22 October 2018

Fragments vs Ras – part 3

Six years ago we highlighted papers from Genentech and Steve Fesik’s group reporting fragments that bind to Ras-family proteins, which are among the best validated but most difficult anti-cancer targets. The fragments bind some distance from the GTP-binding site, but can block Ras signaling by interfering with important protein-protein interactions. However, the most potent molecules reported bound at this site with just ~200 µM affinity, and we concluded by musing that “it still remains to be determined whether this is a ligandable site on the protein.” As reported recently in Nature Communications by Terence Rabbitts and collaborators at the University of Oxford, St. James University, Domainex, and the University of Aberystwyth, the answer appears to be yes.

The researchers screened HRas against 656 fragments, each at 200 µM, using SPR, resulting in 26 initial hits. These were tested again by SPR against active-form protein (bound to the GTP mimetic GTPγS) or inactive protein (bound to GDP). A single compound, Abd-1, was selective for the activated form of the protein, and did not bind when the protein was complexed to an antibody the researchers had previously generated that binds at the same PPI site.


Abd-1 had low affinity and was not particularly soluble, so the researchers looked for analogs with better properties, resulting in Abd-2, which binds to both HRas and KRas. Further growing in the direction taken by the Fesik group did not lead to significant improvements, but a breakthrough occurred when the researchers grew off a different region of the fragment, towards what looked to be the wall of the small pocket. As Trevor Perrior mentioned at the DOT meeting last month, this led to the opening up of a new channel and a substantial boost in affinity for Abd-5. Further growing allowed the researchers to trim off the right-hand portion entirely, leading to Abd-7, with mid-nanomolar activity and good ligand efficiency. Crystallography revealed that, despite the conformational changes, the core of Abd-7 still binds in the same location as Abd-2.

Not only did Abd-7 bind tightly to KRas, it also inhibited the pathway in cell-based assays (albeit at 100-fold higher concentrations), presumably by blocking interactions with Ras-effector proteins. The compound also showed low micromolar activity against cancer lines with different Ras mutations in cell viability assays. The researchers note that “the observed discrepancy between affinity (in vitro Kd) and efficacy (IC50 in cells) is a known challenge that can be addressed through chemistry.” Other possible challenges include metabolic stability and oral bioavailability, neither of which is discussed. Nonetheless, the paper reveals that this site in Ras family proteins is ligandable. It is also a useful reminder that proteins can be remarkably plastic, and sometimes the best route forward really is by slamming into what appears to be a solid wall.

1 comment:

  1. Hi Dan,

    it is always great to see new ligand for Ras. It is not an easy thing to find good ligands so these crystal structures are really good to see. I must say howerver, I am concerned abou the affinity values of Abd-5,5 and 7 quotes as 220, 38 and 51 nM by NMR CPMG relaxation dispersion methods. The experiments seem to be just CPMG all be it using the PROJECT_CPMG pulse sequence. There appears to be a fundamental misunderstanding of how to generate a Kd form this experiment. A plot of ligand intensity varying with protein concentration does not measure a property of the complex but just how well the CPMG works. Increasing or decreasing the CPMG delay will change the 'Kd'. The affinity is very important as samll molecule nM binders of Ras are not known. This key bit of data should have been interogated by the reviewers and this data should not have been published. Unfortunately this cast doubt on the mechanism of the other assays.
    The obvious question is why did they not use SPR to detemrien the Kds as it can easily detemrine Kds in this range.
    We really need better peer reviewing of biophysical data in my opinion!

    ReplyDelete