Drugging Disordered Proteins by Conformational Selection to Inform Therapeutic Intervention

Drugging intrinsically disordered proteins (IDPs) has historically been a major challenge due to their lack of stable binding sites, conformational heterogeneity, and rapid ability to self-associate or bind nonspecific neighbors. Furthermore, it is unclear whether binders of disordered proteins (i) induce entirely new conformations or (ii) target transient prestructured conformations via stabilizing existing states. To distinguish between these two mechanisms, we utilize molecular dynamics simulations to induce structured conformations in islet amyloid polypeptide (IAPP), a disordered endocrine peptide implicated in Type II Diabetes. Using umbrella sampling, we measure conformation-specific affinities of molecules previously shown to bind IAPP to determine if they can discriminate between two distinct IAPP conformations (fixed in either α-helix or β-sheet). We show that our two-state model of IAPP faithfully predicts the experimentally observed selectivity of two classes of IAPP binders while revealing differences in their molecular binding mechanisms. Specifically, the binding preferences of foldamers designed for human IAPP were not fully accounted for by conformational selection, unlike those of β-breaking peptides designed to mimic IAPP self-assembly sequences. Furthermore, the binding of these foldamers, but not β-breaking peptides, was disrupted by changes in the rat IAPP sequence. Taken together, our data quantify the sequence and conformational specificity for IAPP binders and reveal that conformational selection sometimes overrides sequence-level specificity. This work highlights the important role of conformational selection in stabilizing IDPs, and it reveals how fixed conformations can provide a tractable target for developing disordered protein binders.