For western blot analysis anti-integrin α5 antibody (sc-10729) was purchased from Santa Cruz Biotechnology (Santa Cruz Biotechnology, Heidelberg, Germany). A300, A300E and A318 are mouse anti-human ADAM17 monoclonal antibodies generated in our laboratory (28). While A300 and A318 bind to disintegrin domain, A300E recognizes the membrane-proximal domain of ADAM17 (29,30).

Mouse embryonic fibroblasts (MEF) and human embryonic kidney 293 cells (HEK-293) were cultured in Dulbecco's modified Eagle's medium (DMEM) high-glucose supplemented with 10% fetal calf serum (FCS), penicillin (60 mg/l) and streptomycin (100 mg/l) (PAA Laboratories, Cölbe, Germany) in 5% CO₂ at 37°C in a humidified atmosphere. The human carcinoma cell lines (MDA-MB-231, Panc89 and Colo357 from Professor Holger Kalthoff, UKSH Kiel, Germany; NCI-H292 from Dr Anja Baumgart, TU Munich, Germany) were cultured in RPMI-1640 medium supplemented with 10% FCS, penicillin (60 mg/l) and streptomycin (100 mg/l).

The coding sequence of the human ADAM17 disintegrin domain (amino acid residue 475 to 580) was cloned into pcDNA3.1 zeo (−). Thereby the signal sequence of the human IL-6R, followed by a His- and a myc-tag was placed in front of the disintegrin domain. HEK-293 cells were transfected using TurboFect (Fermentas, St. Leon-Rot, Germany) according to the manufacturer's instruction and stably transfected cells were selected. The supernatant of these cells was harvested and the disintegrin domain purified by its N-terminal His-tag using a HisTrap column (GE Healthcare, Freiburg, Germany) followed by size-exclusion chromatography (Superdex200 column, GE Healthcare). The identity and purification was confirmed by SDS-PAGE and western blot analysis.

Microtiter tissue culture plates (Sarstedt, Nümbrecht, Germany) were coated with indicated concentrations of the disintegrin domain (28) in coating buffer (0.5 M carbonate-bicarbonate buffer, pH 9.6) at 4°C overnight. Plates were blocked with heat-denatured BSA (85°C for 10 min, 1% BSA in Ca²⁺/Mg²⁺-free PBS) for 1 h at room temperature. In parallel, MEF and carcinoma cells were detached with trypsin. The cells were resuspended in complete medium (DMEM, 10% FCS, 0.1% BSA, 1 mM CaCl₂, 1 mM MnCl₂, 1 mM MgCl₂). Plates were washed, 3×10⁴ cells/well were added and incubated for 1 h at 37°C. In case of the competition assays, cells were first incubated for 1 h at 37°C with the soluble disintegrin domain, washed and then added to the plates coated with the disintegrin domain. Non-adhering cells were removed by washing with complete medium, while attached cells were quantified using titer blue reagent (Promega, Mannheim, Germany). Adhesion was expressed as percentage relative to FCS control, which was set arbitrarily to 100% (31).

The adhesion assay of carcinoma cells to fibroblasts has been previously described (31). Briefly, carcinoma cells were labeled with 10 μM Calcein AM (Invitrogen, Darmstadt, Germany) for 1 h at 37°C. Labeled cells (3×10⁴) were washed three times with wash solution (HBSS, 1 mM MnCl₂, 1 mM MgCl₂, 1 mM CaCl₂), incubated with or without disintegrin domain for 1 h at 37°C and seeded onto 80% confluent monolayers of MEF in 96-well plates for 30 min. In a parallel experiment, anti-integrin β1 inhibitory antibody (clone P4C10, Millipore, Schwalbach, Germany) was added to carcinoma cells instead of disintegrin domain. Not adhered cells were removed by washing, while attached cells were lysed by adding 20 μl of 10% Triton X-100 and qualified by measurement of their fluorescence (480 nm/530 nm) using fluorescence reader (BioTek, Bad Friedrichshall, Germany). Statistical analysis was performed with Student's t-test (http://www.physics.csbsju.edu/stats/t-test_bulk_form.html).

