Part VI: The Updated Preprint on the “Brain Virus” Experiments is Deeply Concerning
Plus: Bait and Switch Number 2
Version 2 of the preprint study by Song et al., published January 21, 2024, helped to clarify a few things, confirming what I had suggested before, but introducing a few really troubling arguments and conclusions that do not seem to be justified. Either way, this research, or the knowledge gained by specific actors, does not negate its potential as a biological weapon - whether covert, overt, or fake.
Version 2’s explanation of what caused the high mortality
The short answer provided in the updated preprint is: GX_P2V(short_3UTR) is intrinsically highly attenuated (in previous versions, they seemed to say the opposite). But it killed all of their humanized mice.
“Cell-adapted mutations can cause 100% mortality in hACE2 mice”
Before we dive into version 2 of the preprint, it’s worthwhile to recall how those mutations emerged: the viral variant GX_P2V(short_3UTR) is the result of passaging GX_P2V, supposedly in Vero cells. This engendered two mutations including a 104ntd deletion. Consecutively, GX_P2V(short_3UTR) was cloned via plaque assays which led to several single ntd changes, including some in the spike gene.
In Version 2, the authors first doubled down on the notion that the mutations could have been responsible for the high mortality they observed, writing,
“the cell-adapted mutations of pangolin coronavirus GX_P2V … can cause mortality in hACE2 mice.”
Note the usage of the word “cause.” As I noted before, the naming of the individual variants in their work is often ambiguous. Nonetheless, in this context, it is pretty clear that the cell-adapted variant referred to above is GX_P2V(short_3UTR).
Yet, later, they clearly emphasized (emphasis added)
“the fundamental nature of GX_P2V(short_3UTR) as being highly attenuated.”
Combining the last two assertions seems to imply a contradiction. However, there is yet one other facet to all of this that in previous posts I mentioned only briefly and due to insufficient information had to put in the parking lot:
Viruses do not exist, independently of a host.
Their virulence and pathogenicity are shaped by their interaction with the host, both at a population as well as individual level.
RNA viruses such as CoVs mutate quickly. Substantial mutations have even been observed in susceptible individuals (see also by book where I discussed this in greater detail in the context of the SARS-CoV-2 Omicron variant).
There is great hype about animal models being able to predict the complex and dynamic interrelationship between viruses and their host. The complexity of this has been greatly under-appreciated.
The updated version of the preprint by Song et al. offers more insights about the special, and largely unknown, animal model they used (at least, how it is portrayed by the authors).
Another cause - involvement of their unique human ACE2-transgenic mouse model
The authors do not say this explicitly, but they seem to have their own understanding of “causation.” The way it is stated above clearly indicates the mechanism responsible for the high mortality - cell-adaptive mutations. Yet, after depicting these as the causative agent, they then go on to describe yet “another” cause.
Shifting gears, Song et al. implicate their animal model as what triggered the high mortality.
Specifically, they blame the unexpected pathogenicity of the virus on the high ACE2 expression in the brain of their “specifically designed human ACE2-transgenic mouse model.”
They further also confirm what they had suggested in their version 1, that these hACE2 mice have “abnormal physiology.”
Do Song et al. argue that it is their “specially designed” animal model that makes the mice sick in and of itself? Are they saying that it is the pathological alterations inherent to the animal model that killed them?
That does not seem to be the case either. In fact, they seem to steer their arguments away from the possibility that the animal model, alone, could be the culprit.
In this context, apparently, it is not the case that agent A leads to phenomenon Z; rather, agent A plus agent B plus possibly many other (unknown/unstated) factors are, taken together, what results in phenomenon Z.
This just points to the fact that in reality, there is hardly ever a single-causative agent for anything in life, and many, if not all things are related. (The complex issue of causation is a different story, see e.g. my book where I wrote an entire chapter on when, ostensibly, per WHO criteria, adverse events following vaccination caused, or did not cause, an observed adverse event).
Nonetheless, the updated preprint takes a strange turn. While it initially appears as if the authors seem to stress that both the viral variant and the animal model together cause mortality, they later undermine this rationale.
Distraction from the fact that the animal model may be seriously flawed - portraying it as the solution
Previous parts of my notes on the Song et al. study have suspected issues with their specific mouse model. Song and collaborators, in version 2 of their preprint, reiterate concerns with this model but then develop a whole new argument. Frankly, the way they do it is rather concerning.
A spin on the animal model
Immediately following the admission of the abnormal physiology, Song and colleagues conclude (emphasis added)
“Thus, the outcomes from the mouse infections in this study have no correlation with human infections.”
But why on earth? This assertion is new in the updated preprint version but has not been justified at all.
The only possible attempts for an explanation I could possibly come up with are:
a) Given that the mouse model seems to cause pathologies in those mice in and of itself, the argument, in context, seems to be that the same would not apply to human infections.
(Note: I don’t buy this, because
The above statement is not a demonstration that humans, when infected with the virus, will not experience any adverse effects.
In addition, recall that Song and colleagues also stressed they believed it was the cell-adapted mutations that caused the high mortality.)
b) Likewise, Song et al. argue that
“Under normal circumstances, both human and mouse brains exhibit low expression of ACE2.”
