I’ve had several requests for comment on this recent PNAS paper, which talks about integration of SARS-CoV-2 sequences into the DNA of human cells. I’m glad to do it, but right off I have to note that a lot of the attention that it’s getting seems (sadly) to be coming from anti-vaccine activists, who are trying to whip up fear that getting the vaccine somehow means that you will be Permanently Corona-ed. This is nonsense, as you’ll see below (skip to the end for this part!)
Here’s a good writeup on the paper itself, and I recommend it for background. The first thing to keep in mind is that the coronavirus is an RNA virus, so there’s a big barrier to getting that integrated into the DNA genome of a cell. You need a reverse transcriptase enzyme, one that runs the usual tape backwards and makes DNA out of an RNA sequence instead. We humans don’t do that ourselves, but there are plenty of viruses that do. And over the untold millennia, we have been infected by many of these, and a good amount of the stuff that they jammed into our genome is still there. This information freaks some people out when they first learn it, but it’s true: somewhere around 5 to 20% of our genomic DNA appears to be detritus from ancient retroviruses. Somehow the “Intelligent Design” folks seem to skip over this – go figure. Now, it’s not all in good shape, because many of these things happened a long time ago. But some of it may be involved in human disease, in some cases possibly through continued expression of pieces of viral proteins.
You can see evidence of this in retrotransposons, which likely have a retroviral origin and can at any rate act like built-in retroviruses. We have a lot of those in our DNA, because they just keep on copying themselves. A particularly well-studied case is the LINE1 sequences. There are a lot of these in the genome, and most of them have deteriorated to the point that they are inactive. But every human still has a few dozen of them that can be expressed into active proteins, among them a reverse transcriptase that then turns around and makes LINE1 DNA and inserts it back into the genome. In mammals, these sequences seem to have undergone evolution inside our cells, showing the history of a long-term “arms race”.
So our cells don’t have any need for a reverse transcriptase themselves, but we have one running around thanks to LINE1. And the PNAS paper under discussion presents evidence that under some conditions, this enzyme can pick up coronavirus RNA during an infection, make DNA off that sequence, and insert it back into the genome of a cell. In the paper, though, those conditions include modifying the cell lines so that they have even more LINE1 than usual, and this is one of the reasons that the extension of these results to real-world infections has been challenged. An earlier version of the paper also took incoming fire because the human/virus chimeric sequences they detected can also be produced artifactually, but this latest version seems to have addressed many of those concerns.
And it’s worth noting that viral infection might actually de-repress LINE1 activity, so this mechanism can’t be ruled out yet (nor does it apply only to the current coronavirus – it could take place with other RNA viruses as well). If it does happen, it could lead (perhaps) to autoimmune consequences through production of these mixed human/viral protein fragments, and it could also confound diagnostic tests if they’re pointed towards viral sequences that have “turned endogenous”. All this remains to be seen, and I think it’s worth investigating in general. But as of yet, we have no hard evidence that this process is taking place in coronavirus-infected human patients at all.
It’s also important to note that even under the cell-culture conditions used here, the authors are just seeing some variable-length insertions from one end (the 3′ one) of the viral genome. This process does not produce infectious virus. And it’s also important to note that authors themselves say that these results in no way imply that the mRNA vaccines cause integration of Spike protein into a cell’s DNA, either. The mRNA in the vaccines does not look like the 3′ end of the viral genome – they have completely different “untranslated regions” (UTRs), and for what it’s worth, the Spike protein itself is nowhere near the 3′ end of the actual viral genome. A vaccination is enough to set off your immune system for future action, but it’s really not like being infected with the virus, either in the scale of it all or in the many details.
I sympathize with the researchers who are frustrated with the way that this work has been picked up by the anti-vaccine folks. That stuff drives me crazy, too. At the same time, I think it’s a hypothesis that’s worth investigating for viral infections in general – unlikely, but worth a look – and I think it’s a shame in general that there are people who will seize on anything they can for the anti-vaccination crusade. If not for this paper, they’d be going on about something else. . .