Sensitizing NMR

OF LATE, the nuclear magnetic resonance community has been buzzing about a technique that can increase NMR sensitivity by 100-fold or more and is opening up new ways to follow biochemical reactions in vitro and in vivo. Called dynamic nuclear polarization (DNP), the technique can be used simply to speed up familiar experiments—data collection that historically took overnight can be done in mere seconds—but it is also making NMR a newly powerful tool for identifying reaction intermediates, probing enzyme kinetics, and imaging in vivo.

The power of NMR lies in its high resolution; it enables researchers to see small differences in chemical environments. NMR is also noninvasive and nonperturbing—it won't harm whatever sample or organism you're scanning. At the same time, NMR is "horribly insensitive," says Lucio Frydman, a chemistry professor at Weizmann Institute of Science, in Israel. NMR insensitivity can be a particular problem for studies of biochemical systems, which are often limited to low concentrations—typically, hundreds of scans must be averaged to bring a signal out of the noise.

NMR basically involves putting a sample in a magnetic field, where the spins of nuclei with odd numbers of protons or neutrons will line up with or against the field direction. Because of the energy difference between these states, proportionately more nuclei align with the field. The population difference leads to a nuclear spin polarization, which is measured by NMR.

Because the difference in energy between spins aligned with or against the magnetic field is very small, the difference between the two populations is typically a mere one nucleus in 100,000 for hydrogen. For other nuclei the ratio is even lower. Increasing the sensitivity of NMR therefore entails coaxing more nuclei to align with the magnetic field so as to increase the spin polarization.

Dynamic nuclear polarization tackles that challenge by transferring the larger polarization of electron spins, such as those found in stable radical compounds, to nuclear spins through irradiation with high-frequency microwaves. The nuclei of the target species then become dynamically polarized, and their NMR signals are enhanced anywhere from 50- to several hundred-fold. Additional signal enhancements can be obtained by doing experiments at temperatures down to 1 K.

Originally proposed by Albert W. Overhauser in his 1951 doctoral thesis, DNP was first used experimentally by Charles P. Slichter in 1953 at very low magnetic fields. Interest in the technique is surging now because of the recent development of commercial high-power gigahertz microwave sources.

One area in which DNP-enhanced solid-state NMR has made an impact is in the study of bacteriorhodopsin, a membrane protein that uses a retinylidene cofactor to harness light energy for pumping protons out of cells. Chemistry professors Robert G. Griffin of Massachusetts Institute of Technology and Judith Herzfeld of Brandeis University and colleagues used the technique to help distinguish and characterize intermediates in the pump cycle (Proc. Nat. Acad. Sci. USA 2008, 105, 883). The research team combined the samples of ¹⁵N-labeled, membrane-bound bacteriorhodopsin with the nitroxide biradical TOTAPOL, irradiated the mixture with light to produce desired pump cycle intermediates, and then cooled everything to 90 K. The group then irradiated the samples with 250-GHz microwaves to transfer the electron polarization from TOTAPOL to the nuclei of bacteriorhodopsin.

Focusing on the ¹⁵N NMR spectra of the second and third intermediates of the bacteriorhodopsin cycle, dubbed K and L, the researchers found indications that the protonated Schiff base of the retinylidene loses contact with its counterion in K and establishes contact with a new counterion in L. Concurrently, low-energy, single-bond torsion in the polyene of the retinylidene in K converts to high-energy, double-bond torsion in L. Thus, the researchers concluded, the chromophore initially stores the energy of absorbed photons electrostatically, then transforms it into torsional energy as the system moves from K to L. The energy is subsequently released when the Schiff base proton is transferred and the connectivity of the bacteriorhodopsin active site changes, although the original chromophore conformation is not regenerated until later in the cycle.

THE GROUP also saw indications that four different L states of the protein actually exist, and the researchers have further investigated them through two-dimensional NMR experiments. Their results indicate that only one of the L states is functional, and the other three decay back to a resting state. Griffin says that the enhanced signals available through DNP are critical to the studies because the researchers can trap only 5 to 25% of the sample in the L state. "We're looking for small signals to distinguish between the species," he says.

Texas A&M University chemistry professor Christian Hilty is also using DNP to investigate enzyme reactions. Hilty has coupled an NMR magnet with a stopped-flow system to study enzyme reaction kinetics and intermediates. His general approach is to freeze an enzyme substrate with a polarizing agent such as a trityl radical, irradiate the solution with microwaves to transfer polarization to the substrate, and then thaw the solution quickly by mixing with hot buffer before using the stopped-flow apparatus to mix the substrate and enzyme together in an NMR tube within the magnet.

In contrast to Griffin's bacteriorhodopsin experiments, which dynamically polarize protein nuclei, Hilty's approach hyperpolarizes just the substrate nuclei. Hilty and graduate student Sean Bowen initially studied the hydrolysis of Nα-benzoyl-L-arginine ethyl ester by the well-known enzyme trypsin (Angew. Chem. Int. Ed. 2008, 47, 5235). They were able to use the NMR spectra to calculate a catalytic rate constant of 12.1 s^-1, which agrees well with the rate constant calculated from ultraviolet-visible spectroscopy data.

