November 4, 2021
We are here with Dr. Ed Griffen, Technical Director and co-founder of Medchemica, a company specializing in Molecular Design and Analysis. Ed is also one of the core team of the COVID Moonshot project.
Good Morning, Ed. Can you tell us a bit more about your career?
My academic training was as a synthetic organic chemist. I started work in the pharmaceutical industry with AstraZeneca pharmaceuticals, as a team leader, and a medicinal chemist, initially in the arthritis and pain areas. Then I spent quite a period working in anti-infectives, and then I moved into oncology.
In all of these disease areas I had a strong interest in exploiting computational chemistry as much as possible. During my career at AstraZeneca, I was involved in various programs and projects to improve how we worked. I actually had a secondment into our computational chemistry group to do some method development, went back, then applied that in medicinal chemistry.
And then in 2012, because of a reorganization with AstraZeneca, three of us decided to spin out a company (Medchemica) to exploit some of the tools that we've been developing, and I've been doing that and medicinal chemistry consultancy ever since. My consultancy work is broad ranging but has a focus in the anti-infective area. That was essentially what got me connected to the COVID Moonshot, when it started back in 2020.
Thank you. That's really an impressive career. Can you tell us a bit more about the COVID Moonshot project?
The very beginning of the COVID moonshot was when the sequence of the COVID Main protease (MPro), which is a protease involved in making essential viral proteins so that the virus can replicate, was sent from China to the Diamond Light Source group in Oxford. They rapidly cloned and crystallize the protein. They ran a fragment screen and they released around 90 fragments co-crystallized in about six weeks from receipt of the sequence. When those fragments were released, the group in Oxford (Frank von Delt and Martin Walsh and teams) - then said "what can we do with these fragments?" They started discussions with Alpha Lee at PostEra, John Chodera at Memorial Sloan Kettering, and Nir London at the Weitzman Institute. They all worked together a bit before and said, "well, why don't we go and actually try and discover a drug".
I was actually at a conference with John, giving one of the keynote speeches, and by the time I got home on the 14th of March, 2020, there was swirling about, "we need a medicinal chemist who do we know". I got roped in early on, and the team decided that "we don't want to find fragments, we don't want to find ligands, we want to find a drug, so we need a drug hunting team", which is very different beast to just a purely academic exercise, and so I got pulled into that. Then over the last 15 months now, we've built up the teams. It took about six months to find good ligands, and then it's taken us another year to turn those into things that have got oral bioavailability, a good safety profile and show good cell activity.
So, we're now closing in on identifying compounds that could go forward for clinical trials.
An impressive achievement. You say that over 100 Scientists are now somehow involved or have been involved in the initial phases. Also the funding is not conventional. So how was the project funded at the beginning, and how that evolved?
So, to start with the project was funded basically by philanthropy, and grant money that was repurposed for a better cause from the various academic institutions. Also, in particular PostEra's investors were keen to support this to start with.
We've had support from Oxford University, Weizmann Institute and Memorial Sloan Kettering in particular as organizations. So, that was the early phase of the funding. We've then also had various other medium sized grants, and had a huge amount of supporting in-kind. One of the in-kind supporters is CDD.
We've also got very significant in-kind support from Novartis, who is donating a huge amount of testing for us. Takeda is supporting us with testing. UCB have been very generous with both medicinal chemistry design and computational chemistry support. In the next phase, we've formed an alliance with DNDi (Drugs for Neglected Diseases initiative) and they've adopted us as a project. We have a lot of real chemistry and biology support from DNDi, as well as key supporting functions like communications, legal, and we've secured funding for the end of Lead Optimization and the pre-clinical phase from the Covid-19 Therapeutics Accelerator.
It would be interesting to learn more about the drug candidate selection process.
It's really quite a classic straightforward direct acting anti-infective project. The things you care about are, does it work in cells? Have we got oral exposure, because we want an oral drug that you could take as soon as you got early symptoms, maybe after an early test. But looking forward, you might even be able to say well we want to take this prophylactically or in parallel with being tested. So, it's going to be an oral drug of course, because you can't have people using IV drugs in the community and critically it's got to be extremely safe.
