October 14, 2002
The NIH X-Ray Diffraction Interest Group
web site: http://mcl1.ncifcrf.gov/nihxray
Item 1: Topic Discussion - Trends in Crystallography
SUMMARY: Structural genomics is supposed to deliver 3-D structures for all building blocks of biological macromolecules; molecular modeling should be able to organize these building blocks into 3-D structures. However, structural genomics and molecular modeling together cannot provide extensive information on any biological process where intermolecular interactions and signaling are involved, not to mention that for any modeled structure, a real structure is the best and final validation. Modern crystallography, armed with many newly developed and advanced tools, will be doing even better mapping the reaction trajectories where dramatic conformational changes of biomolecules often occur and studying macromolecular assemblies and signaling pathways where intermolecular interactions always dictate.
Lothar Esser (NCI): As usual, I read our Newsletter with great interest. The 26th Newsletter contained the first reports on the XIX IUCr meeting held in Geneva. However, I was surprised that the sole discussion was about structural genomics and its Damocles' sword quality for traditional macromolecular crystallography. As someone who did not attend this meeting, I could not help but ask myself whether we are at a point where new developments in crystallography focus entirely on high-throughput tools (robots, automatic beamlines, crystal imaging, automatic map interpretation). Has there been anything new that would benefit the crystallographer that is not involved in the mass production of crystal structures?
As I see it, in addition to solving lots of structures, questions as to the detailed 'how' will challenge crystallography. Crystallography, if it wants to continue to play an essential and interesting role in structural biology, needs to answer questions about the dynamics of processes. We have seen cases where the dynamics of an active site loop can be inferred by looking at two or more distinct states, Laue experiments on photo-reactive proteins, structure determinations on apo and holo enzymes and a few more. One direction in which progress might lie is the largely but not entirely unexplored area of diffuse X-ray scattering that might reveal the dynamics of a molecule or (as often) a convolution of lattice effects with molecular dynamics within the crystal. Also some notable work has been done in this field but all in all this is still in the hands of a few experts and has not made its way into mainstream crystallography. In short, I do not share the anxiety that might arise in view of the assembly-line structure production. If nothing more, structural genomics will challenge and enable us to pursue more complex systems and expand the possibilities of crystallography.
Mohana Rao (NCI): Twenty-twenty indicates perfect vision. Here I am dreaming of a distant future and conditions therein. All dreams do not bear fruition. However, without dreams and without goals, life is not worth living! In science, too, it is better to dream.
I am here to dream of the possibilities in the field of crystallography in the year 2020. Before it, it would be interesting to review the strides our field took in the last twenty years. Considering that only main frame computers were then in use and diffraction data were collected on films using oscillation techniques on a rotating anode generator and no graphics systems were available, it is significant to note that structures of three plant viruses were solved. It is the late eighties and early nineties when the "boom" started in the crystallographic market. Diffraction data could be collected more rapidly and to a better resolution using the synchrotron radiation as well as at home. These could be processed on the desk using unix boxes. Automated computer programs for structure solution and refinement too began to make their presence felt, initially in small dribbles and later in larger numbers. After gaining an upper hand over data collection, structure solution, and refinement, the only realm to be conquered remained in the crystal growth. From being an "art", crystal growth soon became amenable to scientific exploration. As of now, crystals could be grown using commercial "cocktail" solutions.
Parallel to all these developments, sequencing, cloning, expression, and purification techniques were expanding almost exponentially. It became possible for crystallographers to choose a particular enzyme and study its ramifications with respect to a disease or a defect. Replacement of methionine sulfur by its selenium counterpart and growing Se-Met derivatives of crystals and location of the Selenium atoms using multi-wavelength anomalous dispersion techniques became one of the means for the "final solution" of the phase problem. The new century dawned with the announcement of the ribosome structures. What an auspicious beginning! The human genome has been laid bare. The frontier has presently shifted from genomics to proteomics. At present, there are countless problems in crystallography awaiting solutions. Looking into the telescope of future, I could see that almost all tertiary structural folds of proteins will be sorted out. A discovery of a new fold in 2020 would be like the discovery of a new comet in space! I could also hear the saying: "If you have a gene, you have a protein; if you have a protein, you have a crystal; if you have a crystal, you will have the data; if you have the data, voila, you have the structure and do what you want to do with it!"
The scientist who solved the "phase problem" was awarded the Nobel Prize quite a few years ago. I could see the computers in 2020 to be unbelievably faster. Diffraction data could be collected even with crystals of extremely small size. I could see the real time reconstruction of substrate and inhibitor binding to enzymes. I could see that many crystallographic problems and their solutions lead directly to the discovery of new drugs. I could see certain types of cancer "conquered" due to the elucidation of crystallographic structures. The days of solving individual protein structures would be like what small molecules are today. I could see that the emphasis will be on multi-protein complexes and nucleic acid assemblages I could see that at least in some cases "tailor-made" drugs to suit individuals will be available due to the knowledge obtained after solving the structure of enzymes isolated from the particular individuals. I could also see that while understanding would increase in the working of enzymes and in the origin, nature, and cure of diseases, fundamental knowledge in the essentials of crystallography would be far less and sketchy in the majority who practise the trade. I could also see difficulties in mining and tooling the desired nuggets of information due to the presence of several gigantic databases, in spite of unbelievably high-speed desktop computers linked to those databases.
