Structure and Function

The masses of sequencing information that now confront genomic scientists raise a huge question: Exactly what do the products of these genes do? About 30 genomes have been completely sequenced, and up to 100 will be done by year's end, perhaps including a roughly finished sequence of humankind's 100,000 or more genes. Sequence data can identify gene products involved in disease, but the challenge facing researchers is far broader than that. Somehow, they must characterize the biochemical functions of all proteins. Prominent American and international scientists considered this problem during a Jan. 11-15 meeting in San Diego titled "Quantitative Challenges in the Post-Genomic Sequence Era." Their discussions revealed a consensus taking shape on how to proceed, despite an awareness that the scope and complexity of the task defy a fully systematic approach. Speakers agreed that examining the three-dimensional (3-D) structure of gene products is a promising way to deduce function. Elucidating the structures of proteins permits comparisons between homologous organisms. This can provide clues regarding protein sequences that, in turn, can suggest potential biochemical functions. This area of study, structural genomics, has begun to flourish only in the past few years. Traditionally, the biochemical function of a protein is determined before either X-ray crystallography or nuclear magnetic resonance spectroscopy is used to analyze structure. "The traditional role of 3-D structure doesn't go away; it has a new role that complements the goals of biology," declares Gaetano T. Montelione, a professor of molecular biology and biochemistry at Rutgers University. "There are still a lot of critics [wondering] about whether this is a viable scientific endeavor," he acknowledges. "When we talk about structure, we're all very clear about what we mean," he adds, "but when we talk about function, it's a little more nebulous." Function must be characterized on different levels, Montelione explains, beginning with atomic mechanisms and biochemical functions, then rising to the phenotypical and cellular levels. Yet thorough analyses of function must integrate all these levels. Photo courtesy of Armand Tepper, Leiden University, The NetherlandsA two-layered histidine Three Goals Montelione identifies three major goals of structural genomics. The first, which is the basis for the others, requires a systematic analysis of the 3-D structures of all proteins within a given genome. That will require development of high-throughput technology, analogous to DNA sequencing. Already, the technology exists to create facilities with a potential output of one protein structure determination per day.1 The second goal is a systematic analysis of the folds involved in the process of proteins reaching their native state, the conformation needed for them to do their jobs. Based on such a "Folding Periodic Table," comprising the estimated 1,000 to 1,500 fold families in nature, models could be made from which the rest of a structure could be extrapolated. It has been estimated that one structure from each of 10,000 domain sequence families will have to be determined in order to sample all of the fold families. The third goal is to use protein models to infer the biochemical functions of novel proteins that then could be tested by assays--in other words, a structure-based, functional genomics. The technical challenge is to make function quantitative, declares biophysicist John Moult of the Center for Advanced Research in Biotechnology, University of Maryland. Accordingly, functions can be subdivided into components that are quantifiable. These components include binding partners, the modification of partners, and regulation. Other components include control of expression, concentration, and localization, as well as the phenotypic effect, which can be determined by knockouts. Until now, researchers have focused on individual proteins or systems, but to see the bigger picture, there must be large-scale integration of information, notes Shoshana Wodak, a Free University of Brussels biophysics professor. That requires linkage of functional interactions between genes and proteins, as well as the metabolic, regulatory, and signal transduction pathways, and the physical sites themselves. All this is necessary because the functions of genes and proteins are context dependent. "But only a small fraction of these networks are charted," she notes. Moreover, "This needs to be represented in the computer in a way that we can analyze it, using programs." Photo courtesy of Armand Tepper, Leiden University, The NetherlandsOxygenated active site of limulus hemocyanin, a protein closely related to tyrosinase Drug Targets Of course, a primary objective of determining protein function is to make more and better drugs. Currently,about 6,000 pharmaceuticals have been created, notes Gregory A. Petsko, a Brandeis University professor of biochemistry and chemistry. Each drug required the synthesis of an average of 6,001 compounds and took an average of 13 years and $360 million to develop, he adds. Thus far, only 440 drug targets have been identified for people. "That is less than one-half of one percent of the human genome." Petsko believes a good deal of "cognitive dissonance" exists regarding structural and functional genomics. For example, "The higher the organism is, the tougher it is to know what is a gene and what isn't." Not enough people are working on gene recognition methods, he says. Moreover, clone supplies need to be developed for humans, mice, and fruit flies, to avoid the need for frequent polymerase chain reaction amplification. Regarding function, he notes, "In a eukaryotic cell, you can't know what a gene does until you know where it is." Molecular histology will be one of the most important skills of the next two decades, he predicts, particularly since most genes code for more than one function. Despite the complexity of the field, there is hope that more simple, empirical laws underlying these systems and interactions will be discovered, as in the case of physics. "I think it's wise to look for simplicity," states Harold A. Scheraga, Cornell University professor emeritus of chemistry and chemical biology. Homology modeling may provide important clues to such principles, but comparisons of amino acid sequences are slow and might be misleading, because those sequences can be arbitrary. He counsels an emphasis on analyzing amino acid properties, which he believes will lead to correlations that will greatly improve homology modeling and increase the chances of finding underlying laws. "My view is that biology is in such a primitive state that we just do not have enough data to get to the simplicity," offers Sung-Huo Kim, a chemistry professor at University of California, Berkeley. "I think biology is really in a hunting and gathering mode, and I liken it to days when better observation will lead to better insights," says Stanford University physics professor Steven Chu. Even apparently simple reactions are proving to be more complex than was thought when looked at from different viewpoints with new tools. But such tools also may reduce the need for tortuously clever experiments to uncover what's going on. "I'm actually all for hunting and gathering with a new set of eyes, so we don't have to be so smart." California Institute of Technology chemistry professor Harry B. Gray sees no alternative to gathering structural information as a necessary prelude to determining how proteins work. "At this stage, I don't know what to do except go through this exercise," he says. "The structure is just the beginning." David Eisenberg, a chemistry and biochemistry professor at University of California, Los Angeles, looks back upon theories of energetics, stereochemistry, and Darwinism developed in the 19th century, and he is struck in particular by the influence of Darwin's ideas on molecular biology. He even sees principles of thermodynamics, for example, as less prevalent than families of sequences and structures that derive from Darwin-ism. "If only we could extend this [influence] to energetic principles in biology, we'd be so much better off." "I think we're just at the beginning of combining these 19th century principles," says Peter G. Wolynes, a professor of chemistry, biophysics, and physics at University of Illinois, Urbana-Champaign. He suspects that a new law will emerge at a higher level by combining the lower-level principles. Meanwhile, he says, "I don't think we have to wait for the perfection of calculations to understand the basic outlines of how things work." Steve Bunk (sbunk@uswest.net) is a contributing editor for The Scientist. 1. G.T. Montelione, S. Anderson, "Structural genomics: keystone for a Human Proteome Project," Nature Structural Biology, 6:11-2, 1999.

Structure and Function

Interested in reading more?

Become a Member of