The brain-virus experiments do the opposite they promise and are highly counterproductive
Even within their framework, the purported benefits of GoF-type work are not only disproved but effectively undermined
This is Part III of the “brain virus” series on the work by Song and colleagues. Here, I will scrutinize the argument that such type of work has imported benefits. I will demonstrate that within the framework of this work, based on their own terms and what is publicly available, some substantial pitfalls and concerns demonstratively contradict each of the expected benefits of GoF-type work.
For the sake of the argument - issues to be put in the parking lot
There are a few concerns, to begin with, that I don’t have enough information on, and that I, therefore, cannot comment on in detail. These include:
Isolation of the original virus. Song and collaborators told us that the original GX_P2V sample was originally cultured from lung-intestine mixed samples of a pangolin captured in anti-smuggling operations in 2017. They point out they never isolated this variant.
Isolation of the various mutants, done via successive plaque assays. Thus, isolation in this context means that the virus is passaged through two successive plaque assays.
The particular hACE2 mouse model that was used is largely unknown and has never seen public scrutiny.
Sample size: The preprint by Song et al. is based on one experiment involving only 4 test mice raising issues about the degree to which the presented results hold true in general. (But this shortcoming is not meant as an invitation that such experiments should be repeated elsewhere!)
What has been published is incoherent and confusing. It is difficult to see, what variant of GX_P2V (if any) first caused high pathology in the hACE2 mice employed in the preprint study.
Issues with whole genome sequencing. The resolution level and accuracy of the procedure are unknown.
Despite these important open issues, we know that something killed all of the infected mice. It may not be the virus, all by itself. Some may think they never isolated a virus to begin with. Others may feel that the information presented is unreliable, incomplete, and not trustworthy. After all, why suddenly have such a rapid change in pathogenicity? As indicated before (Part II), much could indeed be explained by an inappropriate animal model. Or was it unprofessional handling or other practical issues that contributed to the killing of the mice? Or do we know of any potential failed results, previous attempts, or ambiguities in how the animals were selected, maintained, or withdrawn from experiments?
Regardless of what may have been going on, let us put these challenges and open questions in the parking lot for now. For the benefit of the doubt, let’s just assume the above concerns are immaterial.
Thus, we ask, within their framework, is it possible to justify the purported benefits and predicted goals of GoF-type work? As will be seen, even if there were no issues at all, according to what is disclosed, we get the opposite of what is promised.
The creation of several novel vulnerabilities
Pathogens are frequently exchanged between laboratories. The new WHO/IHR requirements would also specifically mandate the sharing of their relevant information, creating a scaling and explosion of biorisks in line with the number of experiments done. Obviously, the more (potential) pathogens are being handled and manipulated, the greater the danger. One particular risk is exemplified by the Song et al. work as follows.
A potentially lethal confusion between GX_P2V and GX_P2V(short_3UTR)
As identified above, there is no clarity in the published literature about GX_P2V(short_3UTR)’s pathogenicity in various animal models, including several humanized mice models. According to the original article by Lu and collaborators, this variant did not cause significant disease in various in vitro and in vivo infection models. By contrast, the preprint by Song and collaborators, referring to this work, claims that it was the GX_P2V variant that did not cause disease in humanized mice. In other words, in follow-up experiments, some might use GX_P2V(short_3UTR) as a control virus, assuming it is the harmless one, whereas others might think GX_P2V is the one that is attenuated. Such inconsistencies are troubling, because
If I misread the publications, then the same may happen to others who may follow up with such work. What is worrying, however, is that it is not clear whether or not GX_P2V(short_3UTR) causes significant disease in which animals, including various humanized mice.
If the misunderstanding is not on my side, but rather an error and oversight by Song and colleagues, then this is even more inauspicious.
Notably, one of the two variants in question is ostensibly highly lethal, and the other one is characterized as not disease-causing. I don’t think I need to explain the biorisk implications of accidentally or deliberately switching these variants.
It would not be the first time that humans have made a mistake; unintentionally switching cell lines or pathogens, and confusing names and identities is a common mistake that will be difficult to avoid and could turn into a practical nightmare under the new WHO/IHR policies.
The artificial genesis of something more nasty
It is difficult to grasp why something should first be made worse so that some benefits could be derived from it. This question has been brought forth by many opponents of GoF experiments. The counterargument has been that lab experiments that make pathogens more pathogenic ostensibly serve to enhance science. The knowledge gained by those experiments is allegedly invaluable.
The fallacy of this viewpoint in terms of the pangolin CoVs experiments should be obvious from the following. The original variant was first clearly identified as not being a human pathogen, so much so that some even thought it could serve as a live attenuated vaccine against SARS-CoV-2.
