The latest antivaccine conspiracy theory: The COVID-19 outbreak is due to a failed SARS coronavirus vaccine

Antivaccine activists are incredibly consistent in two things. First, they are conspiracy theorists par excellence. This first observation should not be surprising given that antivaccine views are strongly grounded in conspiracy theories, particularly what I like to call the central conspiracy theory of the antivaccine movement, namely that “they” (the CDC, big pharma, doctors, etc.) “know” that vaccines cause autism and all the adverse health effects falsely attributed to them by antivaxxers but are covering up the studies and data showing that. I’ve been writing about this aspect of antivaxxers for as long as I’ve been blogging, beginning the Robert F. Kennedy, Jr.’s conspiratorial take on the CDC Simpsonwood conference and more recently with Del Bigtree’s cherry-picking of quotes from the World Health Organization Vaccine Safety Summit a couple of months ago. Second, to antivaxxers, it’s always, always, always the vaccines. A recent example occurred when antivaxxers tried to blame the vaccine strain of measles for the deadly measles outbreak in Samoa late last year. Now, with the recent outbreak of serious coronavirus disease that started in Wuhan, China, antivaxxers are looking for another way to blame vaccines. [Note added March 23: the coronavirus is now named SARS-CoV-2, for “SARS coronavirus-2, and the disease it causes COVID-19, for “coronavirus disease 2019.”) Yesterday, they blamed the flu vaccine and viral interference. Today, former scientist turned antivaccine “scientist” James Lyons-Weiler is claiming that 2019-nCoV, the coronavirus strain behind the outbreak, is in reality the result of an attempt to make a vaccine that went wrong.

Because, of course, it’s always vaccines. Always. At least he could have been a bit more original, but originality is foreign to antivaxxers. Their conspiracy theories always involve harm caused by vaccines; so whenever there is an outbreak of a new disease you can always count on them somehow finding a way to blame vaccines, be they existing vaccines like the influenza vaccine or experimental vaccines, like the supposed coronavirus vaccine.

Conspiracy theorists generally don’t want to be perceived as conspiracy theorists; so they often go to great lengths to convince you that their conspiracy theory is reasonable. One strategy they employ to achieve that is to consider, seemingly very carefully, alternate explanations for the phenomenon being “explained” by their conspiracy theory. Lyons-Weiler does exactly that in his post, examining four different “explanations” for the 2019-nCoV coronavirus outbreak:

  1. Natural coronavirus related to bat coronaviruses, Not a recombined virus.
  2. A recombined virus that naturally picked up a SARS-like spike protein in it N-terminus (3′ end) of the viral genome.
  3. A recombined virus made in a laboratory for the purpose of creating a bioweapon.
  4. A recombined virus made in a laboratory for the purpose of creating a vaccine.

Like any good conspiracy theorist, he examines each “hypothesis” in turn and then—surprise! surprise!—concludes that his hypothesis conspiracy theory is the most likely to be true. Let’s start with #1 (Natural coronavirus related to bat coronaviruses, Not a Recombined Virus.) Lyons-Weiler begins by citing a Lancet paper hot off the presses yesterday on the nucleotide sequence of the 2019-nCoV coronavirus. In it, the authors reported the results of next generation sequencing of samples from bronchoalveolar lavage fluid and cultured isolates from nine inpatients, eight of whom had visited the Huanan seafood market in Wuhan, obtaining complete and partial 2019-nCoV sequences. They then did phylogenetic analysis, comparing the 2019-nCoV sequence to those of other coronaviruses in order to examine the evolutionary history of the virus and try to infer its likely origin.

Their findings:

The ten genome sequences of 2019-nCoV obtained from the nine patients were extremely similar, exhibiting more than 99·98% sequence identity. Notably, 2019-nCoV was closely related (with 88% identity) to two bat-derived severe acute respiratory syndrome (SARS)-like coronaviruses, bat-SL-CoVZC45 and bat-SL-CoVZXC21, collected in 2018 in Zhoushan, eastern China, but were more distant from SARS-CoV (about 79%) and MERS-CoV (about 50%). Phylogenetic analysis revealed that 2019-nCoV fell within the subgenus Sarbecovirus of the genus Betacoronavirus, with a relatively long branch length to its closest relatives bat-SL-CoVZC45 and bat-SL-CoVZXC21, and was genetically distinct from SARS-CoV. Notably, homology modelling revealed that 2019-nCoV had a similar receptor-binding domain structure to that of SARS-CoV, despite amino acid variation at some key residues.

