Thanks for the tip! A lot of it is the same level of evidence as the Boston Globe story, centered on Sax and Charness, as was a story in another outlet today (I already forgot which). But now the Charness case study appears to be online, time to take a look.
*edit: I spoke too soon. The Post story adds a Pfizer statement hinting at some unpublished evidence that the placebo group also had rebounds (I doubt this), several newly-fished anecdotes, a rehash of my analysis in Pt 2 that these rebounds exceed normal noise in testing, and a comment section presumably full of people blaming the repeal of the CDC mask mandate lol.
This is like being a fly on the wall in a place where the REAL science is taking place. I can't contribute a lot to the discussion but am very grateful to all of you.
This is a great series, thanks for putting it together.
Your explanation of part of the viral replication process was really helpful to me.
Does the polyprotein contain a bunch of pieces and tools that, once separated assemble into a virion?
It occurs to me that, in individuals vaccinated against spike only, flagging the cell as "sick" via MHC would probably need to find bits of spike to present. But, perhaps the spike proteins are still tied up inside the polyprotein when the protease inhibitors are active. Thus, nothing is gained from the Paxlovid "pause" because the infected cells aren't being flagged and destroyed.
This could be different in a natural infection, if the whole virus is taken apart and antibodies evolved for many different pieces. In computer terminology, you could "execute" the viral RNA in a sandbox to find out what proteins are created and prepare to recognize them; point being this produces a lot more ways to identify an infection, including precursors and byproducts (e.g. nsp5), rather than just relying on "finished" spike only.
Everything in the polyprotein “package” (it’s actually two packages but that’s not very important since nsp5 works on both) is non-structural proteins (nsps) that are tools to help the virus do its work in the cell. The rdrp and other RNA-replication/preparation associated nsps are essential, others are enhancing, in that they suppress the cell’s immune response or organize resources in the cell, essentially making a factory that churns out the “sub-genomic” proteins that don’t get read by the cell’s ribosomes until after some weird reassembling of the RNA molecule (which copy of it, at this point? I’m a bit unclear on this) leads to the rdrp kicking out discrete mRNA instructions for the structural proteins including the spike. So the spike is not tied up, it’s just that the whole viral RNA molecule doesn’t promote reading the non-Orf1 / polyprotein parts and those parts have to be processed by the nsp machinery before they are substantially expressed. So their products, essentially everything after Orf1 which starts with the spike, show up later. This might not mean the Paxlovid pause always happens before spike is being expressed, but probably at least sometimes, and those are the cells which will (in some other percentage) restart replication unless a non-spike-based (innate) immune response takes them out. This could include intracellular innate immunity. So it really shouldn’t be a 100% rebound rate, and doesn’t seem to have been in the trial which was all unvaxxed (“natural” infection, with the possible monoclonal antibody confounder); and if this turns out to be what’s happening in the vaxxed, I would definitely look at innate immune suppression.
*edit: I meant to include a link to the nsp portfolio here, note that their number scheme skips the normal nsp11 and so add +1 to their nsp11 and above to make it consistent with most publications https://zhanggroup.org/COVID-19/index.html#table1
I would imagine that if the viral load is substantial at the time Paxlovid is taken, once Paxlovid is absorbed, any cells subsequently infected by the virus floating around will get stuck in pause mode, so there would be a fairly large number of paused cells ready to unpause as the Paxlovid disappears over time. Because of the exponential decay of the Paxlovid, the resumption won't "turn on" all at once, but it certainly has a head start, unless the infected cells are destroyed in the interim (or the bits of virus decay or are destroyed).
Do you know whether antibodies are typically formed against NSPs in a natural infection? Because that would be a possible distinction between vax and unvax. Are NSPs sometimes displayed to flag an infection? It seems plausible that in many cases the polyprotein is partially unpacked but not completely.
Potential innate immune suppression is another possible reason for differential performance. I still don't understand what is known for sure about that question. The role of pseudouridine would explain some degree of innate suppression, although as I understood it, the suppression occurred intracellularly, suppressing TLRs within a cell that was transfected. But all the transfected cells should have been destroyed within a few weeks.
One thing that does seem to be a possible issue is that the stabilized mRNA has a lot more psuedouridine than is typically used naturally in mRNA. Is excess pseudouridine metabolized and new created when needed? Or does an excess float around for a while causing problems?
I guess it's not clear in the first place that there is a differential between vaxed and not - maybe it just leads to recurrence 1-2% of the time. I now have a vaxxed relative currently taking paxlovid (maybe 2 of them) so I guess I will get some first hand evidence. If one or both have recurrence then that's suggestive that the rate is higher than a few percent.
Right, the virus has a head start in that every instance of Paxlovid "working" is an instance of leaving a potential future restart from unopened polyproteins, and maybe the math on how much viable viral RNA is still around is pretty brutal but this wouldn't be much of a "problem" for the virus since even a 1/1000000 successful restart rate is akin to going a mere "6 turns back" in autosave.
I don't know if anyone's even looked at the antibody response for nsps. So as for whether it is occurring, I could only guess, and would guess that the nsps are not as immunogenic in terms of adaptive immunity and might be more TLR / natural IgM domain, "oh this looks like a bacterial or viral RNA unwinder or protein chopper of some sort." Maybe some conserved immune recognition with other coronaviruses.
Still, it's an interesting question whether anyone's even looked, maybe it's mentioned in some monoclonal antibody dev research. The preoccupation with spike and N protein seems to go back to SARS-1, and then N turned out to be pretty useful for tracking infection when SARS-2 came around.
