Those of us who routinely counter antivaccine disinformation on social media have noticed a new antivax trope popping up again and again and again. It’s the claim that the spike protein produced by the mRNA in the Pfizer/BioNTech and Moderna COVID-19 vaccines (and now the adenovirus-based AztraZeneca and Johnson & Johnson vaccines) is highly toxic in and of itself outside of its association with the coronavirus infection itself. The intent is obvious: To portray COVID-19 vaccines as more dangerous than the infection. This new narrative by antivaxxers has led to their pointing to cherry picked studies with exaggerated findings, such as a study purporting to show that spike protein might cause pulmonary hypertension in the future or fantastical claim based on no science that those vaccinated with these vaccines “shed” spike protein, which then, miasma-like, poisons surrounding people to make them sick and, if they are women, screw up their menstrual cycles and even cause miscarriages.
They’re still at it.
Most recently, antivaxxers have been pointing to a study from the Salk Institute, UC-San Diego and Xi’an Jiaotong University in China. Behold Alex Berenson, the “pandemic’s wrongest man,” crowing about the study:
A few of my readers have asked me about this study, and I keep coming across antivaxxers crowing about it on Twitter. Misunderstanding a study like this, a study that the vast majority of them obviously haven’t read, much less understood, to spread antivaccine disinformation does get on my nerves after a while. So I responded.
Before I go on, the most obvious refutation to the claim that spike protein from COVID-19 vaccines is dangerous remains simple. Over 240 million doses of these vaccines have been administered thus far in the US alone, many more worldwide. If, as Berenson claimed, spike protein were so very, very toxic by themselves outside of the context of viral infection and at the concentrations produced in the body by the vaccines, we would have expected to see way more evidence of a safety signal related to vascular and clotting problems than we have. Right now, for instance, the J&J vaccine has been associated with a rare kind of clot at a frequency of roughly one or two-in-a-million. That’s not good evidence of horrific vascular toxicity due to COVID-19 vaccine spike protein. Of course, this leaves the question of whether, even theoretically, spike protein itself might be toxic, but as a practical matter, at least in the case of the level of spike protein made by the vaccines, the question has been answered with a resounding no.
But back to Alex Berenson’s claim.
It turns out that this study on a preprint server has been published in Circulation Research. It also turns out—surprise! surprise!—not definitely not to be “smoking gun” evidence for Berenson’s claims. Unlike the case of many papers cherry picked by antivaxxers to support their claims, it’s not that the paper is horrible, either. It’s not. It’s pretty decent, actually, at least as a preliminary, primarily observational study. Even more amusing, in it the authors expressly describe how their work actually demonstrates why vaccines that use spike protein as the antigen are so effective. One of them has also appeared on Twitter to take antivaxxers to task for misusing their study. (I’ll save these tidbits for near the end of the post.) It just amuses me that the antivaxxers citing this study have obviously not read the study itself, nor have they considered the background science and knowledge behind the study. They’ve just read the press release. What do you expect, though? They’re antivaxxers.
Since it’s the Salk Institute press release that’s doing most of the damage providing fodder for antivaxxers, let’s look at it first, and then I’ll look at the study:
Scientists have known for a while that SARS-CoV-2’s distinctive “spike” proteins help the virus infect its host by latching on to healthy cells. Now, a major new study shows that they also play a key role in the disease itself.
The paper, published on April 30, 2021, in Circulation Research, also shows conclusively that COVID-19 is a vascular disease, demonstrating exactly how the SARS-CoV-2 virus damages and attacks the vascular system on a cellular level. The findings help explain COVID-19’s wide variety of seemingly unconnected complications, and could open the door for new research into more effective therapies.
As the press release itself acknowledges, the concept that COVID-19 is primarily a vascular disease is not new. The observation that COVID-19 patients suffered from clotting problems led to this sort of speculation early in the pandemic.
While the findings themselves aren’t entirely a surprise, the paper provides clear confirmation and a detailed explanation of the mechanism through which the protein damages vascular cells for the first time. There’s been a growing consensus that SARS-CoV-2 affects the vascular system, but exactly how it did so was not understood. Similarly, scientists studying other coronaviruses have long suspected that the spike protein contributed to damaging vascular endothelial cells, but this is the first time the process has been documented.
Maybe, but let’s look at the paper itself. The first thing that those of you with access to the paper will notice is how short it is: Three pages, one figure. That’s because it’s not a full research paper, but rather a research letter. As a result, there’s no detailed Methods section, and the results are very briefly described (much too briefly, for my liking). To be honest, for some of the experiments, due to the brevity of the paper, I had a bit of a hard time making heads or tails of what, exactly, the investigators did. I’ll do my best trying to explain, however.
