Thread by @wanderer_jasnah

okay, here we go. so as i've said numerous times, DEFUSE could not have given us SARSCoV2 b/c:
1) there is no evidence that WIV had a viable progenitor sequence (something very clearly stated in the ODNI report)
2) it was not funded, i.e. the grant proposal went nowhere

3) it clearly proposed doing FCS insertions into Spike on PSEUDOTYPED LENTIs, i.e. PSEUDOVIRUS

but here's the thing. even if we assume that WIV had a progenitor seq, & taking the most mendacious misinterpretation of DEFUSE possible, it *still* couldn't've resulted in SARSCoV2.

first off, let's just assume that the reason they're doing this on this one particular viral sequence instead of, say, a SARS Classic or WIV1 backbone is b/c in every other case either:

- the virus had too short of an S1/S2 loop. it's not just the cleavage sequence that matters, but how accessible P4-P1 are to the active site of furin, which is narrow & deep.
indeed, deletions of the QTQTN loop extension attenuate cleavage: pnas.org/doi/10.1073/pn…

& adding a minimal FCS in a different sarbecovirus isolate results in no cleavage:

https://twitter.com/PeacockFlu/status/1674083909997404161

- adding in a usual, canonical FCS, b/c this is what every researcher would do, resulted in destabilization of Spike to the point where egressed virions lacked S1: journals.asm.org/doi/10.1128/mb…

btw, we'd expect these results to be published, b/c they'd be notable, interesting results. this'd be the starting point, nobody would start w/random virus isolates w/o first trying the ones they're familiar w/first.

but okay, after all this let's say WIV found a viral isolate that had a Spike w/all the correct starting properties. the first attempt for an FCS anyone'd attempt would be a canonical sequence like RRSRR, not PRRAR.

furthermore, such a canonical sequence in D614 Spike would *NOT* result in a more fit virus: -- so now you're left w/optimizing the FCS by pessimizing cleavage.eprints.gla.ac.uk/274617/1/27461…

in 2019's level of understanding of CoV fusion proteins, this would have been a nonsensical thought. but let's say that, okay, it's not known what the result would be, but how about just trying a bunch of cleavage seqs until we get improved cell entry?

first we need to switch away from VeroE6+TMPRSS2s to, at minimum, a cell line where the endosomal entry pathway isn't available, or else this'll be a complete waste of time. at minimum, something like Calu-3s.

& now you're doing a selection screen on something where fitness is *incredibly finicky*. there are also other constraints that need to be considered for any of this to work, but those are still unpublished results so i won't get into that here.

but ultimately...

...if WIV had a SC2 progenitor sequence & like, idk, a TARDIS, they would indeed eventually be able to engineer a virus where preprocessing resulted in improved aerosol transmission in a Syrian hamster model.

along the way they would have, *by necessity* discovered/elucidated enough new discoveries about CoV fusion proteins that they'd be able to singlehandedly fill up an entire NIDO's worth of talks, at NIDO's high talk quality, *singlehandedly*. i am not exaggerating.

but the notion that they took a random virus, tried a day's worth of experiments & got SC2 on their first few tries is just *ABSURD*. if mendacious-DEFUSE were somehow implemented, the 1st few *YEARS'* of results out of it'd be a lot of headscratching & negative results.

anyone who thinks otherwise quite plainly has never even set foot in a wetlab.

to elaborate a bit on exactly how conditional Spike preprocessing actually improving fitness is, well, if an FCS results in an easy fitness increase that's not heavily conditional, we'd see insertions all the time.

& if it was so aggressively selected against in nature under any circumstances, it would result in a fitness *DECREASE* in the lab too, over WT virus.

as it turns out, both of those are sort of true: deletions of the FCS, the QTQTN loop extension, or both, happen reasonably often: -- particularly in a D614 Spike.journals.asm.org/doi/10.1128/jv…

& we see acquisition of preprocessing repeastedly in merbecoviruses, for instance: sciencedirect.com/science/articl…

but we can actually go for a far sillier example. in what is my fav. slide from NIDO2023, from @whittakerIDlab's FIPV talk, we have isolates of FECV from various organs from a cat (which was persistently infected w/FECV but did not develop FIPV) sequenced across S1/S2.

