Src Stimulates Fibroblast Growth Factor Receptor-2 Shedding by an ADAM15 Splice Variant Linked to Breast Cancer

Cell-cell interactions are critical for cellular communications during development and in diseases such as cancer. Membrane-anchored proteinases have emerged as key regulators of cell-cell interactions because of their ability to cut and release membrane proteins, a process referred to as protein ectodomain shedding (1). This, in turn, affects the functions of molecules such as the proinflammatory cytokine tumor necrosis factor, or ligands of the epidermal growth factor (EGF) receptor, including transforming growth factor α and heparin-binding EGF. Membrane-anchored metalloproteinases of the ADAM (a disintegrin and metalloproteinase) family are frequently responsible for protein ectodomain shedding events and have been implicated in a variety of physiologic and pathologic processes, including heart development, neurogenesis, and cancer (1).

Recently, increased expression of ADAM15 (ADAM15A), and in particular a splice variant of ADAM15 with an inserted binding site for Src (ADAM15B, see Fig. 1A), was linked to poor survival in node-negative breast cancer (2). This raises questions about how ADAM15 and the ADAM15B splice variant might contribute to the pathogenesis of breast cancer. Because ADAM15 (3–5) is catalytically active (6), we hypothesized that the inserted Src binding site might influence the catalytic activity of ADAM15. This, in turn, could affect the processing and regulation of membrane proteins with potential roles in cancer. To evaluate the catalytic activity of ADAM15A and ADAM15B, we focused on ectodomain shedding of the fibroblast growth factor receptor 2iiib variant (referred to as FGFR2) because this receptor can be cleaved by ADAM15 (6) and also because it has been implicated in the development of breast cancer (7), so that its proteolytic release from cells could conceivably affect its role in mammary tumorigenesis.

Cell lines and reagents. Simian virus large T-antigen–immortalized Src^−/− and wild-type control mouse embryonic fibroblasts (mEF) were previously described (8). COS7 and MCF-7 cells were from American Type Culture Collection. All cells were grown in DMEM with antibiotics and 5% FCS. All reagents were from Sigma-Aldrich unless otherwise indicated. PP1 was from Biomol; PP2 and PP3 were from Calbiochem. Dasatinib was kindly provided by Mark Moasser (UCSF Medical Center, San Francisco, CA). Anti-ADAM15 antibodies were described previously (3). Anti-phosphotyrosine, clone 4G10, was from Millipore, and anti-Src (sc-18) from Santa Cruz.

Expression vectors. The expression vectors for human ADAM15A and ADAM15B, alkaline phosphatase (AP)–tagged FGFR, c-Src, v-Src, c-Src(K295A), c-Src(E378G), c-Src(Y527F), and MAD2 have previously been described (2, 8–10).

Site-directed mutagenesis. Site-directed mutagenesis was done with the Stratagene QuickChange kit following the manufacturer's protocol.

Cell culture, transfection, and stable transfection. Fibroblasts and MCF-7 cells were transfected with Lipofectamine 2000, and COS7 cells with Lipofectamine, as described (6). Stable cell lines were selected in medium containing either 50 to 200 μg/mL hygromycin B or 25 to 100 μg/mL zeocin (Invitrogen).

Ectodomain shedding assay. Cells were transfected with FGFR2-AP and washed with OptiMEM 1 d after transfection; fresh OptiMEM with or without inhibitors of Src-kinases was added and incubated for 1 to 4 h, as indicated (6). AP activity in the supernatant and cell lysates of at least three identically treated wells was measured by colorimetry as described (6, 11).

RNA isolation and real-time PCR. Total RNA isolation, cDNA synthesis, and quantitative PCR were as described in ref. 12. Primer pairs for human ADAM15 and s28rRNA were A15F397 (CACACTGGAAGTGGCCCTCTTGC) and A15R564 (GGCAATCGAGGCAGCAAATGTGCC) and h28SF (GATCCTTCGATGTCGGCTCTTCCTATC) and h28SR (AGGGTAAAACTAACCTGTCTCACG).

Immunoblotting and immunoprecipitaton. Cells were lysed on ice in TBS Triton X-100 (1%), 1 mmol/L EDTA, protease inhibitor cocktail (Roche). Identical amounts of protein were separated on 10% SDS-polyacrylamide gels and transferred onto polyvinylidene difluoride membranes (BioTrace, Pall Corporation). These were blocked in 3% skim milk in TBS, incubated with primary antibodies and then with peroxidase-conjugated goat anti-mouse or goat anti-rabbit secondary antibodies (Promega), and exposed using the enhanced chemiluminescence detection system (Amersham Biosciences) and a Chemdoc image analyzer (Bio-Rad). For immunoprecipitations, cell lysates were precleared with Sepharose beads for 1 h at 4°C, then the primary antibody and protein G-Sepharose were added for 2 h at 4°C. The beads were washed and resuspended in 2× sample loading buffer with 5% β-mercaptoethanol. Bound proteins were eluted at 100°C for 5 min and separated by 10% SDS-PAGE.

Cell adhesion assay. MCF-7 cells (1 × 10⁵) were seeded in triplicate in 96-well plates in serum-free medium and incubated for 1 h at 37°C in 5% CO₂. After decanting the medium, the attached cells were washed with PBS, fixed, and stained; absorbance was read at 630 nm as described (2).

Statistics. Statistical analyses were done using the Statistical Program SigmaStat 3.1 software (Erkrath, SYSSTAT) as described (13).

