After having been at this blogging thing for nearly 18 years, I’m occasionally surprised to see something that I haven’t seen before. Granted, this particular bit of deceptive reference to basic science is just a variant of what I’ve seen many times before when antivaxxers deceptively invoke studies of animal models to claim that vaccines Do Bad Things. But zebrafish? I’d never seen antivaxxers reference zebrafish studies before. The closest I ever came to seeing zebrafish used by antivaxxers—and the only time the word “zebrafish” has ever even appeared in this blog—was in passing as a sarcastic aside in a discussion of some of Gwyneth Paltrow‘s Goop grift. So to encounter a zebrafish study being cited by a prominent “new school” COVID-19 antivaxxer named Jessica Rose was an unexpected treat.
Here’s what I mean. After introducing herself as an avid surfer who worries unduly about how much exposure to bacteria and viruses she experiences while looking for a tasty wave. Granted, she does make a point about surf spots being polluted by effluent and run-off and how one should wait 72 hours after a heavy rain before paddling out to areas where effluent is known to be present and problematic, but then she pivots to invoke the humble zebrafish:
This article entitled “Toxicity of spike fragments SARS-CoV-2 S protein for zebrafish: A tool to study its hazardous for human health?” was published in March, 2022 in the journal Science of the Total Environment. The authors use a model for studying the effects of spike on humans ‘using Zebrafish as a tool to assess the harmful effects of SARS-CoV-2 in the aquatic environment’. They found that the spike protein from SARS-CoV-2 is highly toxic to Zebrafish.
The reason every single person on Earth needs to be concerned about this (and should have been years ago) is because of conserved homologies between humans and other species. For example, genetic homology between zebrafish and humans is highly conserved whereby ~70% of human genes can be found in our stripy little friends.4 Our respective ACE-2 receptors are 72% sequence similar.
Before I directly address Rose’s misinformation, let me just first say that zebrafish, which are fresh water fish in the minnow family measuring less than two inches long and found in the wild in rivers and ponds in India, are really cool. They’re used as experimental models to study a lot of things relevant to human health and disease. I’ll even cite the article Rose cited that explains why we use zebrafish to study disease, sometimes even in preference to mice:
While mice are evolutionarily more similar to humans because they are mammals, zebrafish have several advantages over their furry competitors.
One important advantage of zebrafish is that the adults are small and prefer to be housed in large groups, or “shoals”. As a result, they require much less space and are cheaper to maintain than mice.
In addition, zebrafish breed more readily (as often as every ten days) and can produce 50-300 eggs at a time. In addition:
Zebrafish embryos are also laid and fertilized externally, which allows them to be easily manipulated in a variety of ways. In vitro fertilization can be performed if necessary. The one-cell-stage fertilized eggs can be easily injected with DNA or RNA to permanently modify their genetic makeup in order to generate transgenic or knock-out zebrafish lines. Working with mice in this way is much more complicated. Mouse embryos develop inside the mother, and to access and manipulate them the mother would have to be sacrificed. To keep the embryos alive after fertilizing or injecting them, they would need to be transplanted into another female mouse, as well.
Furthermore, zebrafish embryos are clear, which allows scientists to watch the fertilized eggs grow into fully formed baby fish under a microscope. Their transparency also enables the visualization of fluorescently labeled tissues in transgenic zebrafish embryos. Mouse embryos are not clear and develop inside the mother, so the observation of live embryo development like that in zebrafish is not possible.
Zebrafish have even been used in vaccine safety research, although I am not quite convinced of their claimed advantages. One useful feature of zebrafish is that their embryos are easily injected with gene expression vectors, drugs, and chemicals, and various effects on development can be observed directly. Transgenic zebrafish, in which genes are knocked out or added, are easier to produce than transgenic mice. Diseases that have been modeled in zebrafish include Duchenne muscular dystrophy, melanoma, and a number of others, as well as tissue regeneration, and the NIH maintains a zebrafish core to assist researchers.
So zebrafish are a useful experimental model; of that there is not doubt. However, it’s not perfect, and Rose gets a bit too enthusiastic waxing poetic about how 72% homology between the human and zebrafish ACE receptors, which are the proteins to which SARS-CoV-2, the coronavirus that causes COVID-19, binds to enter cells, is practically identical. (It isn’t. It’s definitely similar, but not identical, and one should understand that the same sorts of problems translating findings from zebrafish findings to humans as often occur translating findings in mice to humans.
