A number of people have asked me about this paper, so let's have a look at it. It's from two researchers in Sweden, and it talks about the effects of the coronavirus Spike protein on DNA damage repair mechanisms. That means we first need to take a look at those, so that we all know what's being discussed. (Update: for the rest of this post, I am taking this paper at its word, but here's a detailed look at its experimental methods, which also need to be taken into account!)
1. DNA Repair In General
Human cells actually have a whole suite of DNA repair enzymes - that's obviously been a favored innovation over the (many, many) years of eukaryote evolution. There are naturally a number of ways that DNA can get damaged, during replication, during gene expression, and just under the normal conditions of life down there in the nucleus. Here's a concise review of them. There are five major categories, so here's a quick overview of each: you have base excision repair (BER), which covers a whole list of enzymes (DNA glycosylases, endonucleases, lyases) that recognize and revert changes to single nucleotides, but specialize in ones that aren't dramatic enough to noticeably distort the whole DNA helix. Deamination, alkylation, and oxidation of nucleotides are some of those changes, this category further subdivides into "short patch" and "long patch" repair mechanisms. Then there's nucleotide excision repair (NER), which takes care of some larger problems that definitely show as an oddly-shaped stretch of DNA. This include UV light damage (which crosslinks some of the nucleotide structures, as well as other DNA alkylations from larger reactive compounds like benzo[a]pyrene or some chemotherapeutics. There are two categories of these, the "global genome" repair pathway and the transcription-coupled one, each with a suite of enzymes involved, and both of them cut out the leison at both ends and fill in the gap. A third category is mismatch repair (MMR), which really kicks in after DNA replication and seeks out base mismatches (pairs that don't fall into the normal A-T and C-G framework) and so-called "insertion-deletion loops". There's a long list of enzymes involved here, too, led by a group of MutS and MutL homologs, with plenty of downstream helpers. This one also is an excision-and-fill-back-in process.
A fourth repair category is homologous recombination (HR) and the fifth is non-homologous end joining (NHEJ). Those are often discussed together because they're targeted at the same sort of problem, an outright-double-strand breakage in the DNA. Ionizing radiation is particularly known for causing this kind of damage, but there are plenty of other causes as well. Both of these pathways exfoliate into several related ones when you study them closely - for example, HR can involve double-strand breakage repair (DSBR), synthesis-dependent strand annealing (SDSA), or single-strand annealing (SSA) repair mechanisms, and the list of enzymes and other proteins involved is quite long. NHEJ, for its part, kicks in when the break is clean and none of the 5' bases have been trimmed off (resected), and it's really one of the major pathways at all points of the cell cycle. HR, for its part, is only active during a couple of cell cycle periods, because it needs a homologous DNA molecule nearby to use as a template. If everything is still functional at the broken ends, NHEJ will just connect (ligate) the ends back together, and it can do some minor repairs to get things ready for that. This doesn't always work perfectly - NHEJ is notorious for leaving some little deletions and insertions at the repair site. If both strands are broken and resection has happened, though, it's HR or bust for getting things fixed, which means that they may not get fixed at all, with consequences that are never good. Some of these pathways bring things back to an indistinguishable state from the original one, while others often produce "crossovers" of one strand to another (and indeed, this is a deliberate feature in antibody and immune cell receptor production, when you want a huge number of mix-and-match variations).
Whew. Just be glad I haven't gone into the detailed mechanism for each of these; we'd be here all weekend for sure (those links for each repair category have some helpful diagrams to show your the differences, if you're interested). And I have even left some pathways out, like translesion synthesis and interstrand crosslink repair, some of which are handled by several of these pathways operating simultaneously. But that gives you some idea of the major categories, and some idea of the heavy evolutionary backup that has accumulated. Some of these pathways go waaaaay back to things like amoebas, and bacteria and viruses both tend to have their own suite of repair enzymes that fall into several of these bins as well. They're important.
2. Coronavirus (and Spike Protein) Effects
Now for the coronavirus connection. When the current coronvirus (or any virus) infects a human body, there are a huge number of nasty changes that start up. Viruses tend to have many means of hijacking cellular functions and many ways to interfere with the normal immune response - you can see how that would be a very strong evolutionary advantage indeed. There's been a lot of research how SARS-Cov-2 accomplishes all this. The new paper linked above is looking at the above-mentioned NHEJ pathway, since it's so important in the antibody response. All these DNA repair mechanisms would be expected to be taking place in the nucleus (except for the ones that are active in repairing mitochondrial DNA), so you would expect to need viral proteins down there in the nucleus to mess them up. And indeed there have been several coronavirus proteins that have been reported in the nucleus, and this paper adds the Spike protein to that list.
And it seems to be affecting DNA damage repair down there - just the full-length Spike, interestingly, not the S1 or S2 subunits by themselves. This was investigated through several DNA-damaging routes (gamma irradiation, doxorubicin treatment, or exposure to hydrogen peroxide), and in cells that were engineered to overexpress the full-length Spike or just its subunits. DNA damage was greater (that is, repair was less efficient) after these treatments in full-length-Spike-expressing cells, but not the others. The DNA repair enzymes themselves seemed to be present in their usual amounts, so their expression or degradation didn't seem to be affected. Looking at other protein levels, though, the evidence is that the Spike's presence affects key proteins (BRCA1 and 53BP1) that recruit the repair machinery to checkpoints in the chromatin structure. Both HR and NHEJ pathways seemed to be affected, lowered substantially but not wiped out (which makes sense considering those two proteins' involvement, actually). And when they engineered the antibody recombination machinery, V(D)J recombination, into cells along with Spike protein expression, that recombination was lowered by 50%. This too suggests inhibition of the NHEJ pathway, although that effect was not as large as one might have expected from the earlier NHEJ activity reporter assay.
