How would you characterize your general research area?

One, I’m still continuing methods development in structure determination, which includes X-ray crystallography, and less so NMR at this point. I’m also still trying to push the limits of methods, particularly computational methods to handle difficult problems. Right now that’s primarily for large complexes of macromolecules, where diffraction data might be limited, and membrane proteins, which are also often limited by resolution. We’re trying to bridge the gap of crystallography on the one hand and cryo-electron microscopy on the other hand. Both methods are beginning to converge in a way. Cryo-EM typically starts at about 20 to 30 Å. Now there are crystal structures that have been solved even at 4.7 Å resolution. With both methods coming from opposite ends, we are covering the mid-range of resolution, which is very important for very large systems because of crystal quality and poor diffraction.

Secondly, over the past eight years we’ve moved into the structural biology of proteins that are involved in neurotransmission—in particular, proteins involved in synaptic vesicle fusion that releases neurotransmitters, such as glutamate. This is the step that occurs in the presynaptic neuron upon depolarization. Neurotransmitters then bind to receptors on the post-synaptic membrane. We’ve been focusing primarily on proteins involved in this presynaptic process and have solved quite a few important structures there. For example, in 1998 we solved the structure of the SNARE complex—a heterotrimeric protein complex involved in vesicle membrane fusion.

  How long have you been working in this area and how did you become interested in it?

As far as crystallography is concerned, I became interested in it from the computational point of view through developing X-PLOR and applying simulated annealing to solution NMR structure solution and crystallographic refinement. It was roughly 1985 that I became interested in that and I have been working on it since then, and on the more biological application that we’re doing right now since roughly 1995.

  What were or are some of the greatest challenges in performing your work?

Certainly these days the greatest challenges are large macromolecular assemblies of complexes on the one hand and membrane proteins on the other. The difficulties are primarily at the biochemical end in obtaining diffracting crystals, getting stable complexes that are suitable for crystallization, or finding conditions under which they crystallize.

There are still computational challenges dealing with those systems, especially if the crystals don’t diffract to a high resolution. There’s also a more practical challenge when many related structures are to be determined, in particular for rational drug design based on crystal structures. The challenge there is actually to do very fast, high-throughput structure determination. The bottleneck is still the refinement part, despite the simulated annealing procedure. It still requires a lot of manual intervention and inspection of electron density maps. That’s not so much a scientific problem, it’s more an artificial intelligence problem. It’s really fascinating because a human being is so much more efficient at recognizing features in these maps than a computer is, and we haven’t been able to train a computer to do this, especially at lower resolutions.

  How rapidly has the state of knowledge in your field evolved in the last decade and what were some of the key discoveries that furthered that advancement?

If you look at the Protein Data Bank, there’s really been an explosion of solved structures. It’s been almost an exponential growth over the past 10 years, and it’s made possible by three things. One, crystallographers have embraced molecular biology as a tool to obtain recombinant protein, sometimes with various mutations to obtain better crystals. Another major factor has been the availability of synchrotron light sources to observe X-ray diffraction data with unprecedented quality and also to be able to easily change the wavelength of the X-ray radiation. This has made possible what we call MAD phasing and it has revolutionized how we solve structures. Third is computational development. The availability of these improved refinement techniques and systems such as CNS, which are a comprehensive approach to crystallography, have made it easier to solve structures.

  What is the implication of your work for the future of your field and allied fields?
Where do you predict the state of knowledge in your field will be in 10 years?

Considering my two interests, computational methods development and neurotransmission, with computational methods development, we still hope we’ll make an impact by enabling people to solve more difficult problems—such as limited data and very large molecules—than we can at this point. In our neurotransmission interest, we hope that knowledge of all these proteins and complexes and how they interact with membranes will be the key to developing new therapeutics to more selectively control neurotransmitter release, which could be quite helpful for treating neurological disorders. At this point, drugs are primarily targeted at the receptors and they’re often nonselective and nonspecific.

  What advice would you give to those entering a research career in general?

In crystallography, I would say that this field is still alive, despite all the efforts for high-throughput crystallography and automation. There are always new challenges. It’s by no means a dead field. It’s just different than it was 10 or 15 years ago. The problems are just much bigger. For a doctoral dissertation it’s now probably insufficient to simply solve one structure of a protein domain; you also have to put it into some biological context or you have to work on a more challenging structure. The bar is higher than in the past, but then the tools make it much easier to solve a structure.

My other recommendation is to have a multidisciplinary approach and not just learn one technique. My own career has been maybe an example where I started out in theoretical physics and mathematics and switched to biophysics for my Ph.D. and then computational chemistry and biochemistry. I like to use techniques from very different disciplines and often the most exciting things happen when you mix disciplines—for example when you use computational methods to apply to biological problems. I think students should be broad- rather than narrow-minded.

  What would you like the general public to understand about your work?

In the post-genome era, where we know the genomes from humans and many other organisms, it’s important to realize that although these sequences contain extremely valuable information, they don’t tell us the whole story. They don’t tell us what these proteins look like or how they interact, how they might change their conformation, or how they might interact with their environment. All this needs to be explored. We’re essentially working on this next step—getting three-dimensional shapes for many of these proteins and complexes.

Axel. T. Brunger, Ph.D.
Howard Hughes Medical Institute
Stanford University
Palo Alto, CA, USA