Investigating the role of tissue-engineered fibroblast-derived extracellular matrix in oral squamous cell carcinoma progression

Introduction: Communication of cells with the surrounding extracellular matrix (ECM) is the foundation that regulates tissue homeostasis (Walker et al., 2018). During cancer progression this regulation is determined by interactions between tumour cells, stromal cells (such as fibroblasts and leukocytes) and other components of the microenvironment including the ECM (Weaver and Gilbert, 2017). The ECM in oral squamous cell carcinoma (OSCC) forms its own tissue-specific microenvironment capable of eliciting biochemical and biophysiological cues, playing key roles in cancer progression (Pantanker et al., 2016, Kumar and Hema, 2019). This architectural change is associated with increased ECM deposition, known as desmoplasia, predominantly generated by cancer associated fibroblasts (CAF) activated from resident fibroblasts by signals such as TGFβ-1 and SDF-1 secreted by OSCC cells (LeBleu and Kalluri, 2018, Calon et al., 2014). Although considerable evidence exists demonstrating a role for CAF in OSCC, little is known about their influence tumour: ECM interactions. These complex interactions are not recapitulated in 2D culture and therefore effort is required to tissue-engineer more complex models of the oral mucosa to study the behaviour of oral cancer cells in response to the CAF desmoplasia observed in OSCC progression. Objective: The aim of this research was to generate novel tissue-engineered 3D constructs using normal oral fibroblast (NOF) and CAF derived ECM, as native scaffolds and utilise these to develop full-thickness oral mucosa models that accurately model tumour: ECM interactions in models of OSCC progression. Methods: Responses of NOF and CAF to TGFβ-1 were assessed in short- and long-term cultures using qPCR, immunofluorescence and Western blotting. Culture conditions were optimised within a transwell culture system to stimulate production of native NOF- and CAF-derived ECM deposition. The composition and physical properties of the ECM was assessed by histology, Western blotting, transepithelial electrical resistance (TEER), scanning electron microscopy (SEM), and metabolism assays. Collagen width, linearisation and elongation in NOF- and CAF-derived matrices (DM) were also analysed using second harmonic generation microscopy (SHGM). Mechanical alterations were assessed using atomic force microscopy (AFM). The adhesion, migration and proliferation of cancer cells (H357) were assessed by live-cell imaging microscopy. Stromal-targeting drugs; Scriptaid (HDAC inhibitor) and Batimastat (MMP inhibitor) were used to investigate NOF and CAF responses in 2D and as NOF- and CAF-DM. Full-thickness oral mucosa models were produced by culturing normal (FNB6) or OSCC (H357) cell lines seeded onto 28-day developed fibroblast-derived matrices (FDM). Results: TGFβ-1 stimulation of NOF and CAF elicited elevated αSMA, FN1-EDA, COL1A1, and VCAN abundance up to 14 days in vitro, with CAF expression showing superior amplification of transcripts and protein abundance compared to NOF. CAF constitutively expressed αSMA observable for at least 7 days in vitro. Optimisation of conditions supported the long-term deposition of ECM by NOF and CAF over a 28-day culture period. NOF generated an organised matrix with a thickness of 109 ± 24.08 µm compared to CAF which produced a significantly thicker (370 ± 13.44 µm 350 µm) highly irregular matrix (p <0.05). Immunoblotting for αSMA, FN1-EDA, VCAN, COL1A1, and LAMB-γ5 was significantly increased in CAF-DM compared to NOF-DM (p <0.001). The physical features of the ECM, assessed by SHGM, showed significant increases in collagen width, length, and linearisation in CAF-DM compared to NOF-DM (p <0.01). H357 cancer cells, exhibited elevated migration, proliferation and reduced adhesion on CAF-DM compared to NOF-DM. The suitability of fibroblast-derived matrices (FDM) was assessed to test the effects of stromal-targeting drugs and validated their use as model systems to further assess CAF/ECM modifications. Full-thickness oral mucosa equivalents were generated by culturing FNB6 and H357 cells seeded onto the 28-day developed FDM, resulting in models histologically resembling normal and cancerous tissue ex vivo and are the first of their kind all human-cell derived oral mucosa equivalent models, supported by native ECM substrates. Immunohistochemical analysis revealed differential expression of various ECM components; evidence of altered matrix proteomics during disease progression in OSCC. Conclusion: Using tissue-engineering techniques it is possible to model fibroblast-mediated ECM deposition providing a novel, physiologically relevant in vitro tool to study tumour: ECM interactions in OSCC.