Mice as a Possible Origin of Omicron?
In my search for studies on Omicron I came across this study by Wei et. al., which suggests that mice are likely to be the source of Omicron. I asked my friend about this paper, and she quickly criticized many of the methodology, so I disregarded it for the time being. In fact, part of me was considering to not write about Omicron in general. But then I came across some more studies, all of which throw questions into the actual origins of Omicron.
I’ve mentioned many of these issues throughout this paper, but I’ll collect them all here:
Omicron has a large number of spike protein mutations- On its own this seems in line with what we would expect from a virus; selective pressure would push a virus to mutate in a way that provides better survivability and transmissibility. However, Omicron does present with a large number of its own unique mutations, which raises questions as to why these mutations are present in Omicron and not in prior variants, especially because of large number of mutations.
The spike mutations in Omicron appear to escape most natural immunity and 2-dose vaccination regiments- Again, this is something one would expect with the natural course of a virus. Considering the millions upon millions of doses of the COVID vaccines that have been provided, we can also see how providing such a large number of pressure on the virus in such a manner is likely to produce escape mutations.
Omicron contains the 69-70 codon deletion- This is a bit of a nitpick, but the disappearance and emergence of this deletion seems a little suspect. Granted, there is evidence of this deletion occurring in immunocompromised patients being treated for COVID, but it’s difficult to think that this particular strain disappeared and came back with a vengeance, and far more infectious than anything seen before.
Phylogenetic tree evidence suggest Alpha to be the closest “ancestor” to Omicron- Although Gamma can be taken into consideration as a possible source of Omicron, it didn’t have the same level of circulation as Alpha. Pair that with the first evidence of Omicron coming from the Southern region of the African continent, and the strange presence of the 69-70 deletion, the evidence leans more towards Alpha being a more possible ancestor. This then raises questions as to how one of the earliest variants disappeared and reemerged more than a year later with so many mutations and without being recorded.
Omicron appears to have attenuated affinity towards human ACEII receptors and Serine Protease- Here’s a real kicker. In the case of Delta, its greater infectivity and virulence comes from a mutation within the furin cleavage site that provided it the ability to be more readily cleaved by serine proteases. It seems the opposite has happened with Omicron. Along with that, the mutations in Omicron also don’t appear to provide greater binding to human ACEII receptors. Instead of evolving in favor of both of these interactions, Omicron seems to have gone in the opposite direction. Several of the assays we have discussed suggest the reduced role that ACEII and serine protease play, and a transition to a cell-to-cell method of infection. This runs counter to what we would expect from a standard evolutionary pathway and suggests a more tangential evolutionary line.
Mice models show high viral load in lungs but not in hamster models- This is a bit of a spotty argument, but considering that hACEII transgenic hamster models did not show significantly higher viral load in the lungs compared to standard hamsters, it leaves one to wonder why mice exhibited such high viral load. One possibility is that murine ACEII receptors are expressed in transgenic mice, and if that were the case it would lend evidence to a possible relationship between Omicron’s spike mutations and affinity towards an animal ACEII receptor.
Collected together, it’s hard to argue that Omicron’s evolution follows the typical pattern one would expect with viral evolution under selective pressure from a human most. As indicated above, the cumulative evidence suggests that there may be a possibility that Omicron originated from an animal host and jumped back into humans. More specifically, the evidence appears to point towards mice being the possible original host. This led me to return to the Wei et. al. study and examine that study in detail.
Here, we’ll examine the study and piece together the evidence that may support such a hypothesis.
Note: As we go through the study, remember that the information points towards support/evidence of a hypothesis, and are not conclusive. Remember to keep that in mind that this study provides a hypothetical framework for future studies and research to work off of, and is not definitive.
