Tuberculosis —once thought defeated—is back with a vengeance. What’s the key for killing this killer once and for all?

By Don Wood

In the early 1980s, many medical authorities were confident that tuberculosis—one of the greatest killers in history—was no longer a major public health threat.

A number of highly effective drugs were available to combat the disease, so even if a particular strain of the TB bacterium showed resistance to one drug, another one would be able to knock it out. But that confidence was shattered within a few years as the number of cases of TB began to rise dramatically in some parts of the world.

Many factors contributed to the rise, but one of the most troubling was the emergence of strains of TB that were resistant to many drugs, not just one. Today, some strains, in fact, seem to ignore all the drugs modern medicine can throw at them. Each year, TB kills two to three million people around the globe, mostly in developing countries, and some estimates say as much as a third of the world’s population harbors the TB bacterium (although usually not developing into active tuberculosis unless the body’s immune system is weakened, as happens with AIDS and many other diseases).

Clearly, new drugs are urgently needed.

A consortium of researchers led by FSU biochemist Tim Cross has recently made a major step toward producing the basic scientific knowledge required for creating a new generation of drugs to treat TB and other diseases.

The traditional approach to finding new drugs is to screen a large number of naturally occurring compounds (typically produced by fungi, bacteria or plants) to see which ones have the desired germ-killing properties. It’s a laborious process, and success is by no means guaranteed.

But in the last couple of decades a combination of sophisticated biochemistry, high-tech instrumentation and powerful computers has given rise to a hopeful new approach. If scientists can identify a particular protein associated with a disease-causing germ, perhaps they can design a drug to bind to that protein and disable or kill the germ.

For such an approach to work, scientists must have detailed pictures of how these proteins are put together, atom by atom. They also need to know exactly what these proteins do, and how they do it, in cells. Enter Cross, who is also director of the nuclear magnetic resonance program at the National High Magnetic Field Laboratory. About four years ago, he realized that advances in technology made it a propitious time to launch a major assault on the problem of determining the three-dimensional structures of the proteins found in cell membranes—if only the best expertise and instrumentation out there could somehow be brought together and focused on it.

The membrane proteins—located in the membranes surrounding cells and also in the membranes of internal cellular structures—are particularly inviting drug targets for three reasons. First, they are essential to the cell’s metabolism because they control everything that goes into or comes out of the cell (nutrients in, waste products out, for example). So if a drug disables the protein, it disables the cell. Second, some proteins on the cell’s surface distinguish one type of cell from another and so offer the potential for developing drugs that target only a single type of cell. Third, because they’re on the surface, they are easier for drugs to latch on to than are proteins inside the cell.

But the membrane proteins present a formidable challenge. Over the years biochemists have developed methods for dealing with proteins in the cell’s interior, to the point that determining their 3-D structures is almost routine. But the membrane proteins exist in a greasy environment and are much harder to work with. So Cross conceived a bold plan. “The goal of the project,” he says, “was to develop methodology that would allow us to solve the structures of membrane proteins.”

Since all cells have membrane proteins, the methodology Cross sought would apply to any cell. But he chose Mycobacterium tuberculosis—the germ that causes TB—as the focus of his effort because of its importance for public health and the urgent need for new treatments.

He forged a consortium of more than a dozen top researchers at labs around the country, bringing a number of sophisticated technologies—x-ray crystallography, electron microscopy, high-resolution mass spectrometry, computational biochemistry and Cross’ own specialty of nuclear magnetic resonance (NMR)—to bear on the problem. With an $8.1 million grant from the National Institutes of Health, the project got under way.

While the consortium as a whole has made progress on many fronts, the most promising development to date has come from Cross and his team of NMR spectroscopists. They focused on a deceptively simple-sounding aspect of the problem—sample preparation for NMR analysis. In fact, the problem is anything but simple and has stymied researchers for years.

