Keith Moffat used his background in physics to tinker with tools that light up molecules in motion.
As an undergraduate at the University of Edinburgh in the early 1960s, Keith Moffat studied physics. "Physics, physics, and more physics," he says. But when it came time to graduate, Moffat was looking to expand his horizons. Bill Cochran, a new professor who'd just arrived from Cambridge, suggested he get in touch with the molecular biologists there: "Francis Crick, Max Perutz, John Kendrew, folks like that," he recalls.
"So I wrote a letter asking for a brochure that would tell me the type of research going on," says Moffat. Perutz himself sent back a one-line note asking for two references. With references dispatched, Perutz's next missive, equally brief, said "come for a visit." To make the most of his time, Moffat arranged to meet with...
And go he did. Working under Perutz, Moffat solved the crystal structures of an array of chemically modified hemoglobins. He then went on to pioneer x-ray techniques that allowed him to generate time-resolved snapshots of proteins in motion, capturing reactions that take place over picoseconds in atomic detail. "Keith was one of the first to realize the potential for using short pulses of synchrotron radiation to take freeze-frame pictures of protein structures as they change over time," says Dennis Mills of the Argonne National Laboratory in Illinois. "He pushed for the technology and moved the field forward."
John Helliwell of the University of Manchester agrees. "Keith took the lead in conducting these ultra-fast experiments to monitor conformational changes in protein crystals in real time," he says. Moffat's efforts, along with those of his colleagues and collaborators, says Helliwell, "have come together to break the conventional wisdom that crystallography is a static technique."
The hemoglobin structures that Moffat solved as a grad student were of the standard, static variety. But his postdoctoral mentor—Quentin Gibson at Cornell University—got him thinking more dynamically. Gibson was using rapid reaction kinetics to compare the behavior of hemoglobin in dilute solution, in polycrystalline slurries, and in red blood cells. The study was designed to address the question of whether crystal structures—like the ones for hemoglobin produced by the Perutz lab—were physiologically relevant or merely artifacts of the crystallization process.
The reaction he focused on was one that was quick, simple, and easy to monitor—and to initiate. In addition to taking up oxygen, hemoglobin can also form a covalent attachment to carbon monoxide. The gas binds to the iron atom in heme, but can be knocked off by a bright flash of light. So Gibson loaded his hemoglobin samples with carbon monoxide, flashed them with light, and watched as the gas diffused away, then drifted back. And he showed that the hemoglobin acted the same way, whether it was in solution, in a crystal, or in a cell.
When Moffat arrived in the lab, he says, "I figured: why don't we use x-rays to follow that reaction, to look directly at the changes in structure?" But after a long afternoon spent discussing this research proposal, Moffat says, "We convinced each other that it was impossible. It might be desirable—we weren't arguing about that. But we surmised correctly that, in 1969, it was impossible."
The problems were twofold. "First, laboratory x-ray sources were simply not powerful enough to follow a reaction that occurs in tens of microseconds," he says. "Second, if I want to make a kinetic measurement, I need some means of starting the reaction." In Gibson's experiments, he used a photographic flash. But illuminating all the hemoglobin molecules in an x-ray–ready crystal would require a laser. "Lasers were around, they had been invented, but they were by no means common currency in biophysics labs," he says.
In time, high-powered lasers would become more readily available. But the thing that really made time-resolved studies possible was the use of synchrotron radiation—a highly intense source of x-rays that could be used to probe the structure of macromolecular crystals. And Moffat was in the right place at the right time. In 1970, he joined the faculty at Cornell, where physicists were looking to upgrade their synchrotron to make the circulating current more stable and strong. "And a number of us thought, wouldn't it be nice to build a synchrotron radiation laboratory that could harness the x-rays emitted by the machine, which are regarded as a waste product by high-energy physicists."
"It was a parasitic way of using an instrument that was built by physicists for physicists," says Michael Rossmann of Purdue University, who used Moffat's setup to determine the first crystal structure of a whole human virus. "Keith had the foresight to realize that this machine could be useful for protein crystallography, and he developed the Cornell synchrotron for use by x-ray crystallographers."
