Thread by @wanderer_jasnah

the paper in question, by Tom Gallagher, is finally out: doi.org/10.1128/mBio.0…

i've read a lot of #SARSCoV2 papers this year. i've even read a lot of *good* #SARSCoV2 papers this year. this is hands-down the best one i've read this year (yes even w/4.5 months left this year).

https://twitter.com/wanderer_jasnah/status/1417805476625108994

it was the subject of his @NIDO20201 talk, which due to technical difficulties had to be uploaded later, & so i never did a livetweet. so having read the paper extensively, i'll do a summary of it & some implications.

first, a bit of background: viral fusion proteins exhibit various metastable points along a tradeoff between fusion protein stability & fusogenicity.

a highly fusogenic protein will be excellent at cell entry but be highly unstable in even slightly harsh conditions.

this, btw, is why the S1/S2 FCS is very commonly lost in cell-culture adapted strains of β-CoVs that harbour such a site, it's a known common thing for #SARSCoV2 but also occurs w/HCoV-OC43 (strain "ATCC") & MHV, among others.

in a cell culture, harsh conditions tend to result in viral particles having the Spike transition to a postfusion state much more easily, often before it can infect another cell. the result is highly fusogenic strains often grow much *worse* & are a pain to get to form plaques.

conversely, the opposite seems to be selected when you have high background transmission & sufficiently many non-immune hosts: even if viral particles can be inactivated quickly, hyper-fusogenic ones will more rapidly enter cells & become infectious more easily.

B.1.617.2/Delta is a perfect example of this: P681R results in extremely high cleavage at S1/S2, often before cell exit in trans-Golgi even, resulting in Spikes that form large syncytia & enter cells easily.

but... one thing that i've wondered for awhile is why was it specifically the .2 variant that became so highly dominant when all of B.1.617 contained P681R?

looking at mutations revealed T478K & a bunch of NTD mutations in loop 2 that remodelled the NTD supersite.

i'm skeptical the first has much of an effect. the latter, though... my biggest mistake when it comes to mutations on Spike has been underestimating the effects of the NTD. and this paper illustrates that beautifully...

it looks at specifically three loops in the NTD that are part of the antigenic supersite. an interesting result is that they are *highly* divergent between #SARSCoV2 & SARS Classic, w/the latter having much shorter ones (loop 1 is only 5 residues long there!)

so, what happens when we swap them between the two viruses? how does SARS-1 w/the SARS-2 NTD loops function & how does SARS-2 w/the SARS-1 NTD loops function? in addition, we know that D614G is a highly stabilizing mutation, so let's look at D614G w/SARS-1 NTD loops too.

here's where it rapidly gets fascinating: there wasn't much of a difference for SARS-1. but for SARS-2 there was a dramatic destabilization w/the SARS-1 NTD loops, to the point where there was a *near-complete absence of S1 on VLPs* & thus unviable virus.

HOWEVER! addition of D614G stabilized the resulting Spike sufficiently that D614G SARS-2 w/SARS-1 NTD loops remained viable & was *better at cell entry* than WT D614G SARS-2.

what was now looked at was *how* the NTD loops so dramatically affected stability of S1/S2 given their distance from most other domains. it turns out:
- they affect RBD positioning into "up" states more likely to bind hACE2.

- furthermore, they affect cleavage at the activating S2' site by TMPRSS2 that results in the transition to a postfusion state. this was most dramatic for SARS-2 w/SARS-1 loops, which was ~100x more sensitive to proteolytic activation there than any other of the VLPs tested.

basically, it turns out that by affecting both RBD positioning & prefusion trimer stability, the NTD hypervariable loops are yet another source of variation for stability vs. fusogenicity of Spike.

now, for a couple interesting conclusions: first off, this paper in my mind solidly demonstates that neither NTD mutations nor FCS improvements were likely viable for #SARSCoV2 *before D614G became dominant*; they'd just be too destabilizing otherwise.

interestingly, though, #SARSCoV2 Spike was still relatively stable compared to at least SARS Classic & most likely most other sarbecoviruses w/similar NTDs; the hyperdivergence there IMHO is quite possibly why an FCS there, once added via template switching, *was not deletrious*.

thirdly, B.1.617.2/Delta has a very dramatic remodelling of loop 2 of the NTD by deletions at 156, 157 & a R158G substitution. recent results by @veeslerlab shows a pretty spectacular refolding of the Delta NTD:

https://twitter.com/veeslerlab/status/1425599671867641861

coincidentally, it starts to resemble the SARS Classic one quite a bit more. imho, this very likely contributes to the enhanced fusogenicity on top of P681R & is quite likely how B.1.617.2/Delta has such a dramatic increase in both infectivity & replication kinetics.

lastly... i'm actually now of the opinion that there *is* no global optimum for Spike b/c the metastable dynamics involved mean that different conditions result in different optimality curves.

for instance, B.1.617.2 is a highly fusogenic 3-up Spike. it's likely very close to optimal in the situation where there's still large pockets of naive hosts b/c it's *extremely* contagious. will it remain optimal once immunity reaches >95% or will hiding RBDs be favoured then?

(i suspect so, by analogy to endemic HCoVs, but i'm not very confident in guaranteeing that, let alone making any claims on what that might do to transmissibility beyond likely weakening it a bit).

lastly, imho this is a fascinating example of how varying fitness mechanisms interact. the hypervariable loops are located in the NTD antigenic supersite & the remodelling takes out binding by a large amount of NTD-targeting nAbs.

however, what likely resulted in this becoming dominant was the consequence that this remodelling improved Spike fusogenicity.

in other words, it evolved in-host due to antigenic pressure, but became dominant in-population due to transmission effects.

(imho, a lot of people overestimate antigenic pressure as a source of viral evolution while underestimating... most other sources).

anyhow, that's all. i cannot understate how much this paper is worth a read, seriously, go check it out!