In cases like this, I like to compare what the antivaxxer says about a study with what the study actually shows. So let’s see what Jessica Rose says about this study. First, though, let’s see what the study authors say about the reason that they did their study::
Despite the significant increase in the generation of SARS-CoV-2 contaminated domestic and hospital wastewater, little is known about the ecotoxicological effects of the virus or its structural components in freshwater vertebrates. In this context, this study evaluated the deleterious effects caused by SARS-CoV-2 Spike protein on the health of Danio rerio, zebrafish.
The authors justify the need for the study thusly:
On the other hand, a field still little explored refers to the possible environmental impacts (direct and indirect) of the current outbreak of COVID-19. Despite the increase in contaminated household (Gautam and Sharma, 2020; Urban and Nakada, 2021; Zand and Heir, 2020) and hospital (Abu-Qdais et al., 2020; Sangkham, 2020; Wang et al., 2020b) waste generation, so far, there is no information on the ecotoxicological effects of SARS-CoV-2 or its structural components on freshwater vertebrates. Therefore, these facts justify the urgent need for studies in order to assess the deleterious effects caused by SARS-CoV-2 virus on the health of aquatic organisms which already suffer as a result of various anthropic activities
The reason for the study is clear. The authors, who are numerous and mostly based in Brazil, were concerned about SARS-CoV-2 in rivers and lakes based on the finding of the virus in freshwater near human habitation where COVID-19 is spreading. Whether or not this is a reasonable fear, to me at least, is debatable given the exceedingly low concentration of virus in wastewater, but I won’t question the utility of the study beyond this. Instead, I will question its methods, interpretation, and applicability to reality.
Rose, of course, immediately dismisses the researchers’ reason for the study and pivots to her bête noire, vaccines:
I don’t want to downplay the fact that SARS-CoV-2 is also problematic from this point of view. Because of the furin-cleavage site, the spike protein once bound to ACE-2, is cleaved and this yields 3 free S1 subunits per spike trimer, able to impose whatever subsequent damage they may. For someone with a high viral replication rate and high viral load, this will necessarily imply more free spike.We demonstrated, for the first time, that zebrafish injected with fragment 16 to 165 (rSpike) [aka ‘their spike bit’], which corresponds to the N-terminal portion of the protein, presented mortalities and adverse effects on liver, kidney, ovary and brain tissues.Oh dear. This confirms two things for me.
- The spike can inflict severe damage on other beings in our ecosystem.
- The spike protein on its own, no virus required, can bind ACE-2 to impose its damages via intracellular signaling cascades. In fact, “77% of the human ACE2 residues of the interface are similar in zebrafish ACE2 sequence”. (Reference #1)
Rose lists some of the findings of the paper and concludes:
The authors also raise the point that I have been screaming about for almost two years and that is the real possibility of the spike-ACE-2 binding event and the resultant intracellular cascade (cell signaling) of the cell and all the immunological mediator activation that would accompany this to induce a complete biological disaster. I have been looking at this from the point of the view of the Renin Angiotensin Aldosterone System (RAAS), but the systemic implications for spike-ACE-2 interaction/intracellular signaling cascade/activation events are astronomical. Why wouldn’t they be?
Here’s the problem, though. As is often the case in studies like this, the amount of spike protein used was massive. As one commenter on Twitter mentioned:
How large was the spike protein dose? Let’s go to the Materials and Methods:
We performed 2 intraperitoneal (IP) inoculations of a solution containing 1 μg purified rSpike diluted in 10 μL of inoculation buffer (7 M urea, 50 mM Tris-HCl pH 7.5, 200 mM NaCl, and 1 mM EDTA). A group of control animals received injections containing only the dilution buffer. Another control group was challenged by a lysate of bacterial fragment of E. coli BL21(DE3) extract.
By way of comparison, the Novavax COVID-19 vaccine contains 5 μg of recombinant spike protein, and it’s injected into humans, let’s say a 70 kg adult.) Obviously, whether or not 2 μg of recombinant spike protein—between the two doses—is a large dose or not depends on the weight of the organism into which it’s injected. How much does a typical zebrafish weigh? Adults weigh around 0.5 g, give or take, which means that the equivalent dose for a 70 kg human would be ~0.14 g, or 140 mg, or 140,000 μg—double that, actually given that the zebrafish got two doses. In other words, the authors used a huge dose of spike protein compared to anything humans ever encounter.
