Activation of the GPR35 pathway drives angiogenesis in the tumour microenvironment

Abstract

Objective: Primary sclerosing cholangitis (PSC) is in 70% of cases associated with inflammatory bowel disease. The hypermorphic T108M variant of the orphan G protein-coupled receptor GPR35 increases risk for PSC and ulcerative colitis (UC), conditions strongly predisposing for inflammation-associated liver and colon cancer. Lack of GPR35 reduces tumour numbers in mouse models of spontaneous and colitis associated cancer. The tumour microenvironment substantially determines tumour growth, and tumour-associated macrophages are crucial for neovascularisation. We aim to understand the role of the GPR35 pathway in the tumour microenvironment of spontaneous and colitis-associated colon cancers.

Design: Mice lacking GPR35 on their macrophages underwent models of spontaneous colon cancer or colitis-associated cancer. The role of tumour-associated macrophages was then assessed in biochemical and functional assays.

Results: Here, we show that GPR35 on macrophages is a potent amplifier of tumour growth by stimulating neoangiogenesis and tumour tissue remodelling. Deletion of Gpr35 in macrophages profoundly reduces tumour growth in inflammation-associated and spontaneous tumour models caused by mutant tumour suppressor adenomatous polyposis coli. Neoangiogenesis and matrix metalloproteinase activity is promoted by GPR35 via Na/K-ATPase-dependent ion pumping and Src activation, and is selectively inhibited by a GPR35-specific pepducin. Supernatants from human inducible-pluripotent-stem-cell derived macrophages carrying the UC and PSC risk variant stimulate tube formation by enhancing the release of angiogenic factors.

Conclusions: Activation of the GPR35 pathway promotes tumour growth via two separate routes, by directly augmenting proliferation in epithelial cells that express the receptor, and by coordinating macrophages' ability to create a tumour-permissive environment.

Keywords: angiogenesis; colorectal cancer; primary sclerosing cholangitis; receptor characterisation; ulcerative colitis.

Figure 1

GPR35 increases tumour size and release of proangiogenic cytokines. (A) The area of intestinal adenomas from Apc ^min mice either wild-type (Gpr35^+/+ ) and globally deficient for GPR35 (Gpr35^-/- ) (left panel) or from Gpr35^+/+ and Gpr35^-/- mice exposed to AOM and three cycles of DSS (right panel) were compared. N=7 tumours per genotype. (B) CXCL-1, VEGF and IL-1β levels measured in AOM/DSS tumour tissue of Gpr35^+/+ and Gpr35^-/- mice. N=6 tumours per genotype. (C) Mice conditionally deficient for macrophage GPR35 (Gpr35 ^ΔMΦ) were crossed with APC^min mice and tumours were counted when the mice reached an age of 16 weeks. N=14 mice per genotype. (D) Gpr35 ^ΔMΦ mice were exposed to AOM and DSS. Colon adenomas were counted in Gpr35 ^ΔMΦ mice and in their control Gpr35^fl/fl . N=11 to 13 mice per genotype. (E) Representative images showing H&E stain of tumours from Gpr35^fl/fl and Gpr35 ^ΔMΦ mice. (F) The area of intestinal adenomas from either Gpr35^fl/fl ; APC ^min and Gpr35 ^ΔMΦ; Apc ^min mice (left panel) or from Gpr35^fl/fl and Gpr35 ^ΔMΦ mice exposed to AOM/DSS (right panel) were compared. N=9 tumours per genotype. (G–I) CXCL-1 (G), VEGF (H) and IL-1β (I) levels measured in AOM/DSS tumour tissue of Gpr35^fl/fl and Gpr35 ^ΔMΦ mice. N=6 tumours per genotype. All data represented as mean±SEM. Statistical significance was calculated using Mann-Whitney U after Kruskal-Wallis testing. * p<0.05, ** p<0.01. AOM, azoxymethane; APC, Adenomatosis Polyposis Coli; DSS, dextran sodium sulphate; IL-1β, interleukin 1β; n.s., not significant.

