Chemistry Nobel Laureates Helped Develop Tools to Study Large Biological Molecules

John Fenn, Koichi Tanaka, and Kurt Wüthrich will receive the 2002 Nobel Prize in Chemistry for helping to develop tools for the study of large biological molecules. Fenn, a professor of chemistry at Virginia Commonwealth University in Richmond and professor emeritus of Yale University, and Tanaka, an R&D engineer with Shimadzu Corp in Kyoto, Japan, will share half of the prize for “their development of soft desorption ionization methods for mass spectrometric analyses of biological macromolecules.” Wüthrich, a professor of molecular biophysics at ETH Zürich, will receive the other half for “his development of nuclear magnetic resonance spectroscopy for determining the three-dimensional structure of biological macromolecules in solution.”

Too big to fly

Mass spectrometry (MS) determines the mass of an ionized molecule from the mass-to-charge ratio, thus providing an important and sometimes sufficient clue to its identity. ¹ Nuclear magnetic resonance (NMR) helps decipher the three-dimensional structure of a molecule by observing its nuclear spins.

Until the mid-1980s, neither MS nor NMR was widely applied to analysis of biological molecules. The stumbling block was the large size of most biomolecules of interest. Proteins, for example, can range from a few thousand atomic mass units (daltons), such as for the hormone insulin, to more than 5 million Da for large enzyme complexes.

For MS, the challenge was to convert such large and thermally unstable molecules into the gas-phase ions required by the technique. As Fenn quips, “It was like teaching an elephant to fly.” Fenn adapted the concept of electrospray ionization (ESI) to do the job, and Tanaka showed that laser desorption could work.

Today, one of the most important applications of ESI and laser desorption is to protein identification. Using enzymes, researchers can break a protein into constituent peptides and analyze the fragmentation pattern. This pattern acts as a fingerprint to be matched against a protein database. As a check, each peptide can be further fragmented using tandem MS.

Biological macromolecules also posed a challenge for NMR. How could one sort through the NMR signals from so many atoms and deduce any useful information about the molecule’s structure? The methods developed by Wüthrich helped lead the way.

For more than a decade now, NMR has served as an alternative to x-ray crystallography, especially for structure determinations of those biological molecules that must remain in solution or that can’t be crystallized. By May 2002, NMR had been used to determine about 20% of the roughly 15 000 atomic coordinate sets deposited in the protein data bank. Both MS and NMR are expected to play a role in the growing field of proteomics—the study of how proteins and other substances interact in a living cell.

Electrospray ionization

The phrase “soft desorption ionization methods” used in the citation for Fenn and Tanaka refers to the gentle methods needed to get large biomolecules off a surface and into the gas phase. In ESI, a solution of the molecules to be studied (known as analyte molecules) effuses from the tip of a small hollow tube into a chamber of inert gas at or near atmospheric pressure, as depicted in the top panel of the figure at right. An intense electric field at the needle tip disperses the emerging liquid into a fine spray of charged droplets and drives the droplets toward the end wall.

Solvent evaporation increases each droplet’s surface charge density until Coulomb repulsion overcomes surface tension and the droplet breaks into many smaller droplets. These offspring evaporate in turn and break into still smaller bits. Although the exact process is still debated, the end result is the production of droplets containing only one solute molecule and some of the charge. Some of these charged molecules then pass through a slit for analysis by a mass spectrometer.

Lord Rayleigh predicted the initial droplet breakup in 1882, and John Zeleny demonstrated it in 1917. In 1968, Malcolm Dole recognized that continued evaporation would lead to charged molecules of solute. ² Unfortunately, the experiments he and his colleagues tried didn’t work very well.

Fenn, who earned a PhD in physical chemistry from Yale in 1940, learned about Dole’s paper soon after joining Yale’s chemical engineering department in 1967 (he remained there until he became emeritus in 1987). Fenn had just arrived from Princeton University, where he had studied free jet expansion of a gas into a vacuum to produce intense molecular beams. Fenn realized one place where Dole had gone wrong: Dole had filled his chamber with an inert gas flowing in the same direction as the analyte ions, but a counterflowing gas is required. That’s because the solvent molecules, which are cooled by evaporation, will tend to condense on the analyte ions unless swept away by an inert gas.

