From genes to drugs in Erice 2002

Erice, Sicily, May 2002

The genomics revolution is here – are you ready? Participants at the 33^rd Erice course From Genes to Drugs via Crystallography directed by N. Borkakoti of Mediver UK Ltd learned that this revolution is having a profound impact on the field of drug design. Genomics provides boundless data regarding sequences and potential drug targets; the challenge for structural scientists will be to extract information and knowledge from the data.
How? Some of the answers provided by the speakers at this school are scale-up and high throughput. Scale-up to use the mounds of sequence data and the increasing number of known macromolecular structures to predict function, or scale-up to enable the determination of every macromolecular structure coded by a genome, or scale-up to screen every available compound in either a real or virtual database as a target ligand for the development of drugs. High throughput is a familiar concept when it comes to screening potential ligands for a new macromolecular target, whether through biological testing or through docking computations. High throughput has been a requirement for the methods that were developed to analyze the explosion of sequence and structural data in recent years. Now structural genomics groups are developing high throughput and automated technologies for macromolecular structure determination because these groups seek to complete the structural library of a genome. The automated determination of protein structures and the subsequent abundance of structural information will have many implications for our community with regards to the quality of the structures, the methods for extracting meaning from the vast array of structures, and the way macromolecular crystallographic research is done, funded and published; hence the question at the beginning of this article – are we ready?

The 33^rd Crystallography Course was held by the Int’l School of Crystallography, Ettore Majorana Foundation and Centre for Scientific Culture in Erice, Sicily, May 23 - June 2, 2002. The program began with a historical review of ideas from H. Hauptman (Buffalo, USA), of crystallographic milestones from J. Glusker (Philadelphia, USA), and of drug design from P. Goodford (Oxford, UK) and then covered the components of the pathway from a target gene through sequence, function and structure determination to identification of ligands and effective drugs, beginning with an overview from T. Blundell (Cambridge, UK), Director of the School.

A major component of the drug design pathway presented at the meeting was structural genomics, which builds on the advances in molecular biology, including multiplex cloning, to study the structures coded in a genome. Many consortia have been established world-wide to address this challenge. The consortia bring together molecular biologists, bioinformatics specialists, crystallographers and others to identify target proteins and produce their structures. Groups are variously organized to develop high throughput technologies, to focus on either functional targets or on characterization of new folds, or to compile the entire structural library of a genome. Both tightly coupled and open collaborative large-scale structural genomics consortia, as well as single laboratories focused on a specific target function, were represented at the meeting. Presentations outlined the advances in technology for high-throughput cloning and crystallography, techniques for crystallization of membrane proteins, and the contributions of other physical methods: NMR, mass spectroscopy, and cryo-electron microscopy to structural genomics.

Structural genomics projects promise a profusion of structural data that will match the enormity of the sequence database. Bioinformatics scientists are faced with the challenge of decoding the relationships among sequence, structure and function. Advances in prediction of new folds from sequence, function from structure, and function from sequence were presented and opened the question regarding which of structure, function and sequence is the unique determinant. Another feature of the volume of data available is the several thousand protein binding sites tabulated in the PDB; these data allow groups to identify the common features of a 'beautiful binding site' and to seek such sites in new structures generated by the genomics information explosion. These promising methods will be important to the extraction of information from the structural data that will help in the prediction of function from sequence and structure.

Ligand design and screening provided another focus for the course. Speakers covered the many computational approaches to mapping binding sites and screening large libraries of potential drug compounds with docking algorithms that identify novel chemical classes and serve as excellent generators of ideas. Several different computer screening approaches were presented along with examples of successful applications. Difficulties regarding the development of effective scoring functions for ranking hits and the computation of bioavailability (adsorption, distribution, metabolism, and excretion) were reviewed, and the relative value of virtual high throughput screening (HTS) vs. HTS by biological assay was debated. In each case, the goal is the generation of novel starting points for further drug development. Throughout the discussion, the importance of the quality of the crystallographic structural information was highlighted since it is the basis for the development of interaction potentials and scoring functions and for the identification of binding sites.

An exciting set of presentations outlined the results of drug design efforts and the range of macromolecular targets that can be drug targets. High resolution structural work on Factor Xa, lysosomal cysteine proteases and IMP-1 (a metallo-β-lactamase) demonstrated how powerfully crystallographic results can influence the drug design process. Results on large structures – summarized in a mini-symposium on the ribosome and in the description of the inhibition of picornavirus - demonstrated the importance of these large structures in the development of therapeutic agents.

The lecture portion of the course was supported by an excellent set of workshops covering protein crystallization, databases (Cambridge Structural Database, Protein Data Bank), structural checking (WHATCHECK), docking (DOCK, FlexX), bioinformatics, and metal ion geometries. In addition, teams of participants were challenged by the Director to develop flow charts that mapped the pathway between genes and drugs. The flow chart exercise highlighted the complexity of the drug design process; participants used a variety of images - finding a needle in a haystack, wandering the complex street map of the ancient city of Erice (where the Edelweiss Hotel and the Edelweiss Restaurant are not even on the same street!), the ball passing pattern in a football match – as representations of the pathway from genes to drugs. Finally, J. Bernstein’s (Beer-Sheva, Israel) presentation regarding the pitfalls of patent law and the risks inherent in testifying as an expert witness added a new dimension to the scientific discussions of designing drugs.

Much of the lecture material presented at the course, as well as the flow charts, will be available on the course website, www.geomin.unibo.it/orgv/erice/downloads.htm. The course was timely, instructive, provocative, and fun thanks to the superb direction of N. Borkakoti and, as always, the outstanding local organization provided by P. Spadon, L. Riva de Sanseverino and their many assistants deserves special mention. Director Borkakoti’s program provided a fine summary of the state-of-the-art in drug design and the local organizers made sure that the cultural and scientific activities of the conference were enjoyed by each participant. The crystallographic community is privileged to have access to the fine conference facilities at Erice and truly blessed that Lodovico and Paola are such gifted organizers.

Penny Codding