Figure 2

GPR35 deletion in myeloid cells reduces angiogenic potential. (A) Adenomas from Gpr35^fl/fl and Gpr35 ^ΔMΦ mice exposed to AOM/DSS stained for CD31⁺ endothelial cells and CD206⁺ M2 macrophages. N=5 each genotype, representative confocal microscopy. Scale bars 100 µm. (B) FACS analyses of CD31⁺, CD68⁺, CD163⁺, CD206⁺ and Ly-G6⁺ cells in Gpr35^fl/fl and Gpr35 ^ΔMΦ tumour tissue. N=8–12 Gpr35^fl/fl tumours from 8 to 12 mice and N=7 Gpr35 ^ΔMΦ tumours from mice. (C) VEGF levels in M0, M1 and M2 BMDM from Gpr35^+/+ and Gpr35^-/- mice. N=6 per genotype. (D) CXCL-1 levels in M0, M1 and M2 BMDM from Gpr35^+/+ and Gpr35^-/- mice. N=6 per genotype. (E) VEGF levels in supernatants of M0, M1 and M2 human iPS cell-derived macrophages. N=6 each genotype. (F) CXCL-8 levels in supernatants of M0, M1 and M2 human iPS cell-derived macrophages. N=6 each genotype. * p<0.05, ** p<0.01. AOM/DSS, azoxymethane followed by dextran sodium sulphate; iPS, inducible-pluripotent-stem; n.s., not significant; WT, wild type.

Figure 3

GPR35 deletion in macrophages and neutrophils reduces tube formation. (A) Tube formation assay with murine SVEC endothelial cells incubated with Gpr35^+/+ and Gpr35^-/- M2 macrophage supernatants. Branch length, number of junctions and total mesh area were determined using ImageJ’s angiogenesis tool. N=18. (B) 2 H-11 control tube formation assays with control RPMI/10%FBS and RPMI/10%FBS containing 30 ng/mL VEGF. No macrophage supernatants present. Branch length, number of junctions and total mesh area were determined using the Image J’s angiogenesis tool. N=14 C. 2 H-11 tube formation assays with supernatants of wild-type (WT) and knock-out macrophages in the presence of blocking anti-VEGF antibody. Branch length, number of junctions and total mesh area were determined using Image J’s angiogenesis tool. N=14. (D) HMVEC tube formation assays with human microvascular endothelial cells. Endothelial cells were incubated with supernatants of iPS cell-derived macrophage supernatants (WT and T108M risk variant). Branch length, number of junctions and total mesh area were determined using Image J’s angiogenesis tool. N=10. (E) VEGF receptor 2 phosphorylation of 2 H-11 cells treated with WT or Gpr35^–/– M2 supernatants. All data represented as mean±SEM. Statistical significance was calculated using Mann-Whitney U after Kruskal-Wallis testing. HMVEC, human microvascular endothelial cell; iPS, inducible-pluripotent-stem; n.s., not significant.

Figure 4

Macrophage GPR35 controls vascular sprouting by interacting with the Na/K-ATPase. (A) Aortic rings from Gpr35 ^+/+ or Gpr35 ^–/– mice embedded in collagen matrix. Number of sprouts counted daily until day 9 of the experiment. Aortic rings were either embedded in media containing 30 ng/mL of VEGF (full circles) or left in OptiMEM media without growth factors (open circles) N=9 mice per genotype. (B) Vascular adventitia was removed from aortic rings from Gpr35 ^+/+ or Gpr35 ^–/–. Number of sprouts on day 9. N=6 mice per genotype. (C) Aortic rings from wildtype animals (Gpr35^fl/fl ) or mice lacking GPR35 conditionally on their LysM⁺ cellsGpr35 ^ΔMΦ. N=13 for Gpr35^fl/fl and 23 for Gpr35^ΔMΦ mice. (D) KYNA (100 µM) was added to the aortic rings of Gpr35 ^+/+ or Gpr35 ^–/– mice and vascular sprouts counted on day 9. N=5 mice for each genotype. (E) CXCL-17 (20 ng/mL) was added to the aortic rings of Gpr35 ^+/+ or Gpr35 ^–/– mice and vascular sprouts counted on day 9. N=6 mice for each genotype. (F) Anti-VEGF antibody (100 ng/mL) was added to the aortic rings of Gpr35 ^+/+ or Gpr35 ^–/– mice and vascular sprouts counted on day 9. n=6 for each genotype. (G) Blocking CXCR2 pepducin x1/2pal-i3 (3 µM) was added to the aortic rings of Gpr35 ^+/+ or Gpr35 ^–/– mice and vascular sprouts counted on day 9. n=6 for each genotype. (H, I.) ouabain (100 µM) was added to the aortic rings of Gpr35 ^+/+ or Gpr35 ^–/– mice (H) and to the aortic rings of Gpr35^fl/fl and Gpr35 ^ΔMΦ mice (I) and vascular sprouts counted on day 9. N=6 for each genotype. (J, K) pNaKtide (1 µM) was added to the aortic rings of Gpr35 ^+/+ or Gpr35 ^–/– mice (J) and to the aortic rings of Gpr35^fl/fl and Gpr35 ^ΔMΦ mice (K) and vascular sprouts counted on day 9. N=9 for each genotype. All data represented as mean±SEM. Statistical significance was calculated using Mann-Whitney U after Kruskal-Wallis testing.* p<0.05, ** p<0.01. n.s., not significant.