The testing cascade is everything you would expect, so we've got enzyme assays, multiple cell assays. Cell assays are challenging, because you're dealing with a known human airborne pathogen. In terms of how you have to handle it, it's a BSL 3 category organism, which means there aren't that many places in the world where you can get your cell testing done if you're not in a large pharmaceutical company, and even if you are it's still challenging. We've been very lucky to establish good relationships with multiple labs with BSL 3 facilities so we tested multiple cell lines, and then it's a classic in vitro and in vivo ADME/T cascade, looking for oral bioavailability, good absorption, making sure you've got no toxicities that would be threatening to actually using a drug in the community. If you're giving it to a very wide population, you're worried about things like drug-drug interactions because obviously your potential patient group, particularly as we know, COVID is particularly severe in older patients, are highly likely to be taking other medications so you can't be altering the effects anticoagulants or anti-psychotics or oral contraception any of the other critical drug-drug interactions that you worry about. It's a complicated cascade but to a very simple purpose. It is not conceptually difficult, it's just there's a lot in it.
And so the question that I was anticipating you asking me which I'm going to answer anyway...
What is the benefit of having CDD Vault? So, you could ask me the question, why did we come and talk to me within a month of starting?
I immediately knew that we would have a complicated cascade, we would have to test in multiple places because we need a wide range of data and beyond that at various points we've lost labs for a month because people had got COVID, or they've been locked down. Trying to run a drug hunting project in the middle of a pandemic - it's a complete pain in the neck.
We have around about 12 different centers generating data for us. I never met our chemistry team in person. There are about two dozen of us who input into chemistry design, and we've been spread all over. At the start of the project, we had a huge number of contributors all over the world, and we still do.
We had to have one place where people can see the data, and it has to be in a form where everybody can understand it because we can't all get together in a room and hack it out. We have to have a central repository, and it has to be appropriately formatted so that people aren't asking questions about it and don't make some errors of interpretation. Those of us who'd worked on big drug hunting projects before were absolutely convinced that we had to have a robust system. We've got some outstanding computational chemists working on the project, some very good software developers. But they needed to be doing things that were unique. Building a chemical and biological data database has been done before. Why would you do it again when you can buy an off the shelf solution?
It is great to hear about the CDD Vault being used by the whole COVID Moonshot community, and the benefits it brings to networked projects. Are there any other critical challenges that you see, not just in the past, but in the future as well?
There are normal drug hunting challenges, for instance hitting unexpected toxicity. Many things can go wrong in preclinical research, and then clinical research, we know how to deal with those so we can mitigate those professionally. You worry about technical challenges around how we run the clinical trials. Phase one, pretty straightforward for normal compound. By the time we reach Phase two, I suspect that COVID-19 will still be endemic -unfortunately- so it is realistic to presume multiple sites running the trial.
The team challenges are, curiously, not as bad as you might believe in a massively distributed team. That's because everyone is bought into the long term goal of actually being part of the solution to this pandemic.
The new challenge that I haven't run into before is the politicization of the entire project. We are aiming to produce a drug that is as absolutely as cheap as possible, and broadly available, irrespective of what territory you live in - high-, middle- or low-income country. One of the COVID moonshot goals is "accessible therapy for all". If we manage to deliver a drug, it's it will aimed at global access. I can't recall a drug hunting area that has had so much political heat associated, probably, since I started my career when the first AIDS therapy drugs were coming out.
All our enzyme data is published, completely in the open. Anybody in the world can pull that off our website. We don't publish the cell or ADME/T data because people are desperate for therapies, and one of the things that we are concerned about is the consequences of putting too much data into the public domain of new compounds until they have had proper safety testing. People could be injured very badly. There's a huge disclaimer on our website that basically says, "These are research compounds, they are not drug, under no circumstances should you take these as therapy, until they've been completely tested".