Fenn and Michael Labowsky repeated Dole’s experiment with a counterflowing gas, but were discouraged by the low ion currents. In the early 1980s, Fenn and Masamichi Yamashita tried again but with smaller molecules, and they met with success. In a subsequent series of studies, ³ done with Chin Kai Meng, Shek Fu Wong, and Matthias Mann, Fenn showed that ESI could produce intact ions with molecular weights up to 17 500 Da. Fenn could have done the same experiment in the late 1960s if only he had realized that the larger ions carried multiple charges; larger charge allows one to use detectors that multiply the charge signal, hence compensating for the low currents. Fenn’s group then found they could produce intact ions of proteins with molecular weights of 50 kDa, with no upper limit in sight.

Laser desorption

As work was proceeding on ESI, other researchers were trying to produce intact ions of large organic molecules, guided by the idea that heating big molecules very rapidly might vaporize the molecules before they had time to decompose. ⁴ To deliver such short bursts of energy, researchers tried fission fragments, beams of ions or atoms, and laser beams.

Some researchers started embedding the analyte molecules in a matrix of another substance. A matrix can help disperse the (strongly interacting) individual analyte molecules. And, as the matrix explosively evaporates during bombardment by particles or photons, the matrix molecules can lift analyte molecules off the surface or transfer charges to them. The bottom panel of the figure at left illustrates the process of laser desorption.

Michael Barber and his group, ⁵ for example, used matrices to develop fast atom bombardment into a useful MS tool for biomolecules up to 10 kDa. Although such early work could not reach masses that were much higher, the Royal Swedish Academy of Sciences noted, it did help lead to eventual success.

By 1983, Franz Hillenkamp (now at the University of Münster) and Michael Karas (now at the University of Frankfurt) had begun a systematic study of the physics of laser desorption and were working their way up to increasingly larger molecules. They used a matrix composed of small organic compounds, ⁶ which absorbed energy directly from the laser, protecting the analyte from disintegration. Hillenkamp and Karas had analyzed masses up to 3 kDa by 1987, when Tanaka made the surprising announcement at a meeting in Takarazuka, Japan, that he and his colleagues had analyzed molecules as large as 35 kDa. ⁷

Tanaka had joined Shimadzu Corp, a manufacturer of mass spectrometers and other analytical instruments in 1983, right after he had earned a degree in electrical engineering. His first assignment was to help develop an instrument to analyze ions removed from a surface by laser irradiation. For their next project, Tanaka and a colleague, Satoshi Akita, sought to use the experience they had acquired and turned to the need for mass analysis of biological materials.

The procedure followed by Tanaka and his coworkers differed from the approach of Hillenkamp and Karas primarily in the type of matrix used. Tanaka’s colleague Yoshikazu Yoshida suggested mixing the biological macromolecule with a fine cobalt powder because cobalt has a very high rate of heat absorption. To disperse the analyte evenly in the powder, the researchers blended the two ingredients in glycerin. Besides Akita and Yoshida, Tanaka credits Yutaka Ido and Tamio Yoshida for their help.

Shortly after Tanaka and his group announced their results at the Takarazuka meeting, Hillenkamp and Karas tried out their own approach, unchanged, on a heavier molecule, successfully performing a mass analysis on a 67-kDa molecule. ⁸ Their approach soon acquired the acronym MALDI—matrix-assisted laser desorption/ionization. Because of its high sensitivity and its adaptability, MALDI has become the standard for laser desorption. In 1990, Hillenkamp demonstrated that MALDI-MS can also be done with wavelengths in the infrared, compatible with diode lasers, if suitable matrices such as glycerol are used. ⁹

Nuclear magnetic resonance

The basic concept of NMR is easy enough: Put a sample in a strong, static magnetic field, and the nuclear spins will tend to align with that external field. Then perturb the spins by applying a radio frequency (RF) wave of just the right frequency and measure the absorption. The resonant frequency will be changed by interactions with the neighboring atoms. To measure such “chemical shifts,” one scans a range of frequencies, looking for absorption peaks. Each frequency peak contains information about the environment of a given spin. For organic molecules, one looks at the spins of hydrogen nuclei (protons); the common carbon and oxygen nuclei have no net spins.

