Current research in the Green Lab is focused on two primary biological systems: (1) signal transduction through the heterotrimeric G-protein pathway; and (2) the initial steps in the recognition of target cells by the HIV-1 virus. Each of these systems is being used to explore two distinct general problems of biological interest. The G-protein signal transduction pathway is our focus for explorations of the specificity of protein–protein interactions, while HIV–cell recognition is our prototypical system for understanding the role of glycosylation in modulating the interactions of proteins. In each system, sub-projects are focused on specific questions.
Rational design of mutants that remove unsatisfied polar groups from the core of the virucidal lectin cyanovirin-N has led to a series of ultra-stable variants.
Over the past decade, a number of carbohydrate-binding proteins from diverse sources have been identified as having virucidal activity against HIV. While the details of the inhibitory mechanism are not fully known, a key component is the association of these lectins with carbohydrates on the surface of the heavily glycosylated envelope glycoprotein gp120. The anti-viral activity of these proteins provide a promising new direction for the development of novel virucides; as it is simple serendipity that these lectins inhibit HIV infection of new cells, it is likely that rational protein engineering could be used to develop more potent variants. Our work in this area focuses in two main areas. The first is the application of molecular modeling techniques, such as molecular dynamics simulations and continuum electrostatic analysis, to provide structural models for the mechanism of action. Secondly, we are actively pursuing the computer-aided design of novel variants both with enhanced stability and with enhanced affinity and specificity against particular carbohydrate targets. While our design work is driven by rational, computer-driven approaches, we addtionally have an effort to express and characterize our designs.
Many interacting proteins in mammalian cells belong to families of structurally-related families; the subunits of the G-protein heterotrimer are a good example, with over 20 α, 6 β and 13 γ variants in the genome. This raises the question of how cognate associations are specified. While some specificity can be derived from differential expression and sub-cellular localization, it is additionally clear that some specificity exists at the molecular level. We are actively pursuing a deeper understanding of the biophysical underpinnings of this specificity with two parallel approaches. First, we are using the tools of molecular simulation to decompose the energetic contributions to binding of each residue in the trimer; this work involves large-scale molecular dynamics simulations on the NY Blue supercomputer, followed by detailed analyses of both energetic and dynamic properties. Secondly, we are directly considering the specificity of subunit association by exploring the sequence-space of allowable interactions using the tools of computational protein. Using the Dead-end Elimination and A* algorithms, coupled with a hierarchical energetic-evaluation scheme, we are able to enumerate all low-energy states for a given complex — analysis of the distribution of sequences in this space gives direct insight into the specificity of association.
It is impossible to fully explain the function of any component of a non-linear signaling network solely by considering its behavior in isolation. Thus, while molecular-level studies can give profound insight into the origins of signaling specificity, to tell the complete story the network itself must be considered as well. To this end we have implemented a differential-equation-based dynamical model of G-protein-mediated signal transduction, and are using this to explore the relationship between promiscuous and/or non-specific biophysical interactions and subsequent responses by the system. We are additionally working to use a probablistic network model (Bayesian Network) to more completely define the network of possible molecular interactions in these pathways, taking experimental observations from the literature as an input.
By optimizing the radii used in continuum electrostatic calculations with carbohydrates, remarkable agreement can be obtained with solvation free-energies computed with explicit solvent free-energy perturbation (FEP). J. Phys. Chem. B 112: 5238-5249 (2008).
The research in the Green Lab is biologically oriented — we are primarily motivated towards solving the underlying problems of biology in the glycobiology of viral—host cell recognition and in the specificity of G-protein signal transduction. In many cases, current methods can be directly applied to these problems, but we additionally continue to work to improve these methods and to develop new techniques. Among the methodological work that we have done is: (1) the development of optimized parameters for the continuum electrostatic analysis of systems containing carbohydrates; (2) the development of a rigorous statisitical treatment of hierarchical methods in protein design; and (3) the development of methods for residue-based-energetic decompositions using a Generalized-Born formalism. In the last case, we have also undertaken a detailed comparison of the the Generalized-Born and Poisson–Boltzmann solvent models in describing the energetic contributions of specific residues.
This page is maintained by David F. Green <dfgreen@stonybrook.edu>. Last updated: Tue Jan 10 12:19:48 2012Copyright © 2003–2009 David F. Green. All rights reserved.