Part IV - Is the "lethal brain virus" the result of a lab mistake?
Can the findings about the high mortality to a large part be explained by inconsistencies of the study?
Initially, I thought my discussion about the brain virus study by Song et al. would consist of three parts “only.” However, there are several ambiguities with the study. The authors portray a viral variant as 100% pathogenic on humanized mice, in sharp contrast to the findings of another group of researchers which determined this virus as not harmful at all. Additional concerns have emerged raising the question about their potential consequences.
Along these lines, Part 4 of my analysis tries to understand if the brain virus could be the result of some mistakes done in the lab or during analysis, fostered by some of the gaps identified here.
Why ask this question?
In recent weeks, this study by Song et al. has been vehemently criticized on various social media platforms. Rumors have emerged that the new killer virus has been intentionally fabricated by the CCP, and hypothetical links have been made with Disease X which repeatedly has been “promised” by the WHO. Others believe it’s just a hoax and fear-mongering.
In the spirit of true and independent science, it is imperative to take a step back and scrutinize all reasonable options.
Dissecting the study by Song et al. has revealed several inconsistencies that seem difficult to ignore. Thus, three possible scenarios seem to be:
1. The ambiguities and gaps did not influence the genesis of the novel killer virus.
2. The high pathogenicity of the obtained virus is the result of some lab mishaps brought about by unrecognized issues and gaps.
3. The inconsistencies and gaps only contributed to the high pathogenicity of the virus, but additional key factors were overlooked
In the former case, the researchers were lucky (if that is the right word in this context!). The latter case requires additional scrutiny and will be discussed separately.
Thus, in this part, I am going to consider the experiments themselves, asking if they were the result of certain ambiguities which then lead to weighty mistakes.
The seeming confusion between GX_P2V and GX_P2V(short_3UTR)
As I noted before, the naming of the individual variants in both Lu et al., and Song et al. is ambiguous. This is not only of academic interest alone but could have serious biorisk implications.
To quickly summarize the importance of this concern:
1. Song and colleagues stress that the original GX_P2V virus was not a human pathogen; likewise, the group around Zheng-Li Shi found it to be rather harmless when tested in several animal models.
2. In the preprint, referring to their own previous work, Song et al. argue that they, too, had identified the attenuated nature of GX_P2V.
3. Yet, consulting this previous publication reveals that the variant that was tested was not GX_P2V but GX_P2V(short_3UTR).
Reviewing the list of authors in the preprint study by Song et al. and the one by Lu et al., confirms that this work had indeed been done by the same group of researchers. This makes the above even more intriguing. This is so critical, because:
The preprint study underlines the fact that GX_P2V did not cause obvious disease in hACE2 mice.
Surprisingly, Lu et al. never mention that the animal model they used included humanized mice; the only two animal models they described were golden hamsters and wild-type BALB/C mice, underlining that for those, GX_P2V(short_3UTR) was attenuated.
It does not seem plausible that Song and collaborators - in their preprint study - could not remember what they had done in their own previous work. The significance is the following:
If Song et al. had written that GX_P2V(short_3UTR) (rather than GX_P2V) was NOT pathogenic for the humanized mice they had previously tested, then this preprint which features the 100% lethality of this very same variant for their hACE2 mice, would not make sense!
Even if the preprint study employed a different mouse model, nobody would believe such a radical change in pathogenicity.
Obscure and unreported mutations obtained by cryptic cell-culture experiments?
It is unclear if the apparent confusion between GX_P2V and GX_P2V(short_3UTR) tarnished the experiments done by Song and collaborators.
Is it possible that the ambiguous nomenclature of the variants led to experimental errors?
Song et al. were aware that their findings were inconsistent with those obtained by Zhengli Shi et al. They tried to explain this away by arguing, on the one hand, that Zhengli Shi’s group “tested the virulence of GX_P2V in two different hACE2 mouse models,” but then immediately contradicted themselves in the same paragraph, asserting that “Zhengli Shi et al tested the GX_P2V(short_3UTR) variant that we reported.”
I spent a lot of time trying to distill when and how the seeming confusion between the naming of the viral variants could have emerged. What I found is that there was substantial ambiguity about the naming of all the various variants in the relevant publications by Zheng-Li Shi et al., Lu et al., and Song et al., respectively.