As long as RNA is single strand I don't think there is a intracellular immune response issue but I'll double-check. Otherwise the pseudouridine resists normal RNA degradation, but that's more "optimization" than evasion. (*edit: ok, there's a bit of overlap into antiviral RNA breakdown evasion https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3241635/) For my part I try not to get too elaborate with the innate immune suppression question; it could just be a result of injury and immune exhaustion from the transfection, nothing fancy. Akin to how diabetes, which is inflammatory, also suppresses innate immunity just by promoting a lot of vascular and tissue damage all the time.
There's also still the potential Omicron (BA.1 / BA.2) interaction, with the latter sibling potentially explaining why these anecdotes didn't emerge earlier.
My take-away was that there are several intra-cellular mechanisms that are on the lookout for specific mRNA sequences that commonly occur in antigens. However, these sequences occasionally are needed to be used in legit host mRNA.
In order to prevent these "false positives", there is a mechanism for substituting pseudouridine strategically around the sites where these false positives occur. The presence of pseudouridine apparently disables these checks and flag it as an exception to pass through. I really don't know much about this but it makes sense that if you are looking for "bad" sequences, they will inevitably show up in "good" mRNA.
By encoding the vaccine mRNA using pseudouridine it thereby evades these various protection mechanisms, but the question is whether it also has a more widespread or subsequent impact?
Right, that would be the part about TLR7 and TLR8, so there is an immune element but it's just intrinsic to ssRNA, not "viral" ssRNA, so it's a bit redundant to the problem of avoiding normal mRNA breakdown. It also might not apply in vivo since the LNPs don't release the RNA payload into endosomes. (Also, from a vax perspective you might want a little TLR activation, even if it shuts down translation / protein output more quickly, as it would give and adjuvant effect. And in Study No. R-20-0112 (https://www.tga.gov.au/sites/default/files/foi-2389-06.pdf) BioNTech only compared the pseudouridine (modRNA) RBD and full spike (the final version) vs a non-mod mRNA for the RBD, so who knows if a non-mod mRNA for the full spike would have performed just as well.) There are also some mRNA sequence considerations, strings of nucleotides that might lead to a diminishment of host expression, but again this is more like "optimization" than pure immune evasion, and is more about alternate codon spelling than nucleotide modification.
I have to feel that the long term consequences of your body making a toxic spike protein and then presenting alongside other proteins (natural) will lead to de-immunization because the body would eventually start viewing spike as "self". This is basically immune tolerance in a nutshell.
Its unclear to me whether this is an effect of Paxlovid - or the mRNA vaccines.
Right, that would make a lot of sense and is yet another reason constant boosters are crazy. For the moment my money remains on tolerance not happening based on published antibody kinetics after boosters, but there could be a situation where T Cell immunity is desensitized to the spike even though B Cells / antibodies keep ramping up, due to the former not having been as heavily primed to begin with, who knows. Also as far as Paxlovid my guess is that there isn’t a lot of spike around when the “needs more uncleaved Orf1” moment that defines the pause occurs, so what probably makes the difference in that assumption is innate immunity.
The trial document says both previously infected and vaxxed were excluded. Seems weird, maybe they thought there wouldn't be enough severe outcomes to show efficacy, the rationale might be somewhere in the paperwork
I'm guessing this is done as some position of sanctimony like "we are helping the unvaccinated, even if they won't help themselves!" That would at least explain the high cost as well- "The unvaxxed can receive treatment, but they will have to pay."
I think that's because there really is no rationale. For Molnupiravir and their press briefing they just went "well, the unvaxxed are in bigger danger so let's just give it to them as top priority!" What I believe is happening is that these therapeutics are supposed to be considered less intrusive. Essentially just grab and go. At Biden's SOTU he mentioned a "test to treat" protocol in which someone gets a PCR test and has a prescription filled at the same time.
I guess all I can say is that I really don't know why they are doing what they are doing. I do hope you didn't read through all 318 pages of that protocol!
it would be interesting to see what happens to those Ab titers over time. But - I think its unknown how the body will respond to this synthetic spike and immune challenges in the future..
Well, just like you now I'm starting to consider how it may be pretty obvious. I stated that we may not have used protease inhibitors alone to treat viral infections. For HIV a cocktail usually entails a nucleoside analogue, integrase inhibitor, protease inhibitor, essentially it is a full cocktail of drugs which are intended to target many facets of the virus. It would make sense that the virus may have a few built-in workarounds to go against clogged channels.
But you've spent a lot of time digging deep and piecing things together and it really shows that you don't need to be some PhD as long as you are a critical thinker, and it clearly shows in this case Brian.
I guess we still need to wonder how long the immune dysfunction is, but I am unsure to what extent that is occurring. Ironically, I saw that The Naked Emperor's Newsletter posted an article in which they believe that long COVID may partially be blamed on reduce immune function. Considering that some of the symptoms of vaccine side effects are similar to that of Long COVID I'm actually curious if there are concerns over immune dysfunction. There's also consideration into whether the host cells can recognize uncleaved proteins and target them for removal, but I suppose that would require even further investigation.
Edit: this was the article that The Naked Emperor posted:
The OUTRT assumes that host cell protein degradation essentially does the work of the nsp5, with endogenous cysteine proteases attacking the same cleavage motifs and freeing the nsps, just more slowly than nsp5 would do. So even if this is part ot protein breakdown it rescues the replication process before it can prevent the same. Since the virus doesn’t pack nsp5 into the virion there’s got to be some overlap, though I haven’t looked up the lit on endogenous cysteine protease motifs.