It’s actually worse than that, though. Look at the methods:
The DNA fragment coding for the SARS-CoV-2 Spike protein fragment from 16 to 165 (rSpike) was amplified by PCR using SARS-CoV-2 cDNA transcribed from the RNA isolated from the second patient, strain HIAE-02:SARS-CoV-2/SP02/human/2020/BRA (GenBank accession number MT126808.1).
I’ll translate. Here’s a map of the spike protein peptide sequence:
The peptide fragment used by the investigators is only part of the NTD (N-terminal domain) region, rather than the most important part of the spike protein for purposes of infecting cells, the RBD (receptor-binding domain), the part of the protein that binds to ACE2 receptors. This has two consequences. First, it means that the functional concentration of spike is much higher using the recombinant protein, at least of the NTD domain, compared to using full length spike. Second, the part of the spike protein responsible for the ability of SARS-CoV-2 to bind to ACE2 receptors, activating them and allowing the virus to gain entry to the cell, wasn’t even tested here! Moreover, most neutralizing antibodies to spike target RBD, not NTD. That’s not to say that NTD is not important. It is. (For example, mutations in NTD are thought to have a significant role in immune escape.) However, it puzzles me to no end that the authors didn’t make the more obvious choice of making an RBD fragment, which, some might recall, was the choice of antigen in some early attempts to generate a COVID-19 vaccine.
There’s also a difference between injecting spike protein into the muscle and into the peritoneum. Injecting intraperitoneally is very much like a direct intravenous injection. That’s why we use it in animal models. (Indeed, I’ve used intraperitoneal injections many times to test experimental therapeutics in mouse models of cancer.) Protein with adjuvant injected into muscle generally stays there and, if absorbed, the absorption occurs slowly. That doesn’t even consider that Novavax is new and the vast majority of vaccines administered have been mRNA-based vaccines that, as I’ve pointed out, produce very little recombinant SARS-CoV-2 spike protein, with the amount that gets into the bloodstream being infinitesimal.
Right off the bat, you can (I hope) see that the investigators were using a very artificial system with a huge dose of spike protein. There’s nothing inherently wrong with that if you’re doing a preliminary experiment looking for a large effect (i.e., screening for any detectable effect), but such studies always leave open the question of whether their findings are actually relevant to reality. For example, if the authors were concerned about spike protein toxicity from runoff containing SARS-CoV2 in wastewater, why did they inject it instead of putting it in the water in which the zebrafish were swimming?
Rose cites a key finding thusly:
They found that the difference in survival rate between their spike bit-injected fish and control fish, was statistically-significant.The relative risk of death in the period studied between the groups was significant (chi square = 79.70; p < 0.0001).The authors also found evidence of (temporary?) deleterious effects of their spike bit on the ovaries and testes of the fish.
Let’s take a look at the actual survival curve from the paper:
There’s no need to look at all the details. Panels A and B simply illustrate how the gene for the spike protein was inserted into a plasmid and used to make the protein. Panel C is the money panel, with the black line being the group injected with spike protein, Control #1 indicating the group that received buffer alone with no spike protein, Control #2 being a mix of different and unrelated recombinant proteins. The first thing I notice about this figure is that the buffer alone is actually fairly toxic relative to everything else. The second thing I notice is that the spike protein decreases survival only 10% more than the buffer alone does.
The authors reported:
Two bioassays were carried out to analyze the toxicity of the rSpike. The first injection of the rSpike generated high toxicity to the fish (Fig. 1c). Therefore, the assay was repeated by adding different control groups to confirm whether the toxicity findings were specific to the rSpike (Fig. 1c). In the first bioassay, after the fish were injected with rSpike, the survival rate was 78.6% during the first seven days (Fig. 1). The lethality was significantly increased when compared to naive control and fish injected with protein buffer (control 1), where the survival rates were 100% and 90%, respectively (Fig. 1c). Nonetheless, after a second injection, the rSpike injected group maintained the plateau survival rate, with no further increase in lethality in the treated group. Therefore, a second assay was conducted by adding different control groups in order to confirm that the toxicity findings were specific to the rSpike, and also the presence of antigens was accessed. The Kaplan-Meier survival analysis confirmed rSpike injection presented a lower survival rate compared to the two previous controls used (Control naïve and protein buffer) and compared to females injected with Escherichia coli extract or a culture medium mixed of two purified recombinant proteins (PilZ protein from Xanthomonas citri, and a N-terminal fragment of LIC_11128 from Leptospira interrogans Copenhageni) (Control 2) (Fig. 1c). The survival rate was maintained after the second injection for the next seven days. The relative risk of death in the period studied between the groups was significant (chi square = 79.70; p < 0.0001).
