NIH's Ad Bax Takes NMR Spectroscopy to New Demensions

NIH's Ad Bax Takes NMR Spectroscopy to New Dimensions

Over the past decade, analyzing the structure of proteins has become considerably more than an academic pursuit. Pharmaceutical companies now routinely employ a protein-structure group, with the goal of elucidating the structure and properties of proteins relevant to their drug-development programs. But until recently the only way to determine a protein's structure was first to crystallize it and then analyze the diffraction pattern of X-rays passing through the crystal. While nuclear magnetic resonance (NMR) could be used to determine protein structure, it appeared until early in this decade that the method could succeed only with the very smallest proteins. That a protein such as interferon-(gamma) could be studied by NMR seemed unimaginable-until it was done and published in 1992 by a team of researchers led by Ad Bax of the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health IH), Bethesda, Maryland. Bax has been the driving force behind a revolutionary use of NMR to reveal the structure of proteins.

Ad Bax

"My big goal is to study intact integral membrane proteins by NMR," says Ad Bax of the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland. "It would yield an enormous amount of biomedical relevant information."

Evidence of just how influential Bax has been is his standing as the number-one cited chemist between 1981 and June of last year, according to figures recently compiled by ISI's Research Services Group. Bax's total of 21,000 citations, averaging a whopping 142 per paper, is a full 7,000 more than his closest competitor. Four of his papers published between 1981 and 1990 have now been cited over 1,000 times each; these papers continue to average more than 100 citations yearly. (For a list of Bax's more recent high-impact papers, published since 1990, see the table on page 4.) And in the previous issue's ranking of authors who have published the greatest number of high-impact papers in the physical sciences over the last seven years, (see Science Watch, 8[6]:1-2, November/December 1997), Bax placed fifth.

Born in 1956 in the Netherlands, Bax attended the Delft University of Technology, where he studied applied physics and concentrated on the design and development of control software for a new generation of NMR spectrometer. In 1981 he earned his Ph.D. from the same institution-while doing most of his work, however, at Oxford University in the laboratory of Ray Freeman. He went on to spend two years as a postdoc at Colorado State University on solid-state NMR before moving in 1983 to NIH, where he began as a research associate in the Laboratory of Chemical Physics. Since 1991, Bax has been chief of the section on Biophysical NMR Spectrometry at the Laboratory.

Bax gave Science Watch correspondent Gary Taubes his views on the developments in NMR spectroscopy from his office at NIH in Bethesda.

When you joined NIH in 1983, did you have any inkling that you could someday use NMR to decipher the structure of large proteins?

There was some very interesting work coming out just around that time by the Wüthrich, Clore & Gronenborn, Kaptein, and several other labs that indicated the feasibility of NMR structure determination for smaller proteins. But to me it was rather far-fetched to imagine one would be able to solve three-dimensional structures of proteins 20 kilodaltons or larger. Nevertheless, I thought maybe we could have a stab at seeing whether we could study small regions in such proteins. As a postdoc in Colorado, I had developed some experiments that allowed me to study individual sites in a transfer RNA. This suggested to me that it wasn't fundamentally impossible-although I never anticipated getting complete structures for those complex molecules.
The big breakthrough came in 1987-88, when we combined the idea of heteronuclear isotopic labeling-that's using nitrogen-15 and carbon-13-with the old idea of doing three-dimensional NMR, introduced by several other groups.

First, can you explain what you mean by heteronuclear isotopic labeling, not to mention three-dimensional NMR?

What we do, when producing the protein to be examined in E. coli, is to include stable isotopes in the growth medium-for example, nitrogen-15 ammonium chloride-so that the protein produced by the bacteria will include this stable isotope, which is NMR-visible. That means we get a signal for this nucleus, whereas nitrogen-14, for instance, the most common isotope, does not give a useful NMR signal. And by having those nitrogen-15 isotopes in our protein, we can increase the dimensionality of our spectra; that means make them a lot less overlapping than they would be if they were just two-dimensional NMR spectra.

Overlapping?

Typically we have, for example, 1,000 hydrogens in a protein, and in principle one could expect up to 1,000 x 1,000 interactions, or a million interactions between those pairs of protons. Fortunately, only the ones close in space will interact with one another, but it still adds up to roughly several thousand interactions. Each of these cross peaks, the interactions between proton A and proton B, corresponds to a spot in a two-dimensional NMR spectrum. This spot has a given size. Think of it as a coin. You can put it on a piece of paper, but if you try to put a thousand coins on a piece of paper, some will be on top of one another. The only way to be able to separate them is if we have some other dimension. If you can vertically space them by some other kind of property, you will then be able to position all of them without having them touch one another. Then they don't overlap anymore.

How does the nitrogen-15 help?

We can use the fact that the nitrogen-15 chemical shift is typically different for all the different nitrogens in the protein. We can separate pairs of protons according to the resonance frequency of the nitrogen for pairs of protons that involve at least one backbone amide proton. That's the chemical name for the nitrogen in a peptide bond. That was used in our early experiments as the third dimension. We can also include carbon-13 and then separate the hydrogens that are attached to carbons. They can be separated in a similar manner, by spreading them according to the carbon-13 frequency.

Is it possible to go to higher dimensions and get even more information?

Yes, we actually extended this technique to four dimensions. That was the next big step, in 1990. We then looked at pairs of protons and separated them not just by the frequency of the nitrogen atoms attached to one proton, but also by the frequency of the carbon or nitrogen attached to the other proton.
We applied that to calmodulin, which is the system we used to develop many of these methods. It's a calcium-binding protein that's involved in the regulation of several dozen different enzymatic processes in the body. It's 17 kilodaltons.
continued