That's actually interesting. So that would mean that the recognized sequence of amino acids is conserved between SARS-COV2 and humans? I wondered if it may be akin to restriction enzymes which recognize specific nucleotide sequences and clip accordingly. Evolutionarily, that would provide some ability to say "clip our proteins here and it will work out for us" while simultaneously stating "clip here and you can kill off your enemy". Although I suppose our genomic makeup with exons and start/stop portions are important as well.
OK, the OUTRT totally survives the autocleavage problem. Uncleaved nsp5 turns out to behave the same way as my assumed endogenous proteases, i.e. slowly. Once they are cleaved, they work quickly, probably tons more compared to bound. So unless all nsp5 is pax’d, Paxlovid just makes it take longer until all nsp5 (Pax’d in a cell) is cleaved and all polyprotein in turn. Literally just delays the inevitable.
Just to clarify are you saying that the uncleaved protease is able to still function? That would mean that it would still have to fold into the proper conformation and that seems a bit difficult to imagine.
I guess the idea was that the covalent interaction would just hold out long enough that the virus just dies out? I suppose we would have to wonder about the activity of endogenous nucleases and whether they would get rid of the mRNA and RNA after a set period, but then we would have to factor in downregulation or suppression of cellular functions from some of the virus' proteins as well.
Extremely difficult to imagine, which is why I felt safe assuming an endogenous protease here, but if there is one, no one has looked for it as far as I can tell. The coronavirus nsp5 / 3CL-pro motif might in fact be a reinvented wheel.
What Muramatsu, et al. show (link in my other comment) is that it still auto-cleaves (and therefor cleaves according to the operative motif) with ~20 pro-form residues on either end, but at a slower rate. By varying the concentration of already matured or not nsp5 added to pro-form, they show that not-matured is much slower than matured at cleaving. But it still cleaves. This effect might be orders of mag more punishing when there's still nsp4 on one end and nsp6-11 on the other but still leads to a self-accelerating effect as every slowly-freed nsp5 increases the rate of freeing other nsp5s.
In this case it doesn't matter if each individual Pax bond is permanent (and so Pax'd nsp5s don't contribute to cleavage whether still poly'd or free), because non-Pax'd nsp5 still do (and so ramp up the cleavage rate as eventually freed). There's other more complex models where we could imagine Pax has easier access to freed nsp5 (essentially assumed in the first version of the OUTRT) and this would lead to a pause exactly at the "critical mass" of freed nsp5 (every freed nsp5 would cross the critical mass, only to be Pax'd and drive the math back down below the line) which would require constant replenishment of Pax. That's fun to imagine, but the new simplified model doesn't require anything beyond what Muramatsu, et al. demonstrated in the lab and so I'll go with that. And "does the RNA survive? let's assume at least in a small % of cells" is already in the OUTRT v1.
Right, the spike cleavage sites are all examples of repurposing human proteases, and this means employing the same motifs. Likewise, the signal peptide on the spike exploits human transmembrane signaling motifs. There's no reason for the virus to reinvent these wheels as long as local proteases are present in the right quantity or context. For S1/S2 cleavage, the viral "swarm" might even benefit from local exhaustion of proteases, so that some S1 is released into circulation and creates an inflammatory "diversion," but the rest stays in tact until released virions are next to another cell in another host. Or, it could be that late stage co-evolution favors dropping the cleavage site altogether, as all our endogenous coronaviruses and some influenza viruses seem to have done.
But for the nsps released from the polyprotein, there's only one relevant location of action (inside the currently-infected cell) and so exhaustion of endogenous proteases would seriously defang the virus. So it makes sense that nsp5 is worth packing along in the genome. But there's zero reason not to also use human C protease motif as a redundancy, in fact my theory assumes it is required in order to get work going. Wait let me just look it up finally. Anyway point being that when the Pfizer trial protocol calls nsp5 "essential," no, it's not, and that's why the viral rebound happens, per the OUTRT.
Well it looks like the stupid thing auto-cleaves - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7164132/pdf/FEBS-280-2002.pdf - That's astounding as far as the coronavirus biological talent show goes, but renders superfluous the endogenous protease piece of the OUTRT. It seems more likely that in some percentage of cells the RNA molecule keeps "meat-shielding" the cell with polyproteins until Pax it exhausted. It's essentially the same math but breaks my illustrations in Pt 1.
Brian, I greatly appreciate you digging so deeply into this topic.
Paxlovid was tested on on unvaccinated people only. Now they are giving it to vaxxed people, whose reaction to it, and timeline of development of immune response, is very possibly different from unvaxed.
As your graph shows, paxlovid failed even in some unvaxed people , but seems to be failing much more often (this is just a hunch) in real mostly vaccinated people.
This could be a story not just of Paxlovid failure. This could be a story illustrating vaccine failure of not allowing vaccinated people mount rapid (within 5 days) immune response.
It could also be a story of Omicron not clearing as quickly as Alpha/Delta.
And thanks for sharing it to twitter. Maybe someone could send it to all those confused experts in Boston, though that would probably prompt a B&R, haha.
I'm at the moment not sure about the "much more" part, since even 4-to-8 of 97 (as in the FDA fig 2) is enough to drive a huge, hyuuuge observed rebound "trend," as well as effectually cap the increase for the vaxxed at 10x. But as hinted in the self-quote of my reply to Bartram, an influence on innate immunity could definitely tilt the scales toward even more common rebound.