To them, this seems to be proof that in general, the expression of ACE2 in humans is NOT high. And the latter is obviously meant to demonstrate that humans would therefore be safe, should they be injected with the cell culture mutated virus.
Are they saying their virus is attenuated because normally you have a low expression of ACE2 in the human brain?
If it was the high ACE2 expression that killed the infected mice, how do you know that lower ACE2 expression does not cause ANY problems?
After all, ACE2 plays important roles in CoV infection. SARS-CoV-2 relies on it (among others) as a cell receptor to invade human cells; but apparently, it then also usurps the normal function of ACE2.
Another shift in research objectives
And here is the kicker. Based on their arguments, Song and colleagues strongly believe that (emphasis mine)
“[GX_P2V(short_3UTR)], which shares a certain degree of homology with SARS-CoV-2, may be valuable in assessing the effectiveness of broad-spectrum COVID-19 vaccine candidates against unknown future variants.”
Surprisingly, this shifts the focus again; rather than as a vaccine candidate itself, the aim now seems to be the use of GX_P2V(short_3UTR) as a proxy for SARS-CoV-2.
Additionally, they suggest that
“our lethal mouse infection model presents no obvious inflammatory responses in the main affected organs, the lungs and brain, thereby providing an alternative model to evaluate antiviral drugs' efficacy in inhibiting viral replication in vivo.”
This, too, is a shift from previous objectives. Now Song et al. are asserting that their untested mouse model, which may have killed the test mice, is a perfect surrogate model to be used in all sorts of future experiments, specifically for evaluating the efficacy of drugs and vaccines against SARS-CoV-2.
How do they come to this conclusion? Apparently, in their view:
The virus itself is harmless.
Their hACE2 mouse model is a perfect animal model for these types of experiments because it causes specific pathologies in those animals.
It is those pathologies that predispose the mice to getting killed when infected with the virus.
But since humans do not have these pathologies, the model is perfect for conducting drug and vaccine tests for human use. (Never mind the mouse model was intentionally endowed with human features.)
At present, it is a mere hypothesis that the animal model ingresses ONE specific vulnerability, absent in humans, and that this alone is what allows the virus to kill those mice.
In reality, there is a complex interrelationship between a pathogen and its host. (But this does not seem to be part of the theoretical framework employed by Song et al.)
Bait and switch number 2
In Part 4, I described the potential scenario of a first bait and switch that shifted the orientation from vaccine development to a new scary virus.
But Song and colleagues may have soon realized that things still don’t add up. They finally indicated it was not the virus per se that killed the virus. In the new preprint, they reiterate that it all could be linked to the animal model instead.
Yet, they may only have learned about the inherent flaws of their specific animal model after they had done all the experiments and published their results.
Rather than concluding that their animal model may have been flawed or otherwise inappropriate, the authors turn the issue on its head via another seeming bait and switch. This is done by the series of the following arguments. In essence, these are:
“A great animal model exaggerates specific features that you normally do not see in real human infection”
The point that, in essence, a great animal model is one that is unrealistic seems a bit exaggerated. However, it is very much in line with how it is presented by the authors.
Song et al.’s differentiation between human infections and those of their test animals
This is based on the following line of thought:
“It is very likely that the high pathogenicity of GX_P2V C7 in our hACE2 mice is due to the strong expression of hACE2 in the mouse brain.”
“Under normal circumstances, both human and mouse brains exhibit low expression of ACE2.”
“Thus, the outcomes from the mouse infections in this study have no correlation with human infections.”
Concerns and Counterarguments
Some may find the above conclusion extremely troubling, for various reasons.
1. ACE2 expression in the brain: While Song et al. refer to studies that suggest “normal” human brains exhibit low ACE2 expression levels, this has meanwhile changed.
In earlier years of the pandemic, the distribution of ACE2 in the brain was controversial. Yet, later studies mostly contradicted purported low ACE2 expression patterns (which in part may be a result of detection methods that are significantly sensitive to different forms of the enzyme). For example, findings in 2021 pointed to ACE2 expression being downregulated in the brain with Alzheimer’s disease (AD) whereas more recently, Reveret and collaborators identified higher levels of soluble ACE2 in the brain of AD. They also discovered that ACE2 protein is predominantly localized in microvessels in the mouse brain but more frequently found in neurons in the human brain.
Additionally, Lukiw and colleagues confirmed that ACE2 is very widely expressed in multiple human cell- and tissue types and organ systems. Importantly, they also identified “significant ACE2 expression in different anatomical regions of the brain and CNS,” noting “high ACE2 expression” in the amygdala, cerebral cortex, and brainstem and “highest ACE2 expression level” in the pons and medulla oblongata in the human brainstem. Furthermore, according to their Fig. 2, tissue-specific patterns of ACE2 expression also revealed substantial ACE2 expression levels in the fetal brain.
2. Human infection and disease presentation, while impacted by ACE2 expression, is shaped by much more than this factor alone. It is not clear how Song et al. could possibly envision that the entire pathophysiological context and immune history might not be relevant at all.