But then, via lab experimentation that lasted a few short years, variants thereof were reported as highly lethal to humanized mice. So, according to Song and colleagues, they made the harmless virus much more harmful. But why would this be helpful? If you wanted to use an attenuated virus as a novel vaccine, what is the point of first turning this virus into a potential killer? Once you have enhanced its virulence or pathogenicity, you have achieved the opposite of what is needed for such a vaccine.
With live vaccines, a major difficulty has been to find a way to attenuate them stably. But if the opposite is happening and a candidate virus is pushed, through lab experiments, to evolve into more troubling variants, you are indeed getting ahead of what could have happened in nature - but tragically, it would be exactly the sort of viral evolution we wanted to avoid!
Knowledge gained and digital information sharing
Even if the novel virus itself never escapes a lab, the new insights gained can be used in a different or even clandestine context.
It is unclear what the impact of these potential killer variants is on humans. Suppose the research work was done in high-security laboratories. Even if that was the case, the knowledge obtained can easily be outsourced. Nonetheless, there is no evidence that the work has been done in the highest-level biosafety setting. Overall, their potential “spill-over” - notably, from a lab - to the human population is certainly not zero.
In addition to the concern raised by John Campbell that these novel brain viruses are in effect biological weapons, it is important to note that published, unpublished, and covert experiments of this type may not only involve the obtained viral mutants per se.
Substantial insights can be gleaned from such work which then could be utilized for the deliberate genesis of novel pathogens via the host of available bioengineering mechanisms. Apropos, in recent weeks, thanks to USRTK, we have seen much more evidence of how SARS-CoV-2 was likely “stitched together” from individual parts; note that one of the key components - the FCS carrying a Moderna-patented sequence - could indeed have been developed in a Moderna lab as first suggested by Ambati et al. and then further justified and extended in my follow-up publication.
Instead of the promised benefits, the GoF-type experiments are counterproductive
There are clear reasons why the type of work done by Song and colleagues belies any potential benefits of GoF(type) work, even if one ignores any inherent challenges and inconsistencies. In fact, within the very framework of their analysis, the publicly available information clearly shows that the opposite is the case for all of the promised benefits:
“To help predict spillover events”
Song et al. suggest their work might be of great relevance to predicting future spillovers. However, some of the major flaws of this argument are:
The authors themselves described their finding of the 100% mortality as “surprising.” Shouldn’t this be proof that their models and frameworks are radically incapable of predicting real-life developments?
It is important to recall that the progenitor virus GX_P2V was not a human pathogen. That strain was tested in several animal models, including three mouse models, and 7 different cell lines, but always showed limited pathogenicity. Importantly, this variant was able to infect wild-type BALB/c mice, human ACE2-transgenic mice, and human ACE2 knock-in mice, albeit without causing severe disease.
But then Song and collaborators claimed that a variant thereof was 100% lethal for the humanized mice they used. If there is indeed such a jump in pathogenicity, then that was precisely the result of lab experiments because this is how these variants came about. There is no indicator that such a rapid change would automatically have happened in nature, let alone in humans. Logically, it does not make sense that a virus that was confirmed non-pathogenic for all the animals it had been tested on, could by itself become a perfect killer.
The sad and tragic reality is that instead of solving the problem, laboratory work on pathogens creates much greater ones, which however we do not even know how to assess.
Work done in the laboratory is used to model spillovers from nature - but always creates new problems to begin with. In Song et al.’s case, as is customary for such work, these viruses had previously been dug out from their hidden niches, and transported into urban settings; consecutively, they were analyzed and “worked with” in a laboratory setting, forcing them to evolve in different cells lines, etc. However, Song and colleagues never reported any field work or study of how the pangolin viruses evolve in their natural habitat, especially if left alone. Ironically, the reports on the “dangerous viruses” all only emerge after they have been taken from their normal environment and manipulated in a lab.
Spillovers - from nature only? Spillover events have traditionally been defined and modeled for naturally occurring pathogens. However, given that pathogens can now be synthetically manipulated/engineered, it is impossible to ensure they will remain confined to the laboratory setting. Via accidents or deliberate releases, engineered pathogens have, and will, find their way into the open environment. (Of course, much more could be said about this. Those who have been working on biosafety issues have long known that the number of accidents/releases from laboratories far eclipses the “true” spillovers from nature. But this is a story in itself!).
Predicting spillover events - using lab manipulations as proxies: The capacity to engineer and lab-manipulate viruses has exploded in the last few decades. Albeit, spillover modeling generally ignores any laboratory involvement at all, thereby attempting to merely model a wee tiny fraction of potential developments and phenomena in a natural context only.
