Thread by @wanderer_jasnah

b/c i apparently seem to like doing this a little too much, i'm going to be livetweeting this whole summit below!

looks to be an absolutely *phenomenal* set of talks.

https://twitter.com/lajollaic/status/1445793952301400065

oh gods this is happening at the same time as #VRBPAC? lol oops. (not really of any interest to me but i imagine at least a few people would want to see both).

okay, here we go! opening comments from co-chairs Hilde Cheroutre & Li-Fan Lu right now.

first talk will be the opening keynote by @VirusesImmunity, talking about Immune Responses to #SARSCoV2!

huge variation in #COVID19 disease outcomes: everything from asymptomatic to fatal in the acute case, & everything from acute but self-limiting disease to long-haulers.

main outcome determinants are age & sex but nevertheless heavy variation also *within* those determinants.

talk will be divided into 4 parts:
- how does timing of nAbs affect #COVID19 disease outcomes?
- how do various mutations in variants impact vaccine-induced immunity?
- what are the adaptive immune determinants of protection to #SARSCoV2?
- possible mechanisms of Long #COVID19.

first section: nAb timing (the paper for this is: nature.com/articles/s4159…).

higher anti-S & anti-RBD antibodies are produced in hospitalized pts than mild disease pts -> severity-dependent increase in Ab response (replicated widely at this point).

nevertheless, viral load was maintained at *HIGHER LEVELS* in deceased pts!!!

this holds even at the peak of the Ab response when looking at deceased vs. hospitalized discharged pts.

so what gives? well, taking a closer look shows people w/lethal #COVID19 had ***DELAYED*** Ab response.

splitting pts into early & late neutralizers by development of nAbs before or after day 14 from symptom onset. late neutralizers had higher viral load & mortality likelyhood compared to early neutralizers.

no obvious correlation w/age, sex or BMI.

next topic: variants & vaccines!

mRNA vaccines elicit robust Ab responses to Wuhan-Hu-1 Spike regardless of prior infection status (after dose 1, prior infection status is easily determined but it normalizes after dose 2 between exposed & naive).

likewise, mRNA vaccines induce robust T cell immunity towards both ancestral & in this case, P.1/Gamma Spike antigens, both CD4+ & CD8+.

(robust activation occurred for both ancestral & P.1 Spike).

looking at variant mutations in Spike (using LIVE VIRUS ASSAYS), there's not much loss in neutralization from Wuhan-Hu-1 to Alpha, but once you start going to Delta onwards, there's a progressive loss of neutralization as you add mutations.

two of the most impactful mutations are L452R & E484K.

INTER-INDIVIDUAL VARIATIONS IN NAB RESPONSES ARE LARGER THAN TRENDS IN VARIANT-SPECIFIC NEUTRALIZATION OVERALL.

(generally, if you mounted a very strong, robust titer against Wuhan-Hu-1, neutralizing ability was maintained across variants. if you never mounted any nAb response, this persisted across variants).

prior infection generally improved cross-neutralization titer, likely by acting as effectively a third dose / third round of affinity maturation.

next topic: @BenIsraelow's correlates of immunity against #SARSCoV2 results! (paper: science.org/doi/10.1126/sc…).

using the AAV-hACE2 mouse model, first thing was looked at was muMT, i.e. mostly-B-cell knockouts, clearance was delayed by 7 days but successful clearance by 14 days.

however, Rag-/- cells lacking both B & T cells were completely unable to clear #SARSCoV2.

mice lacking CD4+es or CD8+es were able to clear the infection as normal, but mice lacking *BOTH* had trouble clearing, some mice were unable to clear infection.

muMT mice lacking CD8+es mostly wouldn't clear the virus, interestingly *some* muMT mice lacking CD4+es wouldn't clear the virus either. muMT+CD4-/-+CD8-/- mice had viral clearance kinetics equivalent to Rag-/-.

next up, doing adoptive transfer of serum or T cells into Rag-/- mice showed that nAbs were sufficient for clearance, whereas T cells only were *usually* sufficient but not guaranteed to clear the virus, but *did* reduce titers consistently.

next up, looking at mRNA vaccine response: there was a pretty direct inverse correlation between IgG level or neutralization titer in mRNA-vaccinated mice & how much weight they lost.

-> higher titer implied less weight lost.

finally, mRNA-vaccinated mice were challenged either w/B.1 or B.1.351, but first were either depleted of CD8+es or not.