And conclusions:

2019-nCoV is sufficiently divergent from SARS-CoV to be considered a new human-infecting betacoronavirus. Although our phylogenetic analysis suggests that bats might be the original host of this virus, an animal sold at the seafood market in Wuhan might represent an intermediate host facilitating the emergence of the virus in humans. Importantly, structural analysis suggests that 2019-nCoV might be able to bind to the angiotensin-converting enzyme 2 receptor in humans. The future evolution, adaptation, and spread of this virus warrant urgent investigation.

Translation: This virus probably originated in bats, but investigators can’t rule out its going through an intermediary, such as an animal sold at the seafood market in Wuhan where the outbreak originated and a number of non-aquatic animals such as birds and rabbits were also on sale before the outbreak. The authors noted that the genome “sequences of 2019-nCoV sampled from nine patients who were among the early cases of this severe infection are almost genetically identical, which suggests very recent emergence of this virus in humans and that the outbreak was detected relatively rapidly.”

So there is an inserted sequence that is different from the bat strains examined. Lyons-Weiler beliefs it came from a DNA vector pShuttle-SN, that was used in China in the 1980s to create a more immunogenic coronavirus. Let’s just say that he doesn’t make his case very well, but let’s take a look. First, he dismisses explanation #1 thusly:

Option 1. Natural coronavirus related to bat coronaviruses, Not a Recombined Virus.

Evidence for: Phylogenetic clustering with Bat coronaviruses.

Evidence against: Low bootstrap support (N=75) and presence of a INS1378.

Status: Falsified hypothesis. Test: Survey coronviruses in animals in the wild.

Lyons-Weiler might be correct here in that there probably to be some sort of evolutionary or recombination event resulting in the seemingly inserted sequence, but I wouldn’t call the hypothesis “falsified.” He’s leaping to that conclusion because he wants to blame a vaccine.

Next up:

Option 2. A recombined virus that naturally picked up a SARS-like spike protein in it N-terminus (3′ end) of the viral genome.

Evidence for: The INS1378 codon bias similar to snakes ($)

Evidence against: Insufficient match in database search to other known CoV spike proteins (Ji et al., 2020)

Status: Speculative hypothesis. Unlikely.

Test: Find an isolate that matches 2019-nCoV in the wild and reproducibly independently isolate the virus from a wild animal (a match will confirm).

What is “codon bias”? First, let’s explain what a codon is. A codon is the three nucleotide sequence that specifies a single amino acid in the protein that is encoded by a given gene. Because there are more three nucleotide combinations than there are amino acids (64 different codons—61 if you remove the stop codons that tell the ribosomes to stop making protein—versus only 20 amino acids used to make proteins), the genetic code is said to be “degenerate,” meaning that multiple codon sequences can encode a single amino acid, as this chart shows:

Note that “U” is uracil, which is what RNA uses instead of “T” or thymidine, which is what DNA uses. Basically, in terms of the genetic code, functionally U = T. Codon usage bias is a term that refers to a bias in which codons are used to specify a given amino acid. When multiple codons can encode a given amino acid in a protein, organisms have a bias in which codon(s) are used to encode that amino acid. In other words, the codons used to specify a given amino acid are not used randomly. Some are used more frequently than others, hence the term “codon bias” or “codon usage bias.”

What Lyons-Weiler is referring to is this paper, which claims that there was a coding bias consistent with snakes. Lyons-Weiler is also correct that this particular paper is not very good and makes an extrapolation that’s difficult to justify.

As noted in this WIRED story:

“It’s complete garbage,” says Edward Holmes, a zoologist at the University of Sydney’s Institute for Infectious Diseases and Biosecurity, who specializes in emerging RNA viruses, a class that includes coronaviruses like 2019-nCoV. Holmes, who also holds appointments at the Chinese CDC and Fudan University in Shanghai, is among a number of scientists who are pointing out—in virology forums, science Slacks, and on Twitter—what they deem to be major flaws in the paper, and calling on the journal to have it retracted. “It’s great that viral sequence data is getting shared openly in real time,” says Holmes. “The downside is then you get people using that data to make conclusions they really shouldn’t. The result is just a really unhelpful distraction that smacks of opportunism.”