Figure 5

Lack of GPR35 results in reduced MMP levels. (A) MMP levels in Gpr35 ^+/+ and Gpr35 ^–/– adenomas. Tissue extract MMPs were activated with APMA. Area under the curve was analysed using image J. N=6 tumours per genotype. (B) MMP levels in M1 and M2 BMDM supernatants from Gpr35 ^+/+ and Gpr35 ^–/– mice. Supernatants were APMA-activated. area under the curve was analysed using image J N=9 mice per genotype. (C) MMP levels in M2 macrophages pretreated with ouabain (100 µM) or pNaKtide (1 µM). Area under the curve was analysed using image J N=6 mice per genotype. (D) Total (inactive zymogen and active enzyme) MMP2 and MMP9 levels in tumour tissue from Gpr35^fl/fl and Gpr35 ^ΔMΦ mice. (E) MMP2 and MMP9 levels from M2 murine BMDM supernatants. (F) Total (inactive zymogen and active enzyme) MMP2 levels from supernatants of human iPS cell-derived macrophages. Macrophages were polarised to M0, M1 and M2 macrophages. All data represented as mean±SEM. Statistical significance was calculated using Mann-Whitney U after Kruskal-Wallis testing. * p<0.05, ** p<0.01/ APMA, aminophenylmercuric acetate; MMP, matrix-metallo-proteinase; n.s., not significant; WT, wild type.

Figure 6

Lack of GPR35 decreases endothelial transmigration. (A) Right panel: Gpr35 ^+/+ and Gpr35 ^–/– monocytes transendothelial migration towards conditioned media from Gpr35 ^+/+ and Gpr35 ^–/– M2 macrophages in the lower compartments of migration chambers. Untreated endothelial cell monolayer. Left panel: adhesion of Gpr35 ^+/+ and Gpr35 ^–/– monocytes to 2 H-11 murine endothelial cells. Untreated endothelial cell monolayer. N=6 mice per genotype. (B) Left panel: transendothelial migration of monocytes through monolayers of 2 H-11 murine endothelial cells exposed to conditioned media from Gpr35 ^+/+ and Gpr35 ^–/– M2 macrophages. right panel: adhesion of Gpr35 ^+/+ and Gpr35 ^–/– monocytes to 2 H-11 murine endothelial cells treated with conditioned media from Gpr35 ^+/+ and Gpr35 ^–/– M2 macrophages. N=6 mice per genotype. (C) Upper panel: VCAM-1 expression in 2 H-11 cells treated with conditioned media from Gpr35 ^+/+ and Gpr35 ^–/– M2 macrophages pretreated with control RPMI, ouabain (100 µM) or pNaKtide (1 µM). Lower panel: VCAM-1 expression in Gpr35^fl/fl and Gpr35 ^ΔMΦ AOM/DSS tumours. (D) ERK1/2 phosphorylation of 2 H-11 cells treated with conditioned media from Gpr35 ^+/+ and Gpr35 ^–/– M2 macrophages pretreated with control RPMI, ouabain (100 µM) or pNaKtide (1 µM). (E) Src and ERK1/2 phosphorylation in Gpr35 ^+/+ and Gpr35 ^–/– AOM/DSS tumours. N=3 tumours per genotype. (F) Numbers of macroscopic and microscopic tumours of either g35i2 or vehicle treated mice with AOM/DSS induced tumours. N=8–9 mice/group. (G) AOM/DSS tumour tissue from g35i2 or vehicle treated mice probed for Src and ERK1/2 phosphorylation. N=3 tumours per genotype. (H) CXCL-1, VEGF, MMP2, MMP9 and inflammatory infiltrate in tumour tissue from g35i2 or vehicle-treated wild-type mice. N=6 per genotype for CXCL-1, VEGF, MMP2 and MMP9 ELISA assays, and n=8 for histological assessment of inflammatory cell infiltrate. All data represented as mean±SEM. Statistical significance was calculated using Mann-Whitney U after Kruskal-Wallis testing. AOM/DSS, azoxymethane followed by dextran sodium sulphate; n.s., not significant.