Transmembrane collagen XVII, an epithelial adhesion protein, is shed from the cell surface by ADAMs

Abstract

Collagen XVII, a type II transmembrane protein and epithelial adhesion molecule, can be proteolytically shed from the cell surface to generate a soluble collagen. Here we investigated the release of the ectodomain and identified the enzymes involved. After surface biotinylation of keratinocytes, the ectodomain was detectable in the medium within minutes and remained stable for >48 h. Shedding was enhanced by phorbol esters and inhibited by metalloprotease inhibitors, including hydroxamates and TIMP-3, but not by inhibitors of other protease classes or by TIMP-2. This profile implicated MMPs or ADAMs as candidate sheddases. MMP-2, MMP-9 and MT1-MMP were excluded, but TACE, ADAM-10 and ADAM-9 were shown to be expressed in keratinocytes and to be actively involved. Transfection with cDNAs for the three ADAMs resulted in increased shedding and, vice versa, in TACE-deficient cells shedding was significantly reduced, indicating that transmembrane collagen XVII represents a novel class of substrates for ADAMs. Functionally, release of the ectodomain of collagen XVII from the cell surface was associated with altered keratinocyte motility in vitro.

Fig. 1. Schematic representation of collagen XVII and the epitope regions recognized by antibodies. The full-length collagen XVII is a type II integral transmembrane protein, which is cleaved to yield a soluble ectodomain. White bars, collagenous sequences; light gray bars, non-collagenous sequences; black line, the cell plasma membrane. Dark gray bars, epitope regions of the domain-specific antibodies. Endo-2 recognizes the intracellular domain, NC-16A the juxtamembranous extracellular NC16a-domain, and Ecto-1 (Schäcke et al., 1998) and Ecto-5 the distal C-terminus of the molecule. The lower panel shows immunoblots of keratinocyte cell layer (C) and media (M) with the above antibodies. Ecto-1, Ecto-5, and NC16A recognized both the 180 kDa full-length collagen XVII and the 120 kDa soluble ectodomain. In contrast, Endo-2 recognized only full-length collagen XVII.

Fig. 2. Shedding of collagen XVII ectodomain. (A) Time-chase experiments with normal human keratinocytes. The cells were surface biotinylated, chased with biotin-free medium and cultured for 96 h. Aliquots of culture medium and cell extract were immunoprecipitated with antibody Ecto-1 or Endo-2 and analyzed by protein blotting with streptavidin–alkaline phosphatase. The lower panel shows the densitometric analysis of the signals. Note that the intact ectodomain was stable in the medium for at least 72 h. (B) Shedding was stimulated by phorbol esters and IL-1β. Keratinocytes were cultured for 24 h in the presence of 100 nM PMA or IL-1β (5 ng/ml). Aliquots were drawn at regular intervals and analyzed by immunoblotting with Endo-2 and NC16A antibodies. The signals were analyzed densitometrically with TotalLab1D (Phoretics) software.

Fig. 3. Furin, BMP-1, MMP-2, MMP-9 and MT1-MMP are not directly involved in the processing of collagen XVII. (A) Purified native collagen XVII was digested with recombinant human MT1-MMP, MMP-2, MMP-9, furin and BMP-1. The digests in the absence (–) or in the presence (+) of 5 mM EDTA were immunoblotted with antibody Ecto-1. Furin and BMP-1 did not cleave collagen XVII. MMP-2, MMP-9 and MT1-MMP generated a 140 kDa fragment which was distinct from the authentic 120 kDa fragment in culture medium (Med). (B) Ectodomain shedding in MMP-2-deficient gastric carcinoma cells (left panel) and MT1-MMP-deficient murine keratinocytes (right panel). In both cultures the 120 kDa ectodomain was generated similarly to controls. C, cell extract; M, medium. Immunoblotting was with a mixture of antibodies Endo-2 and NC16A (left panel) or Endo-2 and MO-NC16A (right panel).

