Oncogenic PI3Kα variants reveal graded conformational spectrum with mutation-specific cryptic pockets

Abstract

Cancer-prone alleles exhibit single hotspot mutations. However, the combination of a cancer hotspot and a weak or moderate mutation (the one-two punch hypothesis) produces same-allele double variants with a significantly different and potentially graded clinical phenotypic spectrum. Oncogenic PI3K variants, which are also associated with benign tumors and neurodevelopmental disorders, offer statistical support for this model. Using atomistic molecular dynamics (MD) simulations, we revealed that PI3K variants with single and double mutations exhibit expanded conformational profiles. Double mutations significantly shift the conformational ensembles toward the active form--a more pronounced effect than a single mutation. These double mutants facilitate nSH2 release, iSH2 shift, and A-loop protrusion in solution, promoting PIP2 substrate recruitment at the membrane. We discovered that observable, potentially drug-targetable cryptic pockets are PI3K mutation-specific. Ab initio discovery of these pockets through simulations is essential for AI-aided virtual drug screening. A key challenge is that a single drug is often ineffective against PI3K variants due to their diverse conformational spectra. To address this, we propose a conformational selection strategy involving a combination of allosteric drugs for variants with graded conformational spectra, particularly those with strong double mutations; we identified such potentially targetable cryptic pockets in double mutants conformers.