At this rate I wouldn't be surprised if they finally submit to COVID being seasonal, only to release annual vaccines and prophylactics. Both Molnupiravir and PAXLOVID were supposed to be prophylactic/early treatment so I won't be surprised if that's what comes next.
It still seems a bit incoherent to push a vax and a therapy. This is why I’m surprised that Paxlovid actually works on a molecular level, I assumed it was just more corporate welfare and never took a look at the trial until now. Meanwhile I assumed Molnupiravir was going to be blamed for cancers in the vaxxed that were going to happen anyway, but once again this isn’t coherent with the Paxlovid push. So for now things don’t look very grand-plan, Great Reset-y, but that’s always in flux.
There's an argument to be made that antigen tests may be faulty in the sense that they may be picking up on the nucleocapsid proteins from noninfectious viral particles. I've stated that antigen tests may be best during a symptomatic infection but nearly a week or two out it really doesn't make sense.
Based on my experience with the lateral flow antigen tests, I tested negative for months before and after my infection (thanks to mandatory testing at work due to my vaccine status), and tested positive for eight days during my COVID-19 infection. The most positive tests (as in, they were almost immediately positive, and the control line barely visible) were actually around day five after symptom onset, by which time my symptoms had mostly abated. I later learned that at least for Omicron, peak viral load is between days 3 and 6 after symptom onset (and the CDC claims we can exit quarantine at day five). It's a small sample size, but based on my experience, the at-home tests have been surprisingly accurate. A resurgence of positive tests 10+ days after symptom onset after repeated negative tests, along with a rebound of symptoms, indicates renewed replication IMO.
The issue with such a test is that it doesn't measure active infection, it measures if "stuff" is there. It's why during a symptomatic infection we can correlate the two. Someone has symptoms, there is a presence of something there as shown by PCR or antigen tests, and thus we will argue that the symptoms are caused by the virus. The issue is that these tests don't tell whether you are dealing with an active infection. You may well have cleared any active virus, but the virus' proteins and RNA may still be around. Swabbing for leftover material may turn out positive for said material but doesn't indicate the actual nature of the material and whether it is still infectious.
I actually believe Brian' assessment is correct or at least along the right lines, however I draw criticisms with how closely we can associate them to the tests. Ironically, positive tests, followed by PAXLOVID and negative tests eventually leading to positive tests would argue that the tests are actually not very sensitive.
I don't disagree with your general premise, but if someone tests negative for days, or even a week, and then begins testing positive again along with a recurrence of symptoms, I think there's little chance it's just picking up leftover viral debris (very different from PCR tests where a person can test positive for many weeks even after symptoms have resolved). That's not the typical course of events for a COVID-19 infection. The rapid antigen tests don't necessarily need to be terribly sensitive, they'll still give you a good idea of the course of the infection, as viral load naturally increases and then declines toward resolution. The averages I read for Omicron are peak viral loads at 3-6 days after the start of symptoms, and clearance around day 9. Paxlovid appears to be messing with the natural course of the infection in some people, causing viral load to drop and remain low for over a week, only to peak again at two weeks. That up and down and up again pattern is abnormal, it should be up and down and gone (assuming a healthy immune system and an infection allowed to run its course).
Right, antigen test sensitivity thresholds are a good correlate for symptomatic / presumed infectious levels of replication; the problem is more that they still have a high baseline false positivity rate. I was going to include the swirly graphic here and a discussion about same in Pt 2 but that didn't make it to the final cut https://chs.asu.edu/diagnostics-commons/blog/covid-19-test-accuracy-when-too-much-good-thing-bad
It's balance of sensitivity and accuracy, along with when the test is administered. PCR tests are good for very early stages of the infection, but when you give them to millions of people you are inherently going to get a decent number of false positives by virtue of all inherent false positivity rates of any test.
The two tests also don't relate well to an active infection. We know that there is stuff there, but is it infectious? We can't tell that because that's not what these tests are intended to do, which is why I don't find any worth in using them after someone's symptoms are gone.
I consider the reinfection studies that use PCR+ as a proxy for infection as essentially a post hoc validation that PCR+ is pretty selective. Especially https://doi.org/10.1093/cid/ciab234 where there wasn't a single re+ (after 90 days from first +) despite some presumed level of screening testing since it was a healthcare center. Obviously if you *only* test when the virus isn't spreading, you will have 100% false positives.
I was recently digging into the kids hepatitis/adenovirus lead and it turns out adenovirus PCR assays, which have taken over for diagnosis in the last decades, are well-validated (https://pubmed.ncbi.nlm.nih.gov/32349251/) even though you would expect that everyone has "a little" adenovirus action going on in the GI at all times given how many types there are.
Washington Post just published an article about the viral rebound effect with people getting sick from the same infection again (not a reinfection) - https://www.washingtonpost.com/health/2022/04/27/paxlovid-second-case-covid/
Thanks for the tip! A lot of it is the same level of evidence as the Boston Globe story, centered on Sax and Charness, as was a story in another outlet today (I already forgot which). But now the Charness case study appears to be online, time to take a look.
*edit: I spoke too soon. The Post story adds a Pfizer statement hinting at some unpublished evidence that the placebo group also had rebounds (I doubt this), several newly-fished anecdotes, a rehash of my analysis in Pt 2 that these rebounds exceed normal noise in testing, and a comment section presumably full of people blaming the repeal of the CDC mask mandate lol.
