Targeting of Cell Surface Proteolysis of Collagen XVII Impedes Squamous Cell Carcinoma Progression

Abstract

Squamous cell carcinoma (SCC) is one of the most common skin cancers and causes significant morbidity. Although the expression of the epithelial adhesion molecule collagen XVII (ColXVII) has been linked to SCC invasion, only little is known about its mechanistic contribution. Here, we demonstrate that ColXVII expression is essential for SCC cell proliferation and motility. Moreover, it revealed that particularly the post-translational modification of ColXVII by ectodomain shedding is the major driver of SCC progression, because ectodomain-selective immunostaining was mainly localized at the invasive front of human cutaneous SCCs, and exclusive expression of a non-sheddable ColXVII mutant in SCC-25 cells inhibits their matrix-independent growth and invasiveness. This cell surface proteolysis, which is strongly elevated during SCC invasion and metastasis, releases soluble ectodomains and membrane-anchored endodomains. Both released ColXVII domains play distinct roles in tumor progression: the endodomain induces proliferation and survival, whereas the ectodomain accelerates invasiveness. Furthermore, specific blockage of shedding by monoclonal ColXVII antibodies repressed matrix-independent growth and invasion of SCC cells in organotypic co-cultures. Thus, selective inhibition of ColXVII shedding may offer a promising therapeutic strategy to prevent SCC progression.

Keywords: carcinoma; collagen; invasion; proliferation; proteolysis.

Figure 1

Enhanced ColXVII Expression and Shedding during SCC Invasion and Metastasis (A–C) Human SCC tissue cryosections were stained (n = 16). (A) Staining for podoplanin and ColXVII (NC16a-1). D, dermis; E, epidermis; T, tumor. (B) Double staining for histone H3 phosphorylated at serine 10 (H3S10p), a marker of mitotic cells and for ColXVII (NC16a-1). (C) Double staining of membrane-anchored full-length ColXVII (NC16a-1) and its released ectodomain (HK139). Nuclei were stained with DAPI. (D) Total skin lysates from healthy donors or SCC patients were subjected to western blot analysis. Representative blot of normal skin (NS) and two SCC samples (1 and 2). Protein of the 180 kDa ColXVII and its 120 kDa ectodomain (Ecto) was detected with the polyclonal antibodies Endo-2 and NC16A, respectively. Staining for ERK1/2 confirmed equal loading. (E) Immunohistochemistry of 13 FFPE samples from aggressive primary cutaneous SCCs and their corresponding metastases (regional lymph node and/or soft tissue metastases) using the ectodomain-selective ColXVII antibody NC16a-3. Nuclei were counterstained with hematoxylin. Representative stainings of four SCC patients: SCC #1: primary tumor: capillitium/corresponding metastasis: capillitium; SCC #2: primary tumor: central capillitium/corresponding metastasis: preauricular locoregional; SCC #3: primary tumor: lower lip/corresponding metastasis: submandibular lymph node; SCC #4: primary tumor: forehead/corresponding metastasis: preauricular lymph node. Scale bars, 50 μm.

Figure 2

ColXVII Expression Is Required for SCC Growth and Invasion (A) Immunoblot of cell lysates and culture media for ColXVII in SCC-25 Ctrl and Kd cells using NC16A antibody. (B) Immunoblot of cell lysates and culture supernatants for the re-expression of the murine full-length form of ColXVII in Kd cells (Kd-FL) using EC-7 antibody. GAPDH was used as loading control. Shedding of ColXVII was confirmed by the presence of the 120 kDa soluble ectodomain of ColXVII (Ecto) in the culture medium. (C–E) Ctrl, Kd, and Kd-FL cells were subjected to (C) cell aggregation, (D) colony-forming efficiency, and (E) invasion assays. (C) Left panel: representative pictures of cell aggregates grown in suspension for 5 days on anchorage-independent conditions. Right panel: quantification of the cell aggregate diameters (one-way ANOVA: ***p < 0.001; n = 3). (D) Left panel: representative pictures of colony formation on 6 cm culture dishes after 7 days, followed by crystal violet staining. Right panel: quantification of the number of colonies (one-way ANOVA: **p < 0.01; ***p < 0.001; n = 3). (E) Left panel: representative pictures of invasion assay after 24 hr using Matrigel-coated transwell inserts, followed by crystal violet staining. Right panel: quantification of the number of invading cells per filter (one-way ANOVA: ***p < 0.001; n = 3). Data are shown as mean ± SEM. Scale bars, 200 μm.

Figure 3

ColXVII Induces Tumor Growth of SCC-25 Cells In Vivo (A) SCC-25 Ctrl, Kd cells were injected subcutaneously into NOD/SCID mice. Tumors were resected after 9 weeks and embedded in Tissue-Tek O.C.T compound or in paraffin. Left panel: representative pictures of SCC-25 xenografts and quantification of the tumor weight as mean ± SEM (used animals in two independent experiments: Ctrl, n = 10; Kd, n = 8; Mann-Whitney U test, **p < 0.01). Right panel: H&E staining on paraffin-embedded sections and immunofluorescence on cryosections. (B) Kd and Kd-FL cells resuspended in Matrigel-PBS (1:1) were subcutaneously injected into NOD/SCID mice 24 hr after immunosuppression with busulfan. Tumors were resected after 4 weeks and analyzed as described before. Left panel: representative pictures of tumor xenografts and quantification of tumor weight as mean ± SEM (used animals: Kd, n = 7; Kd-FL, n = 8; Mann-Whitney U test, *p < 0.05). Right panel: H&E staining on paraffin-embedded sections and immunofluorescence on cryosections. Scale bars, 50 μm.

