Current Status
The field of X-ray crystallography as applied to biological macromolecules is even more vibrant than in 1991 and continues to have a major impact on the advancement of biology at the molecular level. Knowledge of three-dimensional structures of key macromolecules fundamentally changes biological research in associated areas. However, there has been a dramatic change in the interaction of structural biology with other fields of molecular biology in the last five years. Structural results are now more eagerly sought and better understood by biologists whose primary research does not involve crystallography or NMR. The great engine of molecular biology continues to roar, producing ever more interesting and complex materials for structural study. In concert with this, crystallography has advanced to the point where structural work can now keep pace with other molecular studies and is no longer the rate-limiting step in understanding the molecular mechanism of a biological function. Thus, the predictions of growth and demand made in the 1991 BioSync report, which were controversial at the time, have been resoundingly borne out. These and other qualitative changes in the field are discussed below. The conclusions presented below derive from comparison of the user surveys of 1991 and 1997 and on the deliberations of the BioSync Committee.
More crystallographers are producing more structures. The number of research laboratories continues to expand, as evidenced by the large fraction of investigators who have been running independent laboratories for five or fewer years (about 30% of the total, Table A-4). Assuming equal penetration of the structural biology community by the 1991 and 1997 BioSync surveys, the number of independent investigators is now 25-30% larger than in 1991. Another factor that has had an increasing impact on the field of biological crystallography is the support of the Howard Hughes Medical Institute (HHMI), which has grown from an initial group of about six investigators in structural biology to about 20 today. The growth in the number of crystallographic investigators can also be obtained from the enrollment of the American Crystallographic Association (ACA). The ACA, long a bastion of small molecule crystallographers, is now strongly influenced by biological crystallographers. The membership of the Macromolecular Crystallography Special Interest Group now includes more than half the total membership of the ACA. This proportion continues to rise.
At the same time, the number of new crystal structures has grown at a substantially faster rate than in 1991, demonstrating the greatly increased productivity of the community. Four times as many new crystal structures of macromolecules were published in 1996 as in 1990 (Table 1), far outpacing the growth in number of investigators. In another measure of crystallographic output, the number of entries deposited in the Brookhaven Protein Data Bank increased similarly in the same time period (512 new entries deposited in 1991 vs. 1437 in 1996, Fig. 1). Of these the considerable majority represent the result of protein crystallography and provide a benchmark for the activity in this field. Although many of these structures were determined in the users' laboratories, an increasing number have made use of synchrotron radiation and, if access to synchrotrons had been more available or quicker, these numbers could have been doubled.
Figure 1. Depositions in the Protein Data Bank. All entries deposited in the Brookhaven Protein Data Bank are included. The vast majority of these are coordinate sets from crystallographic experiments. Source: Protein Data Bank web site (http://www.rcsb.org/pdb/).
Synchrotron radiation has penetrated more deeply into the body of crystallographic results today than in 1991. More than 40% of the new crystal structures published in 1995 were determined using diffraction data from synchrotron sources. In 1990, this number was 18% (Table 1). This represents a major shift in the behavior of crystallographers and a major increase in the impact of synchrotron radiation in macromolecular crystallography. The reported use of synchrotron radiation per research group has doubled since 1991 (averaged over all respondents, five days per group in 1990 compared with eleven days per group in 1996, Table A-6). The attitudes of crystallographers towards synchrotron radiation have also changed. In 1991, one-third of respondents to the BioSync survey did not use synchrotron radiation because they did not need it, although virtually all respondents thought synchrotron radiation would be important for their future research. In 1997, the future has arrived. Only 7% of crystallographer respondents do not currently use synchrotron radiation because they don't feel they need it (Table A-7), but all think they will depend on synchrotron radiation in the future (Table A-8).
