The approach described herein will utilize computational and chemical intuition built from work that the team can bring together, including previous work on WDR5 and RBBP4 on design of compounds that affect structural integrity of homo-multimers and on protein folding in general. We would begin by identifying compounds de novo that would be expected to bind to Thr2356 and two other anchor residues in the "donut" region. Thr2356 has been reported in natural variant T->I as showing decreased WD domain homodimerization (Uniprot Q5S007). We would endeavor to design compounds based on this three-point pharmacophore that would contain a central core allowing for modular design. Compounds would be docked based on this pharmacophore for binding prediction using MM and using biased (to keep run times down) short-run MD simulations to demonstrate disruption of the homodimer. Based on hit-to-lead in silico of several hundred of such compounds, we would then take the best of these compounds and search the Enamine database for the most structurally similar compounds from this subset (prioritizing good ADMET properties). This hybrid methodology is being selected in order to avoid the requirement of synthesis; all IP from the de novo compounds will also be available. Up to 10,000 compounds selected to bind to this pharmacophore will be docked and the best 500 subjected to MD evaluating disruption of homodimer. We will propose the top 100 for the next stage based upon these results.
