Statistical mechanics of helix bundles using a dynamic programming approach

BIOT 226

Adam Lucas, alucas@maxwell.ucsf.edu1, Liang Huang2, Aravind Joshi2, and Ken A. Dill3. (1) Pharmaceutical Chemistry, UCSF, 600 16th street, San Francisco, CA 94143, (2) Department of Computer and Information Science, UPenn, (3) Department of Pharmaceutical Chemistry, UCSF
We develop a statistical mechanical theory for the equilibrium folding of helix bundle proteins and peptoids. A longstanding problem has been how to treat local interactions in chain molecules, such as those that dominate helix-coil processes, together with non-local interactions, such as those that dominate collapse processes. We use a dynamic programming algorithm to efficiently compute the partition function for a 3-helix bundle homopolymer, modeled on a cubic lattice.

We determine energy parameters to match experimental thermodynamic parameters and denaturation curves. Our energy function is based on three parameters, a helix interaction energy, a hydrophobic contact term and a three-body interaction energy. A three-body interaction term is necessary to capture calorimetric two-state cooperativity. We find a helix to hydrophobic interaction energy ratio of 1.3:1. Interestingly, the model predicts the balance of forces that explains why two-helix bundles are not observed in proteins, and yet are stable states of peptoids.