Deciphering the selectivity of CBL-B inhibitors using all-atom molecular dynamics and machine learning

We employ a combination of accelerated molecular dynamics and machine learning techniques to unravel the dynamic characteristics of CBL-B and C-CBL, and how their configurational changes conferring the binding affinity and selectivity of their ligands. We demonstrate that the activity and selectivity against CBL-B and C stem from subtle structural disparities within their binding pockets, and dissociation pathways. Our predictive model for dissociation rate constants (koff) demonstrates a moderate correlation with experimental IC50 values, effectively aligning with two available experimental koff values. Moreover, the binding free energies calculated using MM/GBSA highlight the ΔG distinction between CBL-B and C-CBL. By employing a regression strategy on dissociation trajectories, we identified key amino acids in binding pocket and along the dissociation path responsible for activity and selectivity. These amino acids are statistically significant in achieving activity and selectivity and correspond to the primary structural discrepancies between CBL-B and C-CBL. Through microsecond-scale replica exchange molecular dynamics coupled with generative model of molecular generation and ensemble docking, we accomplish comprehensive simulations of the complete apo-holo-apo transformation cycle. This approach provides an enabling first-in-class drug design technology based on apo-to-holo structure transformation.