Database Crystallizes Protein Engineering Studies

NIST now offers crystallographers a way to let their computers do the work with a database of crystal structures WASHINGTON - Ask a biologist about crystallization - the process of precipitating crystals out of solution - and you're apt to hear it described as a black art requiring patience and a special touch. But a databank that became available last month is expected to make the process more efficient, less time-consuming, and perhaps one day even less magical. The National Institute of Standards and Technology (NIST) is selling the databank, the brainchild of NIST fellow Gary Gilliland, as a software program that anyone with a personal computer can use. The unique product is the first standard reference data set on biotechnology that's available from the institute. Load it up and it will give you crystal data and crystallization conditions for more than 1,000 crystal forms of 600 biological macromolecules. By the end of the year, it will contain the same data for 2,000 crystal forms of more than 1,000 macromolecules. Though it just became available in mid-April, demand is expected to be heavy. Already, NIST officials have received some 500 requests for more information after conducting a brief demonstration last summer at the Third International Conference on Protein Crystallization in Maryland. Some researchers - sight unseen - have already given it rave reviews. "I think it's a super idea. It's fantastic," says Janet Smith, chairman of biological sciences at Purdue University and head of the American Crystallographic Association's special committee on biological macromolecules. As a child Gary Gilliland didn't try to blow up his garage with his chemistry set. Nor did he, as former Surgeon General C. Everett Koop once admitted, try to operate on cats. "I didn't do anything exciting when I was young," he says with a smile. Instead, the 41-year-old Gilliland, a NIST fellow, has had a steady, lifelong interest in science that spurred him to major in chemistry at Ohio University and, several years later, to earn a doctorate from Rice University. That interest outshone the lure of full-time work with computers. It also survived a stint in the military. Gilliland served with the Army in Korea from 1968 to 1970, then returned to computer business school for a year. "But I didn't enjoy that," he decided, so it was back to Ohio for a career in science. Gilliland joined NIH in 1979 to work with David Davies, who was studying the structure of immunoglobulins and other immune proteins. Gilliland's job involved purifying and trying to crystallize proteins. In 1983 Gilliland moved to Nutley, N.J., to set up a crystallization lab for Hoffmann-LaRoche Inc., then returned south to Gaithersburg, Md, for a year with the Genex Corp. before joining NIST in 1986. When not hard at work on proteins, Gilliland spends time with his wife and two daughters, ages 8 and 15. And when not spending time with them, he pursues what he considers another complex occupation: golf. --J.Z. Gilliland started compiling crystallization minutiae in 1980 while working with immunoglobulin at the National Institutes of Health's National Institute of Diabetes and Digestive and Kidney Diseases. He says he's happy that the product has finally been released. "It involved a lot of work, there's no doubt about it," he says. Chemists have been growing crystals in their labs for hundreds of years. But it was only in this century that scientists found the means - X-rays - to peer inside a crystal's molecular and atomic structure. Crystallization and X-ray crystallography are now essential to biotechnology and the study of proteins. By precipitating protein crystals out of solution and bombarding them with X-rays, scientists can plot the protein's atomic structure in three dimensions. Opportunities for crystallization scientists in industry have grown rapidly over the past decade as companies try to design new drugs. Some of these drugs, such as renin inhibitors for treatment of hypertension, may be on the market in a few years. Yet scientists need crystals if they're going to do crystal studies, and that's where the bottleneck is. Sufficient quantities of protein can be made, but there are many proteins no one knows how to crystallize. Those that scientists know how to make may require months to precipitate into crystal form, and even then a scientist can't be sure the process will work. Gilliland wanted to eliminate some of that guesswork. He reasoned that it would help to know if someone had already crystallized a similar substance, and how it had been done. If a team of researchers found a crystal in their protein soup the morning after combining reagents, he thought, wouldn't it be great to know if somebody had already done so? "I began collecting data," he said. "And then I just started collecting all the data." Gilliland eventually developed his original version of what would become the NIST databank. He attracted the attention of officials at the Protein Databank at Brookhaven National Laboratory, which agreed in 1984 to distribute Gilliland's data, in simple text file form. But Brookhaven didn't have the staff to keep Gilliland's work updated and maintained adequately. Things began to improve in 1986, when Gilliland went to work for NIST's predecessor, the National Bureau of Standards. In 1987, Gilliland joined forces with colleague Dorothy Bickham to design a new interactive software program and update his crystallization data. Each database entry has two sections. One includes macromolecular information, such as name, molecular weight, and the source of the macromolecule. The other section contains crystal data, such as the number of molecules in the asymmetric unit (the smallest section from which an entire crystal can be produced), crystal density, crystallization method, and macromolecule concentration. The data are based on an extensive search of the literature. That search includes what are known as hidden sources - papers that, by their titles, give no indication they contain crystallization data. The Brookhaven database picks up where Gilliland's leaves off, concentrating on the three-dimensional structure of a crystal after X-ray study. A scientist trying to make a crystal could search Gilliland's database for a similar molecule on which to base the experiment; then, after obtaining a crystal and the coordinates of its components, the scientist can check with Brookhaven to see if there are any data to help analyze and classify the crystal's structure. "They're totally complementary," says Brookhaven databank director Thomas Koetzle. He adds, "It's a good system and I hope it will have a lot of users. We've used it ourselves." NIST plans to send information about the databank to everyone on its mailing list, as well as advertising in a relevant scientific publication. And Gilliland isn't finished. The next step, he says, is an expert system for crystal growth. So far, no one has been able to derive any first principles of crystallization from the literature. The size of the molecules is one stumbling block. It can vary from 20 to 100 angstroms, compared with 2 for an average sodium chloride molecule. Also, there are many complex ways that a crystal's surface interacts with its inner structure. Gilliland plans continual updates of the databank. Its increased size may mean switching it to an on-line database in a few years. Eventually, he'd even like to make it possible to link the system to robots. Eli Lilly and Co. is already considering incorporating the databank into its automated crystallization system. "I think it will be extremely helpful, extremely useful," says Noel Jones, a senior research scientist and protein crystallographer at Lilly. For Smith, who met Gilliland when she worked at the Naval Research Laboratory in Washington, the system will save on long-distance phone calls. "My use of it [the database] in the past has been basically calling Gary up on the telephone and saying, `Hey, has this been done?' " Now she and others can order a database on their own. For more information about the databank, write to: National Institute of Standards and Technology, Building 221, Room A323, Gaithersburg, Md. 20899; or call Joan Sauerwein at (301) 975-2208. Jan Ziegler is a freelance science writer based in Washington, D.C.

Database Crystallizes Protein Engineering Studies

Interested in reading more?

Become a Member of