Multi-State Coarse Grained Modeling for Intrinsically Disordered Peptides

Many proteins display a marginally stable tertiary structure, which can be altered via external stimuli. Since majority of coarse grained (CG) models are aimed at structure prediction, their success for an intrinsically disordered peptide's conformational space with marginal stability and sensitivity to external stimuli cannot be taken for granted. In this study, by using the LKα14 peptide as a test system, we demonstrate a bottom-up approach for constructing a multi-state CG model, which can capture the conformational behavior of this peptide in three distinct environments with a unique set of interaction parameters. LKα14 is disordered in dilute solutions, however it strictly adopts the α-helix conformation upon aggregation or at a hydrophobic/hydrophilic interface. Our bottom-up approach combines a generic base model, unbiased for any particular secondary structure, with nonbonded interactions which represent hydrogen bonds, electrostatics and hydrophobic forces. We demonstrate that by using carefully designed all atom potential of mean force calculations from all three states, one can obtain a CG model, which behaves intrinsically disordered in bulk water, folds into an α-helix at an interface or a neighboring peptide, and is stable as a tetramer.