Investigating Targetable Mechanisms of Cisplatin Resistance in Osteosarcoma Using Functional Genomic Screening and Multiomic Profiling of in Vitro Models.

Abstract

Introduction: Resistance to chemotherapy is an independent predictor of relapse and lower overall survival in osteosarcoma. However, mechanisms of chemotherapy resistance are poorly understood with no biomarkers in clinical use, and no effective medical therapies to offer for chemorefractory disease. We sought to identify and model mechanisms driving resistance to cisplatin in osteosarcoma, which could be potentially targeted to enhance tumour response to chemotherapy. Method(s): We utilised the Calabrese Set A P65-HSF CRISPR activation (CRISPRa) library to functionally interrogate the whole genome of a cisplatin-sensitive osteosarcoma cell line, to identify genes that mediate cisplatin response. We then designed a custom CRISPRa library covering 1271 genes which we utilised to screen low-passage patient-derived osteosarcoma cell lines to identify mechanisms of resistance that are highly conserved. We validated our findings against transcriptomic and clinical data from the 'TARGET OS' paediatric osteosarcoma dataset, and other human osteosarcoma cohorts. We also compared our results against multiomic studies on our four unique in vitro models of cisplatin resistance which have been characterised through whole genome sequencing, RNA sequencing and proteomics. Result(s): Our whole genome CRISPRa screen identified 1319 genes that mediate cisplatin response. Of the enriched genes, >30 are potentially 'druggable', including genes involved in DNA repair, the NRLP3 inflammasome, WNT/s-catenin signalling, glutathione transferases, PI3K/AKT/MTOR and EGFR signalling. Ten of the genes we identified using our functional genomics approach have been associated with poor outcomes when overexpressed in human osteosarcoma cohorts, including the cellular stress and autophagy regulator HMGB1, glutathione transferase GSTP1, the multidrug-resistance efflux pump ABCB1, and ESPN, a potent activator of PI3K signalling. Furthermore, custom pooled screening demonstrated that many of the resistance mechanisms identified through our whole genome screen are conserved across different patient-derived osteosarcoma cell lines. Complementing our findings, our cisplatin resistance models demonstrate recurrent mutations in genes involved in differentiation and development including the epithelial to mesenchymal transition (EMT) pathway and WNT/s-catenin signalling. Conclusion(s): By integrating functional genomic screening with clinical data and our in vitro modelling techniques, we have comprehensively identified potentially targetable molecular drivers of cisplatin resistance in osteosarcoma. Many of the genes, or associated signalling pathways, are overexpressed in human osteosarcoma tissue, which adds translatability to our findings and highlights these genes as potential biomarkers. Future work will utilise drug screening and in vivo murine models to identify targeted therapies that could be used to synergistically to enhance tumour response to chemotherapy.