Molecular Informatics Department, BioFocus plc

The annual Cutting Edge Approaches to Drug Design, jointly organised by the RSC Biological and Medicinal Chemistry Sector and the RSC Molecular Modelling Group, was held at the SCI lecture theatre in London on Wednesday 19^th March 2003.

The Genomic Scale:

As in previous years, the event was aimed at exploring the current state of the art, highlighting key issues in the field.  Professor Graham Richards started the day off with a keynote lecture highlighting the impact of the human genome project on the number of potential protein targets for drug design studies.  He discussed recent developments in the distributed computing project for virtual screening of small molecule candidates against many of these protein targets - this has now expanded to include more than 2 million PCs; effectively a 100 teraflop machine!

Neera Borkakoti continued the theme on the impact of the genome project, focusing on developments in the industry to process and analyse the volume of data.  The role of Bioinformatics in the drug discovery process has now reached an exciting watershed.  With the potential to resolve the key genes in the disease pathway drugs can be selectively targeted, thus reducing the likelihood of undesirable side effects.  However, realising the logical culmination of this goal, that of individual patient-tailored drugs, is still some considerable way off.  James Mills then highlighted some of the more immediate issues that need to be addressed.  For example, as a result of the human genome project, it has become clear that there are thousands of potential target proteins.  However, very few of these have been successfully taken through to market with new drugs (ca. 2‑4 annually).  Part of the problem appears to be in identifying whether potential target proteins fulfil ‘druggable’ and clinical profiles appropriate for development.  It was proposed that filters should be developed to score proteins for suitability.  Furthermore, these filters must be amenable to the genomic scale.  To exemplify the point, a novel approach developed by Pfizer was applied to the entire yeast genome.  As a result, 166 new potential anti-fungal drug targets were identified.

Improving Drug Design:

Similarly, David Rees examined the lack of increase in drugs launched on the market, in relation to the substantial increases in R&D spending.  He suggested that the quality of initial chemical leads has been a factor.  For example, issues associated with their chemical structures can affect development into drug candidates.  In the past, rational design approaches coupled with HTS have been applied to address these chemical issues.  However alternative screening approaches are now becoming viable.  For instance, crystal-soaking methods can routinely be applied to detect weak binding small molecular fragments.  The rationale behind this strategy is that when subsequently linked together, these can often generate significantly stronger potency compounds.  As a result of this form of approach, Astex has developed a number of novel lead compounds against kinase and protease targets that are now in pre-clinical optimisation.

Walter Ward introduced the impact of enzyme thermodynamics on drug design.  Fundamentally, an enzyme is not a single molecular target, since a number of different conformations exist simultaneously.  The exact distribution however is controlled by thermodynamics.  In assay screens, test compounds may bind to several of these forms.  However, the observed potency depends not only on its affinity, but also on the relative abundance of the target enzyme forms.  Consequently in drug design studies, it is preferable to obtain the structure of the intermolecular complex that is most responsible for biological activity.  In short, knowledge of kinetics is required to understand biological activity.

Mike Tarbit moved on to look at the state of the art of computational models for predicting ADME properties of lead compounds.  Typically, multiple pharmacokinetic characteristics need to be optimised in order to yield tractable drug candidates that are safe.  However, in vitro and in vivo studies are typically costly and time consuming.  Consequently, models able to predict these key properties at earlier stages in the drug discovery process are now seen as essential to help improve the quality of lead compounds.  While much work still needs to be done to improve the predictive quality, future software tools accessible in the lab will be capable of optimising compound design both in terms of SAR and ADME in parallel.

Searching Smarter:

Looking to selection methods, David Manallack discussed the quality of pharmacophore screening methods currently available.  Of interest is the development of methods to take some account of protein active site information, as part of the model generation.  This is particularly useful when limited information of the binding site negates the use of virtual docking studies.  The performance of one such improved pharmacophore modelling package, Quasi2, was examined in comparison to the current stable of tools.  The results indicated that the ‘extended pharmacophore’ features used by this approach could lead to improved enrichment in virtual screening hit sets.

To design better libraries for screening, Roger Crossley discussed the emerging and exciting trend to reclassify proteins, not in terms of evolutionary or phylogenetic space, but instead in one more meaningful to drug design:  chemogenomic or ‘drug recognition’ space.  He argued that such a change leads to a potential mindshift in understanding SAR profiles.  For example, new proteins can be rationalised and targeted much more quickly, as close relatives in this latter space offer the opportunity to transfer selected SAR features.  Furthermore, selectivity issues can become much more apparent when understood in these terms.  These claims were subsequently illustrated with a selection of case studies on the superfamily of 7TM G-Protein Coupled Receptor proteins.

To finish the day, Brian Cox undertook a review of current understanding of ion channel proteins, from function to disease pathways and the challenges to drug design.  While knowledge of the structure of the subunits is important, he demonstrated that the impact of the pharmacology of these ubiquitous class of proteins was also important.

In summing up, the event provided a timely overview of changing trends across the industry, attracting in excess of 100 delegates from both industry and academia.  The high turnout demonstrated that the event continues to be well regarded in the annual conference calendar.