/FL and Gsdmd^∆Hep mice (Scale bars, 40×, 200 μm; 400×, 50 μm, left). Statical analysis of CXCL1 expression were calculated from five fields per liver (n = 5/group, right). ****P < 0.0001. D Representative liver CXCL1 staining from ST-RILD mice treated with AAV9-control and AAV9-GSDMD-FL adeno virus (Scale bars, 40×, 200 μm, left). Statical analysis of CXCL1 expression were calculated from five fields per liver (n = 4/group, right). ****P < 0.0001. E Detection of the neutrophil infiltration in liver of mice treated with isotype IgG antibody and anti-CXCL1 antibody. The percentage of cells with CD11b⁺Ly6G⁺ with flow cytometry ⁽left) and its statical analysis were presented in both ST- and LT-RILD cohort (right; n = 3 mice/group, ST; n = 5 mice/group, LT). **p = 0.008, ST: isotype IgG vs anti-CXCL1 Ab; *p = 0.03, LT: isotype IgG vs anti-CXCL1 Ab. F Representative liver HE images from ST-RILD mice (Scale bars, 200×, 50 μm). G–I Representative liver macroscopic images (G), ɑ-SMA staining (H, p = 0.0002) and sirius-red staining (I, p = 0.0039) in cohort of LT-RILD mice. (Scale bars, 400×, 50 μm). Statistical analysis of ɑ-SMA and collagen area were present in the right, 5 fields were observed per mice. (F-I, n = 3/group). J Hepatic mRNA levels of Col1a1 (p = 0.0169), Tgfb1 (p = 0.044) and Timp1 (p = 0.0478) from LT-RILD mice. K Imunoblot analysis of COL1A1, Fibronectin and ɑ-SMA. n = 5/group for J-K. L Serum levels of ALT, AST, and ALB (n = 3/group, isotype-ST; n = 5/group, anti-CXCL1 Ab and isotype-LT). ST-isotype IgG vs anti-CXCL1 Ab: ALT, p = 8.27E-07; AST, p = 0.000219; ALB, p = 0.0068. LT-isotype IgG vs anti-CXCL1 Ab: ALT, p = 9.72E-07; AST, p = 1.52E-05; ALB, p = 0.0016. p values were using two-way ANOVA with Fisher’s LSD multiple comparisions test (C, L) and unpaired two-sided Student’s t test (B, D, E and H–J). Data are expressed as mean ± SEM. No data were excluded from the analyses. Source data are provided as a Source Data file.

Fig. 5. GSDMD regulates hepatocytes release CXCL1 to recruit neutrophil.

A Immunofluorescence visualization of CXCL1 (green), HNF4α (red), ɑ-SMA (red) in liver sections from Ctrl and IR-treated mice (n = 5/group). (Scale bars, 200×, 100 μm). B Double-immunostaining for CXCL1 and GSDMD-FL in primary hepatocytes 24 h after irradiation. Confocal analysis demonstrates coexpression of CXCL1 and GSDMD-FL in hepatocytes receiving radiation. Scale bars, 200×, 50 μm (n = 5/group). C Cell viability was assessed by Annexin-V⁺/PI⁺. D ELISA assay of CXCL1 release in culture supernatants from Gsdmd^FL/FL and Gsdmd^∆Hep hepatocyte accepted irradiation or not. IR: Gsdmd^FL/FL vs Gsdmd^∆Hep, p = 0.0004. C, D each sample was from 3 biologically independent replicates. E Immunoblot analysis of caspase-1, cleaved-caspase-1, GSDMD-FL/N and CXCL1 expression from Gsdmd^FL/FL primary mouse hepatocytes treated or untreated with 20 μM Ac-YVAD-cmk, analyzed at 12 h following 8 Gy irradiation (3 biologically independent replicates). F ELISA assay of CXCL1 release in culture supernatants from hepatocytes pretreated without or with 20 μM Ac-YVAD-cmk. IR: Vehicle vs Ac-YVAD-cmk, p = 1.42E-07. G, H Quantification of neutrophil migration. Quantification of neutrophil migration in response to irradiated Gsdmd^FL/FL hepatocytes pretreated with anti-CXCL1 antibody or not (G), p = 0.003; Quantification of neutrophil migration in response to irradiated Gsdmd^∆Hep hepatocytes pretreated with CXCL1 RP or not (H), p = 0.002. F–H, n = 3 biologically independent mice per group. P values were using two-way ANOVA with Fisher’s LSD multiple comparisions test (D, F) and unpaired two-sided Student’s t test (C and G, H). No data were excluded from the analyses. Data are expressed as mean ± SEM. Source data are provided as a Source Data file.

Fig. 6. GSDMD deficiency inhibits the activation of the stat5a/CXCL1/neutrophil pathway.