Table 1
Macromolecular Crystal Structures1 1990-1996
| Year: | 1990 | 1991 | 1992 | 1993 | 1994 | 1995 | 1996 |
| New crystal structures: |
| 109 | 127 | 165 | 204 | 352 | 394 | 460 | |
| New structures with synchrotron radiation: |
| 19 | 30 | 44 | 50 | 100 | 158 | 202 | |
| Percent: | 18% | 24% | 27% | 25% | 28% | 40% | 44% |
| Journals |
New structures with synchrotron radiation/
Total new crystal structures |
| Structure | -- | -- | -- | 3/7 | 20/50 | 33/67 | 34/68 |
| Nature | 4/22 | 8/18 | 15/38 | 14/29 | 25/42 | 36/51 | 25/38 |
| Nat Struct Biol | -- | -- | -- | -- | 8/26 | 15/43 | 22/56 |
| J Mol Biol | 4/18 | 4/21 | 6/33 | 12/39 | 8/45 | 10/42 | 27/59 |
| Biochemistry | 4/15 | 1/7 | 1/8 | 0/18 | 2/34 | 11/40 | 11/46 |
| Science | 1/12 | 5/26 | 2/20 | 4/25 | 8/30 | 13/33 | 19/31 |
| PNAS | 1/6 | 1/14 | 3/14 | 2/25 | 6/27 | 8/26 | 8/25 |
| Cell | 0/2 | 0/1 | 1/4 | 3/9 | 5/12 | 10/25 | 12/23 |
| Acta Cryst | 2/5 | 5/8 | 2/4 | 2/10 | 2/17 | 4/18 | 6/22 |
| EMBO J | 0/5 | 2/7 | 4/5 | 2/13 | 9/22 | 9/18 | 17/25 |
| J Biol Chem | 1/6 | 0/9 | 1/13 | 1/10 | 2/15 | 4/13 | 5/20 |
| Protein Sci | -- | -- | 2/3 | 2/6 | 0/10 | 1/7 | 7/21 |
| Other | 2/18 | 4/16 | 7/23 | 5/13 | 5/22 | 4/10 | 9/26 |
1Source: Macromolecular Structures, 1991-1997, eds., W. A. Hendrickson & K. Wüthrich, Current Biology Ltd., London. All published crystal structures of biological macromolecules are abstracted in Macromolecular Structures if they meet the criterion of crystallographic uniqueness, i.e., they are not isomorphous with previously reported crystal structures. Approximately half of the abstracted structures were determined by molecular replacement. Not included are new ligand states, mutants, etc. that crystallize isomorphously with previously published structures.
New technologies are responsible for the growing prominence of synchrotron radiation in macromolecular crystallography. In 1991, crystallography with synchrotron radiation was still in an "heroic" mode because extraordinary results only came from experiments that were expensive, labor intensive and more prone to failure than experiments in the home laboratory. In 1997, the "heroic" mode is largely a thing of the past, thanks to many technological advances that have dramatically improved the effective use of synchrotron radiation.
Growing reliance of researchers on synchrotron radiation has resulted in new activism to gain access to suitable facilities. The most significant change in behavior has been self-organization of leading research groups to participate actively in the operation and construction of beamlines in order to guarantee access to synchrotron radiation. This trend was pioneered by the HHMI in providing beamline support for MAD measurements at the NSLS, which has also benefited many crystallographers outside the HHMI umbrella. More recently, five different groups of researchers have organized themselves to take over management and/or fiscal responsibility for beamlines in order to insure that they will have experimental time for their projects. These researchers are neither synchrotron experts, instrument specialists nor methodological aficionados, but are primarily structural biologists with competitive projects requiring synchrotron radiation. Such a change in behavior could not have happened during the "heroic" mode of synchrotron experimentation that existed in 1991.
Crystallography has joined the mainstream of molecular biological research tools. In the 1991 BioSync report, a strong "latent" demand for crystallographic synchrotron beam time was predicted for biologists whose primary research does not involve crystallography. The magnitude of the demand would depend on the support such researchers received at synchrotron sources. In 1997, it is clear that this prediction has been fulfilled. Many molecular biologists are adding crystallography as an experimental tool in their laboratories because of their growing dependence on structural information and the growing ease of obtaining it. The crystallographic expertise usually comes from postdoctoral associates. Following this trend, the Institute of General Medical Sciences at NIH recently began to supplement research grants of biologists with salary support for postdoctoral associates who are trained in crystallography. U.S. synchrotron facilities routinely receive applications for beam time from leading biologists who are not collaborating with established crystallographers. However, adequate support for these non-expert users remains beyond the means of nearly all synchrotron facilities. Thus, in 1997, use of synchrotron radiation for crystallographic experiments by non-specialists is limited to biologists who are able to recruit and pay for crystallographic expertise within their research programs.
Equally dramatic is the change within crystallography laboratories since 1991. The clear separation between expert and non-expert users of crystallographic synchrotron stations is markedly blurred compared with the situation in 1991. Freed from much of the arduous work formerly required in crystal structure determinations, crystallographers are expanding their experimental repertoire beyond crystallography and today are more accurately described as structural biologists. While macromolecular crystallography is becoming easier and more streamlined, it is as a result being applied to more challenging biological problems. Thus, development and production of biologically interesting molecules for crystallization requires an ever greater effort, an effort that increasingly comes from within the structural biologist's own laboratory. Cross-training at the postdoctoral level occurs in both directions -- crystallographers in molecular biology laboratories and molecular biologists in crystallography laboratories. This is a remarkable and refreshing exception to the tendency for greater specialization in scientific research.