-> all the immunized mice survived, regardless of variant or CD8+ status, all the naive mice succumbed to the infection.

-> T cells *or* B cells can clear a primary infection, one can compensate for the other somewhat, in the absence of adaptive immunity mice become chronically infected.
-> CD8+ T cells are not *necessary* to protect against viral challenge (in vaccinated *or* recovered mice).

fascinatingly, despite not being designed for it, our mRNA vaccines seem to induce some amount of mucosal immunity (paper: thelancet.com/journals/lanin…).

finally, our last topic, Long #COVID19!

two pathways to Long #COVID19: ~70% after severe illness (fibrosis, tissue damage, organ damage?), but also ~10% - ~30% after mild disease exhibit an ME/CFS-like illness.

(interestingly, the latter seems to be enriched in young women, which is a fairly different risk cohort to severe #COVID19 outcomes).

Long #COVID19 after mild disease have symptoms that affect many organ systems: those that have <12 or so symptoms tend to eventually recover, whereas those w/>14 symptoms tend to persist or even get worse.

several hypotheses for pathophysiology of Long #COVID19, including autoreactive lymphocytes & persistent viral reservoirs.

evidence for both of these: persistent staining for N in gut has been demonstrated, whereas autoantibody development is well-documented.

holy shit that is an amazing autoantibody mapping technique.

Are there any treatments that impact Long #COVID19 symptoms?

(also hi @VirusesImmunity!!!)

interestingly, after vaccination ~55% of people reported improvement, ~25% reported no change & ~20% reported worse symptoms.

what could be the relevant mechanisms? well, viral reservoirs could be eliminated, or cytokines produced could divert an aberrant inflammatory response temporarily.

partnership & collaboration w/the #COVID19 Survivor Corps, which made much of this research possible.

and end of an *AMAZING* talk! Q&A now.

@ENirenberg: what is the role of *non*-neutralizing Abs in protection against #SARSCoV2?
@VirusesImmunity: protection from infection largely correlates to circulating nAbs, there are studies showing that non-N but binding Abs also contribute to response.

very difficult to do correlates of protection using T cells b/c of a lack of standardization between labs. standardization would help a lot here.

"Did the lethal cases have any pre-existing conditions or did they fall into a certain age group?"
@VirusesImmunity: lethality correlated w/age, male sex, certain "pre-existing" conditions & autoAbs to type-I IFNs (as first shown by @casanova_lab).

"Do you know how many percent of SARS-Cov-2 infected individuals do not produce any antibody response lasting e.g. >2months? some papers claim non-response as high as ~30% which is hard to believe"

@VirusesImmunity: assay-sensitivity dependent, mild cases tend to produce lower titers so a non-sensitive assay might not detect circulating nAbs, but that doesn't guarantee lack of response & ofc doesn't measure tissue-resident B/T cells which also provide protection.

"Do the variant changes in spike protein affect the affinity to the ACE2 receptors?"
@VirusesImmunity: not measured by our lab, but yes, several labs show improved binding from many RBD mutations in VOCs such as N501Y.

(me: L452R, E484K & N501Y all improve binding by ~order of magnitude -- combining several of these can lead to dramatic improvements in binding).

"Would depletion of CD4+ T cells in vaccinated mice abrogate protection against viral challenge?"
@VirusesImmunity: wasn't studied, but it seems CD4+es primarily act as help for Ab generation & less so as direct effectors.

"Very convincing data about the efficacy of mRNA vaccine. Are there comparative stuies on other vaccination strategy (DNA virus, inactivated virus)?"

@VirusesImmunity: some head-to-head comparisons have been done & are being done right now, we are in the process of looking at different combinations of different prime-boost combos & how this affect VOC immunity & how robust it is -- important b/c mRNA vax supply is limited.

"Thanks. Have you ever adoptively transferred T cells from infected lung into navie mice to look at the protective function?"
@VirusesImmunity: haven't done that experiment, but it would stand to be the case that tissue-resident memory T cells would be protective.

"How do B cells get activated and produce AB in the absence of CD4 T cells?"
@VirusesImmunity: indeed, there is massively reduced titer & breadth of Ab response in CD4+-depleted mice (see supplementary of paper).

"Do you recommend children get vaccinated?"
@VirusesImmunity: just b/c the relative rate of severe disease is lower doesn't mean that children can't develop severe disease, just that it's rarer. plus, we don't know Long #COVID19 rates in children. YES, VACCINATE CHILDREN.

