When developing a drug, it’s often best to start with an exquisitely detailed portrait of your biological target, typically one of the many medically significant proteins in the body.
“When a company has the structure of a protein of interest,” said MIT chemist Robert G. Griffin, “they’ll target it with a large library of a 1,000 or more candidate drugs to see if anything works.”
That, at any rate, is the optimal case, says Griffin, Professor of Chemistry and head of MIT’s prestigious Francis Bitter Magnet Laboratory. But often there is no picture of such proteins at atomic scales. Particularly elusive have been the amyloid fibrils that are formed by protein aggregation in about 25 different diseases, including Alzheimer’s, Parkinson’s, kidney disease, and Type-II diabetes. And while some drugs target amyloids, until recently there’s been no detailed molecular structure of the fibrils.
Why so? Traditional techniques to determine molecular structure don’t work with amyloid, or with many comparable substances. X-ray crystallography can only yield detailed structures of proteins that can form crystals. “To date, amyloid fibrils have not been crystallized,” said Griffin, “because they prefer to form structures with a twisted ribbon morphology” (Figure id).
Griffin has spent years seeking ways to unveil the structures of hard-to-image molecules. And at the core of his efforts is nuclear magnetic resonance (NMR), the predecessor of magnetic resonance imaging (MRI), which is applied in mostly nonclinical settings.
Magnetic resonance methods rely on the fact that many stable atomic nuclei behave like tiny magnets that align in powerful magnetic fields and reorient under the influence of pulsed radio waves in a manner that conveys information about the structure of the surrounding molecule. For his part, Griffin’s specific goal is sharply boosting the capacity of this technology to reveal the structures of complex molecules. Some examples illustrating molecular structures of increasing size and complexity are shown in Figure i.
By combining his approach with traditional methods, for example, Griffin and collaborators recently detailed the first structure of an amyloid fibril (Figure id). Applying the same arsenal of techniques to a protein that resides in cell membranes, his group has shown why drugs that once offered relief from a common type of flu no longer do so.
One key to success has been Griffin’s ability to see new promise in decades-old insights into magnetic resonance. It’s an approach that has encountered skepticism about not only the innovations themselves, but also his choice of targets. When Griffin first focused on amyloid, for example, a leading figure in his field told him not to bother. Amyloid, this doubter said, doesn’t really have a structure. But recent results are validating both Griffin’s approach and priorities.
Magic Angle Spinning
The first approach Griffin focused on was magic angle spinning. This technique, as its name implies, has an abracadabra quality to it: While insoluble samples typically give broad, indecipherably overlapped NMR signals, narrow signals can be obtained by spinning the sample around an axis tilted exactly 54.7° to the field (Figure ic). If you spin at frequencies up to 100 kilohertz (thousands of revolutions per second), you can get high-resolution information about the sample’s chemical features.
However, critical structural information is lost in the process. Most biological molecules are folded, bringing different parts of the molecule together in complicated configurations. Folding is a big issue in drug development, since knowing exactly how a protein folds makes it much easier for drug-makers to choose candidate drugs that will interact effectively with that protein. So, it is very useful to be able to measure distances between nuclei on the two sides of a given fold. But, said Griffin, traditional magic angle spinning “sacrifices the information on distances.”
Early in the 1990’s, Griffin set out to solve this problem. And while his solution also has a magical feel, it works.
His approach is to selectively undo some of the effect of magic angle spinning by beaming radio waves at the spinning target with suitable frequencies, in pulses of precise lengths, at intervals that match the spinner’s rotation frequency. The result is what the parlance of the trade calls dipolar recoupling.
The system works for many kinds of targets. For example, it can provide detailed structures of many polymers. Chemical firms, Griffin noted, “have been using this system for 20 years.” The approach can also yield the structures of some biological molecules, as shown in Figure i.
However, the method remains reliant on the generally weak signals of conventional NMR spectroscopy. With lengthy times required for probing even relatively simple proteins, the structures of many biological molecules remained off-limits to magic angle spinning NMR. For molecules with an amyloid-like level of complexity, therefore, other innovations were required.
Dynamic Nuclear Polarization
In the mid-1980’s, Griffin set out to enhance NMR signals. The key point about nuclear spins is that if you can boost the number of nuclei aligned with the magnetic field – in other words, increase their polarization – those nuclei will give you larger NMR signals. At the time, the broader NMR environment was pursuing ever stronger magnetic fields. (The strength of present-day superconducting NMR magnets may be as high as 20-plus Tesla, hundreds of thousands of times stronger than Earth’s magnetic field.)
Taking another tack, Griffin felt it made sense to try improving on traditional dynamic nuclear polarization (Figure iia). It was known that unpaired electrons, in selected molecules known as free radicals, could boost the polarization of nearby atomic nuclei. So, the researchers’ aim was to strengthen signals from target nuclei in their protein samples by manipulating electrons.
One step was to create specialized free radicals that are especially effective at boosting the polarization of nuclei. And thanks to synthetic chemistry contributions by Timothy Swager’s group, also in the MIT Chemistry Department, the researchers created compounds, known as polarizing agents, which fit the bill (Figure iib).
Yet another need was to create devices that could generate high-frequency microwaves.
How high? In the electromagnetic spectrum, microwaves go from about 1 to 300-plus gigahertz (GHz) (billions of cycles per second). Microwave ovens produce frequencies at the low end: 2.4 GHz. But for the researchers to enhance the ability of electrons in their specialized compounds to polarize target nuclei, they needed microwaves of a much higher frequency, anywhere from 100 to 600 GHz.
It looked like a tough assignment. “When we started, that part of the electromagnetic spectrum was dark,” noted Griffin. “There were no sources.” This meant the researchers would have to create their own devices.
The instrument the researchers focused on is a kind of vacuum tube – albeit one far more sophisticated and powerful than what most of us envision when we think of such a tube.
With a team that is directed by Richard Temkin of the MIT Physics Department, the researchers spent years developing the specialized instrument. Versions of the device, known as a gyrotron, can presently produce microwaves at up to 527 GHz for Dynamic Nuclear Polarization (DNP) experiments. And at such frequencies, the instruments will boost the polarization of target nuclei to hundreds of times their normal levels as shown in Figure iic.
The addition of this tool to the arsenals of those probing complex molecules is making its mark. A German firm has recently been marketing a gyrotron-equipped NMR spectrometer. And while most users of the systems so far have been university researchers, interest in drug and chemical companies is growing and a major U.S. firm will soon become the first to obtain one.
It’s likely that as academic laboratories continue to solve the molecular structures of more key amyloid and membrane proteins, drug-maker interest will accelerate. Presently, five million Americans are diagnosed with Alzheimer’s disease, and expected new cases in an aging population are motivating the amyloid structural studies by in Griffin’s group and other laboratories.