This is like being a fly on the wall in a place where the REAL science is taking place. I can't contribute a lot to the discussion but am very grateful to all of you.
This is a great series, thanks for putting it together.
Your explanation of part of the viral replication process was really helpful to me.
Does the polyprotein contain a bunch of pieces and tools that, once separated assemble into a virion?
It occurs to me that, in individuals vaccinated against spike only, flagging the cell as "sick" via MHC would probably need to find bits of spike to present. But, perhaps the spike proteins are still tied up inside the polyprotein when the protease inhibitors are active. Thus, nothing is gained from the Paxlovid "pause" because the infected cells aren't being flagged and destroyed.
This could be different in a natural infection, if the whole virus is taken apart and antibodies evolved for many different pieces. In computer terminology, you could "execute" the viral RNA in a sandbox to find out what proteins are created and prepare to recognize them; point being this produces a lot more ways to identify an infection, including precursors and byproducts (e.g. nsp5), rather than just relying on "finished" spike only.
Everything in the polyprotein “package” (it’s actually two packages but that’s not very important since nsp5 works on both) is non-structural proteins (nsps) that are tools to help the virus do its work in the cell. The rdrp and other RNA-replication/preparation associated nsps are essential, others are enhancing, in that they suppress the cell’s immune response or organize resources in the cell, essentially making a factory that churns out the “sub-genomic” proteins that don’t get read by the cell’s ribosomes until after some weird reassembling of the RNA molecule (which copy of it, at this point? I’m a bit unclear on this) leads to the rdrp kicking out discrete mRNA instructions for the structural proteins including the spike. So the spike is not tied up, it’s just that the whole viral RNA molecule doesn’t promote reading the non-Orf1 / polyprotein parts and those parts have to be processed by the nsp machinery before they are substantially expressed. So their products, essentially everything after Orf1 which starts with the spike, show up later. This might not mean the Paxlovid pause always happens before spike is being expressed, but probably at least sometimes, and those are the cells which will (in some other percentage) restart replication unless a non-spike-based (innate) immune response takes them out. This could include intracellular innate immunity. So it really shouldn’t be a 100% rebound rate, and doesn’t seem to have been in the trial which was all unvaxxed (“natural” infection, with the possible monoclonal antibody confounder); and if this turns out to be what’s happening in the vaxxed, I would definitely look at innate immune suppression.
*edit: I meant to include a link to the nsp portfolio here, note that their number scheme skips the normal nsp11 and so add +1 to their nsp11 and above to make it consistent with most publications https://zhanggroup.org/COVID-19/index.html#table1
Thanks again for the useful information!
I would imagine that if the viral load is substantial at the time Paxlovid is taken, once Paxlovid is absorbed, any cells subsequently infected by the virus floating around will get stuck in pause mode, so there would be a fairly large number of paused cells ready to unpause as the Paxlovid disappears over time. Because of the exponential decay of the Paxlovid, the resumption won't "turn on" all at once, but it certainly has a head start, unless the infected cells are destroyed in the interim (or the bits of virus decay or are destroyed).
Do you know whether antibodies are typically formed against NSPs in a natural infection? Because that would be a possible distinction between vax and unvax. Are NSPs sometimes displayed to flag an infection? It seems plausible that in many cases the polyprotein is partially unpacked but not completely.
Potential innate immune suppression is another possible reason for differential performance. I still don't understand what is known for sure about that question. The role of pseudouridine would explain some degree of innate suppression, although as I understood it, the suppression occurred intracellularly, suppressing TLRs within a cell that was transfected. But all the transfected cells should have been destroyed within a few weeks.
One thing that does seem to be a possible issue is that the stabilized mRNA has a lot more psuedouridine than is typically used naturally in mRNA. Is excess pseudouridine metabolized and new created when needed? Or does an excess float around for a while causing problems?
I guess it's not clear in the first place that there is a differential between vaxed and not - maybe it just leads to recurrence 1-2% of the time. I now have a vaxxed relative currently taking paxlovid (maybe 2 of them) so I guess I will get some first hand evidence. If one or both have recurrence then that's suggestive that the rate is higher than a few percent.
Right, the virus has a head start in that every instance of Paxlovid "working" is an instance of leaving a potential future restart from unopened polyproteins, and maybe the math on how much viable viral RNA is still around is pretty brutal but this wouldn't be much of a "problem" for the virus since even a 1/1000000 successful restart rate is akin to going a mere "6 turns back" in autosave.
I don't know if anyone's even looked at the antibody response for nsps. So as for whether it is occurring, I could only guess, and would guess that the nsps are not as immunogenic in terms of adaptive immunity and might be more TLR / natural IgM domain, "oh this looks like a bacterial or viral RNA unwinder or protein chopper of some sort." Maybe some conserved immune recognition with other coronaviruses.
Still, it's an interesting question whether anyone's even looked, maybe it's mentioned in some monoclonal antibody dev research. The preoccupation with spike and N protein seems to go back to SARS-1, and then N turned out to be pretty useful for tracking infection when SARS-2 came around.
As long as RNA is single strand I don't think there is a intracellular immune response issue but I'll double-check. Otherwise the pseudouridine resists normal RNA degradation, but that's more "optimization" than evasion. (*edit: ok, there's a bit of overlap into antiviral RNA breakdown evasion https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3241635/) For my part I try not to get too elaborate with the innate immune suppression question; it could just be a result of injury and immune exhaustion from the transfection, nothing fancy. Akin to how diabetes, which is inflammatory, also suppresses innate immunity just by promoting a lot of vascular and tissue damage all the time.