A Identification of promoters and enhancers of CXCL1 with integrative analysis of transcriptomic sequencing and database within the GeneCards Suite (http://wwww.genecards.org/). B Heatmap of expression of the indicated genes from (A) (n = 5/group). C, D Primary hepatocytes were pretreated with or without 25 μM STAT5A inhibitor IST5-002 for 6 h prior to 8 Gy irradiation. After 24 h, the translocation inhibited ability were determined by the WB assay (C); the culture medium level of CXCL1 was assayed by ELISA (D). Vehicle vs IST5: Ctrl, p = 0.008; IR, p = 4.98E-06. E Luciferase activity assay was performed to measure the binding of STAT5A at the Gsdmd^FL/FL or the promoter region with mutations (Mut) of Cxcl1 gene with or without IR in HEK 293 T cell. WT-Ctrl vs WT-IR: p = 6.66E-13. F Quantification of neutrophil migration in response to irradiated hepatocytes pretreated with IST5-002 or not. Vehicle vs IST5: Ctrl, p = 0.003; IR, p = 1.71E-05. G Immunoblot analysis of STAT5A and P-STAT5A from Gsdmd^FL/FL and Gsdmd^∆Hep primary hepatocyte exposed to 8 Gy irradiation. H Protein level of STAT5A and P-STAT5A in cytoplasmic extracts and nuclear extracts from 8 Gy irradiated Gsdmd^FL/FL and Gsdmd^∆Hep primary hepatocytes were determined by Immunoblot. I Co-IP analysis for physical interaction of GSDMD and STAT5A in hepatocytes. J ChIP assays were performed to detect the promoter sites of STAT5A binding to Cxcl1. IgG as a negative control for ChIP assay. Gsdmd^FL/FL: IgG vs stat5a, p = 1.12E-08; stat5a: Gsdmd^FL/FL vs Gsdmd^∆Hep, p = 1.75E-08 C-J, n = 3 biologically independent mice per group. K, L Immunoblot analysis of STAT5A and P-STAT5A from Gsdmd^FL/FL and Gsdmd^∆Hep mice (n = 1, no IR; n = 5, IR). M Immunoblot analysis of STAT5A and P-STAT5A of IR-Gsdmd^∆Hep mice administered with AAV9-Control and AAV9-GSDMD-FL (n = 4/group). p values were using two-way ANOVA with Fisher’s LSD multiple comparisions test (D-F and J). No data were excluded from the analyses. Data are expressed as mean ± SEM. Source data are provided as a Source Data file.

Fig. 7. The therapeutic potential of the GSDMD inhibitor disulfiram in preventing RILD.

A Hepatic protein levels of GSDMD-FL, GSDMD-N, STAT5A and P-STAT5A between DSF and vehicle groups were evaluated by Immunoblot. B Representative staining of CXCL1 in liver tissue from vehicle and DSF-treated mice (Scale bars, 40×, 200 μm). n = 4/group, Vehicle-ST; n = 3/group, Vehicle-LT; n = 5/group, DSF. ST-Vehicle vs DSF: p = 1.64E-07; LT-Vehicle vs DSF: p = 4.38E-07. C, D The neutrophil infiltration was evaluated by flow cytometry (C) and MPO immunostaining (D) in vehicle and DSF-treated mice (Scale bars, 200×, 50 μm). (n = 4 mice/group, vehicle; n = 3 mice/group, ST-vehicle; n = 5 mice/group, DSF-LT). CD11b ⁺ : ST-Vehicle vs DSF: p = 2.77E-07; LT-Vehicle vs DSF: p = 0.002. E Representative images of HE staining in vehicle or DSF-treated ST-RILD mice. Scale bars, 100×, 50 μm. F–H Representative liver macroscopic images (F), ɑ-SMA staining (G) and sirius-red staining (H) in cohort of LT-RILD mice (Scale bars, 200×, 50 μm). Two-sided and Student’s t test. G-H, n = 3 mice/group, vehicle; n = 5 mice/group, DSF. I Immunoblot analysis of COL1A1, Fibronectin and ɑ-SMA. J Serum levels of ALT, AST, and ALB. n = 5 mice/group, vehicle; n = 6 mice/group, DSF. ST-Vehicle vs DSF: ALT, p = 2.49E-08; AST, p = 2.49E-08; ALB, p = 0.0004. LT-Gsdmd^FL/FL vs Gsdmd^∆Hep: ALT, p = 7.76E-05; AST, p = 7.76E-05; ALB, p = 1.44E-05. K Immunoblot analysis of GSDMD-FL/N, STAT5A and P-STAT5A. L-M Supernatants from vehicle or disulfiram pre-treated primary hepatocyte were collected for LDH (L **** p = 3.6E-05) and CXCL1assay (M, **** p = 6.35E-05). n = 3 biologically independent mice per group for (K–M). p values were using two-way ANOVA with Fisher’s LSD multiple comparisions test (B, C, J, L, M) and unpaired two-sided Student’s t test (D and G-H). No data were excluded from the analyses. Data are expressed as mean ± SEM. Source data are provided as a Source Data file.