CD47 regulates collagen I-induced cyclooxygenase-2 expression and intestinal epithelial cell migration

Abstract

Increased epithelial cell expression of the cyclooxygenase-2 (COX-2) enzyme is a characteristic event of both inflammatory bowel disease and colon cancer. We here report the novel findings that collagen I-induced de novo synthesis of COX-2 in intestinal epithelial cells is inhibited by pertussis toxin (PTX) and by an inhibitory peptide selective for the heterotrimeric G alpha(i3)-protein. These findings could be explained by a regulatory involvement of the G-protein-dependent integrin-associated protein CD47. In support of this notion, we observed a collagen I-induced association between CD47 and alpha2 integrins. This association was reduced by a blocking anti-CD47 antibody but not by PTX or a control anti-beta2 antibody. Furthermore, a blocking antibody against CD47, dominant negative CD47 or specific siRNA knock down of CD47, significantly reduced collagen I-induced COX-2 expression. COX-2 has previously been shown to regulate intestinal epithelial cell adhesion and migration. Morphological analysis of intestinal cells adhering to collagen I revealed a co-localisation of CD47 and alpha2 integrins to non-apoptotic membrane blebs enriched in Rho A and F-actin. The blocking CD47 antibody, PTX and a selective COX-2 inhibitor, dramatically inhibited the formation of these blebs. In accordance, migration of these cells on a collagen I-coated surface or through a collagen I gel were significantly reduced by the CD47 blocking antibody, siRNA knock down of CD47 and the COX-2 inhibitor NS-398. In conclusion, we present novel data that identifies the G-protein-dependent CD47 protein as a key regulator of collagen I-induced COX-2 expression and a promoter of intestinal epithelial cell migration.

Figure 1. Collagen I induces COX-2 expression through a PTX dependent G-protein.

(A) Int 407 cells were incubated with or without 500 ng/ml PTX for 2 hours, after which they were plated out onto 10 µg/ml collagen I or 6% BSA (control) coated dishes for 1 hour. Adherent cells were then lysed and analysed for COX-2 expression by Western blotting as previously described. (B) Cells were transiently transfected with empty vector or vectors expressing small inhibitor peptides against either Gα_i1-2 or Gα_i3 before plated onto 10 µg/ml collagen I coated dishes for 1 hour. Adherent cells were the lysed and analysed for COX-2 expression by Western blotting. All membranes were re-probed for actin to ensure equal loading. The accumulated data of the densitometric analyses are given as percent of control and represent means ± SE of four separate experiments. The statistical analyses were performed with unpaired Students t-test; *P<0.05, **P<0.01 relative to the control and compared to collagen I treatment.

Figure 2. Expression and association of CD47 with α2 integrin upon collagen I stimulation.

(A) Whole cell lysates from the intestinal epithelial cell line, Int 407 and platelets, were analysed by Western blotting for their expression of CD47, and re-probed for GAPDH to ensure equal loading. (B) Int 407 cells were pre-incubation with either 500 ng/ml PTX for 2 hours or with either anti-CD47 (B6H12), anti-β2 integrin control antibody or pre-immune control IgG for 20 minutes before being plated onto 10 µg/ml collagen I or 6% BSA coated dishes. The indicated immunoprecipitations, with either an anti-CD47 antibody or control IgG, were performed as outlined in the Methods. Samples were analysed by Western blotting for the presence of CD47 and the α2 integrin. The accumulated data of the densitometric analyses are given as percent of control and represent means ± SE of three separate experiments. The statistical analyses were performed with unpaired Students t-test; *P<0.05, **P<0.01 relative to the control.

Figure 3. Collagen I mediated COX-2 expression and cell adhesion requires CD47.