There's also still the potential Omicron (BA.1 / BA.2) interaction, with the latter sibling potentially explaining why these anecdotes didn't emerge earlier.
My conception of the impact of pseudouridine came from here: https://rwmalonemd.substack.com/p/when-is-mrna-not-really-mrna?s=r
My take-away was that there are several intra-cellular mechanisms that are on the lookout for specific mRNA sequences that commonly occur in antigens. However, these sequences occasionally are needed to be used in legit host mRNA.
In order to prevent these "false positives", there is a mechanism for substituting pseudouridine strategically around the sites where these false positives occur. The presence of pseudouridine apparently disables these checks and flag it as an exception to pass through. I really don't know much about this but it makes sense that if you are looking for "bad" sequences, they will inevitably show up in "good" mRNA.
By encoding the vaccine mRNA using pseudouridine it thereby evades these various protection mechanisms, but the question is whether it also has a more widespread or subsequent impact?
Right, that would be the part about TLR7 and TLR8, so there is an immune element but it's just intrinsic to ssRNA, not "viral" ssRNA, so it's a bit redundant to the problem of avoiding normal mRNA breakdown. It also might not apply in vivo since the LNPs don't release the RNA payload into endosomes. (Also, from a vax perspective you might want a little TLR activation, even if it shuts down translation / protein output more quickly, as it would give and adjuvant effect. And in Study No. R-20-0112 (https://www.tga.gov.au/sites/default/files/foi-2389-06.pdf) BioNTech only compared the pseudouridine (modRNA) RBD and full spike (the final version) vs a non-mod mRNA for the RBD, so who knows if a non-mod mRNA for the full spike would have performed just as well.) There are also some mRNA sequence considerations, strings of nucleotides that might lead to a diminishment of host expression, but again this is more like "optimization" than pure immune evasion, and is more about alternate codon spelling than nucleotide modification.
I have to feel that the long term consequences of your body making a toxic spike protein and then presenting alongside other proteins (natural) will lead to de-immunization because the body would eventually start viewing spike as "self". This is basically immune tolerance in a nutshell.
Its unclear to me whether this is an effect of Paxlovid - or the mRNA vaccines.
Right, that would make a lot of sense and is yet another reason constant boosters are crazy. For the moment my money remains on tolerance not happening based on published antibody kinetics after boosters, but there could be a situation where T Cell immunity is desensitized to the spike even though B Cells / antibodies keep ramping up, due to the former not having been as heavily primed to begin with, who knows. Also as far as Paxlovid my guess is that there isn’t a lot of spike around when the “needs more uncleaved Orf1” moment that defines the pause occurs, so what probably makes the difference in that assumption is innate immunity.
Pfizer data on Pax -- I wonder if they had a vaccine arm? Maybe they knew about this?
The trial document says both previously infected and vaxxed were excluded. Seems weird, maybe they thought there wouldn't be enough severe outcomes to show efficacy, the rationale might be somewhere in the paperwork
I'm guessing this is done as some position of sanctimony like "we are helping the unvaccinated, even if they won't help themselves!" That would at least explain the high cost as well- "The unvaxxed can receive treatment, but they will have to pay."
Wrapping up my read of the Protocol (https://www.nejm.org/doi/suppl/10.1056/NEJMoa2118542/suppl_file/nejmoa2118542_protocol.pdf) and it doesn't seem to provide any rationale. I can see why they might have assumed little real-world need for Pax+vaxx when starting out but it's surprising they didn't expand this later as they did to include mAb treatment.
I think that's because there really is no rationale. For Molnupiravir and their press briefing they just went "well, the unvaxxed are in bigger danger so let's just give it to them as top priority!" What I believe is happening is that these therapeutics are supposed to be considered less intrusive. Essentially just grab and go. At Biden's SOTU he mentioned a "test to treat" protocol in which someone gets a PCR test and has a prescription filled at the same time.
I guess all I can say is that I really don't know why they are doing what they are doing. I do hope you didn't read through all 318 pages of that protocol!
it would be interesting to see what happens to those Ab titers over time. But - I think its unknown how the body will respond to this synthetic spike and immune challenges in the future..
Well, just like you now I'm starting to consider how it may be pretty obvious. I stated that we may not have used protease inhibitors alone to treat viral infections. For HIV a cocktail usually entails a nucleoside analogue, integrase inhibitor, protease inhibitor, essentially it is a full cocktail of drugs which are intended to target many facets of the virus. It would make sense that the virus may have a few built-in workarounds to go against clogged channels.
But you've spent a lot of time digging deep and piecing things together and it really shows that you don't need to be some PhD as long as you are a critical thinker, and it clearly shows in this case Brian.
I guess we still need to wonder how long the immune dysfunction is, but I am unsure to what extent that is occurring. Ironically, I saw that The Naked Emperor's Newsletter posted an article in which they believe that long COVID may partially be blamed on reduce immune function. Considering that some of the symptoms of vaccine side effects are similar to that of Long COVID I'm actually curious if there are concerns over immune dysfunction. There's also consideration into whether the host cells can recognize uncleaved proteins and target them for removal, but I suppose that would require even further investigation.
