W0310: Structurally-biased Combinatorial Design of Antivirals

We have been investigating methods that combine structure-based and combinatorial chemistry approaches to drug design. In these methods, high-resolution structures of target molecules are used to design a general template for a ligand. This template serves as a bias in the design of limited libraries (100-1000) compounds whose diversity is focussed on regions of “chemical space” that are likely to bind the target. Two examples will be presented. In the first we have used the program MCSS to develop a template for molecules that bind to the capsid of poliovirus and related viruses and inhibit conformational changes associated with cell entry. We have developed a template for a lead that differs significantly from most published drugs in this series. We have developed a mass spectrometry based assay that is suitable for high throughput screening crude libraries, and demonstrated that the approach yields several respectable leads (MIC’s in the micromolar range). In the second example we have used the recently reported structure of a complex between a C-terminal fragment of the catalytic domain of the Herpes Simplex polymerase ands its processivity factor UL42. UL42 is an unusual processivity factor in that it binds DNA strongly as a monomer. The C-terminal fragment of the HSV is comprised of two helices separated by a short stretch of extended chain. In the structure the C-terminal helix lies in a deep groove in one surface of UL42. Genetic studies suggest that the contacts involving the C-terminal helix dominate the pol-UL42 interaction. We have developed a template that allows the synthesis of libraries of compounds that display amino acid like side chains in a fashion that closely mimics the positions of the i, i+3, i+4, and i+7 side chains on one surface of the helix, and developed computational tools to optimize the selection of side chains to be displayed. Computational studies suggest that these compounds bind as strongly as the natural peptide.