3. Conundrum about the difference between brain and lung: If the presumed “late brain infection” is only possible in the brain of the test mice because of their high ACE2 expression level (and precluded in humans due to the purportedly low expression of this enzyme in the human brain), then one should at least be able to observe some pathologies in other organs or tissues that have a high enough ACE2 expression.
Paradoxically, Song et al. detected “high viral loads” in both lung and brain tissues in their human ACE2 transgenic mice, which they emphasized were “correlated with the strong expression of hACE2 in these tissues.”
But if the ACE2 overexpression in the brain is such a unique phenomenon, acting as the unique thing that killed those mice, then why did the lungs of infected mice show “no significant pathological alterations” at all?
“The fact that the animal model is unrealistic is a good thing”
Again, this may sound harsh. Albeit, this is at the core of the arguments made by Song and colleagues. Here, their rationale is as follows:
Regarding the hACE2 mouse model they used: “we are notified that these hACE2 mice have abnormal physiology, as indicated by relatively reduced serum triglyceride, cholesterol, and lipase levels, compared to those of wild-type C57BL/6J mice.”
This admission is in both preprint versions.
The assertion that because of the ACE2 expression issue the model is highly beneficial is based on the following:
(a) It is because of the high ACE2 expression in the mouse brain, absent in the human brain, that the human ACE2 transgenic mice purportedly “have no correlation with human infections,”
(b) Those ACE2 expression characteristics in the mouse model also do not "alter the fundamental nature of GX_P2V(short_3UTR) as being highly attenuated.”
Apparently, these arguments want to support the claim that:
Apart from the artificially induced over-expression of ACE-2 in the brain responsible for the observed mouse mortality (which at this point is merely speculation), the hACE2 mouse model should be able to mimic real human infection with GX_P2V(short_3UTR).
Additionally, leaving the ACE2 expression issue aside, viral characteristics such as additional cell entry mechanisms, tissue susceptibility, virulence, and transmissibility should thereby not be altered/influenced by their animal model.
I find this difficult to grasp. Even though ACE2 plays a critical role in the infection of human cells, SARS-CoV-2, for example, relies on co-receptors and additional mechanisms (in greater detail discussed in my book), and infectivity alone is not the sole factor that determines the outcome of an infection.
Finally, turning the ACE2 expression issue on its head, Song et al. seem to have every reason to praise their model, thereby shifting the attention from its inherent flaws to their believed benefits:
Song and colleagues highlight the applicability of their animal model, diverting attention to something else: it caused no inflammation in all the main affected organs, including the brain.
As a result, they conclude, it provides “an alternative model to evaluate antiviral drugs’ efficacy in inhibiting viral replication in vivo.”
Conclusion
The updated version of the preprint started by asserting that mutations obtained during cell culture were the thing that caused the high mortality of GX_P2V(short_3UTR). Later, however, the unexpected lethality of this variant was blamed on the unique humanized mouse model. Finally, the sole causative agent was reduced to the artificially high ACE2 expression in the brain in this particular animal model.
In a strange move, rather than owning inherent flaws of their mouse model (known concerns as well as those still unknown), Song and colleagues, in an apparent new bait and switch, claim that the animal mode used has great benefits (summarized in the figure below). They emphasize it had been specially designed, apparently because of its high ACE2 expression in the brain which, they argue, makes it a great model for antiviral research.
Summary of preprint version 2 by Song and colleagues.
Yet, the claim that for humans, ACE2 expression in the brain is low has meanwhile been proven incorrect, not only for “normal” conditions.
The purported pathogenicity of the GX_P2V variants varies substantially between the animal models that are used, ranging between being harmless to seemingly being perfect killer viruses.
Interestingly, Song et al.s’ conclusion about GX_P2V(short_3UTR) steers away from its previously reported killer potential and merely stresses its homology to SARS-CoV-2 which could make it useful for COVID-19 vaccine research.
Ironically, the series of papers, beginning with the one by Lu et al., then the first preprint by Song et al., and finally the second, now completes the circle, putting us - almost - back to where we started: these experiments originally aimed to characterize GX_P2V(short_3UTR) in different animal models to determine its pathogenicity and potential as a vaccine candidate.
However, the findings by Song and colleagues puts a troublesome spin on it. Suppose GX_P2V(short_3UTR) will indeed be used again (directly or indirectly) for the development of CoV vaccines.
Unfortunately, the Song et al. characterization of their animal model will expect to see mortality or substantial pathology in hACE2 transgenic mice even when infected with the “inherently attenuated” variant.
In the context of vaccine design, related concerns for humans would likely be brushed off and explained away as above.
On the other hand, should GX_P2V(short_3UTR) be investigated for its spillover potentials again, it will be interesting to see if we, once again, will be informed of its pathogenic potential….
(And so, apparently, research - and funding - on numerous GoF-type experiments will never end).
Overall, the updated version of the preprint does not provide a clear confirmation or denial of whether or not GX_P2V(short_3UTR) has spillover risk (i.e., is indeed dangerous), or how this could be described by various 3-letter agencies. Either way, this does not negate its potential as a covert, overt, or fake biological weapon. And there are, unfortunately, many ways how this could be done (but I will refrain from articulating some of the ideas I have about this).