Loss of ADAM9 Leads to Modifications of the Extracellular Matrix Modulating Tumor Growth

Abstract

ADAM9 is a metalloproteinase strongly expressed at the tumor-stroma border by both tumor and stromal cells. We previously showed that the host deletion of ADAM9 leads to enhanced growth of grafted B16F1 melanoma cells by a mechanism mediated by TIMP1 and the TNF-α/sTNFR1 pathway. This study aimed to dissect the structural modifications in the tumor microenvironment due to the stromal expression of ADAM9 during melanoma progression. We performed proteomic analysis of peritumoral areas of ADAM9 deleted mice and identified the altered expression of several matrix proteins. These include decorin, collagen type XIV, fibronectin, and collagen type I. Analysis of these matrices in the matrix producing cells of the dermis, fibroblasts, showed that ADAM9^-/- and wild type fibroblasts synthesize and secreted almost comparable amounts of decorin. Conversely, collagen type I expression was moderately, but not significantly, decreased at the transcriptional level, and the protein increased in ADAM9^-/- fibroblast mono- and co-cultures with melanoma media. We show here for the first time that ADAM9 can release a collagen fragment. Still, it is not able to degrade collagen type I. However, the deletion of ADAM9 in fibroblasts resulted in reduced MMP-13 and -14 expression that may account for the reduced processing of collagen type I. Altogether, the data show that the ablation of ADAM9 in the host leads to the altered expression of peritumoral extracellular matrix proteins that generate a more favorable environment for melanoma cell growth. These data underscore the suppressive role of stromal expression of ADAM9 in tumor growth and call for a better understanding of how protease activities function in a cellular context for improved targeting.

Keywords: ADAM9; collagen type I; extracellular matrix; fibroblasts; melanoma.

Figure 1

Differential expression of ECM proteins in the absence of ADAM9. (A) Analysis of ECM proteins at the tumor-stroma border. Black dotted lines indicate the microdissected areas used for mass spectrometry. t, tumor; s, stroma. Scale bar 50 µm. In the lower heatmap are the relative intensities of protein expression in ADAM9^−/− and WT. (B) Immunofluorescence analysis of collagen type I, collagen type XIV, decorin, and fibronectin (green, nuclei: blue). As a negative control, sections were incubated in a blocking buffer without the primary antibody. Representative pictures are shown (WT n = 5; ADAM9^−/− n = 5). Scale bar 400 µm. Staining was quantified in five different fields from 3 mice each genotype. Densities are shown as the percentage of the measured area; these are on the right. *** p < 0.0005.

Figure 2

Expression of decorin and collagen type I in ADAM9 deficient fibroblasts. (A) Immunoblot analysis of decorin and collagen type I in cell supernatants and lysates from monolayer cultures of ADAM9 deficient or wild type dermal fibroblasts. Actin was used as a control. Underneath, in the table, is the average densitometric analysis after normalization to control (n = 3; * <0.03–4) (B) Quantitative real-time PCR analysis of decorin and collagen type I was performed using S26 as an internal control for normalization of transcripts. Data are mean values ± SEM (average of 6 different fibroblasts cultures per genotype). (C) Immunoblot analysis of decorin and collagen type I in cell supernatants (sp.) and lysates from monolayer cultures of B16F1, ADAM9^−/−, or wild type dermal fibroblasts, without or with stimulation B16F1 conditioned medium (B16F1 c.m.). Underneath, in the table, is the average densitometric analysis of collagen and decorin after normalization to control (n = 4; * <0.01). Ponceau staining of the membrane was used as a control for loading.

Figure 3

Collagen type I is a substrate of recombinant ADAM9 (A) Upper insert, detection of purified ADAM9 by western blot using anti-ADAM9 antibodies. In the lower panel, gelatin zymography analysis shows the high gelatinolytic activity of the catalytic active 55 kDa (sADAM9) and low of the catalytically inactive ADAM9 (sADAM9E/A). (B) Enzymatic processing of fibronectin or collagen type I by recombinant ADAM9s as assessed by immunoblotting. 500 ng of recombinant collagen type I or fibronectin were incubated alone (control, −) or without (−) and with one µg of recombinant sADAM9 or sADAM9E/A (+) for 16 h in substrate buffer. Incubation of collagen with 500 ng MMP-1 used as a positive control. Molecular weight markers are on the right.

Figure 4

Deletion of ADAM9 leads to reduced expression of collagenolytic enzymes and thereby enhanced tumor cell proliferation. (A) MMP-13 and MMP-14 were analyzed by immunoblot in fibroblast supernatants (normalized to amounts of protein lysates) and lysates, respectively. Both ponceau staining of the membrane and actin immunoblot were used as the loading controls. Underneath, the quantification of MMP13 and 14 is shown (n = 5–9 independent fibroblasts isolation each genotype). (B) The proliferation of B16F1 melanoma cells was assessed in the presence of increasing concentrations of purified collagen type I by BrdU incorporation. The graph represents the mean ± SD. p values were calculated for cell proliferation in the presence of collagen type I compared to serum-free untreated control. * p < 0.05.