(4q) Protein Functional Dynamics

Bowman, G., University of California, Berkeley

It is commonplace to assume a protein exists in a single structure when studying protein function or designing novel proteins. However, this single structure approximation often leaves us unable to rationalize the effects of mutations on function, much less to design proteins. Overcoming these limitations will require a rigorous understanding of the structural fluctuations proteins undergo at equilibrium and their role in a protein’s function. Combining computation and experiment to understand protein functional dynamics and incorporate this structural flexibility into design protocols will be the core of my research program.

In this poster, I present some of my initial results on identifying new means of manipulating a protein’s function, which I have generated as a Miller Research Fellow at UC Berkeley. This work builds off of methods I developed for building quantitative maps of a proteins conformational space. These models—called Markov state models (MSMs)—are discrete-time master equation models built from extensive molecular dynamics simulations using a combination of physics, Bayesian statistics, information theory, and network theory. In the past, I have used these models to advance molecular simulations from microsecond timescales to millisecond timescales. Now I am using Markov models to capture a representative sample of the ensemble of structures a protein visits at equilibrium and identify allosteric control points—regions of the protein that are distant from the active site yet can have profound effects on activity in response to perturbations like mutations or small molecules. I show that my methods identify known allosteric sites and identify new sites that I am currently testing via biophysical experimentation.