The dominant negative (DN)-ADAM17 is a construct of human ADAM17 that lacks the catalytic domain (32). The human ADAM17 sequence (starting from amino acid residue 475) is placed directly behind the signal peptide of human interleukin-6 in pcDNA3.1 expressing a gentamycin resistance (32). MDA-MB-231 cells were transfected with the plasmid using TurboFect transfection reagent (Fermentas) according to the manufacturer's protocol. Stable transfected clones were selected and grown in medium containing G418 (PAA Laboratories). The expression of DN-ADAM17 was confirmed by flow cytometry and western blot analysis and clone 5 (DN-5) was chosen for further experiments. The DN-ADAM17 construct contains a C-terminal His- and myc-tag.

Stably transfected MDA-MB-231 cells (DN-5) were cultured in RPMI-1640 medium, harvested, lysed in 250 μl lysis buffer [20 mM Tris-HCl pH 7.6, 150 mM NaCl, 2 mM EDTA and ‘complete’ protease inhibitor (Roche Applied Science, Mannheim, Germany)] and centrifuged for 15 min at 14,000 rpm. Cell lysate (100 μl) was mixed with 50 μl of protein-G agarose (Thermo Scientific-Pierce, Bonn, Germany) and 2.5 μg anti-α5β1 integrin antibodies (NKI-SAM-1, TS2/16; Biozol, Eching, Germany) or 5 μg anti-ADAM17 A300E (28). As control, 100 μl of cell lysate was mixed with 50 μl of protein-G agarose without antibodies. The samples were incubated 2 h at 4°C by gentle agitation, washed five-times with 1,000 μl of lysis buffer containing 1 mM CaCl₂ and prepared for western blot analysis. To detect the immunoprecipitated proteins, 20 μl SDS denatured protein cell lysate was separated on a 12.5% polyacrylamide gel. After separation by SDS-PAGE, proteins were transferred onto a polyvinylidene-fluoride (PVDF) membrane (GE Healthcare). The PVDF membrane was blocked with 5% milk powder in TBS-T (10 mM Tris-HCl, pH 7.6, 150 mM NaCl, 0.1% Tween-20) for 2 h and then incubated overnight at 4°C with rabbit anti-myc tag (1/1,000; Cell Signaling, Frankfurt, Germany) or with anti-α5 integrin antibody (sc-10729, Santa Cruz Biotechnology) in 1% milk powder. After washing, the blots were incubated for 1 h with an anti-rabbit IgG POD conjugate and visualized by chemiluminescence using an ECL kit (GE Healthcare).

Microtiter plates (Nunc, Denmark) were coated with 50 μl (10 μg/ml) per well of human purified α5β1 integrin (gift from Dr Humphries, University of Manchester, Manchester, UK) in coating buffer (PBS, 1 mM MnCl₂) and incubated at 4°C overnight. Then the plates were washed with washing solution (TBS, 1 mM MnCl₂, 0.1% BSA) and blocked with 5% BSA in TBS for 3 h at room temperature. After washing, 50 μl of recombinant disintegrin domain (80 μg/ml) plus 50 μl of mouse anti-disintegrin monoclonal antibodies (A300, A318; 20 μg/ml) (28) or anti-His-tag antibody (1/500; Cell Signaling) were added and incubated for 3 h at room temperature. Blank control was performed with PBS. The plates were washed with washing solution and the antibodies were detected by anti-mouse IgG-POD conjugate (Southern Biotech, Birmingham, AL, USA) diluted 1/3,000 in washing buffer. After washing, 80 μl of BM blue POD substrate (Roche Applied Science) was added to each well for 10 min. Then, the reaction was stopped by adding 80 μl of 0.8 M H₂SO₄. Measurement of the optical density (OD) was performed at 450 nm/540 nm on a plate reader (Tecan, Männedorf, Switzerland).