Fig. 1 portrays some of the different names of the viral variants that have been used in the relevant publications. The top part I have written about before.
In addition to the above, the following statements taken from the preprint by Song et al. raise several questions related to the following:
Identity and purity of the viruses that were studied
Whilst downplaying the disparity of their findings relative to those obtained by the Zhengli Shi lab, Song et al. indicate that they ‘did not isolate the wild-type GX_P2V strain.” Some may think that this is rather concerning, especially when setting out to study the mutagenic potential and characteristics of variants thereof.
Was there further unreported passaging done, and if so, on which variants?
As for genetic analysis, Lu et al. explain that they compared the sequence differences between the first passage of the GX_P2V sample and the eighth passage of the GX_P2V isolate. Furthermore, the preprint seems to introduce another uncertainty, stating that the variants they examined in the preprint study were obtained in the following way:
“To obtain a genetically homogenous clone for animal experiments, we cloned the passaged mutant through two successive plaque assays.”
It seems that “the passaged clone” refers to GX_P2V(short_3UTR), but it does not specify how often it has been passaged. Furthermore, the identifiers (Fig. 1) “first passage,” “early passaged,” and “8th passaged” suggest that there may also have been later passages.
Therefore:
There is a real possibility that some variants were further passaged.
Additionally, the variants were likely assessed by plaque assays in several different cell lines.
However, from what is published, it does not seem clear that all the lab experiments with all the variants are clearly parsed out and accounted for.
If additional passaging or plaque assays were done, they may have accumulated additional, potentially unrecognized, mutations.
The context of examining the viral variants for their capacity as a live attenuated vaccine could have put the virus under substantial evolutionary pressure.
The latter point may be particularly relevant. Some of the related outcomes may not have been adequately reported. Worse, mistakes during the lab experiments, including insufficient isolation or purification, may have resulted in contaminants that could have paralleled what some have suggested was key in the lab genesis of SARS-CoV-2.
Cryptic mutations during lab experiments may have led to the furin cleavage site in a SARS-CoV-2 progenitor
In general, one may think that “arbitrary” mutations obtained in the lab, as the result of whichever mistakes or oversights, would certainly not have major implications. Most would think that such genetic changes are likely harmless or even make the viruses less viable.
There are strong indicators that suggest otherwise. In the context of SARS-CoV-2, an underappreciated example was first proposed by Ambati et al. who identified the presence of (essentially) a 2017 Moderna patented sequence in the genome of this virus. That this could be a chance occurrence is very improbable, not only because of the calculations given by Ambati and colleagues. Additionally, in a follow-up study published in Frontiers, I outlined several specific research goals and mechanisms that could have provided the framework to effect the particular genome integration postulated by Ambati et al.
For example, I propose that specific experiments were done to study and develop novel antivirals. By their very nature, these would have involved exposing certain viruses to the new drugs and assessing whether they could impair important viral mechanisms such as getting into the human cell or even the nucleus. Automatically then, such experiments would have created evolutionary pressure on the viruses and favored the development of viral escape mutants.
There are several key factors in my hypothetical analysis:
1. The first concerns unrecognized genetic contaminants. Possibly, synthetic RNAs that could have played several roles during drug development were not recognized as contaminants in cells that later were infected with various viruses on which the new antivirals were tested. Specifically, for RNA viruses such as CoVs, this very setup could have enabled laboratory recombination events between the RNA contaminants and these viruses. That this could have happened very frequently is further supported by very recent insights into what is driving recombination and why RNA viruses would be especially susceptible to gaining and maintaining such newly acquired genetic material under selective pressure -- it is perfectly in line with the mechanisms that could have resulted in the integration of the novel insert surrounding the infamous furin cleavage site in SARS-CoV-2.
2. The second point I am trying to make in my article is that unrecognized contaminants in the form of synthetically generated genetic material, and their biorisks, are insufficiently understood.
3. Another point, that allows a potential explanation as to why the integration would have resulted exactly in the FCS site insert even though the antiviral experiments that I envision did not need this particular sequence, hinges on underappreciated genetic functionalities. This latter point was described as unexpected also in a foundational NIH-funded study.