But Holmes says that this approach has a number of issues. For one thing, comparing codon bias is a very indirect way of identifying an animal host. For another, it works best when looking at species from wildly different corners of the taxonomical hierarchy. Plant and mammal viruses have really distinct codon patterns. So do insect and bird viruses. But within more closely related groups, it’s much harder to tease apart meaningful patterns—especially when the authors only sampled a few species. “There could easily be other species out there that are more similar than snakes, but we don’t know because they just haven’t been put in the analysis,” says Holmes. He and others are also skeptical that snakes could be the intermediate host, because there aren’t any documented cases of reptiles hosting coronaviruses that can transmit to humans. Their cold-blooded biology is just too different. 2019-nCoV, and its closest relative, SARS, belong to a subgroup known as beta-coronaviruses, which are only known to infect mammals.

Here’s the trick, though. Lyons-Weiler dismisses the possibility that 2019-nCoV somehow picked up a SARS-like sequence naturally based on a paper that made unjustified extrapolations and assumptions, presenting that paper as though it’s the only evidence for this. In fact, he hasn’t ruled out a recombination event with another, unknown coronavirus. As he himself pointed out, RNA viruses mutate rather rapidly.

Regarding option #3 (that 2019-nCoV was a bioweapon), the less said, the better. There is a BSL-4 (highest biosafety level) lab 20 miles from the Wuhan market, and, of course, whenever there’s a new outbreak one of the standard conspiracy theories that always—and I do mean always—crops up is that the organism causing the outbreak is a bioweapon that got loose. Personally, the version of this conspiracy theory that most amuses me is that 2019-nCoV is a weaponized coronavirus made in Canada and stolen by Chinese agents. Canada? Making bioweapons? Seriously?

Of course, the main reason Lyons-Weiler rejects the bioweapon conspiracy theory is because he wants to push the “vaccine strain” conspiracy theory instead:

IPAK researchers found a sequence similarity between a pShuttle-SN recombination vector sequence and INS1378. Here’s a shot of the alignment and the DOT Plot. Here’s the nucleotide sequence at NCBI’s Nucleotide database. Here’s a patent for its use in recombination virology. The pShuttle-SN vector was among many described in a 1998 paper by Bert Vogelstein et al; here is a company where one can purchase the pShuttle-SN vector.

Before I go any further, I did some BLAST searches for the “INS1378” sequence helpfully provided by Lyons-Weiler. Here is the result.

Not surprisingly, it’s all sequences related to the Wuhan market outbreak, although there is one segment that closely resembles (78% identity) the bat sequence:

2019-nCoV searches

I recognized the sort of analysis that Lyons-Weiler had done; so I decided to try to replicate it by aligning the “INS1378” sequence with the pShuttle-SN sequence using the BLAST tool. (Here is the paper in which construction of pShuttle-SN was reported. It was indeed done as the starting point of an attempt to make a vaccine against the SARS coronavirus, using an insert for the SARS coronavirus spike protein cloned into Xho4 and Kpn1 restriction enzyme sites.) My analysis found a sequence 1,182 nucleotides long with 68% identity (which, let me tell you, is not very high). Here’s part of it:

2019-nCoV and pShuttle alignment

This is pretty good alignment, but certainly nowhere near slam dunk evidence for the origin of 2019-nCoV being the SARS-like spike protein coding sequence inserted into pShuttle-SN, especially when it’s almost certainly more likely that this sequence could have arisen naturally, given how many coronaviruses have SARS-like sequences in them. If this particular sequence in 2019-nCoV actually did come from the vector and the version of the SARS spike protein gene that was inserted into pShuttle to make pShuttle-SN, there should be large swaths of 100% match, if not 100% match.

As an actual expert pointed out on Twitter, the match should be close to 100% for the entire sequence:

I’m starting to think I know why Lyons-Weiler no longer runs a bioinformatics core.

