PART II - A rational analysis of the new lethal brain virus experiments reveals unseen inconsistencies as well as substantial covert biorisk implications

The identified inconsistencies and vulnerabilities contradict each of the expected benefits of GoF(type) work

This is a continuation of yesterday’s post about the recent preprint study by Song et al. about a novel brain virus that purportedly killed all of the humanized mice after they had been injected with this virus.

2.1. The potential of dangerous mutations obtained via cell culture experiments

The key question remains to be answered: what could explain the high lethality of the novel viral variant as depicted by Song and colleagues?

Motivations for the Study

It is worthwhile to consider the aims of the study by Song et al., which they described as follows (emphasis added):

• Their starting point is the observation that the implied pangolin CoVs “have not been studied in terms of their adaptive mutations in cell cultures.”

• Specifically, “the adaptative evolutionary changes” of GX_P2V(short_3UTR) “during their laboratory culture remain understudied.”

• As such, the previous pangolin CoV variant was cloned “considering the propensity of coronaviruses to undergo rapid adaptive mutation in cell culture.”

• “Due to the propensity of coronaviruses to undergo adaptive mutation during passage culture, we cloned and analyzed mutations in GX_P2V(short_3UTR), focusing specifically on the pathogenicity of the cloned strains.”

Their emphasis on adaptive mutations of CoVs during lab experiments, and what this may mean in terms of their pathogenicity (or infectivity) cannot be overstated.

Rapid and deleterious evolution during laboratory procedures

As portrayed by Song et al., the high lethality of the pangolin CoV variant was first only seen during their experiments published in the 2024 preprint and related to both the cloned and uncloned GX_P2V(short_3UTR) variants but not to its precursors. To investigate the mechanisms of mortality it is important to recall how these variants were obtained.

Step 1: Serial Passaging

Importantly, GX_P2V(short_3UTR) is the result of GX_P2V from cell culture experiments (passaging) only.

• Notably, this resulted in the two mutations as described, including the 104ntd deletion.

• Strangely, GX_P2V(short_3UTR) was in the prior study by Lu et al. found to not cause serious disease in various in vitro and in vivo infection models. These authors attribute the attenuation of this variant to the 104-nt deletion in the HVR in the 3′-UTR. 

• However, as noted in Part I, Song et al. stress that in both golden hamsters and hACE2 mice it is only GX_P2V that is attenuated, in sharp contrast to GX_P2V(short_3UTR), which in their study caused 100% mortality in the mice.

(The apparent conundrum about this disparity will be further dissected in Part 3).

Step 2: Plaque Assays

Additionally, the specific CoV variant that was at the center of the preprint study (i.e. GX_P2V C7)  was obtained from GX_P2V(short_3UTR) during cloning. Interestingly, the term cloning in this context does not only involve the traditional usage of “cloning,” i.e. to obtain individual organisms with identical genomes. According to Song et al, they cloned GX_P2V(short_3UTR) through two successive plaque assays.

Most common usage of plaque assays:

For viruses that cause visible damage to cells, plaque assays are most widely known for their utilization to determine the quantity of infectious virus in a sample. In addition, plaque assays have also long been used for the preparation of isogenic clonal virus stocks. Specifically, as explained by Chapman et al.:

“Viral samples that might contain genetic variation are applied to cell monolayers at a series of dilutions such that in some of the monolayers one can be confident statistically that all of the virus within one plaque are descendants of a single particle, and thus of as uniform sequence as genetic drift will allow.”

This approach dates back to the work by Dulbecco in 1952 and was initially applied to DNA viruses. Thus, it is expected that viruses within a single plaque will be descendants of a single virus so that serial dilution eliminates prior genetic variation.

Plaque assays as mutagens:

Even though Song and colleagues may have set out to employ plaque assays for the isolation of homogeneous clones, what may have happened in reality raises important issues about these assays, particularly regarding their potential to engender (clandestine) mutations.

• Song and colleagues suspect that the high mortality of the new variant may be caused by mutations during the plaque assays. Specifically, they note that compared to the original sequence of GX_P2V(short_3UTR), GX_P2V C7 has two amino acid mutations in the spike protein.

• They suspect that this mutation obtained during cloning may be a virulence-enhancing mutation.

• Note, however, that this suggestion does not align with the fact that the uncloned variant had the same lethality as the cloned. Nonetheless, as observed in Part I, as we do not know the exact identity and characteristics of the variants in question, it is difficult to say for certain which variant was the first to cause severe brain infection in humanized mice.

• The potential of plaque assays to engender (and even assess) mutations has long been known. Chapman et al. suggest that in vitro evolution experiments under selective conditions could be combined with plaque assays for the isolation of mutated clones of some DNA viruses. Now, if this is feasible for DNA viruses, it seems plausible the same applies to RNA viruses as well, even in the absence of specific evolutionary pressure. This may simply be due to the rapid mutagenic potential of RNA viruses as fostered by their error-prone replication mechanisms.