What all of this means is that there are numerous gaps at all levels, ranging from wrong (conceptual/animal) models, and incomplete or flawed comprehension of biochemical, molecular, genetic, and bioengineering principles, to biosafety concerns engendered by inappropriate purification and other mistakes done in the lab. Human error is and remains a factor that must not be neglected, even in high-security laboratories. Likewise, the ongoing discussion about how a gene is defined and how the phenotype is determined is only a small demonstration of how little we know about even the most foundational issues.
In the context of lab-facilitated evolutionary pressure, we should not be surprised that pathogens could mutate in ways not previously expected.
Potential Implications
The experiments done by Song and collaborators, and how they are reported, are plagued by numerous inconsistencies. Furthermore, there are strong indicators of omissions, and it may be that knowingly or unknowingly, these gaps became the breeding ground for a range of mistakes.
We cannot rule out the involvement of additional mutagens
While it is difficult to see that passaging and cloning could have suddenly enabled a killer virus to emerge, if that was indeed the case, there may be some partial explanations for this:
The group could have done further experiments than explicitly detailed.
Noting that the original agenda of their work was to develop these viruses into attenuated vaccines, it is possible that some unrecognized or unreported procedures in the lab created substantial evolutionary pressure on the virus.
Experimental mishaps such as contaminants may have engendered virulence-enhancing mutations that may or may not have been clearly described.
Potential impact of cryptic mutations
If during the lab experiments, some of the variants inadvertently gained some genetic changes that were either not recognized or reported, then this could have had significant implications.
The variant used for the mice study may have undergone virulence-enhancing mutations without being recognized by the researchers.
The baseline variant that was used to determine the genomic differences of the new variants may harbor unrecognized genetic alterations.
Consequently, the interpretation of their findings could be fundamentally flawed.
The inconsistencies may have corrupted genetic analysis
One of the aims of the brain virus studies was the genetic determination of the individual variants, and Song and collaborators do present detailed genomic data on their viral variants. However, both the individual gaps and more so when in combination, could easily frustrate even the most sophisticated mutational analyses.
Even though the genetic analysis was done via whole-genome sequencing (WGS), the individual gaps could nonetheless lead to wrong conclusions, because of the following:
Issues with de-novo sequencing: WGS attempts to determine the genetic sequence information of an entire organism (or virus). Nonetheless, as the technology is based on stochastic principles, it is inherently prone to errors. Despite much technological progress, it continues to be demanding. This is particularly the case in the absence of a template genome and independent validation by independent labs -- e.g. when dealing with novel viruses and incompletely known viruses as is the case with the novel brain virus.
Even if full genome sequencing is done with sufficient accuracy, it is still susceptible to handling errors. The potential ambiguity of naming the variants described above could, for example, result in the swapping of the individual variants and completely frustrate any real analysis.
Likewise, if the viruses had gained additional mutations that were not expected, then, in case of misidentification, the genetic analyses might be rather useless.
Conclusion
The preprint by Song and colleagues raises serious questions. The claimed results are contradicting those obtained by others. It is not clear if the apparent incoherence is due to inherent gaps and consistencies of the study, or if these are the hidden culprits that made the virus more pathogenic.
Some of the gaps identified in the Song et al. study are summarized in Fig. 2 below.
Fig. 2 summarizes various gaps and inconsistencies in the study by Song et al. These and related issues could have furthered the evolution of viral variants, either as partial or key drivers of the unexpectedly high pathogenicity.
It would be difficult to explain the pathogenicity of the novel variant in light of the results obtained in the Zheng-Li Shi lab. However, the unrecognized involvement of the type of gaps as seen in the study by Song et al. could indeed have made the virus more lethal (Option 1 in the figure) but remained unnoticed or unaccounted for during the presentation of the results.
The very existence of those gaps and ambiguities is deeply concerning. Even if it is impossible to conclude that they indeed led to laboratory mishaps or errors during the initial design and setup of the experiments, and as a result thereof to the surprising increase in pathogenicity of the virus, these same issues may in and of themselves further the evolution of more pathogenic variants.
(Outlook: In part V of this series, I will shift gears, examining Option 3 in the figure, asking whether the purported killer virus is the outcome of some deliberate bait and switch.)