Now, you'll note that three different types of DNA damaging agent were used, and the paper shows that Spike protein expression seems to impair all of them to some degree. Doxorubicin is a DNA alkylating enzyme, which would make you think of the BER pathway, although it can also lead to the double-strand breaks that are repaired by HR and NHEJ. Gamma radiation would seem to be a good route to double-strand breaks as well. Meanwhile hydrogen peroxide is classic oxidative damage, making you think of BER, but it has also been shown to lead to double-stranded breaks. So although these are different mechanisms, they are rather mixed in the end - it would have been interesting to see what the effect of UV light would be, since that one is a bit more selective in its DNA effects (and does not prominently feature double-strand breakage, as far as I know).
The paper goes on to hypothesize that vaccines that produce the full-length Spike protein might do the same thing - that is, lower the activity of an essential step in antibody production. That's worth investigating, and it's not a crazy idea. But there are several gaps that have to be filled in for it to be true. For one thing, quantifying the actual amounts of Spike protein being produced in these engineered cells versus the amounts in cells after vaccination (or indeed, the amounts after coronavirus infection) is a big first step. Second, remember that recombination-intensive events like antibody production and T-cell maturation don't happen everywhere - they takes place in specialized cells like thymocytes and in bone marrow lymphoid cells. We don't actually know the extent to which the coronavirus infects these cell types, nor the extent to which they are affected by vaccination (something the authors also make note of). But to that latter point, it's clear that they are not in any way knocked out of production, because the vaccine does of course produce a vigorous immune response. There's also time course to think of: Spike protein is only produced for a limited time after vaccination, after all. The mRNA gradually breaks down, and the cells that are making the Spike are surely taken out by the T cell response once that gets going. So how long can this recombination problem last?
If vaccination does affect V(D)J recombination to some extent, though (which is as yet unproven), there would be room to wonder if a vaccine that did not recapitulate the whole Spike sequence might be more desirable. But you'd have to balance that with the different antibody (and T cell) profiles that such a vaccine would surely elicit - would it still be as protective against the coronavirus, or against its variants? We have no idea. Animal studies would shed some light on this, but the only way to be sure would be to see what happens in human trials.
3. What About Cancer?
Now, the reason that people are sending me this paper is honestly not because they're wondering about V(D)J recombination. That doesn't even come up. They're worried about cancer, because a number of commenters have taken this paper and run with it in that direction. Note that its authors mention no such thing. But this is part of the rather large number of people who are attuned to every possible worry about the vaccines that might present itself.
Now, there is no doubt that real problems with DNA repair enzymes raise the risk of various types of cancer. In the order given above, defective BER pathways have been associated with some types of breast cancer and other solid tumors, NER defects are associated with several rare conditions, one of which (xeroderma pigmentosum) raises the risk of some types of skin cancer, and MMR defects are particularly prominent in colorectal cancer. Meanwhile, HR pathway defects are associated with breast and ovarian cancer, and given its importance, there are surely other connections yet to be uncovered. And NHEJ, well, that one's complicated. In different situations, it can present as both a suppressor of cancer and a promoter of it, so defects in these pathways can be hard to work into a coherent picture.
But in all these cases, we're talking about severe, ongoing defects in these repair pathways, not partial inhibition of one of them for a few days. Remember, even if this mechanism is operating after vaccination, it doesn’t shut down your double-strand repair forever. Everything comes back. The biggest reason not to worry is the T-cell response mentioned above: the cells that are producing lots of Spike protein after vaccination are the ones that are going to be killed off once the T cells become sensitized to them. Now, I realize that a lot of people are worried about Spike protein circulating around through the body, but remember: the experiments in this paper, even if you’re worried about them, were done in cells that were specifically engineered in their DNA to produce Spike protein constantly - this is a different mechanism than the mRNA vaccinations, which use RNA that breaks down in time. There is no evidence (and no particular reason to believe) that circulating Spike protein after vaccination, such as it is, gets taken up into other cell types and then taken into their nuclei, in sufficient quantity to have any noticeable effects at all. That’s a lot of hurdles to clear.
And as just mentioned, there are situations where interfering with double-stranded repair may even be a good idea. There are in fact already pathways in the cell that actually delay double-strand break repair, for reasons that are not clear, but are probably not perverse. Keep in mind that cancer cells tend to have a lot of genomic instability, and if that goes too far, they will not be able to keep things together. So they actually need DNA repair enzymes, and blocking them may be therapeutic, sending them over the edge into unrecoverable DNA damage (with normal cells being less susceptible because they're not having to repair things so relentlessly). So it is definitely not the case that the level of double-strand break repair inhibition shown in this new paper will necessarily raise the risk of cancer. The situation is way too complicated to make that leap.
So no, *if* this current paper's conclusions are correct, and *if* this NHEJ inhibition via the Spike protein really is operating for a time after someone has been vaccinated - both of these need further work, for sure - I still don't see this as a cancer risk. As mentioned before, the authors of the paper didn't even bother to note this possibility either. To me, this is a nonissue that's been whipped up by people who either don't appreciate the biology involved, or perhaps do appreciate it and don't care. Just so long as worries are raised about vaccines - any weapon to hand.