Biorisk Implications

Based on the above, it seems possible that all sorts of common lab experiments, even without deliberate/stated selective pressure, may introduce mutations that could result in much more worrisome CoVs, and that much faster than openly admitted.

While passaging in lab culture has been suggested by many as being one of the key factors that could have led to the emergence of SARS-CoV-2, this notion has not made it to mainstream media or official recognition.

Yet, now, ironically, the studies on these pangolin CoVs clearly demonstrate how quickly and effectively CoVs may obtain deleterious mutations in a lab - both during serial passaging and even in the context of trying to isolate those viruses as specifically done via plaque assays.

2.2. Causes of the high pathogenicity - is there another culprit other than cell culture experiments?

Other than culturing in the lab, is there another potential cause that may have resulted in the sudden high lethality of the pangolin CoV variant? Song and collaborators do indeed offer a suggestion. However, it does not seem their explanation holds much water.

In their opinion, the hACE2 mouse model that they used (note: I added this qualifier to avoid any confusion, thanks to Ben Haskell’s comment below) is relatively unique. They speculate that hACE2 may be highly expressed in the mouse brain - which may in part explain the high mortality rate. They emphasize that these hACE2 mice may have abnormal physiology as indicated elsewhere.

This idea that the high mortality could in part be blamed on this particular mouse model does not readily align with the fact that hACE2 mice were also used as control. Indeed, hACE2 transgenic mice inoculated with the inactivated virus as well as those mock-infected did not show any clinical symptoms. Therefore, the deaths following infection with the real virus cannot be due to the animal model per se.

(Note added: This, however, is only coherent if the animal model used did not introduce any unrecognized vulnerabilities that made the mice injected with the real virus significantly more susceptible than those injected with the inactivated one - thanks to the insightful comments by Ben Haskell below. And indeed, many problems hinge on issues of the models that are used, as I further elaborate below).

2.3. What does all this mean?

The central role of the lab experiments

While Song et al. made a big deal about the surprisingly high lethality, their own work is incoherent in that their explanations do not align with what has been stated or described in other parts of their article.

• The potential mechanisms proposed by Song et al. for the high pathogenicity of the viruses are: (a) their unique mouse model, or (b) mutations obtained during cloning (plaque assays). Yet, as I described above and in Part I, both of these arguments were found to be wrong or incompletely justified.

• Within the framework of what is published, the only potential mechanism that is left is passaging. (This is not to say that plaque assays, which, from a viral perspective, do have some resemblance to passaging, cannot play a substantial role as mutagens either.)

• Both passaging and plaque assays are common methods that are frequently applied in laboratories. Passaging, of course, has widely been believed the source of SARS-CoV-2 itself; (there are plenty of revelations on this, so I don’t feel the need to go into this any further).

• What we are now left with is yet another confirmation that these common lab practices, when applied to CoVs, have enormous risk potential. Thereby it may be possible to rather easily mutate a number of CoVs and other RNA viruses, quickly turning them into something much more nasty.

Several Open Questions Remain

The above does not resolve the seeming controversy between Song et al. and Lu et al. about whether GX_P2V(short_3UTR) could cause severe disease in humanized mice. Other open questions include:

• Viruses do not exist on their own and do not evolve their virulence/pathogenicity independently of a (specific) host. (I discussed all of this much more fully in my book).

• A troubling aspect of the preprint by Song et al. is that the high pathogenicity of GX_P2V(short_3UTR) was only seen/tested in their very special hACE2 mouse model. By contrast, Lu et al. found that this variant was able to infect, but did not cause disease, in all five of the tested infection models. Notably, these did not include humanized mice as the animal models were limited to golden hamsters and BALB/c mice.

• Thus, there is no clear understanding as to what is the “correct” animal model to prove or disprove the high pathogenicity of these viral variants.

• The experiments by Song et al. relied on only 12 mice, with only 4 infected by the real virus. To prove the characteristics of the novel viral variant(s), this would need to be repeated by independent labs. But do we really want that, seeing the potential risks?

At present, it must be concluded that there are serious issues and open questions about these experiments, not only related to their GoF-type nature but specifically regarding details of this work and what has been reported.

What is frustrating, however, is the fact that these very gaps, unknowns, and open questions make it even more difficult to analyze, oversee, and regulate such work; likewise, all the gray areas further increase the likelihood of cutting corners or for deliberate misuse (indicated in the figure below). For example:

• CoVs, after various lab modifications, can have substantially different characteristics and pathogenicities, depending on the cell lines and animals used.

• The specific details of the experiments can effectively shape the interpretation of the results.

• The ambiguities may help camouflage dangerous experiments.

That the experiments by Song and colleagues contradict any potential benefits of GoF experiments will be further described in Part 3. Also further discussed will be the question as to which variant was the first one to cause high mortality, the conundrum of when and if these variants were so lethal,  and why there are reasons to doubt. I will also analyze what all of this means in terms of biosafety, biosecurity, research funding, and politics.