DDR1-induced neutrophil extracellular traps drive pancreatic cancer metastasis

Abstract

Pancreatic ductal adenocarcinoma (PDAC) tumors are characterized by a desmoplastic reaction resulting in dense deposition of collagen that is known to promote cancer progression. A central mediator of protumorigenic collagen signaling is the receptor tyrosine kinase discoid domain receptor 1 (DDR1). DDR1 is a critical driver of a mesenchymal and invasive cancer cell PDAC phenotype. Previous studies have demonstrated that genetic or pharmacologic inhibition of DDR1 reduces PDAC tumorigenesis and metastasis. Here, we investigated whether DDR1 signaling has cancer cell nonautonomous effects that promote PDAC progression and metastasis. We demonstrate that collagen-induced DDR1 activation in cancer cells is a major stimulus for CXCL5 production, resulting in the recruitment of tumor-associated neutrophils (TANs), the formation of neutrophil extracellular traps (NETs), and subsequent cancer cell invasion and metastasis. Moreover, we have identified that collagen-induced CXCL5 production was mediated by a DDR1/PKCθ/SYK/NF-κB signaling cascade. Together, these results highlight the critical contribution of the collagen I-DDR1 interaction in the formation of an immune microenvironment that promotes PDAC metastasis.

Keywords: Cytokines; Neutrophils; Oncology; Signal transduction.

Figure 1. DDR1 induces liver metastasis in pancreatic cancer.

(A) DDR1 and DDR2 expression were analyzed by western blotting in 2 fibroblasts, 14 primary PDAC cell lines, and 2 metastatic PDAC cell lines, in 3 independent experiments. (B) DDR1 was observed at PDX tumors derived from metastatic or primary human PDAC tumors by IHC staining using anti–human DDR1 antibody and identified using PE Vectra3. Scale bar: 50 μm. The H-score of DDR1 quantification was displayed as DBA signals by inForm software. n = 10, unpaired 2-tailed Student’s t test. **P < 0.01. (C) Cell invasion assay in MDA-PATC 148 cells with knockdown or reexpression DDR1 were used by Matrigel transwell chamber. The invading cells in each chamber were counted under a fluorescence microscope after cultured 18 hours, and the average number of cells was calculated based on the number of cells found in 6 fields per chamber. Data are mean ± SD. n = 5, 3 independent experiments; 1-way ANOVA with Sidak post hoc testing. *P < 0.05; ***P < 0.001. (D–F) Mice were orthotopically injected with MDA-PATC 148 (control, DDR1–deficient or DDR1-reexpression clones) cells for 9 weeks. (D) H&E staining of pancreas and liver section. Arrow: region of tumor. n = 12. Scale bar: 50 μm. (E) Tumor size measurement in pancreas. Unpaired 2-tailed Student’s t test. (F) The numbers of liver-met. n = 12; Fisher’s exact test. *P < 0.05.

Figure 2. DDR1 induces CXCL5 production in pancreatic cancer cells.

(A) Chemokine array analysis in cell lysate and supernatant of MDA-PATC 148 cells with knockdown DDR1. (B and C) MDA-PATC 148 cells with knockdown or reexpressed DDR1 were treated with collagen I for 3 hours. (B) CXCL5 mRNA level by using real-time PCR. (C) CXCL5 protein level by using ELISA. (D) CXCL5 expression in overexpressed DDR1 in 5 pancreatic cancer cell lines. Upper: DDR1 levels were checked by western; middle: CXCL5 mRNA level were detected by real-time PCR; lower: CXCL5 protein levels were analyzed by ELISA. (E–G) Mice were orthotopically injected with MDA-PATC 148 (control, DDR1-deficient or DDR1-reexpression clones) cells for 9 weeks. (E) IHC staining with anti-DDR1 (upper panel) and anti-CXCL5 (bottom panel) antibodies in pancreas. (F) ELISA showed CXCL5 level in plasma harvest from mice. (G) FACS by using anti-CD11b and anti-Ly6G antibodies to determine the presence of CD11b⁺Ly6G⁺ neutrophils infiltration in pancreas. (B–D) Data are mean ± SD. n = 3–4, 3 independent experiments; (B and C) 1-way ANOVA with Sidak post hoc testing; (D) Unpaired 2-tailed Student’s t test. *P < 0.05; **P < 0.01. (F and G) n = 5–10 mice, data performed in triplicate; 1-way ANOVA with Sidak post hoc testing. **P < 0.01; ***P < 0.001. Data show signal after membrane exposed to x ray film for 2 minutes.

Figure 3. The correlation of DDR1, CXCL5, and neutrophils infiltration at tissue microarray (TMA) in PDX tumors.

