STRUCTURE BASED PREDICTION OF DRUG-PROTEIN INTERACTIONS AT ALLOSTERIC SITES: TOWARDS DIRECT INHIBITION OF THE RAS ONCOGENE
Date of Graduation
Cell and Regulatory Biology
Doctor of Philosophy (PhD)
Alemayehu A. Gorfe
James M. Briggs
Richard B. Clark
Carmen W. Dessauer
John F. Hancock
John A. Putkey
Ras is a monomeric G-protein that mediates multiple signaling pathways. When bound to GTP, Ras binds to and activates effectors and when bound to GDP it does not. Somatic mutations occur in ~15% of all human cancers and cause Ras to be predominantly bound to GTP. Recently, several studies revealed that Ras exists in an ensemble of at least two states when bound to GTP. In the state 1 conformation, Ras is unable to properly bind effectors but holds potential for binding ligands. Conversely, at allosteric sites, state 2 conformations readily bind effectors but not ligands.
To incorporate the flexibility of proteins into structure based drug discovery, we developed a protocol called LIgand Binding Specificity Analysis (LIBSA). LIBSA uses ligands to probe the surface of a target protein and identifies low-affinity ligands that bind consistently to allosteric sites. This allows for a small library of compounds to search for allosteric binding sites on fully flexible targets. Moreover, without the requirement of defining a binding site on the target, multiple pockets can be screened simultaneously. Using LIBSA and an ensemble of conformations, two pockets near the
switch regions of Ras, switch 1 (p1) and the core β-sheet/switch 2 (p2), were found to bind ligands with both a high affinity and a high preference. This suggested that simulations coupled with docking experiments could reveal novel binding sites.
As a test of this probabilistic model of binding, we investigated if the anti-cancer activity of Andrographolide (AGP), a medicinal extract from the plant Andrographis paniculata, and three derivatives (SRJ09, SRJ10 & SRJ23) was due to a direct interaction with Ras. Ensemble docking revealed that the SRJ compounds best-targeted p1 or p2. Further examination of the docked complexes using MD revealed that binding was stable only at p1 and not at p2. Comparison of the MD optimized complex to previously determined crystal structures suggested that SRJ23 could inhibit nucleotide exchange on Ras. Experimental validation of this hypothesis revealed that treatment of cells with AGP and the derivatives acutely inhibited nucleotide loading of wild type Ras. Amazingly, prolonged but not acute treatment of transformed cells with SRJ23 and SRJ09 reduced the GTP loading of K-RasG12V thus elucidating a requirement for exchange factors for maintenance of oncogenic Ras signaling.
Computation, Drug Discovery, Exchange Factor, Ras