Edit: this was the article that The Naked Emperor posted:
https://medicalxpress.com/news/2022-04-cases-covid-abnormally-suppressed-immune.html
The OUTRT assumes that host cell protein degradation essentially does the work of the nsp5, with endogenous cysteine proteases attacking the same cleavage motifs and freeing the nsps, just more slowly than nsp5 would do. So even if this is part ot protein breakdown it rescues the replication process before it can prevent the same. Since the virus doesn’t pack nsp5 into the virion there’s got to be some overlap, though I haven’t looked up the lit on endogenous cysteine protease motifs.
That's actually interesting. So that would mean that the recognized sequence of amino acids is conserved between SARS-COV2 and humans? I wondered if it may be akin to restriction enzymes which recognize specific nucleotide sequences and clip accordingly. Evolutionarily, that would provide some ability to say "clip our proteins here and it will work out for us" while simultaneously stating "clip here and you can kill off your enemy". Although I suppose our genomic makeup with exons and start/stop portions are important as well.
OK, the OUTRT totally survives the autocleavage problem. Uncleaved nsp5 turns out to behave the same way as my assumed endogenous proteases, i.e. slowly. Once they are cleaved, they work quickly, probably tons more compared to bound. So unless all nsp5 is pax’d, Paxlovid just makes it take longer until all nsp5 (Pax’d in a cell) is cleaved and all polyprotein in turn. Literally just delays the inevitable.
Just to clarify are you saying that the uncleaved protease is able to still function? That would mean that it would still have to fold into the proper conformation and that seems a bit difficult to imagine.
I guess the idea was that the covalent interaction would just hold out long enough that the virus just dies out? I suppose we would have to wonder about the activity of endogenous nucleases and whether they would get rid of the mRNA and RNA after a set period, but then we would have to factor in downregulation or suppression of cellular functions from some of the virus' proteins as well.
Extremely difficult to imagine, which is why I felt safe assuming an endogenous protease here, but if there is one, no one has looked for it as far as I can tell. The coronavirus nsp5 / 3CL-pro motif might in fact be a reinvented wheel.
What Muramatsu, et al. show (link in my other comment) is that it still auto-cleaves (and therefor cleaves according to the operative motif) with ~20 pro-form residues on either end, but at a slower rate. By varying the concentration of already matured or not nsp5 added to pro-form, they show that not-matured is much slower than matured at cleaving. But it still cleaves. This effect might be orders of mag more punishing when there's still nsp4 on one end and nsp6-11 on the other but still leads to a self-accelerating effect as every slowly-freed nsp5 increases the rate of freeing other nsp5s.
In this case it doesn't matter if each individual Pax bond is permanent (and so Pax'd nsp5s don't contribute to cleavage whether still poly'd or free), because non-Pax'd nsp5 still do (and so ramp up the cleavage rate as eventually freed). There's other more complex models where we could imagine Pax has easier access to freed nsp5 (essentially assumed in the first version of the OUTRT) and this would lead to a pause exactly at the "critical mass" of freed nsp5 (every freed nsp5 would cross the critical mass, only to be Pax'd and drive the math back down below the line) which would require constant replenishment of Pax. That's fun to imagine, but the new simplified model doesn't require anything beyond what Muramatsu, et al. demonstrated in the lab and so I'll go with that. And "does the RNA survive? let's assume at least in a small % of cells" is already in the OUTRT v1.
Right, the spike cleavage sites are all examples of repurposing human proteases, and this means employing the same motifs. Likewise, the signal peptide on the spike exploits human transmembrane signaling motifs. There's no reason for the virus to reinvent these wheels as long as local proteases are present in the right quantity or context. For S1/S2 cleavage, the viral "swarm" might even benefit from local exhaustion of proteases, so that some S1 is released into circulation and creates an inflammatory "diversion," but the rest stays in tact until released virions are next to another cell in another host. Or, it could be that late stage co-evolution favors dropping the cleavage site altogether, as all our endogenous coronaviruses and some influenza viruses seem to have done.
But for the nsps released from the polyprotein, there's only one relevant location of action (inside the currently-infected cell) and so exhaustion of endogenous proteases would seriously defang the virus. So it makes sense that nsp5 is worth packing along in the genome. But there's zero reason not to also use human C protease motif as a redundancy, in fact my theory assumes it is required in order to get work going. Wait let me just look it up finally. Anyway point being that when the Pfizer trial protocol calls nsp5 "essential," no, it's not, and that's why the viral rebound happens, per the OUTRT.
Well it looks like the stupid thing auto-cleaves - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7164132/pdf/FEBS-280-2002.pdf - That's astounding as far as the coronavirus biological talent show goes, but renders superfluous the endogenous protease piece of the OUTRT. It seems more likely that in some percentage of cells the RNA molecule keeps "meat-shielding" the cell with polyproteins until Pax it exhausted. It's essentially the same math but breaks my illustrations in Pt 1.
But Nature is supposed to find the way I assume it will!
Brian, I greatly appreciate you digging so deeply into this topic.
Paxlovid was tested on on unvaccinated people only. Now they are giving it to vaxxed people, whose reaction to it, and timeline of development of immune response, is very possibly different from unvaxed.
As your graph shows, paxlovid failed even in some unvaxed people , but seems to be failing much more often (this is just a hunch) in real mostly vaccinated people.
This could be a story not just of Paxlovid failure. This could be a story illustrating vaccine failure of not allowing vaccinated people mount rapid (within 5 days) immune response.
It could also be a story of Omicron not clearing as quickly as Alpha/Delta.
We should keep pushing it.
I tweeted your article from my backup account
https://twitter.com/TotallyCanc3l3d/status/1518046417461424129
And thanks for sharing it to twitter. Maybe someone could send it to all those confused experts in Boston, though that would probably prompt a B&R, haha.
