Faculty Interaction

Alan Friedman

Departments of Biological Sciences
Purdue University

Brief Project Description

 

X-ray (and neutron) scattering of proteins in solution can provide valuable structural information. While not approaching the level of structural information provided by the classic techniques of crystallography and NMR, solution scattering can provide shape information rapidly on relatively small amounts of protein without the need for the often rate limiting step of crystallization. Under NSF award IIS 0502801, our collaborative team has proposed that shape information from solution scattering can be combined with other kinds of rapid experimental data (e.g. residue-specific crosslinking) for discriminating among alternative predicted models of protein structure.

The information from solution scattering would be greatly enhanced if it were possible to localize one or more amino acids in the shape envelope deduced from the scattering data. We have begun to investigate the use of heavy atom reagents covalently attached to particular amino acids as a probe for the location of these amino acids in scattering experiments. In one experiment, for which we have already collected scattering data at the Advanced Photon Source, two amino acids of the pTfa protein of bacteriophage lambda (a convenient experimental system) were mutated to cysteine and then reacted with an undecagold maleimide cluster reagent. Solution scattering data collected from this sample shows substantial differences from the maleimide only control and the signature of substantial scattering from the cluster of eleven gold atoms.

 

This data is being analyzed through standard methods (P(r) calculations, equivalent to the Patterson self-correlation function in crystallography) and by simulating the scattering expected from this molecule with the undecagold clusters in various locations by rigid-body modeling. Because they require the rotation and translation of two large gold clusters (relative to a fixed protein model) followed by the calculation of atom-to-atom distance histograms these simulations are quite computational extensive and have taxed the resources available at Purdue. Teragrid computing hardware resources and Matlab distributed engine at NCSA were used to complete the initial calculations done at Purdue and extend them to a finer translational and rotational resolution. The code we have developed this simulation is being run at Purdue, NCSA, TACC, and IU. Successful completion of these simulations will point the way to the application of solution scattering with labeled proteins to the rapid elucidation of protein structure. Extension of this method to the larger protein-protein complexes that form the frontier of structural biology can readily be envisioned.

 

Basic scheme of rigid body modeling

Cubic translational grid points are generated around the protein. Protein is fixed and two gold clusters are translated on the cubic grid points and rotated by Eulerian angular system, which gives trial structures. Simulated scattering or pair-distance distribution curves of trial structures are compared with that of template structure by scoring functions.

 

 

Last updated 12/18/07