To compare the catalytic activities of ADAM15A and ADAM15B, we expressed either splice variant together with the FGFR2 in COS7 cells. Increased shedding of FGFR2 was observed from cells expressing ADAM15A compared with cells expressing the inactive catalytic site mutant ADAM15E>A (Fig. 1B,, top). Moreover, overexpression of ADAM15B further increased FGFR2 shedding compared with ADAM15A (Fig. 1B,, top), although ADAM15B protein expression levels were equivalent to ADAM15A (Fig. 1B,, bottom). Similar results were obtained with COS7 cell lines stably expressing comparable levels of ADAM15A or ADAM15B that were then transiently transfected with FGFR2 (Supplementary Fig. S1). In addition, ADAM15B had increased FGFR2 sheddase activity compared with ADAM15A when transfected into Adam9/15^−/− double knockout mEFs, ruling out an indirect effect through activation of ADAM9, which can also shed FGFR2 (ref. 9; Supplementary Fig. S2). Finally, no increase in shedding was seen when catalytically inactive E>A mutants of ADAM15A or ADAM15B were coexpressed with FGFR2 in COS7 cells, despite similar expression levels (Fig. 1C), further corroborating that ADAM15B does not stimulate FGFR2 shedding indirectly, for example, by activating another sheddase. Taken together, these findings show that ADAM15B has increased catalytic activity compared with ADAM15A in cell-based assays.

Because the cytoplasmic domain of ADAM15B contains a Src binding site (green in Fig. 1A) and interacts with Src (2, 14), we tested whether Src-kinase inhibitors affected the catalytic activity of ADAM15B. Addition of the Src-family kinase inhibitors PP1 (Fig. 2A), PP2 (Fig. 2B), or dasatinib (Fig. 2C) reduced ADAM15B-dependent FGFR2 shedding in a dose-dependent manner to levels seen in cells expressing ADAM15A, with an IC₅₀ of 4.5 μmol/L for PP1, 3.9 μmol/L for PP2, and 4.6 μmol/L for dasatinib. These compounds had no effect on the activity of ADAM15A (Fig. 2A–C). Furthermore, the structurally similar but inactive PP3 had no effect on the activity of ADAM15A or ADAM15B (Fig. 2D).

As a direct test for a role of Src in activating ADAM15B, we assessed FGFR2 shedding in Src^−/− mEFs expressing ADAM15A or ADAM15B. We found that FGFR2 shedding by ADAM15A and ADAM15B was comparable in Src^−/− mEFs, whereas ADAM15B was significantly more active than ADAM15A in wild-type mEFs or in Src^−/− mEFs rescued by stable transfection of wild-type Src (Fig. 3A). Moreover, coexpression of a mutant inactive c-Src [Src(K295A)] with ADAM15B in Src^−/− cells did not restore increased shedding compared with ADAM15A, whereas cotransfection of v-Src or two activated variants of c-Src [Src(E378G) and Src(Y527F)] increased the activity of ADAM15B relative to ADAM15A (Fig. 3B).

To determine whether one or more of the four tyrosine residues in the cytoplasmic domain of ADAM15B (highlighted in red in Fig. 1A) are responsible for its stimulation by Src, we mutated each tyrosine to phenylalanine, either individually or in combination with others. We found that mutating the third tyrosine or all four tyrosines to phenylalanine reduced the activity of ADAM15B to that of ADAM15A, whereas mutations of the other tyrosines, including tyrosines 1, 2, and 4, did not (Fig. 4A). When ADAM15A, ADAM15B, or the tyrosine mutants were immunoprecipitated and then blotted with anti-phosphotyrosine antibodies, only the forms of ADAM15B containing the third tyrosine were phosphorylated. ADAM15B lacking the third tyrosine, or all four tyrosines, as well as ADAM15A were not or only weakly phosphorylated, although none of the mutations in ADAM15B affected the coimmunoprecipitation with Src (Fig. 4B). These findings suggest that phosphorylation of the third cytoplasmic tyrosine residue is critical for activation of ADAM15B by Src. Moreover, we found that the previously reported ability of ADAM15B to decrease adhesion of MCF-7 cells compared with ADAM15A (2) can be blocked by the metalloproteinase inhibitor marimastat (10 μmol/L) and the Src-kinase inhibitor dasatinib (10 μmol/L), as well as by mutating the third cytoplasmic tyrosine residue in ADAM15B (Supplementary Fig. S3), so ADAM15B could potentially prevent adhesion and thereby perhaps promote the spreading of tumor cells.

Taken together, these results show that the ADAM15B splice variant, which contains a Src binding site not present in ADAM15A (2), has a significantly increased catalytic activity compared with ADAM15A. Remarkably, the increased activity of ADAM15B depends on Src and on tyrosine 735 in ADAM15B, providing the first evidence, to our knowledge, of tyrosine phosphorylation–dependent regulation of the catalytic activity of an ADAM splice variant by an oncogene. These findings suggest that increased expression of ADAM15, and particularly the ADAM15B splice variant, in mammary epithelial cells could promote shedding and activation of growth factors or receptors that can promote tumorigenesis or, alternatively, inactivate tumor suppressors. The FGFR2 has been implicated in both the progression and suppression of cancer (7, 15), so its shedding from the membrane by increased ADAM15 activity could affect its role in cancer. These results suggest that inhibitors of ADAM15, or of the interaction of ADAM15B with Src, might be useful to treat breast cancer patients with dysregulated ADAM15.

No potential conflicts of interest were disclosed.

Grant support: Novartis (T. Maretzky, C.P. Blobel, and C.M. Overall), the Wellcome Trust, Big C Appeal, and EU Framework Programme 6 Cancerdegradome Project LSHC-CT-2003-503297 (D.R. Edwards).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.