De novo identification of universal cell mechanics gene signatures

Mechanical proprieties determine many cellular functions, such as cell fate specification, migration, or circulation through vasculature. Identifying factors governing cell mechanical phenotype is therefore a subject of great interest. Here we present a mechanomics approach for establishing links between mechanical phenotype changes and the genes involved in driving them. We employ a machine learning-based discriminative network analysis method termed PC-corr to associate cell mechanical states, measured by real-time deformability cytometry (RT-DC), with large scale transcriptome datasets ranging from stem cell development to cancer progression, and originating from different murine and human tissues. By intersecting the discriminative networks inferred from two selected datasets, we identify a conserved module of five genes with putative roles in the regulation of cell mechanics. We validate the power of the individual genes to discriminate between soft and stiff cell states in silico, and demonstrate experimentally that the top scoring gene, CAV1, changes the mechanical phenotype of cells when silenced or overexpressed. The data-driven approach presented here has the power of de novo identification of genes involved in cell mechanics regulation and paves the way towards engineering cell mechanical properties on demand to explore their impact on physiological and pathological cell functions.