Ryan Hisner Profile picture
Oct 22, 2025 12 tweets 5 min read Read on X
I beg to differ! If it is not a sequencing mistake—and it looks clean—one of these BA.3.2 has something completely novel in SARS-CoV-2 evolution: an FCS-adjacent deletion!

One of the two QT repeats appears to have been deleted. I've never seen anything like this before. Image
Work by @TheMenacheryLab looked at a similar, more extensive, deletion. They deleted both QT repeats plus the next AA (∆QTQTN). In Vero cells (monkey kidney cells), it produced extra-large plaques & outcompeted WT virus—similar to furin cleavage site (FCS)-deletion mutants. 2/12 Image
But in human lung cancer (Calu3) cells, the ∆QTQTN-mutant replication was dramatically reduced (2.5 orders of magnitude), and in infected hamsters disease was much milder. 3/12 Image
Oddly, despite reduced illness and demonstrably reduced cell entry in vitro, viral replication was apparently increased in hamsters in the nose and throat (as measured by viral RNA titers), while lung replication was unchanged.

Hard to make sense of. 4/12 Image
What's certain is that the ∆QTQTN deletion reduces (furin) S1/S2 cleavage. The smaller ∆QT in this BA.3.2 (assuming it is real, as it appears to be), likely has a similar effect.

Would love to hear @StuartTurville's take on this.

5/12 Image
Importantly, there's never been a variant with a deletion like this. Though there were a few early reports of FCS-adjacent deletions, I'm skeptical they were real. All were from the same lab, led by a fraudulent scientist (Didier Raoult) who's since had 48 papers retracted. 6/12 Image
Also, while the FCS lab experiments are immensely valuable, they took place on an ancestral spike background. BA.3.2 has (relative to WT) N679R & P681H. P681H has been shown many times over to increase spike cleavage, and N679R almost certainly also has that effect. 7/12 Image
Furthermore, BA.3.2 has huge deletions in the spike NTD (S:1-306). The resulting shorter NTD loops (also in SARS-1) have been shown by @EnyaQing to dramatically increase spike cleavage and infectivity (and instability). 8/12
Old 🧵 on this topic:
Regions around the FCS have been completely reconfigured in BA.3.2. On top of Omicron's H655Y (which increases time spent in intermediate fusion configurations), it has K529N, E554D, A575S, E583D, H625R, N641K, V642G, and E654K upstream & A688D + S704L downstream of the FCS. 9/12 Image
And further upstream, in S2, BA.3.2 has two extraordinarily rare & meaningful mutations—S:K795T & the 2-nuc S:A852K.

So the effect of this new FCS-adjacent deletion on infectivity & tropism may be completely different than on a WT spike background. 10/12
Finally, this deletion, which likely decreases S1/S2 cleavage, may be part of a larger trend in SARS-CoV-2 evolution. Recently there seems to be real selection pressure for mutations that temper S1/S2 cleavage. It's not clear why (incr stability?) 11/12
In any case, while the precise consequences of all this aren't known, one thing is clear: BA.3.2 is very different from past variants, none of which could tolerate a deletion like this. What it means going forward is anyone's guess. 12/12

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Ryan Hisner

Ryan Hisner Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @LongDesertTrain

Jul 7
Quick 🧵on previous variants that resemble the recent Ontario SARS-2 saltation variant that's now been detected in 4 sewersheds in NYC and one in Utah.

The practical impact of these variants ranged from ~zero to modest to major. 1/10

Background 🧵👇
• BA.2.87.1 - Appeared suddenly Sept 2023, spread across South Africa, & was detected in WW in Thailand. Then basically disappeared after Dec 2023 (apart from 1 seq collected in May 2024 from Eswatini). Emerged just when JN.1 was sweeping—bad timing. 2/
• BA.2.83 - One of several BA.2 saltation variants that emerged in late 2022. First appeared in India, then in Denmark, Canada, and Australia. Could not compete with BA.2.75 and BA.5 though and disappeared without making any real impact. 3/
Read 10 tweets
Jul 4
A mysterious, unusually diverse saltation lineage descended from KP.3.1.1 (last seen early 2025) has circulated at a low level for ~8 months in Ontario (see 🧵on next post).

