A Singular Focus

© Wolfgang Kumm / Corbis

As a 29-year-old postdoc with a PhD in physics, I sat among 18- and 19-year-old undergraduates in a molecular biology class in 2001, learning the basics of biology for the first time in my life. As physicists, we see the world in numbers and imagine natural processes by their most elemental components. Atomic particles interact by relatively simple sets of rules that can be distilled into predictive formulas. I thought that, since chemistry is simply a more complicated application of physics, and biology a more complicated application of chemistry, it would follow that biology could be reduced to predictive physical formulas as well. But as I quickly learned in my first biology classes, things are not so simple.

Biology's rules are messy and its molecular interactions are almost impossible to faithfully predict. Still, I wanted a cleaner picture of what was happening. I wanted to reduce the biology I was learning to a physicist's perspective. My goal became to watch interactions molecule by molecule.

It wasn't such an original thought; physicists had been working on techniques to watch single molecules interact for the past 10 years. The great advantage is that it allows biologists to get away from studying reactions done with millions and millions of molecules at a time and deducing their interactions from averages. It's similar to watching the New York marathon being run from an altitude of 10,000 feet. We can see the pack of runners slowly snake through the streets of Manhattan as one object. We can accurately measure the average runner's speed, but we'll never see those rare individuals capable of running the 26 miles in two hours or those that stop for a few seconds at the refreshment stand. Only by watching individual runners can we see the interesting dynamics of a marathon.

With the techniques I've used and helped develop, I've seen energy jump between the overlapping discs of a photosynthetic protein. I've seen replication machinery stutter instead of sail across the nucleotides of a thread of DNA. I've started to envision how a virus particle infiltrates the surface of a cell to start an infection. With colleagues I am now working on extending some of these single-molecule techniques to allow researchers to do large-scale screens of viral entry and develop better antiviral drugs.

My interest in biology started while I was working on liquid helium and photon jumping. I was trained in the field of molecular physics at Leiden University in the Netherlands, where our lab used lasers to learn about the quantum mechanical properties of molecules. Other researchers had shown that at temperatures nearing 1°K, or -272°C, you could actually illuminate just one molecule and look at its fluorescence emission. Cooling molecules to temperatures comparable to those in outer space 'freezes' them in place. As a result, we could study a single molecule for a long time, without it being damaged by the laser light.

Toward the end of my PhD, I started to use this low-temperature laser technique to look at the photon-capturing machinery of chlorophylls from photosynthetic bacteria. The photosynthetic proteins absorb photons from the sun and convert them into chemical energy. Plants and photosynthetic bacteria manage to do this at near-100% efficiency, whereas the most cutting-edge human-made solar cells barely reach 20%. We studied the first steps of this reaction: how light is absorbed by the chlorophyll-containing protein and transported to the complex that uses the photon's energy to convert water and carbon dioxide into food. Even though these problems are biological and chemical in nature, the task of getting photon energy from one point to another is essentially a problem of quantum mechanics. How photons travel from the light-capturing protein complex through space to the reaction center where chemical synthesis takes place is pure physics.

Researchers had studied the light-harvesting complex, called LH2, extensively. Still there was no clear explanation for why one ring of this two-ring structure was better than the other at transferring energy. No one had tried to use single-molecule techniques; in fact, before 1997 nothing more complex than simple hydrocarbons had ever been studied using low-temperature, single-molecule, laser techniques. Many people thought it would be impossible to look at a single LH2, a relative monstrosity with its 18 protein subunits and 27 bacteriochlorophyll rings. By using very dilute samples of the LH2 complex, however, we were able to illuminate them individually with a finely focused laser beam. At a temperature of 1° K, the fluorescence of a single complex could be monitored for hours on end. I remember that once we trained our laser on one particular complex, we went home overnight and were able to study the exact same molecule the next day!

These experiments revealed how the photon energy is shuttled between the different chlorophyll-containing rings in the complex (click here for graphic). In one ring, where nine chlorophylls are loosely stacked in the plane of the ring, the energy jumps between the chlorophyll molecules in a very classical fashion: It transfers from molecule to molecule as a discrete package. On exiting this ring, however, the energy becomes 'captured' by the second ring and is smeared out over all 18 chlorophylls in this more tightly packed ring. Here, the energy cannot be described as a classical, discrete package but must be thought of as a quantum-physical entity that is present on all 18 chlorophylls simultaneously. It's the smearing that holds the secret of efficiency. The energy does not need to hop precisely from one center of each ring to the next; when smeared, it's in the right place all the time.

The paper that reported these findings¹ turned out to be an important one for me. It caught the attention of many researchers in the community, and was cited more than 200 times. More importantly it gave me a little bit of insight into biological systems. Even though I had limited interaction with biologists on this project - they simply delivered the samples they had prepared to me for analysis - it was the first time I'd been bitten by the biology bug. I applied for a postdoc position with a pioneer in the field of single-molecule microscopy, Sunney Xie at Harvard. Xie had started using single-molecule techniques to study biological systems at room temperature, arguably a more biologically relevant condition than 1°K.

My first encounter with biology had given me a sense of the problems with which biologists grapple, as well as some of the limitations of that science. Understanding protein interactions on the atomic level is often what's required to overcome these limitations and to get at the heart of the most biologically relevant molecular events. I thought that if I could develop simpler techniques for watching molecules interact, it would shed light on some of the long-standing puzzles in biology.

The puzzle that most intrigued me was the replisome, the large complex of proteins responsible for replicating both strands of DNA simultaneously. Essentially, the replisome unzips duplex DNA strands and synthesizes new DNA onto each of the exposed strands. The replisome moves in one direction, feeding new nucleotides onto the so-called leading strand in a continuous fashion, while it hiccups backwards to make fragments of DNA on the lagging strand (click here for graphic). I wanted to understand the gymnastics of all the replisomal proteins as they stretch and contort to synthesize a fresh strand of DNA.