Demand for synchrotron beam time continues to outpace supply. The U.S. national capacity for crystallographic experiments at synchrotron sources has approximately doubled since 1991 (Appendix B). However, demand continues to outpace supply by a factor of approximately two. Structural biologists report that the largest impediment to their use of synchrotron radiation today is timely access to the facilities (Table A-7).
Technological challenges remain. X-ray detectors continue to be limiting. Fast, efficient X-ray detectors with high spatial resolution and a bright X-ray beam are equally considered to be the two most important features of synchrotron radiation facilities by survey respondents (Table A-9). In the second tier of importance are adequate computer and networking services and well staffed, user-friendly experimental facilities. This reflects that fact that beamline efficiency could be improved substantially. Today's computer networks are insufficient to handle the data rate and data volume generated in typical, present-day crystallographic experiments.
Emerging Trends
More biologists will tackle more structural problems of greater complexity using the full range of technologies.
There will continue to be an increasing interest in macromolecule structures, at least in part because of the advances in ease and precision of structure determination brought on by the use of synchrotron radiation. At the 1997 Protein Society Meeting in Boston, approximately 50% of the symposium talks were based on the results of structural investigations. Even when the speaker was not a crystallographer, frequent references were made to protein structures determined through collaborations with a crystallographic group. The number of such collaborations will continue to grow. However a more important trend is the growing number of biologists who determine structures by X-ray diffraction within their own laboratories, and the growing number of crystallographers whose research programs are more biological. This integration of crystallographic research into the framework of a modern molecular biology laboratory is a phenomenon of the last five years and is clearly growing rapidly, although the numbers are difficult to project with any accuracy. In some cases, biologists are able to purchase and maintain their own diffraction instruments, but the majority will depend on the use of outside facilities, especially synchrotron sources, for data collection. The need for synchrotron time for non-specialist and less-specialist users will continue to rise dramatically. An excellent example of this trend is the recent publication of the crystal structure of the fibrinogen core by R. F. Doolittle and coworkers (Spraggon, Everse & Doolittle (1997) Nature 389, 455-462).
Another trend in macromolecular crystallography is the shift from structure determinations of single protein molecules to large protein-protein and protein-nucleic acid complexes. Examples include the structures of proteasomes, chaperonins, muscle proteins, nucleosomes and integral membrane protein complexes. In addition, it is worth noting that approximately half the proteins now being sequenced in the genome projects are membrane proteins involved in cell-cell communication. Only about a dozen of these have been investigated by crystallographic methods, and it is clear that the next decade will see a surge in the structure determination of these important molecules as methods are sought for crystallizing them routinely. Membrane proteins and macromolecule assemblies will generally diffract more weakly due to their size and complexity. Thus, effective data collection will require synchrotron radiation.
The genome projects will also generate considerable new crystal structure determination. BioSync is aware of at least four projects to determine structures for all hypothetical soluble proteins encoded by open reading frames in new genome sequences for which connections to the extant structure database cannot be made. All of these projects involve production of selenomethionyl proteins for MAD structure determination with synchrotron radiation. The impact of this new area of research is unknown, just as the impact of the genome projects was unknown a few years ago, but it is likely to be substantial.
A better chemical understanding of biological processes is an emerging trend of the increasing spatial and temporal resolution of structural information. Synchrotron radiation is one of the major technological advances contributing to this trend. Ultra-high resolution crystal structures from synchrotron data will continue to have a major impact in this respect. We are only just beginning to reap the advances of these high resolution studies around 1 Å resolution.
Technological changes in several areas of macromolecular crystallography will have a major influence on the research done with synchrotron radiation.
MAD use of polychromatic radiation. Multiwavelength anomalous diffraction (MAD) offers a generally applicable and rapid method for de novo structure determination of biological macromolecules. This method, in conjunction with cryocrystallography, allows one to collect data and solve the structure using a single crystal, and, in favorable cases, complete the entire process in a matter of days. The high brilliance and tunability of synchrotron radiation is essential for MAD; no laboratory X-ray source can be used as a substitute. The general selenomethionine label for proteins, developed for MAD by Prof. W. A. Hendrickson, is most widely used. Statistical direct methods, developed for small-molecule crystallography, have been used very recently to locate large numbers of Se sites in proteins, demonstrating that selenomethionyl MAD is applicable to much larger protein structures than anticipated. It is certain that MAD will become a major, perhaps even the dominant, method of macromolecular structure determination in the future. Selenomethionyl MAD is a critical part of the massive projects now being planned in structural genomics. Beam time for MAD will be the most limiting synchrotron resource in the coming years.
