An Interview with Alejandro Chavez, MD, PhD

Since joining the Taub Institute, your laboratory has developed a method for studying dozens of neurodegenerative disease models in parallel, and cataloguing the common and unique mechanisms by which proteins implicated in neurodegeneration lead to cellular pathology. Now, you are applying a similar approach to research on the SARS-CoV-2 virus. How did this happen?

Basically, very early on after I got to Columbia, I realized that the same idea that I was using to study multiple neurodegenerative disease models at once could essentially be used to study multiple members of any family of proteins at once. I started thinking about other protein families that would be interesting to study and realized viral proteins might be an interesting area to explore because most viruses don’t actually have an approved therapeutic. So, we started thinking, OK, what do a lot of viruses have in common that there’s historical evidence can be drugged? And that kind of led us naturally to viral proteases. About 50% of viruses encode their own protease in their genome and viruses like HIV and Hep C have been treated rather potently with protease inhibitors, so that gave us hope that maybe we could explore the idea of using protease inhibition as sort of a general strategy for more than just HIV/Hep C. So, we did some experiments, saw that we could detect phenotypes from expressing viral proteases in our yeast system, and that we could do this barcoding idea that we have to look at several of them at once. That is where the project was going when this outbreak in Wuhan, China happened.

Viruses are dependent on the protease; if the protease does not function, the virus is dead, so that’s critical. What I do then is I take that little viral protease and express it in a simple cell, yeast. It turns out when the protease is expressed in yeast it causes them to be sick and not divide. Now, if I take that yeast that’s expressing the protease and I add a compound that blocks the protease’s function the yeast cell is now able to grow. This result tells me that I can use yeast as a biosensor for protease activity: yeast is literally sensing the activity of the protease and if it’s grown in the presence of a compound that inhibits the protease it lets me know that beautifully. So, protease active: no growth; protease inhibited: tons of growth.

Now, I take my yeast cell and give it a little DNA name tag, ‘John,’ and I can associate it with SARS-CoV-2 protease. So, if I see ‘John’ at the beginning of the experiment but not at the end of the experiment that must mean that the protease was active because no ‘John’ survived. On the other hand, if I add a drug that blocks the protease, I see ‘John’ at the beginning and I see ‘John’ at the end, meaning the protease must not have been active. In this way, I can make hundreds of unique yeast cells, each with a different name tag—John, Mary, Jim, Joe—and associate each of them with a different protease—HIV, SARS-2, Zika, MERS, etc. And I can mix them altogether at once, read all the barcodes at the beginning of the experiment and say, ‘Ok, I see all these names, what names remain at the end of the experiment?’ And by doing that I can know if the drug worked on any of those proteases, all at once. Basically, every single experiment I do, I test the activity of multiple proteases and the ability of a single compound to block it, but I’m doing this in thousands of wells, every well gets a different drug.

This concept of using DNA-barcoding of yeast to study multiple proteins at once is similar to the approach we originally developed to study neurodegenerative diseases except, in that case, instead of expressing viral proteases, we are expressing proteins implicated in human neuronal loss and dysfunction within our yeast models.

We start first with FDA approved compounds and we’re doing that because you’re hoping there’s something that’s already used in humans that can block SARS-CoV-2 or other viruses, since these would have a more rapid translation into the clinic. After screening FDA approved compounds, we’re going to branch off into a few other chemical libraries, such as natural product libraries or libraries of compounds that have only been tested in preclinical but not clinical trials. We also have a bunch of compounds that look like they should be protease inhibitors but have never been tested against the dozens of different viral proteases we are studying. We have thousands of those, frankly, so it’s quite a diverse set. You don’t really know what’s going to work, so you kind of try quite a few things and try different classes of compounds to see if we get anything that looks reasonable. And, if we do, then we will try and take it and diversify it, to make it a little better.

I definitely am. I think we’ve learned a lot. The system seems likes it’s actually able to give us expected results, and soon we’re going to start seeing how far we can actually scale that. Our goal is that we don't just want to find a cure for SARS-CoV-2; we really want to know is there a compound out there that is broadly active, because this is the third coronavirus to really cause epidemic/pandemic scale problems. It’s not going to be the last. Wouldn’t it be great if we had a compound that not only inhibits SARS-CoV-2 but inhibits a large portion of coronaviruses? And, so, we think our system is now up and running that we can actually see that, and I’m hopeful that we’ll find something exciting. We already have some compounds that look kind of cool! They’re certainly not drugs yet; they’re not nearly as active as we want them to be, but we’re going to keep chipping away at it!

There’s actually already several compounds that show some activity on SARS-CoV-2. Before SARS-CoV-2 became a thing, researchers were making compounds for another purpose, essentially, and they have since found that some do work on SARS-CoV-2. So, those compounds already have animal data, they understand something about how to make them, some have undergone preclinical even clinical trials, some have been approved for other uses. And so, do I think there will be compounds that are FDA- approved to treat SARS-CoV-2 in 18-24 months? 100%! In fact, you could argue with Remdesivir we have already achieved that goal. So, I think we will definitely have approved therapeutics…whether or not the stuff that is really being created de novo from scratch right now makes it all the way to approval in 24 months, that’s going to be harder. But I think antibodies are moving very fast, that looks very exciting. I think that might actually be something that gets approved pretty quickly. Small molecules might take a little bit more time but I do think we’re definitely going to have quite a few approved drugs within 18-24 months. How much of that is truly entirely novel chemical matter that no one’s ever seen before versus a lot of repurposed stuff? I think it’s hard for me to predict.