Fig. 4. Expression of ADAMs in human keratinocytes. (A) Immuno fluorescence staining of collagen XVII and ADAMs in the skin. Collagen XVII antibody NC16A produced a linear signal at the basement membrane zone (a). TACE (b) and ADAM-9 (c) immunoreactivity was stronger in the lower epidermis, in the neighborhood of collagen XVII, whereas ADAM-10 immunoreactivity (d) was distributed throughout the entire epidermis. Scale bar: 250 µm. (B) Immunoblot of keratinocyte extracts with antibodies to the ectodomains of TACE and ADAM-9, and the endodomain of ADAM-10. TACE and ADAM-9 antibodies recognized two distinct protein bands of ∼120 and 85 kDa, and 110 and 79 kDa, respectively, representing the proform and the active form of the enzymes. ADAM-10 antibody recognized one band of ∼60 kDa, corresponding to the active form. (C) Activation of TACE in keratinocytes was inhibited by a furin inhibitor, but not by hydroxamates. Keratinocytes were cultured for 24 h in serum-free medium with or without 50 µM furin inhibitor decanoyl-RVKR-chloromethyl ketone or 25 µM hydroxamate BB 3103. Keratinocyte lysates were immunoblotted with TACE antibodies. The furin inhibitor, but not BB 3103, prevented the conversion of pro-TACE to active TACE. Molecular weight markers are shown on the left.

Fig. 5. Transient transfections of HaCaT cells with cDNAs for TACE, ADAM-9 and ADAM-10. (A) Immunofluorescence staining of transfected cells. Cells transfected with empty vector (control) remained negative, but ADAM-transfected cells showed strong positive signals. Scale bar: 100 µm. (B) HaCaT cells were transfected with different concentrations of cDNA for TACE (upper panel), ADAM-10 (middle panel) and ADAM-9 (lower panel) and shedding of collagen XVII was assessed with immunoblotting with a mixture of antibodies Endo-2, NC16A and Ecto-1. Densitometric analysis of the signals (expressed as a percentage of the control ± SD; n = 3) showed a dose-dependent increase of ectodomain shedding, with a concomitant decrease of full-length collagen XVII.

Fig. 6. Hydroxamates inhibit increased collagen XVII shedding in TACE-transfected cells. The cells were electroporated with empty vector (control), or with 10 or 20 µg of TACE cDNA. After 24 h the cells were washed, 10 µM BB 3103 or BB 3241 added, and after 6 h the ectodomain in the medium detected with immunoblotting using antibodies NC16A and Ecto-1. Both hydroxamates strongly inhibited shedding in this system.

Fig. 7. Decreased collagen XVII shedding in TACE-deficient cells. (A) Murine keratinocytes from wild-type (WT) and TACE –/– mice were analyzed for collagen XVII shedding in vitro. Cells (C) and media (M) were sampled after 6 h and immunoblotted with a mixture of antibodies MO-NC16A and Endo-2 (left panel). The relative intensity of the signals of three individual experiments is expressed as a percentage ± SD (right panel). (B) Wild-type (WT) and TACE –/– fibroblasts were transfected with human collagen XVII cDNA. After 2 days the cells were analyzed for collagen XVII shedding. Cells (C) and media (M) were sampled after 6 h and immunoblotted with a mixture of antibodies MO-NC16A and Endo-2 (left panel). The relative intensity of the signals of two different experiments is expressed as a percentage ± SD (right panel).

Fig. 8. Increased collagen XVII shedding correlates with decreased cell motility. In wound closure assays cells migrate from the edges of a ‘scrape wound’ in the monolayer (interrupted lines) to cover the denuded space (between the lines). Scrape wounds were made in serum-free cultures, and after washing the cells were allowed to re-epithelialize for 22 h at 37°C. Micrographs of non-fixed cell layers at 0, 11, 16 or 22 h after wounding are shown. Scale bars: 200 µm. (A) Transfection of HaCaT cells with ADAM cDNA negatively regulated cell motility. Transfections were with 15 µg of empty vector (control), 15 µg of TACE, 15 µg of ADAM-10, or 15 µg of ADAM-9 cDNA. (B) Addition of purified collagen XVII ectodomain to normal human keratinocytes inhibited cell motility. After wounding, 1 nM collagen XVII ectodomain was added to the culture, and the cells were permitted to migrate into the denuded space for 16 h. Ectodomain-treated cells did not cover the denuded area as efficiently as control cells.