One of the useful features of fragments is that they can give you unexpected answers to difficult questions. That’s true not just at the beginning of a drug discovery campaign but even after leads have been identified. A nice example of this was published recently in J. Med. Chem. by Steven Taylor and colleagues at Boehringer Ingelheim, and was also presented earlier this year at Fragments 2013.
The researchers were interested in a serine protease called chymase, a potential target for cardiovascular diseases. An initial medicinal chemistry effort had arrived at compound 1, a reasonably potent molecule that unfortunately produced reactive metabolites at the benzothiophene moiety. Crystallography revealed that this moiety binds in the hydrophobic S1 pocket of the enzyme. Traditional SAR thus focused on replacing this moiety with other lipophilic groups.
However, the researchers recognized that their fragment collection offered the potential to discover quite different – and less lipophlilic – replacements, as the average clogP of the fragment collection is 1.48. A thousand fragments were screened, and compound 2 was found to be a weak but ligand-efficient hit. Surprisingly given its relatively polar nature, X-ray crystallography revealed that this fragment bound in the S1 pocket, albeit in a somewhat different manner than the benzothiophene moiety.
Linking this fragment to a close analog of the starting molecule yielded compound 12, and further optimization led to compound 15, with nanomolar potency and improved selectivity against another serine protease, cathepsin G. The crystal structure of compound 15 (blue) bound to chymase was determined. The comparison to compound 1 (red) reveals a number of differences that help to explain the improved selectivity.
In earlier days FBLD was often ignored by project teams once they had identified potent leads. This paper is an elegant example of how an empiricial fragment screening approach can impact a reasonably advanced program by delivering new and unanticipated chemical matter. I suspect we will see more and more examples of this type of fragment-assisted drug discovery.