(A) IHC staining showed DDR1, CXCL5, and Ly6G⁺ neutrophils infiltration at PDX tumors and identified using PE Vectra3. Scale bar: 50 μm. (B–D) Pearson’s correlation showed relationship of DDR1, CXCL5, and Ly6G by using H-score, which quantified the DBA signals by inForm software. n = 82.

Figure 4. Metastatic tumors recruit more Ly6G⁺ neutrophils infiltration than primary tumors.

(A) Ly6G⁺ neutrophils were observed at tumors derived from primary and match liver-met cell lines by immunofluorescence staining using anti-Ly6G (green) and DAPI (blue) with a fluorescence microscope. Scale bar: 50 μm. The number of neutrophils were counted in x20 field, 6 fields per slice. Data are mean ± SD. n = 5 mice, unpaired 2-tailed Student’s t test. *P < 0.05. (B) Ly6G⁺ neutrophils were observed at PDX tumors derived from metastatic or primary human PDAC tumors by IHC staining using anti-Ly6G antibody and identified using PE Vectra3. Scale bar: 50 μm. The H-score of Ly6G quantification was displayed as DBA signals in x20 field, 6 fields per slice by inForm software. Data are mean ± SD. n = 10, unpaired 2-tailed Student’s t test. *P < 0.05.

Figure 5. DDR1-positive pancreatic cancer cells mediated NET formation from neutrophils and enhanced cancer cell invasion.

(A–D) Human neutrophils were cocultured with DDR1 knockdown or reexpression of MDA-PATC 148 or BxPC-3 by Matrigel transwell chamber, with or without NADPH oxidase inhibitor, PDA4 inhibitor, NE inhibitor, and Dase I treatment for 18 hours. (A) NET structures were analyzed by immunofluorescence staining using DAPI (blue), anti-NE (red), and anti-histone H3 (green) mAbs. Scale bar: 50 μm. (B) The NET quantification is displayed as NET histone area (μm²) per x40 field, 6 fields per group. (C and D) The number of invaded cells analyzed by immunofluorescence staining using DAPI and calculated based on the number of cells found in 6 fields per chamber. All the data are mean ± SD. n = 5, 3 independent experiments; 1-way ANOVA with Sidak post hoc testing. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 6. CXCL5 involved in DDR1-mediated NET formation and cancer cell invasion.

(A–F) Human neutrophils were cocultured with DDR1 knockdown or reexpression of MDA-PATC 148 or BxPC-3 by Matrigel transwell chamber, with or without anti-CXCL5 neutralized antibody or recombinant CXCL5 treatment, for 18 hours. (A) NET structures were analyzed by immunofluorescence staining using DAPI (blue), anti-NE (red), and anti–histone H3 (green) mAbs. Scale bar: 50 μm. (B and C) The NET quantification is displayed as NET histone area (μm²) per field, 6 fields per group. (D) Cit-histone H3 expression were analyzed by western blotting. (E and F) The number of invaded cells were analyzed by immunofluorescence staining using DAPI and calculated based on the number of cells found in 6 fields per chamber. All the data are mean ± SD. n = 5, 3 independent experiments. (B and E) P values were analyzed by 1-way ANOVA with Sidak post hoc testing. ***P < 0.001. (C and F) P values were analyzed by unpaired 2-tailed Student’s t test. *P < 0.05; ***P < 0.001.

Figure 7. DDR1-positive pancreatic cancer cells mediated NET formation from neutrophils through a soluble factor secretion and enhanced cancer cell invasion.

(A–G) Human neutrophils were cultured with CCM from cancer cells for 18 hours. (A) NET structures were analyzed by immunofluorescence staining using DAPI (blue), anti-NE (red), and anti–histone H3 (green) mAbs. Scale bar: 50 μm. (B) The NET quantification is displayed as NET histone area (μm²) per field, 6 fields per group. (C) Cit-histone H3 expression were analyzed by western blotting. (D and E) The number of invaded cells analyzed by immunofluorescence staining using DAPI and calculated based on the number of cells found in 6 fields per chamber. (F) Neutrophils Elastase activity were showed in human neutrophils with CCM treatment for 18 hours. (G) The number of invaded cells analyzed by immunofluorescence staining using DAPI and calculated based on the number of cells found in 6 fields per chamber. All the data are mean ± SD. n = 4–5, 3 independent experiments; 1-way ANOVA with Sidak post hoc testing. **P < 0.01; ***P < 0.001.

Figure 8. The correlation of DDR1, CXCL5, and NET-like structure in samples of patient with PADC.

