Pin1, a potential anti-cancer target, catalyzes isomerization around phosphoserine-proline and phosphothreonine-proline bonds. Its binding site is relatively shallow, complicating efforts to discover small, non-peptidic inhibitors. In a recent paper in Bioorg. Med. Chem. Lett., Jonathan Moore and colleagues at Vernalis describe their fragment-based approach to tackling this problem.
The researchers used NMR-screening of roughly 1200 fragments to identify five that competed known inhibitors; Compound 4 (see figure) was the most potent, with an IC50 of 16 micromolar in a functional assay. NMR experiments showed a weaker binding interaction, on the order of 200 micromolar, and surface-plasmon resonance (SPR) experiments were even less conclusive: the compound showed super-stoichiometric binding, indicating that multiple molecules were interacting with the enzyme rather than sitting specifically in the binding site. However, the researchers were able to obtain a crystal structure showing that the molecule binds in a hydrophobic pocket at the active site. Although they don’t mention it in this paper, in public presentations the researchers have reported seeing additional molecules of Compound 4 pile on top of each other, essentially forming a stack on top of the protein. Perhaps, as the authors suggest in the supplementary material, this is an example of a particularly insidious aggregation phenomenon: a legitimate hit that can also form aggregates.
In order to access other parts of the protein, and reduce the propensity for aggregation, the researchers relied on analog screening and modeling to generate Compound 18a, which is a more “three-dimensional” molecule. Further elaboration led to a series of compounds such as 19e, with low nanomolar potency, as well as one-to-one binding in the SPR assay.
Despite their potency, these molecules were inactive in cell-based assays, likely due to high polar surface areas and their resulting low cell permeabilities. To fix this, the researchers replaced the benzimidazole fragment with a naphthyl group, which led to a decrease in biochemical potency but did lead to cell-active molecules such as Compound 23b. Moreover, a crystal structure revealed that this molecule binds in a similar manner to the original fragment 4.
This paper exemplifies another example of fragment-assisted lead discovery: the original fragment morphed from an indole to a benzimidazole to a napthyl group, yet the final molecule still owes a debt to the initial fragment.