Supplementary Figure legends

Supplementary Figure S1. Bar graph shows qPCR quantification of TGFB1 mRNA expression in keratinocyte cell lines isolated from normal human skin (NHK), human RDEB skin (RDEB-K), RDEB-cSCC (RDEB-SCCK), and non-RDEB-SCC (SCC25-K). RDEB-K and RDEB-SCCK expressed slightly, but not significantly, higher TGFB1 transcripts than NHK and SCC25-K.

Supplementary Figure S2. Double immunofluorescence staining of wild-type and RDEB mouse lungs. In RDEB mouse lung pan-cytokeratin staining (green) indicated metastatic colonization of RDEB-cSCC cells, whereas LOX expression (red) was seen in fibroblasts, which were co-stained with vimentin (middle panel). In contrast to RDEB mouse skin, TGFβ1 expression was absent in metastasis-bearing RDEB mouse lung (right panel). This absence highlights the importance of C7 deficiency-evoked tissue destabilization for altered bioavailability of TGFβ.

Supplementary Figure S3. TGFβ regulates matrix properties in RDEB mouse skin fibroblasts. A, Western blot with antibodies to collagen I. Representative blot shows a significant reduction of alpha-2 chain (130kDa) of collagen I in RDEB fibroblasts cultured in presence of SB505124 (10µM) compared to vehicle treated cells. The bar graph shows quantification of collagen I expression from the western blot (n=2). Value represents mean ± SD (**P<0.001). B, Western blot with antibodies to LOX. Addition of recombinant TGFβ1 (10ng/mL) to cultures of keratinocyte and fibroblasts isolated from wild-type and RDEB murine skin increased both the 50 kDa pro-LOX form and the 33 kDa mature LOX. Inhibition of TGFβ1 signaling with SB505124 (10µM) decreased both. Both wild-type and RDEB fibroblasts were more responsive to TGFβ dependent LOX expression and proteolytic activation than wild-type and RDEB keratinocytes.

Supplementary Figure S4. Double immunofluorescence stainings of pan-cytokeratin (green) and LOX (red, upper panels), and of vimentin (green) and LOX (red, lower panels) displayed increased LOX levels in the dermis of injured RDEB patient skin and in the cSCC microenvironment.

Supplementary Figure S5. Stromal stiffness activates mechanosignaling in RDEB tumors. A, The bar graph shows quantification of αSMA-positive areas (pixels) of untreated, tumor-primed and tumors of wild-type and RDEB mice. n=7 images quantified per condition, values represent mean ± SD. **P =0.0019 for wild-type untreated vs. RDEB untreated. **P =0.0024 for wild-type tumor-primed vs. RDEB tumor-primed. ***P<0.0001 for wild-type tumor vs. RDEB tumor. B, Double immunofluorescence staining on wild-type and RDEB mouse tumors with antibodies to integrin β1 and pFAKtyr397. In RDEB mouse tumors, β1 integrin (green) and pFAKtyr397 (red) expression are colocalized (merge, yellow) at the plasma membrane of invading tumor cells (arrows). Immunofluorescence staining with antibodies against pAKTser473 (green) show nuclear, cytoplasmic and extracellular expression of pAKTser473. C, RDEB patient tumors exhibit increased staining of integrin β1 (green), pFAKtyr397 (green), and elevated nuclear, cytoplasmic and extracellular expression of pAKTser473 (green), suggesting activated mechanosensing signaling. Nuclei counter stained with DAPI (blue). Scale bar = 50µm.

3