Fragment-based and structure-guided drug discovery

We have been using fragment-based drug (or ligand) discovery (FBDD/FBLD) methods since 2008 to target various clinically important proteins to validate novel targets and to create chemical tools. We are doing this with various collaborators in Cambridge and beyond, with chemists and biologists who complement our biochemistry, biophysics and structural biology expertise, creating powerful multidisciplinary teams.
One of our guiding principles has bee to target not the active sites, where drug most typically bind, but sites that regulate the proteins, such as protein-protein interaction (PPI) sites or allosteric sites. These are often very challenging to target as a typical PPI interaction site is large and relatively flat, lacking the three dimensional features associated with typical small molecules binding sites targeted by drug developers. One of the key challenges we face in this area is to keep the size of the inhibitor small while achieving high potency. ALlosteric regulators are similarly challenging as the mechanism by which binding to a distant site affects protein’s activity is poorly understood and therefore the development of inhibitors is driven not just by increase of affinity, but by mechanistic understanding of what causes the desired effect.

We use a number of complementary biophysical techniques both to screen fragment libraries and to validate the fragments that have been identified in the initial screen. Structural biology is a key to fragment based approach and we make extensive use of X-ray crystallography to analyse the molecular details of the fragment-protein complexes to guide computational and synthetic design of higher potency chemicals. We typically use significant time in developing suitable experimental platform to enable routine structural analysis. One of the methods we try to use for all of our targets is high-throughput crystallographic screen at Diamond synchrotron XChem facility where in a single experiment we will soak some 1000 fragments into our crystals, resulting in hundreds of structures with (hopefully) tens of well-defined hits.

Some examples of our work in this area include the development of novel inhibitors for RAD51 , targeting the binding site for BRCA2 and (at the same time) the site of RAD51 self association. This inhibitor was a result of merger of fragment hits with native peptides, followed by subsequent medicinal chemistry optimisation. The resulting inhibitors, CAM833, is the only fully characterised inhibitor of RAD51 and suitable for in vitro and in vivo work. One of the special aspects of this work as the extensive protein engineering we did in the development of surrogate protein system using thermostable archeal RAD51 ortholog RadA as the starting point.

Another notable example is with ubiquitous protein kinase CK2α (formally known as caseine kinase 2α) where we identified a cryptic pocket close to the active site, enabling the development of inhibitors with significant specificity. The most effective molecule, CAM4066, uses the so-called αD pocket as an anchor point from which it extends to active site where the “warhead” blocks the active site.

With Aurora A, kinase controlling cell division, we used purely fragment-based methods to target a binding site for TPX2, protein that both activates the kinase allosterically and controls Aurora A subcellular localisation. The lead molecule CAM2602 which blocks this protein-protein interaction and arrests cell cycle, as expected.