Figure 4

Prevention of ColXVII Shedding Suppresses SCC Growth and Invasion (A) Schematic representation of the non-shedding ColXVII mutant with the deletion of 41 amino acids (ΔE510-Y550) within the NC16A domain that includes all ADAM cleavage sites. The deletion completely prevents shedding, while maintaining its adhesion characteristics. The light gray and black bands represent the non-collagenous and collagenous domains of ColXVII, respectively. (B) Immunoblot analysis of cell lysates and culture media of Kd cells re-expressing either full-length ColXVII (Kd-FL) or non-sheddable ColXVII (Kd-NS) using EC-7 antibody. The 180 kDa full-length ColXVII was detected in Kd-FL and Kd-NS cell lysates, while no release of the 120 kDa ectodomain was seen in the medium of Kd-NS cells. (C–E) Kd-FL, Kd-NS, and Kd cells were subjected to (C) cell aggregation, (D) colony-forming efficiency, and (E) cell invasion assays as previously described. (C) Left panel: representative pictures of cell aggregates. Right panel: quantification of cell aggregate size (one-way ANOVA: ***p < 0.0001; n = 3). (D) Left panel: representative pictures of colony formation stained with crystal violet. Right panel: quantification of the colony numbers (one-way ANOVA: *p < 0.05; n = 3). (E) Left panel: representative pictures of Matrigel invasion. Right panel: quantification of invading cells per filter (one-way ANOVA: ***p < 0.0001; n = 3). Quantitative data are expressed as mean ± SEM. Scale bars, 200 μm.

Figure 5

Antibody Blockage of ColXVII Shedding Impairs Tumor Expansion (A) Schematic representation of ColXVII and the NC16a-1 epitope between the ADAM cleavage sites within the NC16A domain. The polyclonal antibody Ecto-4 is directed to the distal C terminus (amino acids [aa] 1,421–1,496). (B) Analysis of the shedding inhibition of NC16a-1. Left panel: immunoblot analysis of subconfluent SCC-25 cells incubated with the metalloproteinase inhibitor marimastat (3 μM, MRM) or with 1:50 or 1:5 dilution of NC16a-1 for 6 hr. Immunoblot analysis of the precipitated media was performed with Ecto-4 antibody. Complete loss of ectodomain release was seen with 1:5 dilution of NC16a-1. Right panel: densitometric quantification of immunoblot signals (one-way ANOVA: ***p < 0.001; n = 3). (C) SCC-25 cells treated with either isotype IgG2a κ (IgG) or NC16a-1 (1:5 dilution) were subjected to cell aggregation assays as previously described. Treatment with NC16a-1 caused strongly reduced aggregate growth. Right panel: quantification of cell aggregate size (Student’s t test: ***p < 0.0001; n = 4). (D) 3D co-cultures with HSC-3 cells on NHFs in a collagen I matrix were treated with either IgG or NC16a-1 in 1:5 dilution (NC16a-1). NC16a-1 treatment reduced HCS-3 invasion substantially compared to isotype IgG-treated controls. Left panel: representative images of H&E-stained paraffin-embedded sections. Right panel: quantification of the invasion depths (red arrows) (Student’s t test: ***p < 0.0001; n = 4). Data are shown as mean ± SEM. Scale bars, 200 μm.

Figure 6

ColXVII Endodomain and Ectodomain Play Distinct Roles in SCC-25 Cells Re-expression of ColXVII endodomain or ectodomain in SCC-25 Kd cells was achieved by retroviral transduction (see Figure S6). (A and B) Ctrl, Kd, and Kd cells either re-expressing ColXVII endodomain (Kd-Endo) or ColXVII ectodomain (Kd-Ecto) were subjected to (A) colony formation and (B) Matrigel-invasion assays. (A) Left panel: representative pictures of colony formation after crystal violet staining. Right panel: quantification of the colony number (one-way ANOVA: *p < 0.05; **p < 0.01; ***p < 0.001; n = 4). (B) Left panel: representative pictures of Matrigel invasion. Right panel: quantification of invading cells per filter (one-way ANOVA: ***p < 0.001; n = 4). Quantitative data are expressed as mean ± SEM. Scale bars, 200 μm.

Figure 7

ColXVII Endodomain Promotes Cell Proliferation (A) SCC-25 Ctrl, Kd, Kd-Endo, and Kd-Ecto cells were grown for 2 days in anchorage-independent manner, and the number of cells per colony was quantified (n = 3). (B) Number of S-phase cells was detected following cell cycle arrest by nocodazole treatment. 2 and 6 hr after nocodazole release (noc-release), cells were trypsinized, ethanol-fixed, and stained with DAPI. The DNA contents were analyzed by flow cytometry (one-way ANOVA: **p < 0.01; ***p < 0.001; n = 3). (C) Subconfluently grown cells were BrdU labeled for 4 hr. After trypsinization and ethanol fixation the cells were incubated with anti-BrdU antibody plus Alexa 488 and subjected to flow cytometry. Left panel: flow cytometric analysis for BrdU-labeled cells. Right panel: quantification of BrdU-positive cells (one-way ANOVA: *p < 0.05; **p < 0.01; ***p < 0.001; n = 3). Quantitative data are shown as mean ± SEM.