Amazing size, ambitious goals

Supported by the National Center for Research Resources of the National Institutes of Health, the National Center for Macromolecular Imaging is currently working on over 25 collaborative projects with institutions all over the world, including seven Alliance partners. A few of the particles that require NCSA's Origin2000 to attain a high-resolution, three-dimensional model are: a lens protein of the human eye, known as alpha-crystallin that can cause cataracts as it ages; fatty acid synthase, which builds fatty acids in the human body; and the herpes simplex virus, which causes a variety of disorders such as cold sores, chicken pox, and even congenital defects.

The tiny size of these particles is amazing, and the team's modeling goals are ambitious. The herpes simplex virus is only 1250 angstroms in diameter—an angstrom is one ten-billionth of a meter, about the size of a single atom. Earlier this year the center's researchers used their own Origin2000 to create a model of the virus with a resolution of 8.5 angstroms. Still, the team would like a model with twice the resolution. (A report on the current herpes simplex model was published in an April issue of Science in collaboration with Hong Zhou at the University of Texas at Houston Medical School and Frazer Rixon at the Medical Research Council Institute of Virology in Glasgow, England.)

Typical electron cryomicroscopy micrograph used in the reconstruction of a herpes simplex virus capsid. The dashed circle indicates one herpes simplex virus capsid.

At only about 140 angstroms in width, alpha-crystallin is ten times smaller than the virus. Researchers—including Phil Baldwin, Irina Serysheva, Steve Ludtke, and Florante Quiocho at Baylor College of Medicine and Mark Petrash at Alliance partner Washington University—have already created a model with a resolution of 20 angstroms and would like to increase the resolution to about 10 angstroms. At 10 angstroms, they should be able to discern not only the 24 subunits that make up alpha-crystallin but also how these subunits are arranged with respect to one another. Using this information, researchers may be able to shed some light on why degradation of the alpha-crystallin lens protein leads to cataracts.

With support from the National Science Foundation, NCMI recently installed a 300 kilovolt electron cryomicroscope, which can image biological particles with maximum contrast and at very high resolution. And the better the images collected with the microscope, the better the models that can be built from those images. The most powerful microscope of its kind in the United States, this instrument is a type of transmission electron microscope that has a highly coherent electron beam and looks at samples kept at liquid helium temperature, or about -270 C.

The amount of computing power necessary to get the kind of resolution that researchers want in their models, however, is beyond the capacity of their own 32-processor Origin2000.

"It's gotten to the point that we can't do it on our own supercomputer," says Chiu. "To push the resolution of our model of the herpes virus capsid particle from 8.5 angstroms to 6 angstroms sounds like a small improvement, but the computational task requires 50 times more effort in terms of data size."