06 June 2012

Fragments versus Ras – Part 2


Practical Fragments recently highlighted a paper from Genentech in which researchers there discovered fragments that block the activity of the prominent oncology target Ras. Illustrating just how much interest there is in this protein, Stephen Fesik and colleagues at Vanderbilt University have just reported results of their own work in Angew. Chem. Int. Ed.

Fesik is famous for SAR by NMR, the first truly practical approach to fragment-based lead discovery. In the current work, the researchers also used NMR (HSQC with 15N-labeled protein) to screen 11,000 fragments, yielding about 140 binders to the GDP-bound form of K-Ras. A number of these were then further characterized crystallographically: of 20 cocrystal structures obtained, all of them were found to bind in the same hydrophobic pocket identified by the Genentech researchers. Fesik and colleagues also noticed a nearby, electronegative cleft, and grew one of their fragments (compound 1) to take advantage of this. This led to compound 12, the most potent compound reported. In addition to binding to the GDP-bound form of K-Ras as assessed by NMR, this compound also inhibited Sos-mediated nucleotide exchange in a functional assay.



Overlaying one of these compounds (blue – similar to compound 12) with the Genentech compound DCAI (red) reveals that while both compounds bind in the same hydrophobic pocket, they make very different contacts.



Of course, it still remains to be determined whether this is a ligandable site on the protein (ie, whether these – or any – molecules can be advanced to high potency). Given the importance of Ras, it’s certain that lots of people are doing their best to find out.

1 comment:

Michael S. Chimenti said...

It is interesting that the binding pocket is not readily apparent in the ligand-free structure, but is formed from a rearrangement of the M67 and Y71 side chains upon ligand binding.

That would seem to argue for checking all available structures of a target or even doing some MD runs to look for persistent, alternate conformations. What do others here think?