(A) Upper and middle panel: IHC staining showed DDR1, CXCL5 expression in PDAC patient samples, and identified using PE Vectra3. Scale bar: 3 mm and 50 μm. Bottom panel: NET-like structures were analyzed by immunofluorescence staining using DAPI (blue), anti-CK19 (white), anti-MPO (green), and anti–citrullinated histone H3 (red) mAbs in samples of patient with PDAC. Scale bar: 20 μm. (B) Pearson’s correlation showed relationship of DDR1 and CXCL5 by using H-score, which quantified the DBA signals by inForm software, P < 0.0001.

Figure 9. PKCθ/SYK/NF-κB pathway involved in DDR1-induced CXCL5 production, NET formation from neutrophils, and enhanced cancer cell invasion.

(A) qPCR results were used to quantify enrichment of NF-κB P65 at the CXCL5 promoter using ChIP assay in MDA-PATC 148 cells with DDR1 knockdown. Data are mean ± SD. n = 3, 3 independent experiments; 1-way ANOVA with Sidak post hoc testing. ***P < 0.001. (B) Phospho-NF-κB P65, phospho-PKCθ, and phospho-SYK were analyzed by western blotting in MDA-PATC 148 cells with DDR1 knockdown. (C) Phospho-NF-κB P65, phospho-PKCθ, and phospho-SYK were analyzed by western blotting in MDA-PATC 148 cells with or without SYK inhibitor and PKC inhibitor pretreatment. (D and E) NET structures were analyzed by immunofluorescence staining using DAPI (blue), anti-NE (red), and anti–histone H3 (green) mAbs. (D) In MDA-PATC 148 cells with CCM from MDA-PATC 148 with IκB super-repressor mutation/collagen I, treatment for 18 hours. (E) In MDA-PATC 148 cells with MDA-PATC 148, with or without SYK inhibitor and PKC inhibitor pretreatment/collagen I, treatment for 18 hours. Scale bar: 50 μm. The NET quantification is displayed as NET histone area (μm²) per field, 6 fields per group. (F and G) The number of invaded cells were analyzed by immunofluorescence staining using DAPI and calculated based on the number of cells found in 6 fields per chamber. (F) In MDA-PATC 148 cells with NCCM from MDA-PATC 148 cells with IκB super-repressor mutation/neutrophils/collagen I, treatment for 18 hours. (G) In MDA-PATC 148 cells with NCCM from MDA-PATC 148/collagen I/SYK or PKC inhibitor, treatment for 18 hours. (D–G) Data are mean ± SD. n = 5–6, 3 independent experiments; 1-way ANOVA with Sidak post hoc testing. *P < 0.05;**P < 0.01; ***P < 0.001.

Figure 10. 7rh treatment reduced NET formation through inhibition of the DDR1/PKCθ/SYK/CXCL5 axis and reduced cancer metastasis.

(A and B) MDA-PATC 148 cells were pretreated with 7rh for 30 minutes and then with collagen I for 3 hours. (A) Phospho-NF-κB P65, phospho-PKCθ, and phospho-SYK were analyzed by western blotting. (B) CXCL5 levels were analyzed by ELISA. Data are mean ± SD. n = 4, 3 independent experiments; 1-way ANOVA with Sidak post hoc testing. *P < 0.05; ***P < 0.001. (C–E) Human neutrophils were cocultured with MDA-PATC 148 and BxPC-3 cells by Matrigel transwell chamber for 18 hours. (C) NET structures were analyzed by immunofluorescence staining using DAPI (blue), anti-NE (red), and anti–histone H3 (green) mAbs. Scale bar: 50 μm. (D) The NET quantification is displayed as NET histone area (μm²) per field, 6 fields per group. Data are mean ± SD. n = 6, 3 independent experiments; 1-way ANOVA with Sidak post hoc testing. ***P < 0.001. (E) Cit-histone H3 expression were analyzed by western blotting. (F and G) Mice were orthotopically injected with MDA-PATC 148 cells, with or without 3 mg/kg 7rh treatment for 9 weeks. (F) Liver metastasis was detected by immunofluorescence staining using DAPI (blue) and anti-CK19 (red) mAbs in liver section. Scale bar: 50 μm. The metastasis quantification is displayed as CK-19 positive signals/per x20 field, 6 fields per group. (G) Neutrophils infiltration was detected by immunofluorescence staining using DAPI (blue), anti-CK19 (red), and anti-Ly6G (green) mAbs in pancreas section. Scale bar: 50 μm. The Neutrophils infiltration quantification is displayed as Ly6G positive signals per x20 field, 6 fields per group. Data are mean ± SD. n = 5, unpaired 2-tailed Student’s t test. ***P < 0.001.