Dynamics in molecular recognition

点击翻译按钮获取中文摘要

Proteins movements form a continuum from bond and angle vibrations, side-chain rearrangements, loop or domain movements through folding. In recent years the appreciation of the importance of dynamics in protein function is growing. Our research in computational structural biology aims to understand molecular recognition, i.e., the factors affecting selectivity and specificity in molecular interactions, through the development of innovative methods and their experimental validation in four research axes: reconstructed metabolic networks, docking simulations, the detection of molecular similarities and the simulation of dynamics across multiple scales in macromolecules. The goals of our research program are: 1. To develop methods to simulate dynamics in macromolecules and apply them to understand protein function. We have recently developed the first normal modes analysis (NMA) method called ENCoM, able to account for the nature of amino acids thus permitting to assess the effect of mutations on dynamics. A second application for a NMA method is in the generation of conformational ensembles. We are developing TENCoM, a version of ENCoM where movements occur in torsional angle space (rather than Cartesian space) to generate geometrically correct conformational ensembles. Initial results show that ENCoM is particularly apt at predicting stabilizing mutations such as constitutively active or inactive mutations in G-protein Coupled Receptors (GPCRs). These programs will be used to study biased signalling in GPCRs and the dynamic effect of distal mutations on ligand binding.2. To account for the dynamic nature of proteins and its effect on molecular recognition. One goal is to introduce full protein backbone movements to our docking program FlexAID using normal modes. This will allow exploring the target conformational space more thoroughly in a computationally efficient manner enabling us to apply FlexAID in the search for allosteric small-molecule inhibitors as well as perform protein-protein docking simulations. Secondly, in our group we develop IsoCleft and IsoMIF, two graph-matching based methods for the detection of local 3D atomic and molecular interaction similarities respectively. These programs can be used to predict binding ligands, protein function as well as potential cross-reactivity targets. Currently, IsoCleft and IsoMIF consider flexibility only very simplistically via thresholds values in pertinent parameters irrespective of the actual local flexibility. Yet, the explicit consideration of flexibility through the use of Gaussian distributions derived from normal modes will permit us to treat flexibility in a realistic manner and thus detect similarities more accurately (e.g. loops).3. To validate experimentally our methods. The experimental branch of our group is responsible for cloning, expression and purification of proteins of interest where predicted binding small molecules can be validated using biophysical methods such as circular dichroism and differential scanning fluorescence. We are currently working on the Germination Protease of C. difficile as well as STK38 and BUB1, two human protein kinases involved in several types of cancer.This research program is unique in Canada and at the cutting edge of computational biology worldwide and will help us understand the effect of dynamics in molecular recognition. Furthermore, the multidisciplinary approach in our group integrating calculations and experiments allows for the formation of high quality personnel ready to work at the interface of chemistry, molecular biology, physics and computer science.