Identification of genes with adaptive expression results in distinct transcriptional states of tumor subclones

Authors

• *S Kumar
• *KM Conners
• JJ Shearer
• J Joo
• S Turecamo
• M Sampson
• A Wolska
• AT Remaley
• MA Connelly
• JD Otvos
• NB Larson
• SJ Bielinski
• VL Roger

Abstract

Cancer growth is an evolutionary process involving cells acquiring mutations, epigenetic reprogramming, and differentiation into distinct subclones. It is assumed that this clonal evolution is driven by the selection of cells exhibiting properties related to growth advantage, like immunoevasion and cell growth, which are phenotypic hallmarks of cancer. Studying the evolution of gene expression is critical to understand the driving selective pressures of these phenotypic traits. This is particularly important in the context of drug response, where treatment-induced effects modulate gene expression.

To investigate the role of selection on gene expression in subclonal evolution, we model changes in gene expression along trajectories defined by the evolutionary tree as Ornstein-Uhlenbeck processes. Our approach differs from prior approaches, like differential expression analysis, by jointly leveraging the evolutionary history of tumor subclones (inferred from mutation data) and single-cell subclone-specific expression data. Applying our model to sublines derived from a B2905 cell line revealed that sublines with different growth rates and immunotherapy treatment responses have distinct patterns of gene expression adaptation. Specifically, sublines that showed resistance to treatment had genes with adaptive expression related to immunoinvasion, whereas sublines that responded to treatment had genes with adaptive expression related to increased growth rate. Further, we observed that tumors derived from the parental line and then given anti-CTLA-4 treatment were enriched with genes identified to have adaptive expression corresponding with their treatment response. Together, our results suggest that the adaptivity identified is associated not just with individual sublines but also with the broader phenotypic characteristics.

Scientific Focus Area: Computational Biology

This page was last updated on Monday, September 25, 2023