Cells (5×10⁵) were washed twice in FACS solution (PBS, 0.05% NaN₃) and stained for 1 h at 4°C with 2 μg/ml of anti-ADAM17 (A300E) antibody directed against the membrane-proximal domain of the enzyme (28). After washing, 2 μl of allophycocyanin-conjugated goat anti-mouse Ig (Jackson ImmunoResearch Laboratories, West Grove, PA, USA) was added. The cells were then incubated for 1 h at 4°C and analyzed with a FACScan flow cytometer (BD Biosciences, USA). Cell populations of interest were gated and analyzed using BD FACSDiva software.

MDA-MB-231 cells were seeded on 24-well plate (70–80% confluence) and incubated overnight at 37°C in RPMI medium containing 10% FCS. The soluble disintegrin domain of ADAM17 was added to a final concentration of 20 μg/ml in serum-free medium and incubated for 8 h. The matrix metal-loprotease inhibitor, marimastat (MM), was used at final concentration of 3 μM. The supernatant was collected and the levels amphiregulin were determined using a sandwich ELISA (DuoSet Kit; R&D Systems, Minneapolis, MN, USA) in accordance with the manufacturer's instructions.

A previous study has shown that ADAM17 interacts with α5β1 integrin (7). To verify the specificity of this interaction, we performed an ELISA with the purified disintegrin domain of ADAM17 and the α5β1 integrin (Fig. 1A). Binding of the disintegrin domain to the purified α5β1 integrin was detected by monoclonal mouse anti-disintegrin domain antibodies (A300, A318) (28) and confirmed by an anti-His antibody. The ELISA data demonstrate that the disintegrin domain and the α5β1 integrin interact directly. Co-immunoprecipitation was used as second approach to confirm the interaction between the disintegrin domain and the α5β1 integrin. Thereby, DN-ADAM17 was detectable in cell lysates from MDA-MB-231 cells stably transfected with the ADAM17 deletion mutant and could be co-immunoprecipitated using anti-α5β1 antibodies (Fig. 1B). In addition, α5 integrin could be detected after precipitation with anti-ADAM17 antibody (A300E) from cell lysate of DN-ADAM17 stable transfected MDA-MB-231 cells (Fig. 1B). Thus, the disintegrin domain of ADAM17 binds to the α5β1 integrin.

To examine the adhesive properties of ADAM17, we performed cell adhesion assays using fibroblasts and the disintegrin domain coated on cell culture plates (Fig. 2). Fibroblasts were chosen because they express a wide range of different integrins including α5β1 (33). The amount of fibroblasts bound to the plates coated with FCS was considered as a positive control and set to 100%. A coating-concentration of 50 μg/ml disintegrin domain leads to a maximal adhesion of MEF cells (Fig. 2A). Further increase in the concentration of the disintegrin domain did not improve binding of the MEFs (data not show). Also human fibroblasts IRM.90 bind to the disintegrin domain of ADAM17 in concentration-dependent manner (Fig. 2B).

FACS (Fig. 3A) and western blot (Fig. 3B) analysis demonstrate that carcinoma cells express high levels of ADAM17. This led to the assumption that ADAM17 may be involved in the interactions between fibroblast and carcinoma cells and raised the question, whether upregulation of ADAM17 in carcinoma cells increases the cell-cell interactions between fibroblasts and carcinoma cells. To address this question two assays were performed. First, we analyzed the binding of carcinoma cells to the immobilized disintegrin domain. Secondly, the interaction between fibroblast and carcinoma cells expressing different level of ADAM17 was evaluated.

Fig. 3C shows that various carcinoma cells (MDA-MB-231, Colo357, Panc89, NCl-H292) bind in a dose-dependent manner to solid phase coupled disintegrin domain. In case of three cells lines (MDA-MB-231, Panc89, NCl-H292) the maximal binding reaches 60% at a coating-concentration of 50 μg/ml disintegrin domain, in comparison to the positive control coated with FCS; while only 20% of Colo357 cells bind to the disintegrin domain coated surface. This difference correlates to some extent with the expression level of α5 integrin of these cells (Fig. 3D).