Nuclear magnetic resonance is a powerful tool for determining the structure of biological molecules. (The word “nuclear” refers to an atom’s nucleus but has nothing to do with radioactivity.) The nuclei of many atoms act like little spinning magnets. Put them in a strong magnetic field, and they wobble in sync while they continue to spin like toy tops. Zap them with radio waves and the neat alignment is disrupted. Turn off the radio waves, and the little magnets flip back into alignment. But before they do this they induce an oscillating voltage in a detector coil, which scientists can analyze for clues to the molecule’s structure. To an NMR spectroscopist, the more powerful the magnet, the sharper the spectrum it produces and hence the more accurately the protein’s structure can be determined. Most NMRs operate with 500- to 800-MHz magnets, so the mag lab’s new, one-of-a-kind NMR spectrometer with a 900-MHz wide-bore magnet is something for spectroscopists to drool over (see related article).

There are a few other 900-MHz NMRs available (including one at the University of Georgia that Cross uses in his research), but what makes the mag lab’s machine unique—and the pride of Cross’ NMR program at the mag lab—is its combination of high power and wide bore. The new magnet’s four-inch bore—the hole in the middle where samples are placed—is twice as wide as any other 900-MHz NMR, so it can test samples nearly 10 times larger in volume.

The machine promises to be a boon not only to Cross’ membrane protein project, but also to biochemical and medical research in general. “There are a lot of unique things one can do with this magnet,” Cross says, “and we’re still developing its capabilities.”

Whatever machine they plan to use, the first hurdle for NMR spectroscopists studying membrane proteins has been preparing a sample, a process more art than science. Each protein presents a new challenge. Over the years spectroscopists have employed a hit-or-miss approach. If lucky, they got a sample that worked. In this stumbling fashion scientists have determined the structures of only about 90 membrane proteins. But any given cell type may have hundreds or thousands of different membrane proteins—M. tuberculosis has about 1,100, for example.

“That’s been the bottleneck,” Cross says.

The first step in sample preparation is to produce a quantity of the protein to be analyzed. A snippet of M. tuberculosis DNA—the bit that codes for the protein of interest—is introduced into the common bacterium E. coli. “The E. coli reads that DNA message and produces the protein,” Cross says.

But there’s a wrinkle to the process. Carbon and nitrogen are two of the most common atoms found in proteins. Unfortunately, their most common isotopes (carbon-12 and nitrogen-14) don’t work well with NMR. So researchers feed the E. coli with nutrients made up of the rarer isotopes carbon-13 and nitrogen-15, which act like little magnets. These isotopes are then incorporated into the proteins the E. coli produces. Only in this way can the carbon and nitrogen show up on the NMR spectrum.

“Next we have to fish out the specific protein we want from the E. coli, because E. coli produces a lot of its own proteins as well,” Cross says.

The next step in preparation depends on how the sample will be analyzed. One choice is to prepare the protein in a lipid environment similar to the environment of the cell membrane. Consortium member Stanley Opella of the University of California, San Diego, and Cross have made progress using this approach, which requires bulkier electronics around the sample, making the wide-bore NMR machine particularly useful. But Cross, Opella and two other consortium members, Charles Sanders, Vanderbilt University, and Frank Sonnichsen, Case Western Reserve University, take an additional tack. They prepare samples in a detergent environment. While lipids join together to form flat layers, detergent molecules form spheres called micelles.

Preparing the detergent environment is the tricky part of the process. To produce a good NMR spectrum, the protein molecule must bind to the detergent micelle, but not so tightly that it can’t tumble freely. “The progress that we’ve made in really good sample preparation has been in developing good detergent environments for these proteins,” Cross says.

The success of Cross’ team came from taking a patient and thoroughly systematic approach to the problem. They prepared samples of a range of different types of proteins, using a number of different techniques on each protein. They could then see which techniques worked best for which types of proteins. Eventually they were able to predict (with reasonable certainty) what technique should be applied to a given protein. “We now know how to make samples for a significant number of membrane proteins,” Cross says. “We will be able to start solving a lot of membrane protein structures. Other research groups will pick up on this technique, too, and start going after specific drug targets.

“We will be submitting a manuscript very soon that will show spectra of 20 proteins from within our consortium. There’s never been anything like that before.”

And Cross has 30 more specimens in the refrigerator, ready to be analyzed.

“This may seem a little bit esoteric, but in fact it is a tremendous development,” Cross says. “I think that we really have made a breakthrough in how we’re going to be able to attack membrane protein structure characterization. It’s going to take some time to develop three-dimensional structures. But we’re absolutely confident at this point that we’re going to be able to solve those structures.”