So now that molnupiravir has shown interesting anti-coronavirus results (if taken early enough) and Merck is asking for Emergency Use Authorization, let’s take a closer look at one of the issues that’s getting the most attention as a possible problem. This comes down to the drug’s mechanism of action. I and others have referred to it as an inhibitor of viral RNA-dependent RNA polymerase, but it’s more accurate to say that it’s a promoter of mutations in that enzyme’s actions. Here’s a paper looking at the details. Molnupiravir itself is a prodrug, and the active compound is β-D-N4-hydroxycytidine, known as NHC – and even more specifically, that nucleoside’s triphosphate, NHC-TP, which is the form in which it’s recognized by the RNA polymerase enzyme.
What happens is that NHC-TP is incorporated into the RNA strand being synthesized by the viral enzyme. That paper linked above shows that it’s generally substituted for a cytosine triphosphate, and thus ends up paired with a G residue on the other RNA strand. Now, the coronavirus has a mechanism to catch when mistakes like this occur (as do many other creatures), proofreading enzymes that go along the oligonucleotide strand checking that everything feels right. These enzymes are exonucleases, and can rip and and replace single nucleotides if they detect something off about the sequence. But when NHC is incorporated into the RNA, the proofreading enzymes don’t appear to recognize that it’s a problem – this resistance to the coronavirus proofreading enzymes was already known even before the current pandemic, because NHC and molnupiravir itself have both been around for a while as potential antivirals. And that’s a big feature to have. NHC itself is not incorporated into nascent viral RNA all that well (normal cytosine wins out most of the time), but when it happens, the mistake never gets fixed. This mechanism makes it harder for a virus to mutate out of being affected by NHC, and it’s long been noted that it is hard to
And that’s a big problem for the virus, because having that N-hydroxy analog in there instead of a plain old C goes on to cause severe problems when it comes time to replicate that now-contaminated RNA strand. The base-pairing event with NHC becomes loose and sloppy, with GTP and ATP both being equally likely. This happens because of the two tautomeric forms that the N-hydroxynucleoside can take: the hydroxylamine form pairs with G, and the oxime form pairs with A. So if you’ve ever wondered if those tautomers that you learned about in sophomore organic chemistry have any application in the real world, wonder no more. If a G residue is brought in, then further extension of the RNA seems to come to a halt – that NHC/G pair gums up the works, and the next residue can’t even be added. So that’s bad for the virus, but if an A residue comes in instead, that’s no good either: the NHC/A pair causes both G-to-A and C-to-U switches immediately downstream. There’s only so much of this sort of thing that the RNA replication process can take, as you would imagine, and the end result is a pile of mutations that make the desired replicated RNA only able to code for a series of nonfunctional hosed-up proteins.
Now all this is great stuff when it happens to a pathogenic RNA virus, and you feel like eating popcorn and cheering while you watch it going on. But what you probably don’t want is for NHC to be taken up by your own nucleic acid machinery. The course of events may well not be the same as with the coronavirus RNA polymerase. It’s already known, for example, that while NHC induces very similar changes in the influenza virus, it sets off a different set of mutations in the replication of another pathogen, RSV. But you’d rather not have to work out what interesting mutations it’s promoting in your cells. As you’d expect, this has been the object of quite a bit of study over the years, and even more in recent months. You can find work on this question going back to at least 1980 for NHC itself and at least 2003 for molnupiravir.
So what’s the verdict? There are several things to think about – first off is of course the direct RNA effects, as seen against the RNA viruses the compound has been targeted against over the years. Fortunately, human and viral RNA polymerase enzymes are different enough for this to be less of a worry. But there’s also a possibility that the ribonucleotide reductase enzyme could produce the 2-deoxy form of NHC, which could then potentially be incorporated into DNA as well. This recent paper appears to demonstrate that this is possible, but it has to be noted that this is in a cell assay (rather than a whole organism) and that the cells themselves were allowed to replicate in the presence of NHC for 32 days. It’s also worth noting that the mutagenic effects of the compound even at its highest concentration (3 micromolar over the 32 days) was not as high as the positive control, which was one minute of exposure to ultraviolet light.