END TALK, that was wonderful. ALSO WHY DID I DECIDE TO LIVETWEET QUESTIONS DOING THE TALK WAS HARD ENOUGH AUGH.

next talk is by Dr.Jane Burns speaking on MIS-C: What Are We Missing?

initially, pediatricians thought they could largely "sit this out", but starting ~May 2020 it became increasingly clear that MIS-C was a thing in some children, which resembled Kawasaki Disease somewhat.

early paper on clinical characteristics of MIS-C: ncbi.nlm.nih.gov/pmc/articles/P…

notable differences in age, platelets, CRP, lymphocyte count & ferritin levels between MIS-C, KD & TSS.

Enrichment of IL-1 & TNFalpha pathways in MIS-C pts: HUGE upregulation compared to KD -> possibly suggests treatments targeting this hyperactive innate response?

looking at differential gene upregulation between MIS-C & KD: 102 genes uniquely upregulated in MIS-C: interferon signalling being by far the largest target here.

finally, development of a diagnostic signature by doing RNAseq for MIS-C, KD, TSS & trying to find a signature via bioinformatics. pathway analysis of differentially-expressed genes showed strong differences in coagulation cascade & platelet activation.

-> were able to reduce this down to a 4-gene signature w/a ~95% specificity, which is currently being verified by hand right now.

also, development of a neural-network based classifier approach on lab data, which is likewise being tested right now.

looking at T cell recognition (by @SetteLab et al), no clear mapping to disease severity but clear signatures for MIS-C vs. KD that were largely HLA-independent. most notable difference were in CCR5 expression on CD8+es.

next, trying to see if IVIg has any benefits in MIS-C (since it's a standard treatment for KD, w/very little understanding of what it does).

turns out its effect is killing IL-1beta+ neutrophils via a Fab-dependent manner that's via PI3K/NADPH oxidase pathways.

interestingly, the killing is quite rapid but neither apoptopic nor pyroptopic? this holds for both MIS-C & KD pt neutrophils treated w/IVIg.

MISTIC trial: looking at adjunctive therapy (infliximab, anakinra, steroids) for those pts given IVIg who have known vascular leak syndrome due to leaky endothelial tight junctions.

-> still more questions than answers, we have no clue what determines susceptibility. still rare (only 4661 pts in USA according to CDC).

-> what is the driver of the intense innate response? is there an immune defect? (no evidence for autoAbs, unclear evidence for SAgs -- characterization of T cell response showed no evidence, normal T cell responses in general, not clear if defect in CD8+ T_{CM} development).

-> finally, what is the best therapeutic response? currently, there's no clear answer here.

"is there concern for combo therapy with anti IL1 and anti TNF, given the high level of infections observed in clinical trials for RA (many years ago)?"
A: no, for pediatric disease this is used commonly, also this is in the acute setting, not longitudinal, so there's no concern.

"any role for autoAbs in MIS-C?"
A: nope, so far all cases of autoAbs were due to contamination w/IVIg. would be nice if it were due to that but there's no evidence for it.

next talk is by John Shyy on S Protein, hACE2 & Endothelial Function!

background on RAAS: hACE2 is a metalloprotease that cuts Angiotensin II to AT[1-7] which binds to its receptor MAS which largely counteracts AT2R effects.

turns out Ser680 of hACE2 is an AMPK phosphorylation site that is heavily conserved. looking at S680A (non-phosphorlated) or S680D (phosphomimetic) hACE2 transfected into HEK293T shows phosphorylation improves stability by affecting ubiquitination!

S680D hACE2 generated increased levels of AT[1-7]!

fascinatingly, S680D (phosphomimetic) ACE2 knock-in AMELIORATED pulmonary hypertension, improved vascular integrity, reduced vascular leakage & reduced aberrant arterial remodelling!

turns out MDM2 is an E3 ligase that is capable of ubiquitinating hACE2 & this contributes to PAH (paper: ahajournals.org/doi/10.1161/CI…).

looked at hACE2-K788R (ubiquitination-resistant) mutants: this ameliorates PAH even in the absence of phosphorylation at S680!

anyway, now looking at expression of #SARSCoV2 Spike via pseudo-Spike expression using viral vector: S exacerbated lung inflammation. looking at expression: pseudo-S increased MDM2 levels, decreased ACE2 & phosphorylated-ACE2 levels likely via impaired AMPK activation.