Having shown that the method works in principle, Hilty is now turning to enzymes that are less well understood. One project focuses on the enzyme uronate isomerase, which catalyzes the isomerization of D-glucuronic acid to D-fructuronic acid in microbes. In collaboration with Texas A&M chemistry professor Frank M. Raushel, Hilty is trying to determine the reaction mechanism of the enzyme. The sugars complicate matters by converting into different forms spontaneously in solution, in addition to undergoing the catalyzed reaction. Because "there are several things going on at the same time," Hilty says, it's critical to do real-time measurements such as those enabled by his stopped-flow system. "It's not sufficient to do the reaction and then measure after the fact what the products are," he adds.

Although Hilty's stopped-flow apparatus does enable measurements to start quickly, there is a delay time during which the samples mix and stabilize in the tube before NMR data can be collected. Right now, the delay time for Hilty's system is about 200 milliseconds. Hilty is designing a new flow cell with the goal of reducing the delay to 10 or 20 milliseconds. Ultimately, Hilty would like to use the technique to observe protein folding in real time. "Many proteins have folding events that occur on a timescale that is observable with our method," Hilty notes, and the information provided by NMR could provide key insights into folding mechanisms. Overall, Hilty says, "we're trying to develop this application into a general way of measuring mechanisms and kinetics and intermediates in processes that are far from equilibrium. That is something that is rather new for NMR."

Meanwhile, Song-i Han, a chemistry professor at the University of California, Santa Barbara, is developing DNP methods to study interactions between biomolecules and macromolecules. By tethering a DNP reagent such as the nitroxide radical 4-amino-2,2,6,6-tetramethylpiperidine-N-oxyl (4-amino-TEMPO) to surfactants or proteins, Han and colleagues can illuminate nearby water molecules to quantify local hydration dynamics (Langmuir 2008, 24, 10062). Whether one is interested in protein folding, protein-protein interactions, or protein-membrane interactions, "water exclusion is a general phenomenon that occurs when molecules come together and bind," Han argues, so studying the water dynamics around such interactions should provide clues to how those interactions work. She is also using her technique to shed light on the mechanism of how water is passively transferred through membranes, in particular whether water is somehow dissolved in the lipid membrane or channeled through transient pores.

DNP can also be used to enhance NMR or magnetic resonance imaging (MRI) in vivo. Although MRI can be used to visualize structure in the human body, positron emission tomography (PET) is typically used with radioactive tracers to visualize functional processes (C&EN, Sept. 8, page 13). Dynamically polarizing nonradioactive chemicals to serve as imaging agents could provide a way to use MRI to observe processes without the need for radioactive substances.

Because living organisms cannot be irradiated with gigahertz microwaves, however, DNP imaging agents must be irradiated and then injected. But hyperpolarization does not last forever, 30 seconds is a relatively long lifetime, whereas common radioactive agents can have half-lives of hours. Warren S. Warren, a chemistry professor at Duke University, is investigating DNP agents that would have a longer hyperpolarization lifetime made possible by the presence of two equivalent atoms.

A molecule with a pair of equivalent atoms—C-2 and C-3 of CH₃COCOCH₃, for example—has four possible spin states. When the molecule is hyperpolarized, three of the spin states can relax easily, but because of symmetry the fourth is protected. "We have shown that we can use this protected state to store polarization," Warren says. "A reasonable estimate for the spin lifetime in such states is many minutes."

Warren and colleagues create the imaging agents by starting with a molecule that has inequivalent atoms, dynamically polarizing it, and then chemically converting it to the desired imaging agent in which the atoms are equivalent. (See how this works for diacetyl at image below) The molecule maintains the hyperpolarization until another chemical reaction, such as a metabolic process, makes the atoms inequivalent again.

DEVISING SUCH chemicals and reaction pathways is not difficult, Warren says, and there are several nontoxic options. Using DNP for in vivo imaging, therefore, could be "something more than a curiosity that's cool in the lab," he says. His group is working with several compounds that they're packaging into temperature-sensitive liposomes that would protect the DNP agent from metabolic degradation. In theory, the liposomes could be injected into an organism and then broken open at a particular time or in a particular location. Some of the reagents Warren is developing are antidepressants that he envisions using to image metabolic pathways in the brain. Other groups are investigating dynamically polarizing pyruvate to monitor metabolic activity in different organs.

UCSB's Han is also experimenting with in vivo imaging. Her approach is to use perhaps the ultimate nontoxic imaging agent: dynamically polarized water. The idea is to anchor TEMPO or other free-radical compounds to a gel filtration matrix through which water flows continuously while being irradiated with microwaves; the resulting hyperpolarized, but radical-free, liquid can then be used for imaging (Proc. Natl. Acad. Sci. USA 2007, 104, 1754). So far, Han and colleagues have used the technique to image water flow in small sample vessels, but Han envisions using the technique with saline or another body-friendly fluid that would cross the blood-brain barrier. "We could accurately monitor blood perfusion," Han says, to pinpoint small strokes that are currently difficult to detect by MRI with imaging agents such as gadolinium.

For his part, Weizmann Institute's Frydman got his DNP instrumentation set up only in the past year and is specifically looking at coupling it to ultrafast multidimensional NMR techniques (C&EN, May 7, 2007, page 61). One of his projects is looking at metabolic fluxes of cancer cells to observe the effects of hypoxia or drugs. "Much of the metabolism couldn't be seen before because of the lack of sensitivity," Frydman says. "Now we could have the sensitivity of PET with the spatial resolution of MRI. That is the promise of DNP."