In brief, the the researchers used a “pseudovirus” that was surrounded by a “crown” of spike protein, like the coronavirus, but did not contain actual virus, dubbed Pseu-Spike by the authors. What is a pseudovirus? A reasonable question. In brief, a pseudovirus is a construct that has the external proteins of the virus of interest. There are a variety of pseudoviruses now, as described in this article in The Scientist:
Among these, researchers turned to models of the pathogen such as pseudoviruses and chimeric viruses that can be studied safely in labs with lower biosafety level (BSL) clearance than required for studying the wildtype version, in an effort to expand the study of the novel coronavirus. Pseudoviruses don’t replicate, rendering them harmless, but by replacing their surface envelope proteins with those of SARS-CoV-2, researchers can glean insights into the ways the pathogen infects cells.
Pseudoviruses were first developed in the 1960s, after scientists began studying a vesicular stomatitis virus (VSV) isolated from cattle. In addition to replicating well in culture, they later learned that its surface protein, VSV-G, facilitates entry into all eukaryotic cells, making the virus a useful vector not only as a pseudovirus but as a ferry to deliver DNA into cells for therapeutic purposes. The first Ebola vaccine was developed using a VSV platform, and more recently, the virus has been engineered to seek out and destroy cancer cells.
HIV-based platforms, which came about in the 1980s, have since replaced VSV as the most common model for developing both pseudo- and chimeric viruses. Unlike VSV’s negative-strand RNA genome that must be transcribed once inside the cell, HIV’s positive-strand RNA genome can instantly begin translation, making pseudoviruses based on HIV faster to produce. HIV-based model viruses have now been used in many of the same applications as VSV, with scientists applying them to the study of diseases such as AIDS, SARS, MERS, and influenza.
Also, compared with natural virus, a pseudovirus can only infect cells in a single round, has broad host range, high titer, and is not easily inactivated by serum complement.
Unfortunately, it is not clear from the paper which of these platforms was used to produce the pseudovirus in the experiments or how that pseudovirus was developed and produced. This is the sort of information that a full-length research paper would describe in the Methods section and it’s important information for determining whether the pseudovirus used was likely to be a good model. In another issue with this paper, the authors also do not describe the “mock virus” that they used as a control or how it was constructed. As a result, I find it very difficult to interpret their results. In fairness, some of this confusion might be because I am not highly knowledgeable about this particular system and don’t have the background knowledge about methodology that the authors clearly assume that the reader possesses. On the other hand, in a paper this in a journal like Circulation Research, which is not a virology journal, and particularly given that this is a paper that was likely to make the news and be misused by antivaxxers after its release, explanatory details that allow scientists from other fields with knowledge of molecular biology (but who are not experts in this field) to understand what was done are critical. A Research Letter does not accomplish this.
What are their results, though?
Basically, the authors took pseudovirus or mock virus and instilled it into the tracheas of Syrian hamsters, three animals per experimental group. Another aspect of this study caught my eye, namely the amount of virus used, 5 x 108 pfu. For those of you not knowing what “pfu” stands for, it stands for “plaque-forming units.” Basically it’s a measure of the number of viable virus particles, virus particles that can infect cells and cause a plaque on a confluent layer of cells. That’s half a billion particles, far, far more of a viral challenge than the amount of virus launching any “natural” infection by SARs-CoV-2.
Using what is a highly artificial system, the authors compared the levels of a whole slate of protein markers related to cell signaling and oxidative stress in the mock- and Pseu-Spike-treated hamsters, as well as the histology of the lungs. I won’t go into detail about all of the markers examined, but rather will step back to take a longer view because it is not important for a lay person to understand all the phosphorylation of this protein or ubiquitination of that protein measured. (It’s also easy to get lost in the weeds of a study like this.) As stated, the authors found signs of inflammation in the alveoli (air sacs) of the Pseu-Spike-treated lungs, including thickened walls and inflammatory cells. They measured the levels of various proteins they deemed relevant:
AMPK (AMP-activated protein kinase) phosphorylates ACE2 Ser-680, MDM2 (murine double minute 2) ubiquitinates ACE2 Lys-788, and crosstalk between AMPK and MDM2 determines the ACE2 level.4 In the damaged lungs, levels of pAMPK (phospho-AMPK), pACE2 (phospho-ACE2), and ACE2 decreased but those of MDM2 increased (Figure [B], i). Furthermore, complementary increased and decreased phosphorylation of eNOS (endothelial NO synthase) Thr-494 and Ser-1176 indicated impaired eNOS activity. These changes of pACE2, ACE2, MDM2 expression, and AMPK activity in endothelium were recapitulated by in vitro experiments using pulmonary arterial ECs infected with Pseu-Spike which was rescued by treatment with N-acetyl-L-cysteine, a reactive oxygen species inhibitor (Figure [B], ii).
Translation: Compared to mock virus, Pseu-Spike altered signaling due to the ACE2 receptor, which is not surprising given that it’s been known for a year now that spike protein latches onto the ACE2 receptor in order to get SARS-CoV-2 into the cell. As a result, there was a lower level of ACE2 in the hamster lung tissue treated with Pseu-Spike, although looking at the Western blots in Figure 1B I am not particularly impressed by the magnitude of the decrease in protein level.
