Sali and colleagues advance understanding of proteins

By UCSF School of Pharmacy Editorial Staff / Fri Jan 11, 2008

School of Pharmacy faculty member and computational biologist Andrej Sali, PhD, and international colleagues have developed new techniques to reveal the architecture of large protein complexes within cells. Their ultimate goal is to see how these complexes interact in real time—however fleeting the encounters. "The better we understand the architecture and interplay of protein communities inside our cells, the better we will understand our biology in health and disease," says Sali.

Their research appears twice in Nature, November 29, 2007, under the titles "Determining the architecture of macromolecular structures" and "The molecular architecture of the nuclear pore complex." The papers were rated by Nature editors as their favorite among 2007 cell biology papers. "Determining the architecture of macromolecular structures" was chosen as one of Chemistry and Chemical Engineering News 2007 chemistry highlights. It also appeared on December 26, 2007 as #1 among interesting papers published in the biological sciences by the Faculty of 1000 Biology. The second Nature paper appeared on the same day as the #2-rated paper by the same group.

"Andrej's work on protein architecture is one of those exciting, defining advances in science," says Kathy Giacomini, PhD, chair of the UCSF School of Pharmacy's department of biopharmaceutical sciences. "The ramifications for our understanding of biology are immense."

Sali and colleagues create technique to reveal the architectures of large protein assemblies

Jeff Norris, UCSF Office of Public Affairs

In a tour de force, a UCSF faculty member and computational biologist - in collaboration with his colleagues at Rockefeller University - has invented a new approach to solve a long-standing protein puzzle.

Imagine a puzzle that teams of scientists around the world have been trying to solve for decades, and the excitement within the science community the day the picture finally becomes clear.

Exciting is what that day and subsequent days have been like for UCSF computational biologist Andrej Sali, PhD, faculty member in the UCSF School of Pharmacy. Sali’s research team collaborated with the research groups of Michael Rout, PhD, and Brian Chait, DPhil, at Rockefeller University to develop a combined experimental and computational approach for determining macromolecular architectures. They have set scientists buzzing by at last figuring out how 456 protein puzzle pieces that make up the nuclear pore complex (NPC) fit together to form a key structure within living cells.

It is one of the biggest, most complicated protein assembly structures that researchers have ever determined at this level of detail. To help solve the structure, Sali and UCSF postdoctoral fellow Frank Alber, PhD, developed integrated modeling platform (IMP) software, which has proved to be a powerful new bioinformatics tool.

In a scientific world where it is becoming easier for researchers to accumulate a wealth of data, scientists are increasingly challenged to make the best use of it all. IMP illustrates the value of developing new computational tools.

Solving this grand protein puzzle, unlike finishing a cardboard puzzle picture, is not an idle accomplishment. IMP eventually might have lifesaving consequences by providing a new source of structural clues to use in molecular drug design, for instance.

By tackling such a large and complex structure as the NPC, and by finishing a puzzle that other researchers could not, Sali and colleagues have demonstrated IMP’s potential. With IMP, a worldwide community of scientists has gained a powerful tool to tackle a broad range of molecular puzzles of their choice.

“To understand the workings of a living cell, we need to know the architecture of its molecular assemblies,” Sali says.

In the cells of organisms ranging from single-celled yeast to humans, the NPC is the major gateway that guides molecular traffic in and out of a cell’s nucleus. Many molecular agents and messengers that pass through this portal help control the activation of genes, in turn shaping the cell’s actions and responses.

The UCSF researchers and their collaborators describe IMP and its use in determining the NPC structure in yeast in two reports in the November 29, 2007, issue of Nature. Online, these two reports have for weeks ranked among the most viewed in the scientific literature.

Sali and collaborators envision that IMP, coupled with recent advances in electron microscopy and other experimental techniques, could serve as the foundation for the unprecedented study of many molecular interactions throughout the cell on a grand scale.

Just as a picture-puzzle solver uses different types of information – colors, sizes, and shapes, for example – IMP translates different types of laboratory and modeling data into a common language to describe rules and constraints that limit the possible spatial relationships among proteins and their parts.

“The most important aspect of our approach is its potential to use simultaneously almost any conceivable type of information to determine structures,” Sali says.

The experimental data input includes measures of possible error or uncertainty. IMP uses these measures to help score structural solutions that meet the constraints. Given enough good data, researchers can use IMP-generated rules to transform a randomly strewn batch of protein puzzle pieces into one or just a few possible solutions.

IMP is highly scalable. Researchers can use it to solve large structures of hundreds of proteins. Or, with greater definition, they can use it to determine the structures of much smaller protein couplings, Sali says.

One established technique for determining protein structure in atomic detail is X-ray crystallography. However, the technique requires crystals of purified proteins, and not all protein combinations form crystals, especially when the structure of interest consists of multiple, different proteins.

Another technique, nuclear magnetic resonance (NMR) spectroscopy, also provides high-definition structural data. But when using NMR, researchers must break proteins into small pieces for structural analysis.

At the other end of the size spectrum, large molecular assemblies comprising many proteins can be outlined accurately using electron microscopy. However, these outlines provide only limited information about how individual proteins are arranged within the structure.

IMP accommodates data from all these standard methods, as well as data from unconventional techniques for determining structure. For instance, to help solve the NPC structure, Rout and Chait developed new protocols for a technique called “affinity purification,” and used them to more precisely determine which proteins are in proximity to each other within the larger protein complex.

IMP also can incorporate modeling data from other computational programs.

“We are making IMP freely available in the hopes that it will be used very broadly as a framework for efforts by other scientists,” Sali says. “Others will be able to develop additional software modules to handle new kinds of data that will work with IMP.”