Thread by @michaelzlin

When my lab set out to develop SARSCoV2 protease inhibtors based on the approved HCV drug boceprevir, we figured SARSCoV2 would persist for decades due to slow vaccination, vaccine waning/escape allowing endemicity, or reintroduction from animal reservoirs. All have become true.

We reported the first SARSCoV2 protease inhibitor based on boceprevir with very high affinity (IC50 ≤ 12nM) in 9/2020. It was our first drug candidate, tested by 4 very smart and dedicated academic scientists, and funded by $20k in FastGrants by @patrickc
biorxiv.org/content/10.110…

In April 2021, Pfizer announced PF07321332 (nirmatrelvir, the active ingredient in Paxlovid) with a similar structure, the main difference being the replacement of the ketoamide reactive group of boceprevir with a nitrile group

Pfizer likes to say they had 300 scientists making and testing 800 structures to derive PF07321332 from the very different PF00835231, which they had made against SARSCoV1 protease ~15 years ago.
cen.acs.org/pharmaceutical…

But it's clearly more likely Pfizer read the public academic papers showing boceprevir binding to SARSCoV2 protease, available since June 2020, then made a few structures based on boceprevir, as we did, to get nirmatrelvir.
bloomberg.com/news/articles/…

Ironic that Pfizer would prefer to claim they had to make and test 800 compounds because they didn't read the literature. In academics that kind of approach would earn you a grant denial. But the public ends up paying for industry inefficiency anyway in the form of drug prices.

Pfizer's selection of the nitrile reactive group (warhead) was important because it improved cell permeability and oral bioavailability. Thus nirmatrelvir can reach very high levels in the blood after oral ingestion. Another company, Pardes, also use the nitrile, based on patents

With SARSCoV2's rapid genomic diversification in humans and other animals, and the posssibility of new coronavirus epidemics, we will need multiple types of protease inhibitors. Hopefully we've now learned that we must be prepared for the future, not react to the past and present

One worry is that the SARSCoV2 pool may contain mutants that resist Pfizer of Pardes protease inhibitors and that may get selected after these inhibitors become widely used. This has gotten some attention recently.

As Pfizer and Pardes (which raised $200M) are covering nitrile-based SARSCoV2 inhibitors, we continue to work on non-nitrile inhibitors. We recently expanded ML1000 to make a series of orally bioavailable drugs (thx to @Stanford_ChEMH, Coulter Foundation, and Fast Grants again).

Today I'll reveal how well our drugs work on SARSCoV2 protease mutants. S144A is a mutation found in the natural viral gene pool that resists Paxlovid (described in FDA's info for prescribers). We've confirmed our drugs work equally well on original and S144A. Here's one, ML1006a

As seen above, ML1006a works better than PF07321332/nirmaltrelvir/Paxlovid on original protease ("wildtype", WT). And Paxlovid loses activity against S144A by ~10-fold, whereas ML1006a is not affected at all. Below is a second example, with another of our drugs named ML109m.

So far it appears that the resistance by S144A is related to the use of the nitrile warhead. When we compare the same structure with nitrile vs a ketoamide warhead (the one originally from boceprevir), we see the S144A resists the nitrile but not the ketoamide.

It remains to be seen if this will hold true for all nitrile compounds and all ketoamide compounds, but these results indicate it's crucial to not rely on one type of reactive group such as the nitrile. Rather, diversification of SARSCoV2 requires us to diversity our drugs too.

And while a nitrile warhead is one way to enhance oral delivery, it's not the only way. We've recently achieved oral bioavailability without nitriles via some imaginative chemistry (patented, sorry Pfizer). This may be a route to drugs better than Paxlovid even on nonmutant virus