A multiple regression model-based chemical biology approach to dissect signal transduction pathways downstream of cytokine receptors.

In this paper, we propose a statistical model-based chemical biology approach to extract new biological insights from the compound activity database. We developed and characterised a small-molecule kinase inhibitor library for their ability to inhibit JAK1, 2 and 3 kinase in vitro. These compounds were also tested for their ability to suppress STAT1/5 phosphorylation induced by GM-CSF, interferon (IFN)-γ or interleukin (IL)-2 in primary human peripheral blood mononuclear cells (PBMC). Correlation analysis between the in-vitro and cell-based potencies of the inhibitors was performed by using a multiple-linear regression model. The regression p value for three JAK kinases confirmed the known individual contributions of JAK kinases to signalling pathway downstream of GM-CSF and IFN-γ receptor. Interestingly, the model suggests a previously under-appreciated role played by JAK2 downstream of IL-2 receptor activation. This study demonstrates the potential use of chemical biology approach in generating biological hypotheses when facilitated with proper statistical modelling techniques.