Open Pandemics researchers answer questions asked on the forum
You asked: “How do you choose the best candidates for the next step? Is it merely the tightest bond, or do you have other criteria?”
Compound selection is a complicated, and regrettably somewhat subjective, process. There are criteria inherent to AutoDock that are used, such as the score and the interactions present, which help ensure key hydrogen bonding is present or the ligand fills the space thoroughly (nature abhors vacuums, but our scoring function tolerates them). After these sorts of criteria are applied molecules are visually inspected by multiple team members. This serves to hedge against known issues in the scoring function (i.e., if it tolerates a bad electrostatic interaction more than we believe it should), and also to help select a set of molecules for purchase that are diverse in both their structure and their interactions with the binding site. When the AutoDock and the rest of us come to a consensus, those molecules are ordered, and biological testing begins.
You asked: “Can you make any estimates on possible side effects, the Achilles heel of most drug design projects I believe. Does the size of the molecule play a role in that, for example.”
This is an excellent question, but still a somewhat premature one at this stage. Compound selection generally starts with hit identification and is followed by lead optimization. We are working on hit identification. Lead-like molecules are those that one is pursuing as the actual identity for a drug, and have issues like toxicology addressed, while hit molecules are those that show initial activity and give a scaffold to optimize on. As we're still establishing the potency and binding mode of our initial hits, our focus is more on characterizing significant interactions with the binding site than on advancing an individual structure as a drug candidate. All of that said, we're hopeful that these will advance to a stage that we are considering drug-likeness more and more.
You asked: “How much of the subsequent testing is done in-house at Scripps, and at what point do you send it out?”
This depends on the target. We send molecules to collaborators that are best suited to test against each target. In the case of spike protein, that work is currently being carried out entirely at Scripps, in collaboration with two other labs at the institute. We even share a graduate student with one of our collaborators! By contrast, for PLpro and Mpro, the compounds immediately go to the UK, where we have collaborators at Oxford doing the wetlab work. And this is how collaboration goes in the academic community, and one of the real highlights of being on a project like this; specialists from around the world get to do their parts and contribute to a larger project.
You asked: “Any estimates on the time frames?”
Time frames are always a tricky question, and as soon as you give one you know with absolute certainty it will be wrong. All compounds ordered thus far are being tested, and internally we are happy with the data we're receiving and excitedly planning next steps. However, our major collaborator for purchasing compounds is Enamine, based in Kyiv (https://enamine.net/). Current geopolitical events have obviously impacted their ability to provide compounds. We hope for their safety throughout the conflict, and encourage you to check out their twitter (@EnamineLtd) if you're interested.