Integrative analysis of proteomic, transcriptomic, and FACS-based surface marker data for ADC target discovery in cancer cell lines

In the preclinical space of cancer drug discovery, multi-omics characterization of cancer models is essential for model selection, understanding mechanisms of action, and early biomarker discovery. We launched a panel of 160 cell lines (CL Proliferation Panel) to study drug responses across a large variety of cancer types and validated their molecular characteristics at both the transcriptomic and genomic levels. Recently, we characterized this set of models at the proteomic level using mass spectrometry, identifying a total of more than 15.000 proteins. In the present work, we aim to investigate the relevance of this new dataset in the daily practice of cancer model utilization. First, we demonstrated the robustness of the established proteomic profiles, showing high concordance among duplicate samples. Using a multiomics comparison and dimensionality reduction approaches, we curated the proteomic dataset to remove outlier and irrelevant protein profiles, ensuring high data quality. Next, unsupervised hierarchical clustering of the proteomic profiles revealed accurate classification of models according to their tumor type of origin, confirming the relevance of the data. Given the growing importance of proteomic information in the context of antibody-drug conjugate (ADC) development, we further assessed the relevance of our dataset for evaluating the expression of top 20 ADC targets such as ERBB2 and TACSTD2, among others. We also established a multi-omics strategy integrating transcriptomic and proteomic data to enhance target characterization. Finally, we explored the relationship between protein expression of the targets and drug sensitivity by analyzing ADC targets in the context of responses to clinically approved ADCs, including trastuzumab emtansine (Kadcyla), and sacituzumab govitecan.