Actually, the fact that a second injection of spike protein doesn’t further increase toxicity leads me to wonder just how toxic spike really is. Think about it. If another dose of spike protein didn’t result in more fish dying, then likely the authors had achieved a maximal toxic dose with the first injection, and that maximal toxic does was really not that toxic given the huge dose of recombinant spike protein used.
Nonetheless, the authors concluded:
Our studies reveled that zebrafish showed inflammatory reaction to SARS-CoV-2 rSpike protein which provoked damage to organs (liver, kidney, ovaries and brain) in a similar pattern as happen in severe cases of COVID-19 in humans and resulted in 78,6% of survival rate in female adults during the first seven days. The application of spike protein in zebrafish was highly toxic that is suitable for future studies to gather valuable information about ecotoxicological impacts, as well as vaccine responses and therapeutic approaches in human medicine. Therefore, besides representing an important tool to assess the harmful effects of SARS-CoV-2 in the aquatic environment, we present the zebrafish as an animal model for translational COVID-19 research.
Color me unimpressed, particularly given all the other issues that I’ve discussed.
It must be pointed out that this study was not about COVID-19 vaccines and never intended to say anything about COVID-19 vaccines. Rather, it was designed to test how much damage that COVID-19 infection might cause to aquatic creatures. On, this makes me wonder why the investigators designed the study the way that they did. If they were concerned about spike protein from the virus in the water being toxic to fish, then why didn’t they simply add recombinant spike protein to the zebrafish water? I can imagine one reason. It would take a lot more spike protein—possibly orders of magnitude more, enough to dissolve in the tanks that house the fish—to do this experiment, and they would have had to test whether the protein was stable in freshwater such as that which zebrafish live. Even so, they should have acknowledged that their model uses a dose of spike protein many orders of magnitude than what any fish might experience in nature just swimming through spike-contaminated water due to SARS-CoV-2 in runoff.
As people on Twitter noted:
I’d go even further. It’s not likely to be spike protein from SARS-CoV-2 that one has to worry about in wastewater, but rather the full functional virus! This leads me to ask: Why not just infect the zebrafish with SARS-CoV2 at defined doses and do the same measurements? After all, the authors make much of their studies showing that spike protein should be able to bind to zebrafish ACE2 receptors. So why not test it? Alternatively, why not put SARS-CoV-2 in the water and see if it can infect the zebrafish? I understand that logistically it might be difficult to do an experiment in which SARS-CoV-2 is in tanks with fish swimming around, but if you really want to study an actual relevant question regarding the presence of the virus in wastewater, why not study an actual relevant question about what effects the virus can have in water?
Which brings up this issue:
Precisely. Scientists reading this paper know that what is being studied is not a real world situation. I knew it immediately just from the abstract. Jessica Rose, who has a PhD, should know it. (In fact, I suspect that she probably does know it.) However, she also knows that her audience doesn’t know it. In other words, the choice about why Rose is promoting this paper boils down to ignorance versus dishonesty. Either she really doesn’t know that this study doesn’t represent anything resembling a real world situation, or she does know but knows that she can convince her audience that COVID-19 vaccines are dangerous by spinning it the way she does. Or maybe there’s a little of both. Who knows?
Whatever Rose’s reasons, she concludes:
I am very glad this got published in Science of the Total Environment (Impact Factor: 10.753) but this should be in Nature (Impact Factor: 42.778), and at the same time…
…this modified spike protein shouldn’t be in NATURE.
I didn’t realize that this journal had an Impact Factor of over 10, and it’s utterly ridiculous that Rose thinks a study like this belongs in Nature. No doubt she thinks it’s a clever pun to juxtapose the journal Nature with a statement that this “modified spike protein shouldn’t be in nature.” In reality, though, what her misuse of this study demonstrates more than anything else is that Jessica Rose does not understand nature when it comes to COVID-19 and the virus that causes it.