S protein & ACE2 S680 hypo-phosphorlyation increase glycolysis in endothelial cells!!!

S680D ACE2 is resistant to changes in both glycolysis & oxidative phosphorylation induced by S1!

(evidence that negative effects are due to mitochondrial dysfunction in hypophosphorylated ACE2 ECs).

oh wow, there's evidence AMPK/MDM2-regulation of ACE2 is aberrant in IPAH pts?!?

that was a fascinating talk, & continues to convince me that i *really* need to brush up on my knowledge of the RAAS system in general.

"Do you think the hACE2 SNP also affects the SARS-CoV-2 infection, besides ACE2 SNPs on EC function regulation?"
A: "good question, a future study would be to look around the ubiquitination site. S680 seems highly conserved so nothing there."

"What's the effect of metformin on covid-19?"
A: "some small clinical trials showing metformin is beneficial, especially in people already taking it for T2D. at the very least, it should never be stopped."

"Do we know why S induces MDM2 so strongly, & also have we looked at variants & if they affect phosphorylation differentially?"
A (to the first): "we don't quite know if S directly might, b/c hypoxia is a *massive* inducer for MDM2. we don't know in general."

A (to the second): "good question, it would be interesting to test whether, say, Delta would reduce ACE2 more than others. would be very interesting to see how this affects endothelitis".

short break, we will be back at 11:15 (or for me, 2:15AM lmao).

okay, we're back! our next speaker is Andrew Ward speaking on Mapping Polyclonal Responses to Seasonal & Pandemic Coronaviruses!

vaccine development has really become quite dynamic now & often starts by optimizing antigens via stabilization / epitope focusing (such as the infamous 2P/6P stabilizations of CoV Spikes).

a lot of work goes into stabilizing the *pre-fusion* stabilization of fusion proteins (two examples given, HIV env gp120 & RSV F). generally, this involves things such as breaking helices via proline substitutions, cavity-filling mutations, modifications of disulfide bonds, etc.

a nice review on proline substitutions, which comes up again & again: ncbi.nlm.nih.gov/pmc/articles/P…

interestingly the first human β-CoV Spike to be imaged well using CryoEM was actually *HKU1* due to ease of expression. looking at what stabilized this, two residues in heptad repeat 1 replaced w/prolines did excellently.

expressing this in SARS Classic & MERS allowed efficient expression & also allowed good CryoEM imaging (work done w/@McLellan_Lab & @BarneyGrahamMD).

2P improves immunogenicity by allowing epitope focusing onto the prefusion conformation & preventing development of Abs towards the postfusion Spike which is an epitope antitarget.

brief overview of EM-based polyclonal epitope mapping (EMPEM) which is truly revelationary in terms of how quickly & accurately it allows for mapping immune epitopes.

(seriously it impresses me to this day, it's *wonderful*).

epitope mapping of #SARSCoV2 Spike regions (in both convalescent & vaccinated sera): main variation is in NTD, consistent robust RBD targeting, S2 responses generally not seen.

brief overview of the usual 4 classes of RBD antibody epitope sites that have been well-documented by now.

total loss of NTD specificities in many VOCs (which doesn't surprise me, very few mutations totally remodel the NTD & there's only one major antigenic supersite).

fascinatingly, Vh3-73-biased broadly-neutralizing RBD antibodies towards SARSr-CoVs can be induced in rhesus macaques: biorxiv.org/content/10.110…

brief overview of CryoEMPEM which can be used to *image* these epitopes (paper: ncbi.nlm.nih.gov/pmc/articles/P…) -- this can be combined w/next-gen sequencing of B cells to *engineer* mAbs targeting a specific structure!!!

(holy shit this is FUCKING AMAZING).

workflow: biorxiv.org/content/10.110… (YOU CAN BET THIS IS WHAT I'M READING OVER LUNCH).

finally, exploring the role of pre-existing immunity to β-CoVs (& it seems that post-pandemic there is indeed backboosting of OC43/HKU1 Ab responses after exposure to #SARSCoV2 Spike antigen).

finally, some epitope mapping of pre-pandemic epitopes to OC43/HKU1 incl. molecular details of the immunogenic surface of HCoV-OC43: biorxiv.org/content/10.110…

"you mentioned that K956P and V957P can stabilize spike. do you have any insight into how the D950N mutation in Delta and other variants might affect the structure and if it could contribute to viral fitness?"

(this is actually a question i think is very interesting fwiw).