Inadequate modeling of species susceptibility and others. CoVs can obviously mutate rapidly. Yet, evolutionary distances have mostly been estimated in terms of sequence-based distance measures which however are not well-suited for this purpose. While structure-based homology models are much better, they are poorly understood; computer simulations reveal troubling disparities to the sequence-based approaches that are commonly used. Now add to this the large genetic changes such as the 104ntd deletion during passaging, or genetic changes through recombination, and gaps and errors quickly escalate.
(The above is not meant to be exhaustive as much more could be said. But again, this may be for another time.)
“Help to infer the origin of SARS-CoV-2”
This topic, in greater generality, would require a much more in-depth analysis. Related to the work by Song et al., it is interesting to note the huge discrepancy that exists between what is being done, or what is doable, in a lab context as opposed to natural evolution.
During the earlier part of the pandemic, there was great hype about identifying pangolin CoVs as the natural progenitors of SARS-CoV-2. The group around Zheng-Li Shi determined that two pangolin CoVs showed 85.4% and 92.4% genome-wide and 92.6% and 90.7% S gene nucleotide identity to SARS-CoV-2, respectively. The former is what now is known as GX_P2V. Yet, these low identities do not make these likely candidates for natural spillover events simply because naturally it would have taken much too long for these strains to mutate into SARS-CoV-2.
However, evolution in a lab is radically different than in nature, not only concerning timescale:
Genetic distances in a laboratory context have a radically different meaning. For example, the infamous DEFUSE project proposed to make viruses with unique features of SARS-CoV-2; notably, as recently revealed by USRTK, the proposal was to first identify and then synthetically assemble CoVs that are up to 25 (!) percent different from SARS. And according to these FOIA data, Ralph Baric had already previously generated several SARS-like chimeras which were 20% different from existing pandemic strains, and that even with specifically engineered receptor binding domains.
The synthetic assembly of chimeras in a lab is a radically different ball game than predicting the proximal origin of a virus via natural evolution. Thus, models that only consider what naturally is deemed feasible, completely miss laboratory reality. To put this into perspective: RatG13, which some believe is the closest relative to SARS-CoV-2 is 4 percent different than SARS-CoV-2 but is still too distantly related for natural evolution.
In addition to being able to generate chimeras that have huge genetic differences from some known viruses, the genetic alterations that can be added in the lab, adding novel features, undermine any modeling attempts. For example, one of the features that makes SARS-CoV-2 more nasty is attributed to the furin cleavage site which in the same group of CoVs has never been identified in nature. However, once knowing the specifics of the FCS (which Ambati et al. and I have suggested how it could have come about), it would not require a heroic effort to have it synthetically inserted into another SARS-like virus.
“Help To Prepare Treatments/Vaccines”
Whichever way one looks at it, the above reveals that these types of experiments are plagued by too many issues to be of any benefit. Apart from the fact that it is impossible to a priori know details of novel humanized pathogens, should they indeed emerge from nature, the experimental setup itself impedes any potential advantages for the development of treatments against such pathogens.
The case of the alleged pangolin CoV “evolution” makes this point very clear.
First, the uncertainty about the actual mortality of GX_P2V(short_3UTR) in various animal models is troubling. If it turns out that it is not this variant itself that is so deadly, but rather the specific mouse model used, then this just highlights the many knowledge gaps that could obtained from such experiments, in contrast to actual clinical reality.
If, instead, the high lethality is caused by the specific mutant (for humanized mice first reported for GX_P2V(short_3UTR)), then this is equally troubling as the deadly mutations were the result of common lab experiments performed on a non-pathogenic variant.
What is particularly troubling about these experiments is that the progenitor of the killer virus was initially studied as a novel vaccine candidate. Let’s think about it! GX_P2V was believed to be so harmless - for humans - that researchers eagerly started “working” with it, ostensibly with the goal of injecting variants of it into a significant number of humans.
The very fact that it ostensibly did not take much to turn a vaccine candidate (GX_P2V) into a potentially highly effective killer virus should once and for all expose the limitations, deficits, and dangers of such lab experiments.
According to Song et al., rapid mutations can quickly turn harmless viruses into something much more dangerous. John Campbell described this as a potential threat to humanity. If it is true, as they are saying, that the high mortality was “surprising”, then why are they doing such dangerous research, and why should we entrust anyone with such work, to begin with?
Some of the above is depicted in the figure below, but this does not capture everything yet ….
The figure shows the contrast between the expected benefits of GoF-type experiments (left) and their actual outcome exemplified by the brain virus experiments, solely treated within the framework of these experiments and expectations.
…. As I have been writing about the brain virus experiments, more and more inconsistencies and open questions have come up, beyond their counterproductive potential. Please look for Parts IV and V where I will go into greater detail, also discussing novel issues that only emerged in the updated preprint version, posted by Song and collaborators on January 21, 2024.