Kavraki Lab

Many large protein complexes undergo extensive conformational changes as part of their functionality. Tracing these changes is important for understanding the way these proteins function. It is not always possible to obtain a high resolution structure for very large complexes. While many conformational search methods explore the motions of atomic resolution protein structures, little has been done to handle the abundance of coarser resolution data available.

Constrained fit of model of mm-cpn (open state) to cryo-EM density

Traditional conformational search methods are impractical for very large complexes due to the amount of computational time involved. Moreover, they cannot be applied to coarser resolution data where structural information may be partial or missing.

To address this problem, we propose a novel computational methodology to efficiently trace the conformational changes in biological macromolecules represented as medium resolution structures. We develop and apply a search method from robotics to structural information. By describing a protein system as a generic robotic motion planning problem, exploring the extremely high-dimensional conformational space becomes computationally tractable. Here, primary structure elements are recast as articulated joints with protein dimensions further reduced by modeling certain stretches of the protein as rigid bodies.

Cartoon “roadmap” of protein conformational space with associated energy landscape

Our method is unique in its ability to conduct a computationally tractable search, using approximate data to obtain approximate but reliable results. The pathways obtained by this method can be useful in understanding protein motion and functionality. To provide a baseline test for our method, we tested a simple prototype of our methods—considering atomic resolution models derived from previously performed cryo-EM experiments (not lower resolution cryo-EM data itself yet)—on Adenylate Kinase and the GroEL monomer. We show that we can produce low energy conformational pathways with accuracy well below the structure’s resolution level. This prototype is a promising first step towards exploring the conformational motion of even larger complexes.

Path for a GroEL monomer

This work is done in collaboration with Dr. Steve Ludtke and Dr. Wah Chiu from Baylor College of Medicine.