Also observed in the Pseu-Spike-treated hamster lung was decreased activity of eNOS, an enzyme that generates nitric oxide, as well as damage to the mitochondria, the “power plants” of the cell. The authors also did the same experiments in cell culture alone using pulmonary vascular endothelial cells (the cells the line the inside of the arteries in the lung), reporting that they recapitulated their findings, although they used spike protein at a rather high concentration (4 μg/ml). They also tested whether similar changes occurred in vascular endothelial cells genetically engineered to make a more stable and less stable version of ACE2. They did, although this is only suggestive, not slam dunk evidence, that it is the spike protein-induced degradation of ACE2 that results in these intracellular changes. The authors also reported that in pulmonary arteries isolated from the hamsters vasodilation induced by a drug called nitroprusside was not affected by Pseu-Spike, but the vasodilation caused by acetylcholine was impaired. Nitroprusside works by breaking down in the presence of oxyhemoglobin to produce, among other things, nitric oxide, while acetylcholine binds to specific protein receptors on the cell surface.
Indeed, I hate papers this short (e.g., some Nature or Science papers, which can be even shorter than this) because I can never quite figure out what the authors did; this is one of the rare cases of a paper that to me screams out for an online Supplemental Data and Supplemental Figures section, and I say this as someone who generally despises the trend to dump all sorts of data into supplemental sections.
My annoyances aside, let’s, for the sake of argument, take the results at face value and assume that this study shows what the authors say it shows, namely that spike protein damages endothelium, “manifested by impaired mitochondrial function and eNOS activity.” and can cause oxidative stress that destabilizes the ACE2 receptor, leading to lower levels of the receptor. The authors themselves note that by decreasing the level of ACE2, spike protein could actually decrease the infectivity of SARs-CoV-2, given that the coronavirus needs to bind to ACE2 to get into cells, while speculating that the dysfunction of endothelial cells could result in endotheliitis, or inflammation of the endothelium that more than makes up for the decreased infectivity.
But here’s the kicker, taken right from the final paragraph of the paper:
Collectively, our results suggest that the S protein-exerted EC damage overrides the decreased virus infectivity. This conclusion suggests that vaccination-generated antibody and/or exogenous antibody against S protein not only protects the host from SARS-CoV-2 infectivity but also inhibits S protein imposed endothelial injury.
In other words, the vaccine could be protective not just against infection by SARS-CoV-2 but also against endothelial injury from the spike protein.
I just want to reiterate again that this is a contrived system. It’s far from a worthless system, as pseudovirus systems have value in studying the role of spike protein in the pathogenesis of COVID-19. However, given the crapton of pseudovirus used in this hamster model, I really question any relevance of this system to vaccine safety issues with respect to mRNA- or adenovirus-based vaccines that produce the spike protein as an antigen. Why? The mRNA or adenovirus from the vaccines does not distribute extensively given that it’s an intramuscular injection, and the spike protein is highly unlikely to attain concentrations in the circulation anywhere near the high levels produced by the model used in these experiments. Moreover, the spike protein from the vaccine is not attached in a crown-like array on a virus particle (or pseudovirus particle), but rather exists as naked single protein molecules, and, as I described before, it’s unclear that in this form spike protein, compared to the “crown of spikes” that gives coronaviruses their name, is anywhere near as effective at causing these downstream effects in cells. Add to that the fact that mRNA, even the modified mRNA in the vaccine, doesn’t hang around very long and therefore doesn’t generate spike protein for very long. (Doubters should consider this: Why do the mRNA vaccines both require a second dose 3-4 weeks after the first dose if, as many antivaxxers claim, the vaccines crank out spike protein indefinitely?)
Indeed, one of authors points this very issue on Twitter:
I’ll conclude, as I have so often, by complaining about the press offices of universities hawking studies like this. The study itself, as I said, is interesting, but not evidence that existing vaccines that use spike protein as the antigen are unsafe or damage the vasculature. Add to that, again, the fact that >240 million doses of these vaccines have been administered in the US alone, all without a signal suggesting that they cause any sort of spike protein-induced blood vessel damage of the sort reported in this model. Unfortunately, the Salk Institute press release appears to have been written without any consideration as to how to avoid making it easy fodder for antivaxxers to use as propaganda. In the past, this sort of lapse was forgivable. In the middle of a pandemic, it is not, particularly given that on social media very few read the actual study while most read the no more than the press release or even much beyond the headline.
In fact, I have an idea for authors and university press officials from here on out. Any study of the spike protein’s role in the pathogenesis of COVID-19 should include in the study itself and in any press releases a strong disclaimer emphasizing how the results of the study do not say anything about the safety of COVID-19 vaccines, for the sorts of reasons I have listed above. Consider it a useful “pre-bunking” of the disinformation antivaxxers use basically any study about spike protein to produce these days.