PROJECT SUMMARY Head and neck squamous cell carcinoma (HNSCC) is the sixth leading cause of cancer death. Current limitations in our understanding of intratumoral cellular heterogeneity (ITH) in HNSCC cause major challenges in patient stratification and development of novel targeted therapeutics. Our seminal single cell RNA-sequencing (scRNA- seq) analysis of oral cavity squamous cell carcinoma (OSCC) revealed a partial epithelial-to-mesenchymal transition (p-EMT) program that is a key feature of ITH in HNSCC. p-EMT is characterized by expression of a unique set of mesenchymal markers, without loss of epithelial markers or expression of classical EMT transcription factors and is associated with invasiveness and treatment failure. We demonstrated that transforming growth factor beta-3 (TGF-β3) from cancer associated fibroblasts (CAF) drives p-EMT via cancer cell TGFβ-induced (TGFBI), a principal marker of the unique subpopulation. As in vitro efficacy of TGFβ inhibitors targeting EMT has not translated to the clinic, there is a clear need for better models of in vivo biology. Current modeling of p-EMT is hindered by the inability of OSCC cell lines in monolayer culture to capture ITH or p-EMT. In contrast, three-dimensional patient derived organoids (PDO) maintain differentiation gradients, recapitulating cancer cell heterogeneity. We will use OSCC PDOs and CAFs to investigate the role of the novel TGF-β3-TGFBI axis in driving p-EMT and the value of p-EMT as a predictive biomarker for response to therapy. First, to determine the mechanism by which cancer cell TGFBI regulates p-EMT to promote tumor growth and metastasis, we will use CRISPR interference (CRISPRi) to knock down TGFBI in OSCC PDOs and evaluate the impact on p-EMT. To model the functional impact of this knockdown, we will orthotopically transplant TGFBI- modulated PDOs into the buccal mucosa of immunodeficient mice, anticipating that TGFBI knockdown will reduce p-EMT, tumor formation, and metastasis. Second, to determine the mechanism by which CAF influence OSCC p-EMT via the TGF-β3-TGFBI axis, we will use CRISPRi to knock down TGF-β3 in OSCC CAFs and co- culture modified CAFs with OSCC PDOs. To assess the impact of TGF-β3-TGFBI signaling on xenograft tumor growth and metastasis, we will then orthotopically co-transplant modified CAFs and PDOs into NSG mice. Third, to determine how p-EMT predicts therapeutic response, we will treat TGFBI-modified PDOs, with or without admixed TGF-β3-modified CAFs, with a series of standard OSCC chemotherapies, as well as inhibitors of the TGFβ pathway. To support the use of p-EMT as a predictive biomarker, we will then treat our large biobank of OSCC PDOs with this same set of therapies and determine the relationship between p-EMT expression and therapeutic response. We anticipate that p-EMT, induced by the TGF-β3-TGFBI axis, will be associated with differential responses to standard and TGFβ-targeting therapies, supporting its use as a predictive biom...