Lyons-Weiler then makes another leap of “logic”:

The very researchers conducting studies on SARS vaccines have cautioned repeatedly against human trials;

“An early concern for application of a SARS-CoV vaccine was the experience with other coronavirus infections which induced enhanced disease and immunopathology in animals when challenged with infectious virus [31], a concern reinforced by the report that animals given an alum adjuvanted SARS vaccine and subsequently challenged with SARS-CoV exhibited an immunopathologic lung reaction reminiscent of that described for respiratory syncytial virus (RSV) in infants and in animal models given RSV vaccine and challenged naturally (infants) or artificially (animals) with RSV [32], [33]. We and others described a similar immunopathologic reaction in mice vaccinated with a SARS-CoV vaccine and subsequently challenged with SARS-CoV [18], [20], [21], [28]. It has been proposed that the nucleocapsid protein of SARS-CoV is the antigen to which the immunopathologic reaction is directed [18], [21]. Thus, concern for proceeding to humans with candidate SARS-CoV vaccines emerged from these various observations.” – Tseng et al.,

The disease progression in of 2019-nCoV is consistent with those seen in animals and humans vaccinated against SARS and then challenged with re-infection. Thus, the hypothesis that 2019-nCoV is an experimental vaccine type must be seriously considered.

This doesn’t even make sense. Yes, in animal models, animals vaccinated with a SARS vaccine then challenged with the SARS coronavirus developed severe disease due to an excessive immune reaction to the virus primed by the vaccine. That is indeed a reason to be very cautious moving to clinical trials of vaccines against SARS, MERS, or 2019-nCoV. It is quite a stretch to think that this observation strongly suggests that 2019-nCoV is likely to have come from a strain made in order to produce a coronavirus vaccine. Even Lyons-Weiler seems to realize this:

The disease progression in of 2019-nCoV is consistent with those seen in animals and humans vaccinated against SARS and then challenged with re-infection. Thus, the hypothesis that 2019-nCoV is an experimental vaccine type must be seriously considered.

Evidence for: Sequence homology between INS1378 to pShuttle Coronavirus vaccine; presence of a SARS-like Spike protein in bat coronavirus, otherwise most similar to bat coronaviruses; low bootstrap value.

Evidence against: Low sequence homology (but highly signifiant). NB these viruses are RNA viruses and they can evolve quickly, even under laboratory conditions.

Status: Most likely.

Test: Determine the nucleotide sequence all laboratory types of coronavirus being studied in China (a match will confirm). Find an isolate that matches 2019-nCoV in the wild and reproducibly independently isolate the virus from a wild animal (a match will falsify).

Seriously. Lyons-Weiler used to do genomics for a living before he turned to pseudoscience. Surely, at some level deep down, he must know what thin gruel this is. In fact, he inadvertently signals that he knows by following up “low sequence homology” with “but highly significant.” Then he concludes that his idea is “most likely” immediately after having dismissed the conspiracy theory that 2019-nCoV is an escaped bioweapon.

None of this stops him from fear mongering about 2019-nCoV:

The implications are clear: if China sensitized their population via a SARS vaccine, and this escaped from a lab, the rest of world has a serious humanitarian urgency to help China, but may not expect as serious an epidemic as might otherwise be expected.

In the worst-case scenario, if the vaccination strain is more highly contagious and lethal, 2019-nCoV could become the worst example of vaccine-derived contagious disease in human history. With an uncharacteristic aysmptomatic prodromal period of 5-7 days, individuals returning from China to other countries must be forthright and cooperative in their now-prescribed 2-week quarantine.

Yes, Lyons-Weiler is actually suggesting that the Chinese sensitized their population with a SARS vaccine and that the vaccine strain of virus escaped from the lab. He soon realizes that that possibility would mean that this outbreak would likely be short and quickly contained, as there is no evidence that China ever conducted a mass vaccination program against SARS. There is, after all, as yet no approved vaccine for SARS. So he has to add that bit about the “worst case scenario” in which 2019-nCoV is more contagious and lethal. The problem is that we already know that it’s definitely not more lethal than MERS (not even close, given that MERS is 34% fatal) and likely not more lethal than SARS. He’s just pulling these speculations out of his nether regions. Whatever the origin of 2019-nCoV, Institut Pasteur has not only isolated and sequenced the responsible strain, but figured out how to grow large amounts of it in culture, paving the way to studies to determine the mechanism for its pathogenicity and to develop an effective vaccine.

James Lyons-Weiler’s speculations and fear mongering just go to show how even a scientist can fall under the spell of antivaccine pseudoscience (or any other conspiracy theory-driven pseudoscience). My guess is that the Lyons-Weiler who once ran a genomics core would have recognized that the “science” that the antivaccine Lyons-Weiler of today is laying down is nothing more than wild speculation that’s based on only the thinnest of scientific gruel and highly unlikely to be true. More’s the pity. Even worse, because he has a background in molecular biology, his conspiracy theory will sound plausible to most lay people.