To prove that carcinoma cells and the disintegrin domain interact specifically and directly, we performed a competition assay using soluble and immobilized disintegrin domain. Pretreatment of Panc89 (Fig. 4A), IMR.90 (Fig. 4B) and MDA-MB-231 (Fig. 4C) cells with the soluble disintegrin domain significantly reduced the binding of these cells to the disintegrin domain coated solid phase. A pretreatment with 30 μg/ml of soluble disintegrin domain results in all three cell lines to a decrease in adhesion down to the level of the negative control (plates coated with BSA). In summary, these results indicate that carcinoma cells express α5 integrin and adhere to the disintegrin domain of ADAM17.

To evaluate the role of the disintegrin domain at a cellular level, cell-cell interaction assays between fibroblasts and carcinoma cells in the presence or absence of the disintegrin domain of ADAM17 were performed. In these experiments, the MDA-MB-231 (Fig. 5A), NCI-H292 cells (Fig. 5B) and Panc89 (Fig. 5C) were labeled with Calcein AM, treated with different concentrations of the soluble disintegrin domain and seeded on top of confluent MEF monolayers. The MDA-MB-231, NCI-H292 and Panc89 cells attached to MEFs in the absence of the soluble disintegrin domain. Pretreatment of MDA-MB-231, NCI-H292 and Panc89 cells with the soluble disintegrin domain significantly reduced the interactions between carcinoma and murine fibroblast cells. Similar interactions were observed between the carcinoma and human fibroblasts IMR.90 (Fig. 5D–F). Thus, ADAM17 mediates cell-cell interactions, which can be inhibited by the soluble disintegrin domain.

In order to evaluate further the adhesive property of the disintegrin domain, we performed cell-cell interaction assays using ADAM17 deficient cells. Whereas ADAM17 deficient cells (ADAM17^ex/ex MEFs) were derived from ADAM17^ex/ex mice (34), ADAM17 silencing in NCI-H292 cells was achieved using an ADAM17 shRNA (35). ADAM17^ex/ex mice were generated by insertion of new exon within the gene of ADAM17. This new exon starts with an in-frame translational stop codon. Homozygous ADAM17^ex/ex mice were viable and showed a dramatic loss of ADAM17 protein in all cell types (34). In line with the above results, the amount of ADAM17 knockdown NCI-H292 (−) cells attaching to wt-MEF (+) was significantly reduced as compared to the wt-NCI-H292 (+) cells (Fig. 6). Furthermore, silencing ADAM17 in MEF, using ADAM17^ex/ex MEFs (−), and in NCI-H292 (−) cells decreased further the cell-cell interaction by about half in comparison to the control (Fig. 6). These results indicate that ADAM17 acts as an adhesion molecule and mediates fibroblast-carcinoma cell interaction.

The interaction between α5β1 integrin and ADAM17 keeps the enzyme in an inactive state (9,10). This finding raises the assumption that the soluble disintegrin domain of ADAM17 should be able to induce shedding by competing with the wild-type molecule for the binding to α5β1 integrin. Thereby the complex between the integrin and enzyme should be dissolved, releasing an active ADAM17.

To test this hypothesis, we measured the shedding of amphiregulin in MDA-MB-231 cells incubated with or without the soluble disintegrin domain (Fig. 7). Amphiregulin, a substrate of ADAM17 (36), belongs to the epidermal growth factor family, and acts as mitogenic stimulator through binding to epidermal growth factor receptors (37). The resulting data showed that treatment of these cells with the soluble disintegrin domain increases shedding of amphiregulin in comparison to control cells treated with PBS. The shedding induced by soluble disintegrin could be inhibited by metalloprotease inhibitor, marimastat. Thus, soluble disintegrin domain binds to α5β1 integrin and consequently activates the wild-type molecule, most likely by releasing ADAM17 out of the inhibitory complex.