Examining the Origin Hypotheses
We’ll start with the 3 hypotheses laid out by the researchers:
“The proximal origins of Omicron have quickly become a controversial topic of heated debate in the scientific and public health communities (Callaway, 2021; Kupferschmidt, 2021). Many mutations detected in Omicron were rarely reported among previously sequenced SARS-CoV-2 variants (Shu and McCauley, 2017; Hadfield et al., 2018), leading to three prevalent hypotheses regarding its evolutionary history. The first hypothesis is that Omicron could have ‘cryptically spread’ and circulated in a population with insufficient viral surveillance and sequencing. Second, Omicron could have evolved in a chronically infected COVID-19 patient, such as an immunocompromised individual who provided a suitable host environment conducive to long-term intra-host virus adaptation. The third possibility is that Omicron could have accumulated mutations in a nonhuman host and then jumped into humans. Currently, the second scenario represents the most popular hypothesis regarding the proximal origins of Omicron (Callaway, 2021; Kupferschmidt, 2021).
The first two hypotheses assume that Omicron acquired these mutations in humans (collectively referred to as ‘human origin hypothesis’ hereafter), while the third assumes that Omicron acquired mutations in a nonhuman species. Based on our previous work in viral evolution (Shan et al., 2021), we hypothesized that the host species in which Omicron acquired its particular set of mutations could be determined by analyzing the molecular spectra of mutations (i.e., the relative frequency of the 12 types of base substitutions)”… “Consistent with this phenomenon, viruses belonging to different orders (e.g., poliovirus, Ebola virus, and SARS-CoV-2) were found to exhibit similar molecular spectra of mutations when evolving in the same host species, while members of the same virus species exhibit divergent molecular spectra when evolving in different host species (Shan et al., 2021). Since de novo mutations can thus strongly influence the molecular spectrum of mutations that accumulate during virus evolution in a host-specific manner, the host species in which Omicron acquired its mutations could be determined by analyzing information carried by the mutations themselves.”
I’ll start with the “human origin” hypotheses and point out a few issues with those.
To start with the “cryptic emergence” hypothesis, I’ve already pointed out a few issues with this hypothesis. Considering the level of surveillance being conducted, one would expect that several concerning mutations would be identified before Omicron made its way into the population. Also, considering that Omicron contains many more mutations than normal, it doesn’t quite make sense that there was no point where it would have been noticed.
With respect to PCR, the Spike protein deletion (69-70 codon deletion) would cause an “S Gene drop” during PCR and would not be detected by typical COVID testing procedures, although this drop could be masked by other strains of the virus. However, most sequencing data utilize molecular typing procedures called Sanger-based typing (SBT) or next-generation sequencing (NGS), highly sensitive sequencing methods that can pick up differences in bases between samples, and should thus pick up on these base pairs quite easily. These typing methods do require PCR, but I also expect to see the “S Gene drop” being taken into account. We did, after all, sequence the Alpha variant even with the gene drop appeared during standard COVID testing, and are continuing to sequence Omicron’s variants in light of the “S Gene drop” seen in Omicron.
This doesn’t mean that this hypothesis is false, however I would argue that the evidence in support of it is the weakest of the 3 hypotheses presented, especially considering that it would require serial passage through many infected individuals and going unnoticed during that time frame.
Similar sentiments can be made for the second hypothesis (immunocompromised patient). This is considered the most popular hypothesis, however it still has many issues.
When examining immunocompromised patients as a reservoir for mutations, one must take into consideration the dynamics between a virus and the host.
An infected individual may be infected with many different strains of a virus, all of which are in competition with one another to replicate and spread. Strains that mutate to have greater binding affinity to host cell receptors or quicker entry into host cells will outcompete strains that do not have similar capabilities. It’s the reason why Delta took over prior strains and gained dominance in the world. Following along that evolutionary line, we can suspect that SARS-COV2 strains should continuously gain mutations in favor of ACEII binding.
However, when in the presence of a host’s immune system, selective forces are at play to rid the infection. SARS-COV2 naïve people (those who have not had natural infection or vaccination) will have no experience going against the virus, and will first rely on the innate immune system before the adaptive immune system can come in and eliminate the virus. In those who have been immunized, they can quickly utilize their adaptive immune system to produce antibodies and potentially rid the virus. Although the adaptive immune system is great at eliminating SARS-COV2 strains that are similar to previous ones, it may have trouble eliminating ones that have mutated extensively. Instead, the immune system operates like a sieve, or a better analogy would be to look at antibiotic resistant bacteria. In that scenario, antibodies may operate like antibiotics and remove pathogens that the antibodies/antibiotics can recognize. However, it will have trouble with those that are resistant/escape these defenses, which can then replicate and take over the host.