But it would be 25 years before Moffat would be able to take advantage of synchrotron radiation to conduct his "dream" experiment. In the interim, he solved the structures of a calcium-binding protein and some pituitary and placental polypeptide hormones—things Moffat deems "structural biology of a more conventional variety."
And he continued to plug away at the synchrotron problem. "Everything had to be done from scratch. Everything," says Moffat. "The means of getting the x-rays, the means of conducting the experiment, the means of recording the data, the means of analyzing the data. We had to develop the hardware, new algorithms, and new software. That took a long time."
The bane—and the beauty—of synchrotron radiation is that the x-rays are polychromatic: they emerge in a rainbow of different energies. Standard crystallography relies on x-rays that are monochromatic. But polychromatic x-rays are much brighter, so exposure times are short—a necessity for doing the sort of ultrafast experiments Moffat was interested in.
But polychromatic diffraction patterns are more difficult to analyze because of the way the x-rays get scattered by the crystals: the spots in the diffraction pattern, which ultimately yield information about the molecule's structure, tend to pile up, one atop another. "And it was felt that the inability to observe these hidden spots was a fatal flaw," says Moffat. But in a landmark Science paper, published in 1984, he says, "we showed that it wasn't."
"Keith and his collaborators came up with a truly novel way of processing the data so they could extract from the patterns much more than anybody thought was possible," says Jack Johnson of Scripps. But the images that came out were static. It would take another 10 years—and a move to the University of Chicago—for Moffat to rig the equipment he needed to make it all come together and make his crystals come to life.
Most of the data collection for the first time-resolved experiments—in which Moffat watched carbon monoxide get ejected from the protein myoglobin—took place at the European Synchrotron Radiation Facility in Grenoble, France. "It was a scientific expedition," says Moffat. "We'd crate all our equipment and take a whole team of people, because we had to run flat out. These are 24-7 experiments because beam-time is precious. Everyone's running around, everyone is fatigued. You have to remember to say, 'Let's go to the cafeteria and have a glass of wine.' Thank heavens for French cafeterias!"
But the long hours ultimately paid off and in 1996, Science published movies (and stills) showing myoglobin in action. "These subnanosecond time results are truly spectacular," says Helliwell. "Keith showed how beautifully you can get time-resolved information and watch things as they're actually functioning."
"When I saw those first diffraction maps come out I was blown away," says former postdoc Sean Crosson, also at the University of Chicago. "You see this mix of structural intermediates a few nanoseconds after you excite the system. If you're interested in protein dynamics and catalysis, and the motion associated with those events, this is the way to get at that. These data address fundamental questions in protein science."
And the data were worth the wait. "When Keith started his scientific career, he had the vision to formulate this research goal: to look into an individual protein and see where atoms move as the protein functions," says collaborator Klaas Hellingwerf of the University of Amsterdam. "Back then, this was far from possible. Now, for several proteins, it has been achieved"—in many cases, by Moffat and his team. They have used the technique to look at the photoactivation of a variety of light-sensitive proteins, including a bacterial photoreceptor that responds to blue light (see feature). Moffat has also worked to build and run a state-of-the-art beamline at the Advanced Photon Source at Argonne National Laboratory—a labor Johnson dubs "an extraordinary community service."
The reactions that Moffat monitors are all jump-started by lasers. So these days Moffat has taken to engineering his own photoreactive proteins, including a histidine kinase that phosphorylates itself in response to light—work published this year in the Journal of Molecular Biology. "That was a cool paper," says Crosson. "They lopped off the oxygen sensor from this kinase and put on a photon sensor." Light modulates the activity of the kinase by a factor of more than 1000, a response Moffat calls "gigantic."
"But the paper goes beyond a simple, 'Whoopee, we made a photoactive protein from something that doesn't normally sense light,'" says Crosson. "They played with the length of the connector helix and came up with a mechanism that starts to speak to how photons control the activity of this kinase. For Keith, it's not sufficient just to make the tool. He really wants to understand how it works, down to the molecular level. That's what makes the paper good."
In addition to optimizing that system, Moffat and his colleagues are also working on other artificially photoactive proteins, such as a tryptophan repressor whose DNA-binding affinity is altered by light. Ultimately, he hopes that light will be seen as a versatile probe for artificially controlling cell biological processes. "Are we there yet? No. Are we getting there? We sure as hell are," Moffat says.