The problem with using biochemical techniques to study protein and DNA interactions is a problem of averaging. Most methods require that you take a tube of many millions of molecules and let them perform whichever reaction you're studying. The process is halted at precisely the right moment, products are detected, and then the steps that took place during the reaction are extrapolated. These procedures provide only averaged values, and many interesting dynamic details are blurred out.

The single-molecule approaches that physicists have developed to address this problem weren't perfect either. To study the interaction between DNA and proteins, a powerful technique called optical tweezing is used. Researchers attach a small spherical particle on one end of a DNA molecule and attach another spherical particle to a protein that is firmly bound to some point of the same DNA molecule. They stretch that single DNA molecule by holding the two beads suspended in a laser beam - much like a "tractor beam" could hold an enemy space vessel in Star Trek - pulling the beads in opposite directions. This way, the DNA molecule is stretched, and the distance between the two beads tells you something about the position of the protein on the DNA. Movement of the protein, for example an RNA polymerase reading out DNA and producing an RNA strand, is visible as a change in distance between the two particles. Even though extremely precise measurements can be made this way with accuracies of better than a nanometer, it's almost impossible to view numerous molecules in multiple laser beams at the same time.

If we used this state-of-the-art technique, we'd have just one shot at initiating replication at our suspended DNA fragment. The chances of having all the relevant protein components and enzymes in the right place at the right time were slim. To look at multiprotein complexes such as the replisome, we needed something that increased the probability of seeing a successful formation of a replication fork. Neither the biochemical averaging, nor the physical "tractor beam" methods would work for this problem.

Then, on a particularly nice summer day in August 2001, I sat watching the sailing boats on the Charles River between Cambridge and Boston. It struck me that I could attach large quantities of DNA strands to a microscope slide at one end, and attach a bead to the other end. By flowing solution across the surface I could float the beads to one side - much like the wind forcing the battalion of anchored boats in one direction - and thus stretch out all the DNA molecules at once. This way we could flood the solution with the proteins important for initiating replication and watch hundreds of DNA molecules instead of one, dramatically increasing the chances of a successful replisome forming in our experiment.²

This design also gave us the ability to measure the length of the DNA during replication, which turned out to be most revealing. When a low amount of stretching tension is applied, the single-stranded DNA coils up into a tight ball, whereas the double-stranded DNA remains fairly taut on its beaded tether. The interesting physics of the problem is that single-stranded DNA is much shorter than double-stranded DNA, because the single-strand immediately coils up tightly. Leading-strand replication produces a double-stranded product, but also feeds out a length of single-stranded DNA before lagging-strand synthesis can add on the matching nucleotides. As the replisome traveled, converting the double- to single-stranded DNA, we could see a net shortening of the DNA molecule. By tracking the position of the beads during replication, we could see the replication machinery converting the DNA and, surprisingly, pause at the end of roughly each 1,000 bases (the approximate length of an Okazaki fragment) as the replisome waited for synthesis to resume on the lagging strand.

Not only did we learn something about how the lagging strand replication keeps pace with the leading strand, but also how the method is generally applicable to the study of many other complex protein machineries that act on DNA. One obvious advantage of our method is that it is both conceptually and technically very simple to implement. As a result, we have built a number of microscopes that can be used to do these single-molecule experiments, instead of having to rely on one or two technically complex experimental rigs. This allows us to study many other protein machineries that act on DNA, such as the ones that are responsible for DNA repair and recombination.

Single-molecule techniques bring us one step closer to making molecular movies of complex biochemical processes. These movies won't have the resolution of X-ray crystallography, but they will be able to probe the dynamics of molecular interactions, instead of providing mere snapshots of intermediates.

This idea for molecular movies isn't limited to DNA-protein interactions. I'm also very interested in visualizing the process by which flu virus enters cells. Of course the technical problems are different; for instance, there is no advantage in tethering a viral particle in a flow of solution.

Scientists have already defined many of the steps of the virus-cell interaction, but the precise sequence and orchestration of events are still cloudy. So while we do have a good sense of what virus particles look like from structural biological approaches, we don't know what the parts do in action. Current protocols to investigate fusion events don't get at the mechanics. By staining viral particles with different fluorescent tags, we can directly image the process, from the first association of viral particle and cell right through to the fusion of virus and target cell membranes (click here for graphic). At the point of fusion, the fluorescent tag escapes from the viral particle, something we see in our microscope as a little flash of light. Using this approach, we've already been able to characterize the delay between the fusion of virus and cell membranes and the pore formation that allows the release of the viral contents into the cell.

These methods are not just useful for understanding the biology. If my lab can scale up our viral entry assay, we will have a high-resolution and high-throughput screening technique for assessing novel antiviral medications. Slowing down of fusion in the presence of certain small molecules can help us identify promising drug candidates. Since our single-particle experiment requires a very small amount of reagents, it is ideally suited to the screening and characterization of large libraries of compounds.

Single-molecule approaches transformed the way I do physics, and they are transforming the way biologists study molecular phenomena. Conventional biochemists and structural biologists are teaming up with single-molecule groups to offer new solutions to difficult problems. I believe that these single-molecule techniques, while quite exotic now, will be part of the routine biochemical tool kit in the near future. And while that could put me out of a job as a developer of such techniques, by then I'll have found something else to capture my interest.

Antoine van Oijen is an assistant professor in the Department of Biological Chemistry and Molecular Pharmacology at Harvard Medical School.