This is one of the greatest concerns with the widespread vaccination campaign, such that a widely circulating virus may come into contact with millions of vaccinated individuals, all of who’s immune systems may operate as a sieve and selectively eliminate strains of the virus susceptible to antibody binding while leaving those who can escape to replicate and spread among the population.
But this isn’t the case for immunocompromised individuals. In many instances these individuals have immune systems that are naïve and continue to be naïve to infections. Their lack of creating a robust immune response in the face of an infection does not provide the “sieve-like” antibodies to select for strains that can escape the immune system. They cannot provide a countermeasure to an infection, and instead they operate as a battleground for strains of SARS-COV2 to compete with one another for replication dominance. Because of that lack of pressure from the host’s immune system, we should expect the earlier principle to be in effect; in immunocompromised individuals infected with SARS-COV2, the virus would be expected to gain mutations that confer greater binding to ACEII, as these strains of the virus would be better at competing against strains with less binding affinity.
The evidence in support of this theory would take into account the presence of some of Omicron’s shared mutations in case studies of immunocompromised individuals. For example, the 69-70 deletion has appeared in some case studies of patients with long-term infection with COVID. The N501Y and E484 mutation (I’ve seen either the E484K or the E484Q mutations emerging) have also emerged in some of these patients as well. It should be noted that many of these mutations work in favor of SARS-COV2 binding to ACEII, which we should expect in these individuals. Another important thing to note here is that mutations that confer greater ACEII binding and escape from antibodies are likely to be mutations in the receptor binding domain (RBD) of the spike protein, which come into contact with both the host receptor and possible neutralizing antibodies.
There is an issue of many immunocompromised patients being provided antibodies and convalescent plasma, which could provide a framework for selective pressure against the virus. Yet even then, many of these patients don’t produce the extensive level of mutations seen in Omicron, and certainly not the ones that seem to reduce binding to ACEII that Omicron seems to exhibit. Many of these mutations seem to be in line with what we would expect, and instead provide greater binding to ACEII.
In fact, it would seem that if one were to choose either between an immunocompromised host or someone with natural immunity/vaccinated, it would be more plausible for Omicron to have emerged from the latter cohort due to Omicron’s ability to escape prior immunity, as well as the presence of millions of immunity sieves that can put selective pressure on the virus to select for those escape variants.
Overall, the evidence doesn’t seem to support the human emergence of Omicron, and Omicron’s ability to escape prior immunity runs counter to what one would expect if we were to expect immunocompromised hosts to serve as reservoirs of viral mutation. Add in the fact that Omicron’s ability to replicate and infect runs tangential to the need for ACEII binding or serine protease cleavage, I find it difficult to argue in favor of the human hypotheses.
Remember though, that this is not definitive evidence that such a thing cannot occur. Many variables could explain the factors that I have outlined above. However, based on the evidence so far the human hypothesis does not seem to hold.
On that note, let’s look more into the Wei et. al. study.
The mutations in Omicron’s spike protein show evidence of strong positive selection forces
Getting into the first section of the results, the writing here is a little confusing so I’ll try to clarify some information from the researchers.
The researchers first created a phylogenetic tree to determine the mutations dominant in Omicron. In order to find Omicron’s ancestor, the researchers utilized a sequencing database called BLASTn, which compares sequencing data from different samples or organism. Using this software, the researchers noted that two variants from the B.1.1 (Alpha) lineage were the closest matched to Omicron, suggesting the Alpha line as being the likely progenitor for Omicron.
The researchers then compared 3 samples from chronically infected patients. We would expect chronically infected patients, such as those who are immunocompromised, to exhibit high levels of mutations especially in favor of binding to the human ACEII receptor. However, the researchers noted that examination of these 3 infected patients, as well as analysis of prior variants did not show the same level of mutations that one would expect. In particular, fewer mutations were indicated within the receptor binding domain. Not only that, but comparisons of sequences between the chronically infected patients, the prior variants, and Omicron suggest that Omicron contains a high level of nonsynonymous mutations.