(A) Int 407 cells were pre-treated with 20 µg/ml of the anti-CD47 functional blocking antibody (B6H12) or control IgG for 20 minutes at 4°C. Whereupon they were plated onto 10 µg/ml collagen I or 6% BSA coated dishes for 1 hour. Adherent cells were lysed and analysed for COX-2 expression by Western blotting as previously described. All membranes were re-probed for actin to ensure equal loading. (B) Cells were transfected with a dominant negative form of CD47 and plated onto 10 µg/ml collagen I or 6% BSA coated dishes for 1 hour. Adherent cells were the lysed and analysed for COX-2 expression by Western blotting. All membranes were re-probed for actin to ensure equal loading. (C) Int 407 cells were transfected with 50 nM siRNA against CD47 or with scrambled control siRNA, for 48 hours. Thereafter cells were plated onto 10 µg/ml collagen I coated dishes for 1 hour. Adherent cells were then lysed and analysed for CD47 and COX-2 expression by Western blotting as previously described. All membranes were re-probed for actin to ensure equal loading. (D) Cells were plated onto 10 µg/ml collagen I, or 6% BSA coated dishes with or without 10 µg/ml TSP-1 for 1 hour. All cells were lysed and the COX-2 expression was analysed by Western blotting. All membranes were re-probed for actin to ensure equal loading. (E) Cell adhesion assay; Int 407 cells were pre-treated with 20 µg/ml of the CD47 functional blocking antibody (B6H12) or control IgG for 20 minutes. Whereupon 250,000 cells were plated onto 10 µg/ml collagen I coated dishes for 1 hour. Adherent cells were measured by their conversion of nitroblue tetrazolium to the insoluble formazan product and the absorbance was measured at 544 nm. The accumulated data are given as percent of control and represent means ± SE of at least five separate experiments. The statistical analyses were performed with unpaired Students t-test; *P<0.05, **P<0.01 relative to the control.

Figure 4. ECM dependent morphological differences modulated through CD47 signalling.

(A) Fluorescent microscope images were taken of cells plated onto cover slips coated either with 6% BSA, 10 µg/ml collagen I or fibronectin for 2 hours. Where indicated cells were pre-treated with 500 ng/ml PTX for 2 hours, 20 µg/ml of the CD47 functional blocking antibody for 20 minutes or 100 µM of the COX-2 specific inhibitor NS-398 for 30 minutes. Cells were fixed, permeabilised and stained with primary antibodies against either CD47 or α2 integrin using either Alexa-488 or -546 conjugated secondary antibodies. (B) as in A but stained with primary antibodies against CD47, α2 integrin and F-actin. (C) as in A but stained with primary antibodies against Rho A and F-actin. Shown are representative pictures of three separate experiments taken at 60× magnification. (D) A graphical representation of the number of cells with membrane blebs from 100 cell samples. The statistical analyses were performed with unpaired Students t-test; **P<0.01 relative to the control.

Figure 5. Collagen I dependent 2D and 3D cell migration is regulated by CD47.

(A) Wound healing assay; Int cells were pre-incubated or not with 20 µg/ml of the CD47 functional blocking antibody B6H12 for 2 hours, IgG or β2 integrin antibody for 20 minutes, or 100 µM of the COX-2 specific inhibitor NS-398 for 30 minutes, thereafter cells were plated onto 10 µg/ml collagen I coated dishes for 2 hours, after which a wound was made in the monolayer, and the cells were allowed to migrate for 18 hours. Pictures of the wound were taken after 0 and 18 hours. The wound closure was measured and is presented as the percentage of wound closure as compared to time zero. (B, C) 3D cell migration assay; (B) Cells were pre-incubated or not with 20 µg/ml of the CD47 functional blocking antibody B6H12 for 2 hours, IgG or the β2 integrin antibody for 20 minutes, or 100 µM of the COX-2 specific inhibitor NS-398 for 30 minutes after which 250,000 cells from each group were allowed to migrate through a 3 mg/ml collagen I gel and across an 8.0 µm micropore membrane for 18 hours. (C) Int 407 cells were transfected with 50 nM siRNA against CD47 (as indicated) or with scrambled control siRNA, for 48 hours. Thereafter 250,000 cells from each group were allowed to migrate through a 3 mg/ml collagen I gel and across an 8.0 µm micropore membrane for 18 hours. Cell migration was examined by staining with crystal violet blue and measuring the absorbance at 590 nm. The data are given as percent of control and represent means ± SE of five separate experiments. The statistical analyses were performed with unpaired Students t-test; *P<0.05, **P<0.01 relative to the control.