Thread by @michaelzlin

The full DEFUSE proposal on gain-of-function experiments on bat coronaviruses is available, and I'd say it's quite shocking. It does not lay out a plan to create SARSCoV2, but does propose to identify and culture natural sarbecoviruses with the ability to infect human cells. A 🧵

There are threads interpreting the DEFUSE proposal as intending on making SARSCoV2. A careful read shows that is not the case. However the intention was to identify natural viruses with features that would help them infect cells. So it may be not much of a difference functionally

The draft proposal and related documents can be downloaded at the link below. We know DARPA rejected the final submitted proposal for being too risky. This draft proposal certainly matches that description. Were the experiments attempted? We don't know.

archive.org/details/2021-0…

I'll jump right to it. Page 518 in the downloaded PDF proposes to sequence wild sarbecovirus "strains" for spike (S) genes, then insert them into the genomes of WIV16 and SCH014, two sarbecoviruses deemed less pathogenic than SARSCoV1 (by Shi Zhengli, who had isolated them).

The subset of "chimeric viruses that encode novel S genes with prepandemic potential" (WIV16 or SCH014 backbone, new S) will be identified based on better replication in human cells or in mice expressing human ACE2. Recall the S genes came from natural strains that were sequenced

The natural strains that provided S genes conferring better replication in human cells will then be rescued. They will be fully sequenced, the sequences grouped into 95%-similar subspecies, and a consensus sequence identified (unclear if an average or "representative" sequence)

The consensus genome for each interesting variant will be synthesized in 6 pieces then stitched together using restriction sites that do not disturb the coding sequence (presumably they mean can disturb the RNA but not the peptide sequence), and real live viruses recovered.

Finally they will see how bad these isolates of wild sarbecovirus variants with spike proteins that aid infection of human cells really are, in human cells and hACE2 mice again.

That does sound like a recipe for potential disaster in any place other than a BSL4 facility.

The proposal explains how you can isolate a virus with an unusual spike feature (the FCS) without deliberately engineering it in. Essentially it proposes to set up a genetic screen to find those rare performance-enhancing spike proteins, then grow up that unusual virus.

This proposal also demonstrates by counterexample why engineering a FCS into a viral backbone without published characteristics never made sense. It only makes sense to engineer changes into a well studied less pathogenic virus, such as WIV16 or SHC014, exactly as proposed here.

That's why I leaned against SARSCoV2 being engineered (as I told @zeynep below). But the DEFUSE proposal now raises another way to obtain the unusual features of SARSCoV2: not by direct engineering, but by screening and amplification of natural strains.

nytimes.com/2021/06/25/opi…

It's all very strange though. The behavior of Shi's team in the early days (taking photos in a restaurant after working hard in the lab, musing with friends about where it came from) isn't what you'd expect from someone with consciousness of guilt. Yet the circumstances align.

And recall that the US IC believed that the Chinese didn't know and don't want to know. I still think it's possible that an unknown SARSCoV2 jumped out of a bag of bat feces (still a research incident). Yet DEFUSE does seem eerily predictive. Mysteries...

https://twitter.com/michaelzlin/status/1383642186273607684

BTW some believe a subsequent part of the proposal to fix broken protease sites is damning by explaining how a furin cleavage site appears in SARSCoV2 and not other sarbecoviruses. However I am not sure that SARSCoV2 would have come from this work if that fix were necessary

The sentence in question is "where clear mismatches occur in these S2 proteolytic cleavage sites of SARSr-CoV, we will introduce the appropriate human-specific cleavage sites". This sentence is lousily written; should have been "where latent cleavage sites may exist, we will..."

and the word "these S2 sites" should be better defined. AFAICT "these S2 sites" most likely refers to potentially latent sites for all proteases previously mentioned as processing S2, trypsin, cathepsin, TMPRSS, or furin.

So indeed a FCS could have been artificially introduced via this process, but then, if that's what happened, recall it means the researchers would have needed to have identified the SARSCoV2 precursor without a FCS as having a more effective spike protein in the previous step

So for the FCS to have been introduced into a newly cloned strain, a SARSCoV2 precursor would have needed to have passed the selection for better spikes in order to have been a newly cloned strain to begin with. I wonder if this is true?

Maybe someone more familiar with SARS2 spike can tell us if SARS2 spike minus the FCS is already capable of mediating improved human cell entry compared to WIV16 or SHC014 spike

Could someone have cloned some new sarbecoviruses without going through the first selection and created cleavage sites on their S proteins without first (1) publishing the baseline natural characteristics of those viruses, or (2) testing the effects in chimeric WIV16 or SHC014?

It's possible, but that seems like a lot to do without publishing or publicizing in the interim. So possible but also possibly not. Again the proposal is eerily predictive but not all the human behavioral factors fit. One must be careful of selective posthoc mental fitting.

Nevertheless the DEFUSE proposal proves that everyone involved in coronavirus research knows that a lab leak is a plausible hypothesis. We already knew that from logical priors and from Zhengli Shi's responses to SciAm and Science.

What DEFUSE additionally proves is that some of those most vehemently denying the possibility of SARS-CoV-2 originating from a lab leak are among the most aware of its plausibility. Ironically they are in the US, where public opinion has to be shaped and can't simply be dictated.

For chimerogenesis and viral genome assembly, UNC (Baric) is listed as primary responsibility, but Shi's lab at WIV had performed those techniques before, so had the ability to do the work using other funds if they wanted to.

Ebright correcting my mistake

https://twitter.com/R_H_Ebright/status/1748134073283838015

Anyway the general point is that someone proposing to find or make viruses that fit the SARS-CoV-2 description doesn't prove that they did it. But it does soundly refute denials, by the proposers and those blindly echoing them, that it was a ridiculous or impossible idea.

Similar take by Alina Chan

https://twitter.com/Ayjchan/status/1748386375458607108