Home | Professor G. R. Marshall | Faculty & Staff | CMD | International Acclaim | Publications | The Center For Computational Biology

  CURRENT PROJECTS

   REVERSE-TURN RECOGNITION AND MIMETIC PROPENSITY

Reverse turns are common motifs in protein structure and have been implicated as recognition elements in structure-activity studies of the peptide hormones, angiotensin II, bradykinin, GnRH, somatostatin, and many others. Despite many efforts to design turn mimetics, their application to SAR studies of peptides has been relatively infrequent due in part to complicated multistep syntheses which limit the incorporation of sidechain groups into the turn mimetic. We propose to continue the design and synthesis of novel reverse turn mimics derived from simple dipeptides. Computational tools will be used to predict the reverse-turn propensities of potential mimetics prior to their synthesis. These tools will be experimentally validated using a simple mode, gramicidin S, into which turn mimics will be incorporated. Detailed structural analyses (NMR, FTIR, CD) will be used to characterize the populations of turn structures in this model peptide. Turn mimics will be incorporated into well-characterized turn-containing peptides followed by extensive structural characterization to assess the local geometry at the mimic and any conformational effects that propagate from the turn mimetic along the peptide backbone. The effect of the local environment on turn propensity for each mimic - solvent, adjacent residue type, peptide length - will be probed in the gramicidin S analogs. In the final phase, detailed information about the turn mimics derived from the earier studies will be applied to understanding the conformational link between bradykinin agonist/antagonists and bradykinin itself through use of receptor mutants. (News Release)

   MECHANISMS OF AGONISTS AND ANTAGONISTS OF ANGIOTENSIN

Angiotensin is a peptide hormone involved in the regulation of vascular reactivity and volume homeostatis. Its receptor is a member of the G-protein couple receptor family well studied because of the pathological importance of the renin-angiotensin system. This project focuses on the chemical synthesis of the AT1 receptor in order to provide sufficient material for biophysical studies to determine the mechanisms by which various ligands stabilize the agonist and antagonist states of the receptor. Chemical ligation of transmembrane segments incorporated into lipid bilayers will be used to construct various fragments of the receptor. Synthesis will be used to incorporate various probes into specific locations in the receptor to facilitate biophysical studies to provide distance constraints for molecular modeling.

   CHARACTERIZATION OF THE RHODOPSIN/TRANSDUCIN INTERFACE

The focus of this proposal is to study the rhodopsin-transducin interface and the mechanism of light-activated signal transduction. Synthetic receptor fragments, peptides and peptide analogs derived from transducin and rhodopsin will be used as specific ligands for high-resolution structural studies (transfer NOE and EPR) when complexed to their respective partners. Receptor fragments will be prepared by chemical ligation and by expression ligation to provide "split" receptors for spectroscopic study. By defining the interacting surfaces during photoactivation, an increased understanding of the molecular mechanisms of rhodopsin-transducin complexation is sought. The bound conformation of the C-terminal complexation is sought. The bound conformation of the C-terminal undecapeptide of the a-subunit of transducin will be used to design non-peptide compounds as potential inhibitors of transduction. The information derived on the three-dimension structure of the rhodopsin-transducin complex can assist in devising approaches for treating human diseases (retinitis pigmentosa and night blindness) associated with constitutively active mutations in rhodopsin.

   SCORING FUNCTIONS FOR PROTEIN PREDICTION

The sequence of a protein contains the requisite information which specifies its unique three-dimensional structure. It should be apparent that a reliable solution to prediction of the spatial structure of proteins based on sequence alone would impact many aspects of molecular biology and therapeutic development, especially in view of the rapid generation of gene sequences by DNA sequence analysis. Prediction methods for generating the 3D structure of a protein based on its sequence alone fall into several categories. There are hierarchical methods which predict secondary structures and then attempt to fold those elements together. There are simulation methods which attempt to fold the protein using models of reduced complexity and then refine the prediction by using them to constrain all-atom models. If one considers molecular dynamics all-atom simulations with explicit solvation, then one is clearly a minimum of six orders of magnitude too slow (nanoseconds versus milliseconds at best) to simulate the folding process itself. It is clearly impossible for a protein to systematically search through the myriad of combinations of possible local conformations of the individual residues (the Levinthal paradox), and experimental data suggests that folding paths are followed with probable nucleation of secondary structure. This suggests a hierarchical approach to the problem; determine the secondary structure elements and then their three-dimensional arrangement.

Significant progress has been made in predicting secondary structure with accuracies approaching 75%. Studies by Dill, Skolnick and others have shown that much of the tertiary structural motif is embedded in requirements for compactness and the sequence of hydrophilic and hydrophobic residues. Current statistical potential functions have limitations which restrict them to two-body interactions. The contact matrix approach (CMA) offers computational efficiencies to explore configurational space and only generate candidates with the correct density. A stochastic approach which recognizes the errors associated with secondary structure prediction, utilizes the CMA to generate a set of plausible candidates, and screens these predicted alpha carbon structures with low-resolution potential functions before refinement of all atom models would appear a logical approach given these observations. Scientists at our Center for Molecular Design (CMD) have made considerable progress in this direction and this proposal focuses on scoring functions for backbone-resolution models of proteins as part of an integrated system for protein structure prediction.

 

©2006 by the Department of Biochemistry and Molecular Biophysics at Washington University School of Medicine.