That brings up a point that I don’t think that the general public appreciates. It’s easy to imagine that the default setting for our cells and our bodies is “no mutations”, so that anything that causes them is a huge change. But that’s not really the case: we are exposed to mutagens constantly. Sunlight is one of them, and there are plenty of mutagenic compounds in the food we eat as well. And I don’t mean the stuff that’s been wrapped in plastic down at the quick-mart with an expiration date of 2048 on it, or not just that, anyway, because fresh, crisp, all-natural broccoli and kale that you just harvested from your organic farm have them, too. I’m not trying to minimize the problem of mutagenicity here – it’s a real concern and it can’t (and shouldn’t) be ignored. But not ignoring it means studying it and not immediately assuming that mutagenic potential slams the door on a therapy.
Molnupiravir has a positive Ames test, for one thing. That’s an assay in bacteria used as an initial screen for mutagenicity, and I wrote about it in the early days of this blog, back in 2002. A positive Ames test does indeed make you ask how much you need the compound under investigation, because it demands more work. And more work is just what’s happened – the FDA has a standard set of tests needed to assay for genotoxicity and mutagenicity, and those (as the link at the beginning of the paragraph says) “provide strong evidence of lack of relevance of the Ames test for molnupiravir“. These are not just cell assays, but tests in whole animals.
One of the is the “Pig-A” assay. That’s the name for a gene that’s involved in the synthesis of anchoring proteins on the surface of several different blood cell types, and defects in it are the cause of a genetic disease called paroxysmal nocturnal haemoglobinuria (PNH). Studies of the gene and the disease also indicated that it might be a useful marker for overall mutation rates, because the gene is only on the X chromosome. That means that there isn’t a compensating copy to obscure mutational effects, and one hit is enough. A wide variety of mutations have been found to cause essentially the same disease phenotype, and the ability to take a small blood sample and monitor the changes in the various cells is a big plus as well. Radiation and a wide variety of known mutagenic compounds all set off positive signals. The Pig-A gene is highly conserved among species, and the protocol has been applied to mice, rats, and other animals.
There’s also the “Big Blue®” assay, which uses transgenic rodents that have had a “shuttle vector” along with the bacterial LacI / LacO / LacZ gene combination scattered through their genomes. There are dozens of copies of this package, distributed on every chromosome, and biologists will immediately recognize the classic “lac operon”. What that provides is a quick colorometric readout assay. The shuttle vector lets you take the rodent genetic material back into E. coli bacteria (using a bacteriophage), and you plate those out onto a growth medium called X-Gal. If that LacI gene (which represses the LacZ one downstream) hasn’t been disturbed, you’ll see nothing. But if LacI has been mutated, LacZ will come to life and express an enzyme (beta-galactosidase, not otherwise found in mammals) that will react with a compound in the growth medium and produce blue spots. The more blue spots, the higher the mutation rate in the animals. This assay also responds positively to known mutagens
Molnupiravir has been through both of these and other animal tests, at longer and higher doses than are used in the clinic, and (fortunately) has shown no signs of mutagenic effects compared to the control animals. This is the only way that you could go into human trials in the first place with a compound like this. The negative mutagenicity results, the window between the toxicity testing doses and the ones needed in practice, and the relatively short treatment duration all make this a feasible drug. It should be remembered that because of its mechanism a drug like this will only target dividing cells. That’s as opposed to ionizing radiation or UV light, which can damage DNA right where it stands, dividing or not.
But for that reason, I would still hesitate to see molnupiravir used in pregnant women or those likely to become pregnant during their exposure – or, for that matter, in children (although that fortunately would seem to be a rare event). I have not seen teratogenicity data for the drug, but the assays to predict the risk of birth defects are sometimes difficult to interpret. The antiviral drug favipiravir also works through RNA mutation buildup by targeting the RNA polymerase, although much less potently than does molnupiravir, and it is known to be teratogenic. It’s been approved in Japan, but with restrictions on its antiviral use for just that reason. I would not be surprised to see the FDA put similar limits on molnupiravir. Many of the intended patients will be past childbearing age, which will help.
With that in place, though, I think that the compound looks like it can find a valuable place in fighting the pandemic. It is still far better not to get infected in the first place, and vaccines are still the most valuable medical weapon we have for that. It cannot be said often enough: we have enough vaccine for every adult in this country, and if every adult had gone out and taken advantage of this we would be in a much, much better position than we are today. But an oral small-molecule drug that can keep people out of the hospital – or the grave – is a good thing as well.