Fragments vs BACE1: again, and sometimes unintentionally

If I had to pick one target as a poster child for fragment-based lead discovery, it would probably be BACE1, a hot but controversial enzyme implicated in Alzheimer’s disease. Indeed, several inhibitors that have entered the clinic trace their origins to fragments (though unfortunately LY2886721 was just dropped due to liver problems). Four recent papers give a good flavor for where things stand.

The first paper, published in Bioorg. Med. Chem. Lett. by Thomas J. Woltering and colleagues at Roche, is not really a fragment story. Rather, it describes how a high-throughput screen against BACE1 identified a fragment-sized hit that had first been synthesized at Roche in the 1970s as part of an analgesic program. Despite relatively low affinity, compound 1 had good ligand efficiency, and its similarity to other reported BACE1 inhibitors suggested how to grow into the so-called S3 pocket of the enzyme. At the same time, the cyclic amidine “head group” was modified to try to modulate the pKa of the molecule and thus improve the pharmaceutical properties, leading ultimately to compound 12, with good biochemical and cell potency and marginal activity in a mouse model.

In a related J. Med.Chem. paper by Hans Hilpert and colleagues, the Roche researchers further optimized this series of molecules, most notably by introducing electron-withdrawing fluorine substituents around the cyclic amidine ring to further tweak its basicity. Compound 89 was potent, exhibited good pharmacokinetics, and showed impressive target modulation in both mice and rats. The authors state that “a compound from this chemical class is currently undergoing clinical evaluation.” Thus, even though this program did not start explicitly from fragment screening, an initial fragment hit ultimately led to a clinical candidate.

The third paper, in Curr. Opin. Chem. Biol., is much more fragment-centric. In this brief but lucid review, Merck researchers Andrew Stamford and Corey Strickland describe how FBLD has played an integral role in developing BACE1 inhibitors and highlight several successful examples; given that MK-8931 is the most advanced clinical candidate for BACE1, they know of what they write. They note that:

Key elements of successful fragment based drug discovery (FBDD) approaches targeting BACE1 have been the use of X-ray co-crystal structures to design optimal starting points for subsequent optimization and an emphasis on ligand efficiency (LE) rather than affinity to drive the discovery of drug-like, brain penetrant inhibitors.

And if this just whets your appetite, check out the 25 page review in a recent issue of J. Med. Chem. by Suresh Singh and colleagues at Vitae Pharmaceuticals, which contains more than 100 chemical structures of BACE1 inhibitors. The review is certainly not limited to fragment-derived molecules, but it does note that these are superior to the peptidic inhibitors discovered using more traditional approaches.