A: "we have not looked at that; one of the problems when talking about stability is this is that these Spikes require a lot of mutations to image properly so it's tricky to say things about the energetics of these mutations."

"Have you followed up your pre-pandemic serum after vaccine or natural infection to see if there is evidence of antigenic sin? Do pre-existing abs to seasonal CoVs affect response to vaccine or natural infection?"

A: "that's actually *really* difficult... some people have found Abs targeting the stem that's conserved but that's mostly it. what we mostly see is a cross-boosting effect where OC43/HKU1-targeting Ab titers go up post-#SARSCoV2 antigenic exposure despite not binding to it."

A (cont): "the mechanism is still unknown, we're still trying to elucidate that."

damn, i wanted to ask a question but couldn't figure out how to phrase it. guess i'll email Andrew Ward later LOL.

anyway that was a PHENOMENAL talk holy SHIT.

finally, our last talk before lunch, Erica Saphire on Antibodies to #SARSCoV2: A Global Collaboration!

some people can't be vaccinated: important to have Ab therapeutics for these people, but which Abs to use, how to pick which ones to develop clinically, what assays to use, etc?

about 371 candiates, sourced from #SARSCoV2 survivors, SARS Classic survivors, immunized animals, engineered in-silico, nanobodies, etc.

what features do they have? which are worth focusing efforts on?

overview of the CoVIC workflow, which is *incredibly* well-designed imho.

which CoVIC mAbs are resistant to VOCs?

let's look at the 7 different classes of RBD-binding Abs (1-3 bind RBM, 6/7 are Class 4 mAbs that bind the inner face, 4/5 are Class 2/3 mAbs that bind the outer face).

paper looking at binding of these mAbs: science.org/doi/10.1126/sc…

RBD-1 fully occupy S, 3 per Spike, one arm each. RBD-2 binds one per S w/both arms. RBD-5 cross-link Spikes together 1:2 or 2:2!!!

this may be one of the coolest slides that i've seen in my life:

interestingly, potent neutralizers in 2a & 2b tend to be completely knocked out by mutations in Beta & Gamma but not by Delta (i think what is happening is the charge reversal at E484K is blocking binding).

as a fascinating proof-of-concept, there's a class 1 mAb that binds better over variants; this isn't a true antibody, it's hACE2's contact surface bound to an Fc fragment. but it's a useful demo.

Spike protein is a metastable target: you want to optimize targeting to the prefusion conformation.

looking at optimization: remove one protein from 2P/6P that's detrimental, add an S2-S2' disulfide bond, add a flexible linker to prevent dissociation. called VFLIP.

VFLIP: has greater yield, greater stability; stays in the right shape, more accurate glycation. "third-gen" stabilization, compared to first-gen 2P & 2nd-gen 6P.

what happens when we immunize w/it?

slightly better neutralization after immunization... but 6 months after immunization it's *notably* better, likely due to further epitope targeting!

that was a *really* cool talk holy shit (yes i know i say this a lot but these talks are just BRILLIANT).

"Do you have some mAbs that are good candidates for clinical trials yet? Would they be tested as prophylactic or therapeutic?"
A: i think we have some targets that are *really* good incl. neutralization down to *very* low doses, currently entering NHP trials.

A (cont): goal is mAbs that can be delivered cheaply even to low-income countries, where they are badly needed.

"In your opinion why few mAbs reached the market (3-4 w/EUA in the US), despite of being one of the few effective antiviral drugs available to date?"
A: mAbs need to be given early in an infection (like antivenom w/snakebite) but this makes recruitment tricky.

A (cont): plus it was tricky to get trials done, also wanted to get trials done ASAP but these tended to be 2a/2b mAbs that are readily escaped. but hey, once you've seen one antibody, you've seen one antibody. more careful trials are being done now.

(can i just say how much i *love* people using the "once you've seen one X, you've seen one X" sentence in biology? b/c it's *so* true. h/t @halvorz).

"Do you have antibodies that prefer K & Q at 484 but do not recognize the ancestral from in your set?"
A: i'm not sure. i don't think so, i'd have to double-check, but it'd likely have to be one of the recent targets since most of our submissions were b/c they bound B.1

wrap-up of the "morning" session & time for lunch (which means for me, grabbing some instant noodles at the nearby 7/11 lmfao). next set of talks starts at 13:00 (or 4AM if you're me).

link to part 2 (afternoon session & closing keynote):

https://twitter.com/wanderer_jasnah/status/1448743866262188040