Phenotype remodelling of HNSCC cells in the muscle invasion environment

Tumour invading muscle in head and neck squamous cell carcinoma (HNSCC) is often associated with destructive growth and poor prognosis. However, the phenotypic functions and pathological mechanisms of muscle-invasive cancer cells in tumour progress remains unknown. In this study, we aimed to investigate the phenotypic functions of muscle-invasive cancer cells of HNSCC and their potential crosstalk with tumour microenvironment.

In this study, we profiled the transcriptional programs of muscle-invasive HNSCC cell populations and characterized their EMT and myoepithelial-like phenotypes. Furthermore, our findings highlight the presence of muscle-invasion as a predictive marker for HNSCC patients. G0S2 as one of the markers of muscle-invasive cancer cells is involved in HNSCC intravasation, probably via VEGF signalling.

We obtained scRNA-seq data (SC) from GSE103322 (N = 18) and GSE181919 (N = 37), spatial RNA-seq data (ST) from GSE208253 and GSE181300 (N = 4), transcriptomics of human HNSCC samples from GSE42743 (N = 12) and GSE41613 (N = 97). Utilizing the TCGA-HNSC dataset, we conducted univariate and multivariate Cox analyses to investigate the prognostic impact of muscle-invasion in HNSCC, with validation in an additional cohort. Through Stutility and AUCell approaches, we identified and characterized muscle-invasive cancer cell clusters, including their functional phenotypes and gene-specific profiles. Integration of SC and ST data was achieved using Seurat analysis, multimodal intersection analysis, and spatial deconvolution. The

Our analyses of the TCGA-HNSC cohort revealed the presence of muscle-invasion was associated with a poor prognosis. By combining ST and SC, we identified muscle-invasive cancer cells exhibiting epithelial-to-mesenchymal transition (EMT) and myoepithelial-like transcriptional programs, which were correlated with a poor prognosis. Furthermore, we identified G0S2 as a novel marker of muscle-invasive malignant cells that potentially promotes EMT and the acquisition of myoepithelium-like phenotypes. These findings were validated through in vitro assays and chorioallantoic membranes experiments. Additionally, we demonstrated that G0S2-overexpressing cancer cells might attract human ECs via VEGF signalling. Subsequent in vitro and in vivo experiments revealed G0S2 plays key roles in promoting the proliferation and invasion of cancer cells. Conclusions: In this study, we profiled the transcriptional programs of muscle-invasive HNSCC cell populations and characterized their EMT and myoepithelial-like phenotypes. Furthermore, our findings highlight the presence of muscle-invasion as a predictive marker for HNSCC patients. G0S2 as one of the markers of muscle-invasive cancer cells is involved in HNSCC intravasation, probably via VEGF signalling.