All in all, the state of the art and science of crystallography will be far better, impressive, and illuminating in the year 2020.
Wei Yang (NIDDK): It has been clear for some time that macromolecular X-ray crystallography is becoming a routine technique thanks to the successes of molecular cloning, protein engineering, and the development of new and robust approaches to the phasing problem including molecular replacement, multiwavelength anomalous scattering, synchrotron radiation sources and direct methods. In the near future a robot may be able to determine a protein crystal structure without much human intervention. Similar to what happened to small molecule crystallography, the demand for experts trained for many years to use the technique vanishes with the logarithmic growth of new structures. So how shall we, who have spent most of our professional lives solving structures by X-ray crystallography, continue to find new challenges in structural biology research? One clear answer is to work on more difficult problems as exemplified by the structures of nucleosome, ribosome, RNA polymerase complex, and membrane proteins. But these big problems may not be a prudent choice for small laboratories. The alternative choice is to increase the breadth and the depth of what traditional X-ray crystallography encompasses. The answer to the breadth is structural genomics, which aims to solve thousands to hundreds of thousands of new structures. The success of structural genomics will depend on developing methodology to identify targets, to improve protein production, and to automate crystallization and structure determination. Structural genomics approaches will keep many trained crystallographers employed and challenged, and the improvement in technology will not only make structural genomics possible, but elevate the level of crystallography in general. The answer to the depth is to be a biologist first and crystallographer second, that is to be interested in and solve a biological problem instead of regarding a three-dimensional structure as the final product. Despite notable successes, the need for structural information remains high, ensuring ample challenges for creative structural biologists as we enter the new millennium.
James Hurley (NIDDK): Xinhua asked me to write a few words on trends in crystallography for the x-ray IG newsletter. This is an ever popular topic for discussion. The thoughts below are distillations of many conversations with NIH and other colleagues.
On the one hand, we have unprecedented successes, from the structures of huge complexes like ribosomes, spectacular advances in structure solution of integral membrane proteins involved in transport and ion conduction, the advances of structural genomics; and on to the more everyday inroads of structural thinking and methodology into nearly all disciplines of molecular and cellular biology and medicine.
On the other hand, this enormous success has bred anxieties about the future among crystallographers. Will large structure factories grab the old role of investigator-initiated structural biology? Will the computational biologists come up with structure predictions so good that they will put us out of business? While this still seems close to science fiction to many of us, the possibility may be scaring off some new talent from our field. Given the recent growth in structural biology, how long can we expect the growth in support to continue? Given the increase in ease and accessibility of crystallography as a tool- thanks to pioneering efforts of so many in our community to constantly improve the technology- does it still mean anything to be a protein crystallographer?
There is plenty to do to extend the recent wave of successes noted above. Large multiprotein complexes still comprise a tiny fraction of the total number of structure solved, although their impact is disproportionate. The solved complexes are enriched for those that naturally occur as stable, abundant complexes, such as ribosomes and viruses. Complexes that are rarer, or are formed and broken in a transient and regulated manner, present a much greater challenge. Advances in co-expression and refolding of complexes consisting of many polypeptide chains can extend the possibilities to a much greater range of projects. The real challenge is in the protein chemistry, of course, not the crystallography itself.
There is no shortage of membrane proteins to solve. Steve White’s web site lists about 65 solved structures of integral membrane proteins. A drop in the bucket if ~25% of the typical proteome is in the integral membrane class. The biggest hurdle again is expression. There is still not even one solved structure of a recombinant mammalian integral membrane protein.
Structural genomics still offers more potential than production. The new goals and funds going into this endeavor have been an enormous spur to technology development, and this is all to the good because the technology and resources are benefiting all of us. The strategies and tactics of structural genomics can also be useful when applied on a smaller scale in a more focused way- what Wayne Hendrickson calls “academic structural genomics”. One missed opportunity in the current large scale structural genomics paradigm is that little is being done to integrate structural and functional genomics. Functional analysis is inseparable from mainstream structural biology- why should the genomic cognates of these field be any different?
Most of the work in most of our labs doesn’t fall into the cutting edge categories listed above, but this doesn’t mean it is any less valuable. Structural analysis is central to basic research throughout life sciences. Many of us identify ourselves as much as members of some other life sciences field- in my case, signal transduction- as with crystallography. Many of us become fascinated with questions in our adopted fields outside crystallography and like doing our own experiments in these other fields. Then the question of strategic balance comes up- how much of your time do you spend doing what you are best and most competitive at, versus how much you spend learning something new and expanding your toolkit.
The ever-greater integration of structural and functional work does pose one concern not mentioned above- one that I think is more real than the anxieties that I posed above. That concern is how to help our non-crystallographer colleagues get the structural information they need. I sense a good deal of frustration among biochemists who need structures solved, but cannot persuade a crystallographer to collaborate, and don’t have the skills or resources to do it themselves. At the IUCr I heard about an interesting approach to this being tried at UTSW. The trick seems to be to create a collaborative facility, not quite an independent lab and also not a traditional core service facility.
In my view the future for structural biology is brighter than ever. There is almost an infinity of interesting problems yet to solve; and the resources available, if not equally infinite, are still growing. One of our biggest challenges is to get this message- that it is not all done- out to the new generation of young scientists.
This site is maintained by Dr. Xinhua Ji (firstname.lastname@example.org) on the NCI-CCR-MCL server (http://mcl1.ncifcrf.gov).