So, you mentioned the benefits of CDD Vault in handling the data, how do you do that in practice?
Once we've decided to make a compound, the structure goes into CDD Vault. Then we can calculate their properties and analyze them. Once we've decided to actually progress the compound to synthesis we flag that so it moves into a different project. Then once it's made, again, it moves into a different project. So, we can track the compounds that were planned for synthesis, and the compounds that actually we made. As soon as compounds are made, they progress into the testing project.
Also, we use CDD Vault to record how much material we have, because actually having the quantity data visible when we want to send it to multiple places is absolutely vital.
All the testing data goes into CDD Vault, there is no data that we gather that doesn't go into CDD Vault, that's one of our rules, so we don't have any passing of Excel spreadsheets around or other nightmares.
We've got a couple of people who upload most of the biology data, we've got calculation data being put in as well. Computational chemists upload their calculations so we can do analysis that way, and analyze calculated versus measured data.
We use the CDD Vault "collections" feature to put together collections of compounds that might be going off for particular ADME/T tests further down the testing cascade. We are now doing -thanks to the excellent CDD support team- some quite complex calculations within CDD Vault. We're doing some crude dose to human calculations that help ranking compounds for the very late-stage evaluation. We also store experimental procedures and we use a bit of the ELN. So, I think there isn't any part of CDD Vault we don't use, to be quite honest.
What will be your advice to anybody thinking of starting a large collaborative project?
My advice to anyone is think about your data before you generate it. It's a lot easier than the alternative. So, we planned out the flow of physical material, and in parallel the flow of data, very early on. Ours is still not a perfect setup, I know we could improve it. If we'd had more time to get set up, we probably would have invested in automating more via the API. That's the only thing that I probably wish that we could do differently, just automate more of the data upload process. It's still a bit manual, which means that if people get ill, or if they're on holiday, sometimes things may get dropped.
So, think about how your data flow early, think about your data formatting. Get somebody ideally who's done it before knows what they're doing, because data management is a subfield in its own right. Think of simple things like: prepare your data in a really clean, consistent way; divide your data right; it's much easier to have data that's very divided up. Don't store your data as ranges.
The other thing is, be disciplined. I said, "all structures, all data goes in one place". Do not allow people any private side-channels, because they think they're being quicker, and then something goes horribly wrong. It's not time saving!
Where's the Covid Moonshot going next?
We're aiming to deliver five compounds into preclinical development over the next three months, and that will be one compound delivered into clinical development, end of next year, probably maybe early 2023, because we've got to scale up to key large scale synthesis to be ready for phase one.
There is a successor to the COVID Moonshot because, you wouldn't be surprised to hear that, a team that got itself together and worked on well on a pandemic virus is now working out how we could build a bigger and better program to work on pandemic preparedness.
At the simplest level, in the last 20 years there have been three or four coronaviruses that have crossed over from other mammals, into the human population. So, by that count, we've got six to seven years before the next one.
As a simple rule of thumb, if something's happened multiple times, you shouldn't be surprised when it happens again. We don't really, as either a scientific community or as humanity, have an excuse for not being better prepared.
As the compounds move into the clinic, the research team is moving into looking at pandemic preparedness so we're looking at pan-Corona potential therapies as well as some other viruses.
At the moment we're putting together our plan for how we could really industrialize our approach in the sense of making it bigger, faster, better, and able to deal with more viruses. So, that's what part of what I'm doing now, and we have CDD Vault as part of that as a central data store, because it's worked once, so we're going to do it again.
The reality is that CDD Vault is a web-based tool that everybody can access. We will still be a very distributed team working globally, and we will need to have access, and in this case, we'll be looking at some BSL 4 organisms as well, so need to have some extremely specialized biologists on board.
There really isn't another solution that fits as well for us.
Related: Dr. Lizbe Koekemoer of Oxford University describes how the COVID Moonshot team leverages CDD Vault to effectively store, manage, and share data among multiple international partners