Synonymous mutations are mutations that can be seen across several variants, such as the N501Y and E484K mutation appearing in several variants. Their presence suggests a similar selective pressure in favor of these mutations. For example, these two mutations, which I have discussed extensively, provide greater binding to the human ACEII receptor and escape from human host immunity. Synonymous mutations are, qualitatively speaking, a sign of consistent selective pressure in favor of a similar receptor.
Taken together it appears that, from these researcher’s perspective, that Omicron’s spike protein underwent selective pressure in favor of a different ACEII receptor, as that would explain the high level of nonsynonymous mutations relative to the other viral strains that were examined.
These observations strongly suggested that the Omicron variant had undergone a strong positive selection for the spike protein that no other known SARS-CoV-2 variants evolved in humans had been subjected to. Considering that the spike protein determines the host range of a coronavirus (i.e., which organisms it can infect), we therefore hypothesized that the progenitor of Omicron might host-jump from humans to a nonhuman species because this process would require substantial mutations in the spike protein for rapid adaptation to a new host.
These results fall in line with evidence that Omicron appears to have reduced affinity to human ACEII relative to Delta and wildtype SARS-COV2 strains.
Omicron doesn’t appear to follow the typical evolutionary line seen in humans
The researchers then suspected that the molecular environment within humans would affect the type of mutations SARS-COV2 exhibits.
The researchers’ use of pre-outbreak and post-outbreak Omicron variants are intended to differentiate between Omicron variants that are likely to have undergone selective pressure in humans, with the expectation that pre-outbreak Omicron variants did not have widespread circulation within the human population and are a better representation of the earliest strains of Omicron and not ones that are highly prevalent. For the sake of clarity and simplicity, think of the pre-outbreak Omicron variants as the first variants to possibly jump from an animal reservoir into humans, while post-outbreak Omicron variants are indicative of a variant that has been passed through many humans and have gained some mutations in favor of human ACEII receptor binding.
Here, researchers noted that the point mutations in Omicron pre-outbreak show some slight differences in the base substitutions when compared to prior variants and post-outbreak Omicron, such as the higher rate of C→A mutations in pre-outbreak Omicron samples.
Researchers then chose 45 random mutations from the mutation spectrum of SARS-COV2 (hSCV2 spectrum) to simulate a “psuedovirus”. When done 100 times, the researchers indicated that these samples did not show statistical significance to the same extent as that of pre-outbreak Omicron samples, suggesting that the mutations seen in these pre-outbreak samples were not likely due to randomness.
Finally, the researchers wanted to exclude the possibility that early mutations in Omicron may affect the mutation landscape of Omicron after spreading globally. When the researchers sampled 100 mutations in post-outbreak Omicron samples and compared them to prior sequences, researchers noted no statistical significance. This would suggest that post-outbreak Omicron mutations undergo the same types of mutations as seen in prior variants (human ACEII selective pressure), again hinting that pre-outbreak variants were under different selective pressure than humans to accumulate the mutations seen within these variants.
The molecular spectrum of these postoutbreak Omicron mutations was not significantly different from the hSCV2 spectrum (P = 0.64, G-test; Fig. 2B and 2C). This finding indicated that Omicron would acquire mutations following the same molecular spectrum as other SARS-CoV-2 variants during its evolution in human hosts. Collectively, these molecular spectrum analyses revealed that preoutbreak Omicron mutations were unlikely to have been acquired in humans.
Pre-outbreak Omicron mutations seem to align more with viral mutations seen in mice rather than humans
To figure out which animal could have served as the reservoir for Omicron, the researchers next collected coronavirus sequencing data specific to several mammals and visualized them to see overlaps in sequencing. The results suggest that pre-omicron mutations were more similar to those seen in mice coronaviruses than in humans, suggesting that the mutations had to arise due to infection of mice.
Consistent with the results of our previous study (Shan et al., 2021), drawing 95% confidence ellipses for each host species showed that the molecular spectra clustered according to their respective hosts (Fig. 3B), likely because viruses evolving in the same host species share the mutagens specific to that host’s cellular environment. In supporting this point, the molecular spectrum of postoutbreak Omicron mutations (which are known to have accumulated in humans) was located within the human 95% confidence ellipse. In contrast, the molecular spectrum of preoutbreak Omicron mutations was within the mouse ellipse, suggesting that the preoutbreak mutations accumulated in a rodent (in particular a mouse) host.