You should follow up on a few of their tweets and let them know, if anything it puts them on notice.
there is a reason why most major breakthroughs in patient care happen outside of the major academic medical centers....
Great discussion!
I'm at the moment not sure about the "much more" part, since even 4-to-8 of 97 (as in the FDA fig 2) is enough to drive a huge, hyuuuge observed rebound "trend," as well as effectually cap the increase for the vaxxed at 10x. But as hinted in the self-quote of my reply to Bartram, an influence on innate immunity could definitely tilt the scales toward even more common rebound.
The sooner this superfluously novel $800 drug gets replaced by an even more superfluously novel $2000 drug, the healthier we all will be.
At this rate I wouldn't be surprised if they finally submit to COVID being seasonal, only to release annual vaccines and prophylactics. Both Molnupiravir and PAXLOVID were supposed to be prophylactic/early treatment so I won't be surprised if that's what comes next.
Tamiflu is interesting since it seems like it's something that is given too late in a disease progression anyways.
It still seems a bit incoherent to push a vax and a therapy. This is why I’m surprised that Paxlovid actually works on a molecular level, I assumed it was just more corporate welfare and never took a look at the trial until now. Meanwhile I assumed Molnupiravir was going to be blamed for cancers in the vaxxed that were going to happen anyway, but once again this isn’t coherent with the Paxlovid push. So for now things don’t look very grand-plan, Great Reset-y, but that’s always in flux.
There's an argument to be made that antigen tests may be faulty in the sense that they may be picking up on the nucleocapsid proteins from noninfectious viral particles. I've stated that antigen tests may be best during a symptomatic infection but nearly a week or two out it really doesn't make sense.
Based on my experience with the lateral flow antigen tests, I tested negative for months before and after my infection (thanks to mandatory testing at work due to my vaccine status), and tested positive for eight days during my COVID-19 infection. The most positive tests (as in, they were almost immediately positive, and the control line barely visible) were actually around day five after symptom onset, by which time my symptoms had mostly abated. I later learned that at least for Omicron, peak viral load is between days 3 and 6 after symptom onset (and the CDC claims we can exit quarantine at day five). It's a small sample size, but based on my experience, the at-home tests have been surprisingly accurate. A resurgence of positive tests 10+ days after symptom onset after repeated negative tests, along with a rebound of symptoms, indicates renewed replication IMO.
The issue with such a test is that it doesn't measure active infection, it measures if "stuff" is there. It's why during a symptomatic infection we can correlate the two. Someone has symptoms, there is a presence of something there as shown by PCR or antigen tests, and thus we will argue that the symptoms are caused by the virus. The issue is that these tests don't tell whether you are dealing with an active infection. You may well have cleared any active virus, but the virus' proteins and RNA may still be around. Swabbing for leftover material may turn out positive for said material but doesn't indicate the actual nature of the material and whether it is still infectious.
I actually believe Brian' assessment is correct or at least along the right lines, however I draw criticisms with how closely we can associate them to the tests. Ironically, positive tests, followed by PAXLOVID and negative tests eventually leading to positive tests would argue that the tests are actually not very sensitive.
I don't disagree with your general premise, but if someone tests negative for days, or even a week, and then begins testing positive again along with a recurrence of symptoms, I think there's little chance it's just picking up leftover viral debris (very different from PCR tests where a person can test positive for many weeks even after symptoms have resolved). That's not the typical course of events for a COVID-19 infection. The rapid antigen tests don't necessarily need to be terribly sensitive, they'll still give you a good idea of the course of the infection, as viral load naturally increases and then declines toward resolution. The averages I read for Omicron are peak viral loads at 3-6 days after the start of symptoms, and clearance around day 9. Paxlovid appears to be messing with the natural course of the infection in some people, causing viral load to drop and remain low for over a week, only to peak again at two weeks. That up and down and up again pattern is abnormal, it should be up and down and gone (assuming a healthy immune system and an infection allowed to run its course).
Right, antigen test sensitivity thresholds are a good correlate for symptomatic / presumed infectious levels of replication; the problem is more that they still have a high baseline false positivity rate. I was going to include the swirly graphic here and a discussion about same in Pt 2 but that didn't make it to the final cut https://chs.asu.edu/diagnostics-commons/blog/covid-19-test-accuracy-when-too-much-good-thing-bad
But these people ARE symptomatic
It's balance of sensitivity and accuracy, along with when the test is administered. PCR tests are good for very early stages of the infection, but when you give them to millions of people you are inherently going to get a decent number of false positives by virtue of all inherent false positivity rates of any test.
The two tests also don't relate well to an active infection. We know that there is stuff there, but is it infectious? We can't tell that because that's not what these tests are intended to do, which is why I don't find any worth in using them after someone's symptoms are gone.
I consider the reinfection studies that use PCR+ as a proxy for infection as essentially a post hoc validation that PCR+ is pretty selective. Especially https://doi.org/10.1093/cid/ciab234 where there wasn't a single re+ (after 90 days from first +) despite some presumed level of screening testing since it was a healthcare center. Obviously if you *only* test when the virus isn't spreading, you will have 100% false positives.
I was recently digging into the kids hepatitis/adenovirus lead and it turns out adenovirus PCR assays, which have taken over for diagnosis in the last decades, are well-validated (https://pubmed.ncbi.nlm.nih.gov/32349251/) even though you would expect that everyone has "a little" adenovirus action going on in the GI at all times given how many types there are.