Seemed a local anomaly, but @solidevidence just spied a weird spike in NYC WW—& it's the same one!
1/14
Here's a post I made a couple weeks ago when another Ontario sequence from this lineage showed up.

2/14
The remarkable diversity of this lineage raises questions about its origin and the conditions in which it spread and evolved.

Notably, the version detected in NYC wastewater is the same as the two most recent sequences from Ontario (long, bottom branch w/2 seqs below).
3/14 Image
Read 15 tweets
Jun 19
Wildly diverse branch of MC.10.1.7 (KP.3.1.1 descendant, died out early 2025) in Ontario.

It's apparently been circulating at a low level for the past year. ≥6 different people are represented here.

I think this must involve one or more chronic infections + transmission. 1/7 Image
• Busy 630 loop: ∆621-622 + NVFQ->IFPMAE at S:641-644.

• 2-nuc muts: W452K + D839K.

• Ultra-rare S2 insertion: ins791P

M:H155N, ORF1a:R3164H, & ORF3a:255-256 deletions cluster in chronics, often w/S:D936H/Y or ORF1a:L3201P. Intriguingly, D936Y was in the last seq.
2/7 Image
More info on this unusual branch can be found on the Github page for undesignated lineages, which Fede Gueli tirelessly and selflessly keeps up to date.

3/7github.com/sars-cov-2-var…
Read 8 tweets
Jun 10
Another fantastic preprint on BA.3.2's propensity for children, this time from @yunlong_cao & co.
They not only confirm the findings of David Ho's lab (that kids have ~0 antibody response to BA.3.2) but dig into the details of exactly why kids are so vulnerable to BA.3.2.

1/4
Two big wins for vaccination & mRNA vaccines:

1. Kids vaccinated before being infected have robust antibodies against BA.3.2

2. Unvaxed adults much more vulnerable to BA.3.2, esp. compared to mRNA-vaxed adults.

Read @yunlong_cao's 🧵 & very readable paper for details. 2/4 Image
There's still one major paradox here I can't wrap my head around: countries with the highest vaccination rates & the lowest proportion of children appear—very low sequencing makes hard conclusions difficult—to have the highest proportion of BA.3.2. 3/4
Read 6 tweets
Jun 4
New data from David Ho's lab showing that while adults & kids have ~equal antibody responses to XFG & NB.1.8.1, children have essentially no neutralizing antibodies to BA.3.2.

This seems to largely solve the BA.3.2 + kids mystery. 1/14 Image
If you've missed the story about how BA.3.2 (a novel, divergent saltation variant) is hugely overrepresented in sequences from children, this was my original (very quick) analysis, which subsequent data extended & confirmed. 2/
More details from this preprint. 50 is the limit of detection (i.e. zero). Nearly all kids under 7 had no detectable nAbs to BA.3.2, despite robust nAb titers against NB.1.8.1 & XFG.

Notably, most kids also had zero nAbs to ancestral D614G or XBB.1.5. 3/
biorxiv.org/content/10.648…Image
Read 14 tweets
Mar 26
So it's clear that BA.3.2 preferentially infects children, something we have never seen before in a SARS-CoV-2 variant.

Why?

The question's baffled me, but after a suggestion from Darren Martin, I think I have an explanation that makes sense.
1/16
I've tried to make sense of BA.3.2's penchant for kids by considering its unique spike: more compact, more closed, & more antibody-evasive than any other variant.

But I think another feature of BA.3.2 is responsible: its wholesale deletion of ORF7a, ORF7b, & ORF8 (∆ORF78).
2/
∆ORF78 is rare but not unheard of; it was in several late XBB variants (GW.5.1.1, FW.1.1, GE.1.2, etc) & a few branches of other variants. I've long thought these late XBB had an advantage in some population subsector, but I didn't suspect kids.
3/
Read 18 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(