Many of pre-outbreak Omicron’s mutations overlap with mutations seen in mice-adapted SARS-COV2
Originally, mice weren’t considered to serve as good hosts for SARS-COV2 until the presence of the N501Y mutation, which showed greater affinity to mice ACEII relative to the wildtype strain of the virus (note the N501Y deletion is not seen in Gamma, adding more credence to Alpha being the proper ancestor). This would suggest that, given the ability to infect mice, SARS-COV2 would then mutate and adapt to produce greater binding affinity to the mouse ACEII receptor.
From here, the researchers examined whether there were overlaps in pre-outbreak Omicron mutations with those found in mouse-adapted SARS-COV2, which would indicate that the mutations seen are due to evolutionary adaptation towards mouse ACEII receptors.
If the progenitor of Omicron indeed evolved in a mouse species before the Omicron outbreak, we postulated that its spike protein likely adapted through increased binding affinity for mouse ACE2.
I’ve included too many figures at this point, especially ones full of statistics (we’ll get to that later). Instead, I’ll point to this figure which highlights the shared mutations between pre-outbreak Omicron and mouse-adapted SARS-COV2 (in red) indicating a high overlap of shared mutations.
One thing to point out here are the mutations within the receptor binding domain (RBD). Remember that these amino acids tend to be more conserved compared to the rest of the spike protein domain because mutations here can either provide much greater binding to the host receptor or near loss of binding affinity- it’s a double-edged sword region of the protein. It’s also the region where mutations are likely to produce the most loss of neutralizing antibodies as well.
Taken together, we can see that many of the mutations seen in Omicron not only reside within the receptor binding domain, but are similar to those seen in mouse-adapted SARS-COV2. This would explain why Omicron seems to exhibit both a loss in binding affinity to human ACEII as well as a large loss in neutralizing antibodies; the move from one receptor to another is likely to confer changes in binding affinity for both the receptor as well as antibodies, all of which can be seen in the prior parts of the Omicron series (Part II for ACEII binding and Part III for reduced antibody effectiveness).
RBD mutations in pre-omicron variants provide greater binding affinity to murine ACEII receptors
We’ve previously indicated that Omicron appears to have reduced binding affinity to human ACEII compared to prior variants. We have also outlined that mutations in the spike protein should tailor the spike towards greater binding to the suspected host animal in question.
If Omicron is expected to have come from mice, we should then expect to find evidence of greater binding affinity towards mouse receptors.
And the researchers provide evidence of that through docking models that measured the binding affinity of several point mutations and compared these measures to a reference RBD sequence. It’s interesting to note that most of this binding affinity appears to be derived from two point mutations, specifically Q493R and Q498R, which the researchers comment have been indicated in prior mouse-adapted SARS-COV2 strains.
The molecular docking-based predictions suggested that the RBD of Omicron exhibited higher binding affinity for mouse ACE2 than that of RBD encoded in the reference SARS-CoV-2 genome, further suggesting an evolutionary history in mice (Fig. 5B). And as expected, the mutations detected in the RBD of the other four VOCs of SARS-CoV-2 as well as those of variants isolated from chronically infected human patients, showed no apparent changes in their binding affinity for mouse ACE2 compared with the reference RBD (Fig. 5B).
And for the very last piece of evidence, the researchers compared the RBD of Omicron and its capability to bind to the ACEII receptor of other mammals to examine if other mammals aside from mice could serve as reservoirs for the progenitor of Omicron. Of course, the results suggest that mouse ACEII showed the greatest binding affinity to pre-outbreak Omicron RBD. I won’t post the image here due to size constraints, but it is worth noting that several mammals, including of course humans and some other ones such as camels show significant binding affinity to Omicron’s RBD, suggesting some conserved structures between each species’ ACEII receptor.
The last post (Part V-2) will provide concluding remarks, limitations of the studies examined in these series as well as this mouse model study, and finally the citations.