Decoding the information contained in NMR spectra is formidable in practice, and the complexity of the spectra increases rapidly with the number of atoms in a molecule. Techniques developed over the years have streamlined data-taking and facilitated analysis. One key development was Fourier-transform NMR. Rather than scan the RF range to look for resonant absorptions, experimenters today apply short, uniform RF pulses to excite all the spins in a molecule simultaneously, and then record the time evolution of the decay of magnetization following the pulse. Fourier transformation of such time-domain measurements then generates the same spectrum as scanning slowly across the frequency domain—but more quickly and with higher sensitivity.

A second key development was the two-dimensional Fourier transform, which enabled one to study selectively a particular coupling between interacting spins. As used today, the technique involves a series of four steps: exciting the spins with a short pulse of RF radiation, letting the resulting spin magnetization evolve, sending in a second RF signal to mix the magnetizations of different spins in certain ways, and detecting the decay of the spin magnetization.

One then takes the Fourier transforms of the signals received in the evolution period and in the detection period. Those transforms are then plotted against one another. Any spin that does not interact during the mixing period will have the same frequency during both the evolution and detection periods, and it will appear as a peak along the diagonal of the 2D transform plot. Off-diagonal peaks arise from interactions between spins: A given spin at one frequency transfers its magnetization to another spin at a different frequency.

Experimenters have by now devised hundreds of pulse sequences for the mixing period to enable them to detect different types of nuclear coupling. One mixing scheme, known as COSY (for correlation spectroscopy), reveals scalar, or J, couplings between spins on neighboring hydrogen nuclei, such as the hydrogen atoms bound to adjacent carbon or nitrogen atoms in the backbone of a protein. Another mixing scheme, known as NOESY (for the nuclear Overhauser effect spectroscopy), reveals dipole–dipole interactions. Such dipole interactions occur for protons that are spatially close but may be far apart in the chemical structure of a chainlike molecule such as a polypeptide; they might arise, for example, in a molecule that loops around on itself.

Wüthrich joined ETH in 1969, five years after he had earned a PhD in inorganic chemistry from the University of Basel in Switzerland, and began working on NMR. Early on, he became interested in the challenge of biological macromolecules. As Adriaan Bax of the National Institutes of Health recalls, “I heard him speak with conviction about his vision way back in 1978.”

To meet the challenge of large biomolecules, Wüthrich and his team made clever use of the 2D NMR spectra. From a COSY spectrum, for example, they could learn that a certain proton has a scalar coupling with another proton. From the NOESY spectrum, in turn, they could find that the second of those two protons has a dipole—dipole coupling with a third proton that is farther along the chain.

By following such links, Wüthrich showed how to step along the backbone of a protein and sequentially assign resonances from the 2D NMR diagram to particular protons in the structure. Based on these resonance assignments and additional NOESY data, Wüthrich and his colleagues extracted the pairwise distances between an enormous number of hydrogen atoms in a given protein. With lots of help from a computer, they could then find structures consistent with those constraints, ¹⁰ such as the one shown in the figure above.

Wüthrich encountered skeptics along the way, but any holdouts were won over in 1986 when his team, using NMR, and a second team, using x-ray crystallography, both found essentially the same structure for a new protein. In 2000, Wüthrich’s team decoded the structure of the bovine prion protein, ¹¹ which is relevant to the study of mad cow disease. Just this summer, the group reported ¹² the solution NMR studies of a structure whose size is 870 kDa.

Laying a second golden egg

ETH’s Richard Ernst, with whom Wüthrich collaborated from 1977 to 1986, earned the 1991 Nobel Prize in chemistry for his role in developing NMR methods (see Physics Today, December 1991, page 19 ). Since then, Ernst says, he has had a guilty conscience because he can think of several others—including Wüthrich—who might have shared the honor in 1991. Wüthrich confesses his surprise at this year’s selection. “I wouldn’t have thought the Nobel Foundation would lay the golden egg so soon again in the same nest,” he remarked.

Wüthrich gives much credit to the researchers who have worked with him at ETH. In particular, he cites Martin Billeter, Werner Braun, Timothy Havel, Allen Kline, Anil Kumar, Kuniaki Nagayama, David Neuhaus, Gerhard Wagner, Gerhard Wider, and Michael Williamson for their achievements during what he calls the “heroic years” of 1978–85.