Data collection from microcrystals. The high brilliance of the new synchrotron sources permits data collection from very small crystals. There is anecdotal evidence of 10-µ crystals having been used for data collection. This development holds great promise for study of integral membrane proteins and macromolecular complexes, systems of enormous biological significance but very challenging crystallization problems. Microcrystals are often obtained relatively quickly, whereas it may take years of effort to grow crystals to a size suitable for laboratory data collection. Also crystals frequently persist in growing as plates or thin needles with only one large dimension. Microcrystallography requires an order of magnitude greater precision in crystal and X-ray beam alignment than is implemented on most of today's crystallographic beamlines. The technology for high-precision alignment exists on microfocus beamlines used for nonbiological research. Its implementation on crystallography beamlines could have great impact.
Time-resolved crystallography. One of the most exciting recent developments in the field of experimental protein science has been the application of synchrotron radiation to obtain nanosecond-long snapshots of a protein as it changes in response to ligand dissociation. The examination of the photodissociation of carbonmonoxy myoglobin and the photochemistry of photoactive yellow protein represent a major advance in time-resolved protein crystallography. All proteins undergo some structural change when carrying out their biological function. However, the measurement and understanding of the kinetics of conformational changes in proteins has hitherto been a largely unexplored area. The use of single-pulse Laue radiation coupled to laser photoactivation permits direct observations of structural changes in the nanosecond time range. Based on the interest generated by this work, it is clear that more studies will be directed toward understanding the kinetic events associated with phenomena where photostimulated triggering is possible. With currently available sources the maximum time resolution will be about 150 psec, and this will enable direct comparison between the time dependence of atomic positions determined by X-ray crystallography and molecular dynamics calculations, and their relation to spectroscopic observations.
High resolution crystallography of macromolecules. Crystallographers are increasingly finding crystals that diffract to much higher resolution than was previously thought to be possible. The reasons for this improvement are several. Better purification procedures yield larger, more strongly diffracting crystals. Cryogenic data collection results in lower background noise in diffraction images. Great improvements in signal-to-noise are obtained with synchrotron radiation due to a cleaner, more parallel and monochromatic X-ray beam. The result is an improvement in effective resolution, in the range of 0.2 Å to 0.4 Å for typical cases, and sometimes substantially more. Crystal structures have been determined to 1.0 Å or better for at least a dozen proteins. Crambin, a small plant protein, probably still leads the pack with diffraction to about 0.67 Å, but there is now an increasing number of larger and more complex proteins for which it is possible to determine structures to what can truly be called atomic resolution. At these resolutions, the structures can be determined by the actual data without need for stereochemical assumptions. Thus, more and more structures are accessible to phasing by statistical direct methods, which are being extended to protein crystals with diffraction data beyond 1.2 Å. The immediate rewards for this improved resolution are clear: more precise atomic coordinates for structure analysis and modeling; better descriptions of active sites, including hydrogen locations for more informed models of enzyme mechanisms; better understanding of protein flexibility through individual anisotropic temperature factors. The impact is likely to be a much improved understanding of the chemistry of biological processes. In addition there will be other observations such as the distribution of bonding electrons whose utility may have to wait for future fundamental research.
Very large unit cells. A recent spectacular example of advances in large unit cells is for orbivirus with a molecular diameter of about 700 Å, and unit cell dimensions of considerably greater than 1000 Å. The highly collimated beams from third-generation synchrotron sources permit the diffraction patterns of crystals with very large unit cells to be recorded.
The growing speed of macromolecular crystallography is well matched to the rapid pace of biological science. The burst in productivity of crystallographic research groups between 1991 and today will continue. The mid-1990s burst was due in major part to the adoption of cryocooling technology, both at the synchrotron and in the home laboratory. Cryocrystallography is probably the largest single factor in increasing the use of synchrotron radiation. The adoption of two new general labels for structure determination - selenomethionine and high-pressure xenon gas - will be major contributors to increasing the speed of crystallography. The widespread use of MAD for direct structure determination will drive an even greater demand for synchrotron facilities and speed structure determination overall. The experimental electron density maps of very high quality that frequently are the result of MAD phasing will also speed structure determination. The emphasis on speed will also drive development of new models for access to synchrotron radiation, including remote data collection and more complete on-site data analysis.
Recommendations
The importance of synchrotron radiation to biotechnology, molecular medicine and all molecular fields of basic biology has come of age. The dependence of macromolecular crystallography on synchrotron radiation has impacted all fields of biology, and leads to the following specific recommendations.