After 3 months, Ahmed has achieved the following goals:
- To design in-direct calpain inhibitors we docked various fragments of the default MOE software fragment library and ranked the results based on their binding energies and analyzed the frequency of binding of certain fragments with certain residues by SeeSAR. We found that there are certain highly favored residues that interact more often, so-called key residues. We then generated a pharmacophore based on fragments of the highest binding frequency with those key residues, and used this pharmacophore to virtually screen ZINC database, where we retrieved a number of hits. Those hits were filtered and then docked, the best binding poses were then selected and clustered according to their binding patterns in the binding site. Interestingly, we discovered that there are 2 main sub-pockets of the binding site. We then classified the best hits as fragment-like and drug-like. Fragment growing was performed for the fragment-like and core replacement for drug-like hits using SeeSAR's Inspirator mode.
- We applied more optimization for some of the hits generated from the last step of fragment growing and core replacement techniques to obtain enhanced predicted Optibrium properties and Hyde estimated affinities. Then, the best compounds were docked by FlexX and sorted to select the best ones. Molecular dynamics simulations were initially done to the apo-protein to compare the effect of the designed compounds on the mobility of the p10 part of the p35 activator protein, as our aim is to fix the structure of p35 by interfering with calpain binding to prevent the generation of the pathological p25. The simulation results of the apo-protein in combination of other docking results analysis gave us clear insights on how to validate the effect of designed compounds. Preliminary simulations were run to the best selected lead compounds. However, these simulations are currently being re-run for longer time to get more knowledge about CDK5/p35 conformational changes elicited by those compounds.
- The synthetic feasibility of the best compounds were assessed using the synthetic complexity score (SCScore), an MIT learned synthetic complexity metric trained on reactions from Reaxys (https://askcos.mit.edu/) , and Spaya AI score which is a metric related to the probability of a disconnection to happen and consequently to the confidence the algorithm has on this disconnection/route; then optimization of the structures were done accordingly. Some were done by running core replacement to the problematic parts using Inspirator, and others through manual modifications interactively assessed by Hyde-estimated affinity using molecule editor mode till we reached some few promising scaffolds with good synthetic feasibility scores. Preliminary synthetic schemes were planned for these scaffolds, and quote requests were sent to chemical vendors to inquire about the essential chemical building blocks. We are currently planning to generate more derivatives to increase our chemical space.