@whittakerIDlab note that the bulk contain an S1/S2 FCS, but isolates from heart ventricle abrogate the cleavage site by mutating away P4, & isolates from nasal biopsy did the same at P1. conversely, isolates from conjuctiva acquired a cleavage loop extension that makes furin cleavage easier.

@whittakerIDlab let me repeat that: in the viral quasispecies derived from one starting virus, from a persistent infxn from *THE SAME CAT*, we have FCoV with FCS, FCoV w/o FCS & FCoV with FCS & a loop extension.

@whittakerIDlab we are talking about a feature where, in order for it to not get readily lost, & for it to actually improve fitness, a whole bunch of different properties need to all be satisfied at once. properties which *WE DID NOT UNDERSTAND* in 2019.

@whittakerIDlab ultimately, all of this is irrelevant, as it concerns a version of DEFUSE that only exists in Joseph Murphy & DREGSTIC's heads.

but even when steelmanned to the point where it's made of SAE 440C Martensitic, the argument *STILL* falls apart completely if you squint at it a bit.

ultimately, what drives me up the wall from folks like Asher, Esvelt & the BSNow clown collective is... okay, yes, it's fairly easy for me to take a random β-CoV reverse genetics system & insert some subsequence into it.

but that's not what we care about. what we care about is inserting some *new feature or phenotype*.

& sure, perhaps in the land of, say, computer viruses, these two are roughly equivalent. but that's not how biology works. time for a bit of a story...

in the early 1990s, some scientists got the idea for a first test of transgene insertion into petunias of a copy of the already-present chalcone synthase, responsible for synthesizing flower pigments, in the goal of making more strongly-coloured petunias.

this was as straightforward & likely to work as it gets, after all, it's just a copy of the already-present gene, so you know the protein'll fold correctly & has all cofactors already present, you use the same promoter that's already there for the native gene, etc.

so they inserted their copy of chalcone synthase & got... plants w/white flowers. sequencing showed the transgene was inserted successfully, yet instead of more strongly coloured flowers, it seemed that there was now *NO* chalcone synthase activity at all!

sure enough, it seemed that while the inserted gene was present & transcribed to mRNA, not only did it not result in any produced protein but now the native gene *ALSO* didn't produce any protein!

anyway, a lot of headscratching & a ton of investigation later, the researchers discovered RNAi: ncbi.nlm.nih.gov/pmc/articles/P…

by sheer coincidence, the transgene was inserted in the opposite direction to various promoters that, when those promoters became active, would produce antisense transcripts that silenced the desired transgene *AND* the native gene.

this is somewhat of an extreme example, but it's very illustrative: in synthetic biology, the devil is in the details. by which i mean each minor detail is yet another devil that will *COMPLETELY* fuck your experiment up if you don't account for it.

@peroxycarbonate ultimately, the mendacious-DEFUSE scenario depends on this being easy enough it could've happened on an early attempt & w/o any of the procedures being extensively documented. *this* is impossible.

& so the notion that we are one accidental GoFRoC-gone-wrong away from supermeaslesebola is just patently laughable.

that's simply not the reality of experimental biology.

@andrewtanyongyi @whittakerIDlab keep in mind that even for D614G, which is recurrent & *VERY* rapidly acquired, you still needed at least a few transmission events. think the earliest G614 isolates were late Jan. -- & that was for something that very clearly was unconditionally more fit in humans.

@emmecola granted, it's a bit weird since they also talk about activation via exogeneous trypsin, but you do that to force fusion by cleaving at S2', independent of preprocessing or even necessarily receptor binding.

@emmecola as for 2), i genuinely have no idea, since i have no clue what could possibly be "low risk" when we're discussing experiments done on lineage B. again, someone would really need to ask Ralph here.

@emmecola thing is, any interpretation i can think of for 2) entails inserting cleavage sequences obtained from sampling into *known* backbones, be they pseudovirus or SARS Classic.

@emmecola which makes sense, if you haven't characterized the backbone it makes no sense scientifically to be modifying it to see what happens.

@emmecola also that's the only interpretation of "high-abundance parental strain" that makes sense -- a random isolate you just pulled out from a field sample wouldn't qualify, the virus from the same lineage that you understand well & have a reverse genetics system at the ready for would.