The Soul of a New Structure-Function Machine

Abstract

It was at the 2001 American Crystallographic Association meeting that we witnessed the first reports from structural genomics (SG) centers and companies. As newly independent crystallographers we had set up our laboratories a mere three years prior. We looked over to the legion of similarly junior colleagues seated around us. We did not utter a word but it was clear we were all thinking the same thing: If these centers and companies can churn out structures that fast, are our small biologically oriented crystallographic labs destined to go the way of the dodo? Would these speed-demons eventually tackle the structural science that we deliberately pursued? We were convinced our nascent research programs were doomed to extinction as less-efficient generators of structural results. We felt like Indiana Jones, running for our lives from the formidable SG rolling boulder.Fast forward six years: we survived. Sure we occasionally got scooped like everybody else, but mostly by competitors in individual laboratories, not by SG centers. We didn't get overrun because we and the SG centers were by-and-large running in different directions. That said, in what ways have the SG centers had an impact on hypothesis-driven structural research?In aggregate, SG centers across the globe have been productive, having already deposited over 6000 structures in the Protein Data Bank (Janin, 2007xStructural genomics: winning the second half of the game. Janin, J. Structure. 2007; 15: 1347–1349Abstract | Full Text | Full Text PDF | PubMed | Scopus (10)See all ReferencesJanin, 2007). Of these about 2700 are from the NIH-sponsored Protein Structure Initiative (PSI), which aims to provide representative folds for most of protein fold-space (http://www.nigms.nih.gov/Initiatives/PSI/). It thus appears that the SG centers are accomplishing the task originally put before them. On a per-structure basis, the SG centers seem more cost-effective than individual labs, especially in the more recent “production” years of PSI-2. The question is: are the SG centers generating “science” more efficiently than hypothesis-driven structural research? That is, if the NIH dollar for structural work is distributed according to the science generated, is the PSI worth the 200–250 individual R01 grants that it costs? With renewal of the PSI just around the corner (2010), and in light of the statistic that currently only 8% of R01s are funded on their first try (down from 21% in 1998) (Couzin and Miller, 2007xNIH budget: boom and bust. Couzin, J. and Miller, G. Science. 2007; 316: 356–361Crossref | PubMed | Scopus (14)See all ReferencesCouzin and Miller, 2007), this is the question of the day.To be sure, all of us in the crystallographic community have benefited from the advances in protein expression, automation of crystallization screens, robotic handling of crystals at synchrotron beamlines, and development of crystallographic software, all of which were generated by the Specialized Centers of the PSI. In particular, the new generation of software has brought macromolecular structure determination almost to the automated level of small molecule crystallography for some, but not nearly all, macromolecular specimens. This brave new era in which the means to obtain high resolution structures is available to an ever-increasing body of apprentice structural biologists is evidenced by the half dozen or so new structures solved by the 50 novice crystallographers each year at the RapiData Workshops run by Bob Sweet of the National Synchrotron Light Source (http://www.px.nsls.bnl.gov/rapidata2007/). Via the PSI, the structural genomics centers have not only made high-resolution structure determination methods faster and easier, but in doing so have also made the field of structural biology more attractive to many more new and established investigators.If the PSI is successful in its mission, it will provide the means to approximately model the individual domains of the majority of proteins for which there exists no crystallographic or NMR structure, based entirely on sequence similarity. This might seem an attractive proposition to an investigator whose research is focused on such a protein, but what does it get her? Perhaps most importantly she gains some idea of approximately where individual residues are located within each domain, which ones are likely to be inside the core, which ones are likely to be on the surface, which ones are perhaps close enough to be interacting with each other within the domain. She might even be able to model how the individual domains of the protein could be oriented relative to each other, and thus build an overall model for the protein. It would surely suggest a myriad of experiments, many of which would reasonably be aimed at validating the derived protein model itself. Would anyone, or the study section of any granting agency, be satisfied with such a model, with all of its “maybes” and “likelies,” in an age when high-resolution structure has never been more accessible by X-ray and NMR methods? It is somewhat ironic that from the PSI were forged the powerful tools that obviate their own existence, by significantly lowering the hurdles involved in pursuing bona fide structural information.Biological structures are sought in order to assist in the understanding of a protein's function. Given the rapidly-growing list of proteins (identified for example by microarray techniques) whose function is of intrinsic biological and medical interest, it seems wasteful of limited financial, instrumental, and personnel resources to solve structures for the sake of structure. Knowledge of a protein's structure brings us only marginally closer to understanding its function. Only in combination with the results of other biophysical, biochemical, and genetic experiments is function elucidated. Altogether this suggests that investigator-initiated collaborative grants which bring together biochemists and molecular and structural biologists is the most efficient mechanism for tackling the role of a given protein in normal and pathological physiology of the cell and organism.