Method to derive structure of protein components in biological nanomachines

By Ruth SoRelle, M.P.H.

Some day a biological researcher will sit down at a computer, extract the maximum amount of data from a publicly available structural database and generate a model of the structure of the protein he or she is studying.

That's not possible yet, but researchers led by Baylor College of Medicine moved a step closer recently when they developed a method that enables them to narrow down significantly the choices of possible configurations in which the molecules can exist.

Predicted model

In a report that appears online in the Public Library of Science Computational Biology journal, Wah Chiu, Ph.D., and Matthew Baker, Ph.D., of Baylor College of Medicine, David Baker, Ph.D., of the University of Washington and colleagues predicted the structure of a small herpesvirus protein using a combination of modeling and cryo-electron microscopy.

The predicted model is well supported by other biological data, said Frazer Rixon, Ph.D., a collaborating virologist from the Medical Research Council Virology Unit in Glasgow, Scotland.

"Traditional structural biological techniques typically look at single domains or small proteins and study their structure and function," said Baker, an instructor in the BCM department of biochemistry and molecular biology. Chiu is a professor in the same department and director of the National Center for Macromolecular Modeling at BCM.

However, Baker said, these techniques are harder to use for studying an assembly of proteins in a normal or biologically active state.

Molecular shape

Cryo-electron microscopy (cryo-EM), when coupled with approaches from computationally based structure prediction, can give the assembly's structure in a state close to that in which it exists in nature. But at the present time, the cryo-EM can only provide a low-resolution view of the proteins within an assembly.

Alternatively, structural models of small proteins can be generated by computational prediction methods based on the physics of proteins.

"The problem with the computational approach is that you generate thousands of models but you have no way of evaluating which one is the right model," said Baker.

However, Baker, Chiu and their colleagues can now evaluate all of these computational models in the context of a molecular "shape" based on cryo-EM.

Useful for drug design?

Exhaustive searching with thousands of models would have taken over two months of continuous calculations on the most advanced computers. However, using special techniques that determine how the potential models "fit" within a molecular "shape," they were able to winnow 20,000 potential structures down to several candidates in less than 10 minutes.

They were able to then identify the most likely structure for the core domain VP26, a structural protein in herpesvirus. Such a model may be useful for rational drug design in the treatment of herpesvirus, which infects as much as 80 percent of the world's population, causing diseases such as chicken pox, mononucleosis and facial/genital lesions.

Others who participated in the research include: Drs. Wen Jiang of Purdue University and William J Wedemeyer of Michigan State University.

Funding for this research came from the National Science Foundation and National Center for Research Resources at the National Institutes of Health.