Dr Nir London, Senior Scientist, Department of Organic Chemistry, The Weizmann Institute of Science, has brought together a wide range of players—from academia in four countries, to biotech and contract research organisations, to specialised software companies—to accelerate the development of a drug against COVID-19. Dr London reveals more details about this endeavour, in interaction with Sanjiv Das
Tell us how the global consortium was formed? How were the partners chosen and why? What does each partner bring to this collaboration?
Our group at the Weizmann is focussed on covalent ligand discovery and chemical biology, which is a specific niche within the field. In the past, we have collaborated with the group of Frank von Delft from the Oxford University and the Diamond Light Source on covalent inhibitor discovery. As soon as the Oxford team expressed the Mpro protease – an essential protein in the virus, which is especially sensitive to such covalent inhibition, they sent it to us, so that we could apply our technology to it.
Indeed, we identified a few dozen hits, and they, in turn, were able to determine the co-crystal structures of these and additional hits. At that point, we had promising starting points but needed to optimise them. A team of like-minded collaborators coalesced around this challenge. PostEra, a California biotech, established a web-platform to disseminate our data and to crowdsource design ideas from chemists around the world. They also have machine learning algorithms to rank the synthetic feasibility of such designs. A team from Memorial Sloan Kettering Cancer Center contributes molecular dynamics simulations via Folding@Home to prioritise the most potent designs.
Enamine, one of the largest chemical vendors in the world, agreed to synthesise these proposed designs at significant discounts and is taking an active part in the project. Other groups, companies and CROs such as UCB Pharmaceuticals, Sai Life Sciences and Nanosyn, have since joined, contributing their expertise and/or services – all with the common single goal to progress a compound to the clinic as fast as possible.
Researchers across the globe are working to find medicines and vaccines against COVID-19. Where have you reached in this process? At what stage is your research? What have been your major findings so far?
So far, we have released publicly more than 90 structures of our target protein bound to various ligands, as well as a compressive list of covalent fragment binders. This is an unprecedented dataset which could significantly progress the design process. Our crowd-sourcing effort resulted in more than 4,000 designs, of which we were able to synthesise more than 300 and test about 150 in a protease activity assay. This work has already produced significantly better hits than the original fragments, but not yet of a potency that can be advanced to cellular studies. More compounds are being synthesised every day and we hope to be able to advance a compound to animal studies within half a year.
How and why is your approach to tackle COVID-19 different or novel? Will it also help to scale up production in huge quantities since that is an urgent need?
First, I want to stress, that in all likelihood, our efforts will not reach the clinic in time to serve as a therapeutic option for COVID-19. A new drug development project can take years. We are embarking on this important mission, first to address a horrible scenario in which no other treatment will be found in a shorter time-frame; and second, to make sure that when the next coronavirus outbreak will emerge, we will be ready. The protease is highly conserved amongst coronaviruses and we believe our effort will result in a broad-spectrum antiviral that would be able to address future outbreaks. If something like this was done for SARS 17 years ago, we wouldn’t be in this bad situation today. In a sense, this is an insurance policy.
We are still in very early stages but have promising initial results. Other approaches to tackle the protease rely on bulky peptidomimetic compounds with poor pharmacological properties. We take a bottom-up fragment-based approach and collect data that enables us to build smaller and more efficient inhibitors. In a larger sense, our approach differs in several important ways from the traditional drug discovery pipeline. First is the fact that everything is shared openly with no patents, intellectual property or commercial considerations. Second, instead of the traditional sequential model, we try to start with as broad as possible funnel, testing more and perhaps also riskier compounds, hoping to reach potency early, with less iteration. One of four major design considerations in this project is that the final compound will be cheap and easy to make. Thus, once it shows safety and potency, scaling up the synthesis and production will not be a major hurdle in distributing the drug to where it is most needed.
Can you elaborate on the techniques and technologies your team is deploying to generate actionable insights and optimise them?
As I mentioned, our lab specialises in covalent ligand discovery. We’ve developed several technologies towards this goal including both computational and experimental techniques. Experimentally, we have established a screening method to detect small fragments (much smaller than the average drug) that are mild electrophiles, that is, not too reactive. We use mass spectrometry to assess whether these are able to irreversibly bind a target protein. We have screened a library of 1,000 such compounds against the viral protease and have found several series of fragments that are able to selectively bind to it. A complementary approach we used is computational covalent docking, in which we model theoretical covalent ligands in the protein active site, and this method predicts which ones will be able to bind irreversibly. This approach has also yielded preliminary hits, that were validated experimentally and that we are currently following up on. To assess progress, we’ve established a high-throughput protease activity assay that guides our iterative optimisation.
Open-science has emerged as an urgent need in these times. Can you explain how are you supporting and facilitating it?
Open-science is one of the main pillars of this project. All of the data we acquire is published online almost in real-time, to allow the scientific community to provide fast feedback and allow them to use this data for other projects. All of the protein structures and chemical structures were made available, a long time before any scientific publication. There is no intellectual property involved and no material transfer agreements. This also includes crowd-sourced designs. This open collaboration allows us to move fast with no bureaucratic obstacles. Our future plans and testing cascade are all available through a public website, and we welcome ideas, feedback and scrutiny. This is one of the unique aspects of this global collaboration.