Pore Picture Construction
By computationally combining incomplete imaging information with bits and pieces of structural data from all sorts of different experiments, researchers have worked out the protein-by-protein structure of an important cellular assembly called the nuclear pore complex.
Like puzzles? Here’s a tough one: Try figuring out the construction of a nearly 500-piece machine without blueprints or a complete picture. Biologists have now accomplished just such a feat, working out the protein-by-protein structure of an important cellular assembly called the nuclear pore complex. Their success depended on computationally combining incomplete imaging information with bits and pieces of structural data from all sorts of different experiments.
“It is as if we use many tiny lights, each of which shines from a different perspective, to illuminate every part of the whole structure,” says Andrej Sali, PhD, a coauthor and professor of biopharmaceutical sciences and pharmaceutical chemistry at the University of California, San Francisco. “We are able to use information from many sources, even sources that haven’t been traditionally used for structure determination.” As described in Nature in November 2007, this gleaning strategy should be helpful in determining the structures of many hard-to-pin-down cellular complexes.
The nuclear pore complex (NPC) is a gatekeeper of the nucleus, a 456-protein assembly in the shape of a thick donut spanning the nuclear membrane. Scientists know the general structure of eight spokes that make up the donut and have a roster of its proteins. But where each protein fits has been difficult to pin down.
The challenge is that no one tool images the details of complexes this size. Electron microscopy reveals the overall shape and outline but not individual proteins. NMR spectroscopy and X-ray crystallography, on the other hand, show individual proteins in stark relief, yet can’t be used on the whole assembly.
In collaboration with two groups of experimental biologists at Rockefeller University led by Michael Rout, PhD, and Brian Chait, PhD, Sali’s team found a way of combing structural information from disparate sources and computationally putting all the bits together to create a low-resolution image of the entire complex. They included experimental data from, for example, affinity purification assays (which indicate interactions between proteins) and ultracentrifugation (which reports on protein shape). At least seven different experimental techniques were used to produce structural data.
With the data in hand, they translated each piece into a “spatial restraint”—a mathematical probability of the structure’s geometry. For example, one restraint might indicate that protein A very likely interacts with protein W. Then the computer, starting with a random configuration of the proteins, moved them step-by-step in a direction that minimized violations of restraints. This process was repeated until the group had acquired 1,000 optimized structures that each satisfied the restraints. (In total, that took 200,000 trials run on 200 CPUs for 30 days.) The small variations between those 1,000 structures were then combined into a slightly blurry final image.
Sali was struck by the simplicity of the final structure. “If you look at electron scanning microscope pictures of the nuclear pore complex, and imagine how many proteins are involved, you think, ‘This is a mess! How did this evolve?’” But once the scientists began analyzing the protein architecture, they noticed a number of symmetries and a simple three-layer architecture: one layer to hold the pore to the membrane, one layer to facilitate transport of molecules through the pore, and a final scaffold layer to hold it all together. “It is not hard to imagine the evolution,” Sali says.
Establishing the protein architecture is also a huge step in coming to a better understanding of how the NPC facilitates controlled transport of molecules in and out of the nucleus. The group of Klaus Schulten, PhD, director of the theoretical biophysics group at the University of Illinois at Urbana-Champaign, is using this structure to study the transport mechanism. “The recipe that the investigators found for combining many experiments into one picture worked so consistently and so coherently across many independent trial predictions that the results must be true,” Schulten says. “Already now the relatively low resolution structure helps us to understand much better how the NPC organizes its complex function.” Chait, Rout, and Sali are now working on a high-resolution structure, with detail down to the atomic level.