Universal Immunity: Why people don’t generate broad spectrum responses to mutating viruses

Hi, my name is ron Najafi. I'm CEO of Emory Pharma. We're honored to have Dr. Jake Glanville here. Jake and I both serve on the biotech advisory board of University of San Francisco. Jake did his undergraduate degree at UC Berkeley in molecular biologist. He went on to Pfizer for a few years and then after Pfizer, he went to get his PhD at Stanford. He is one of the preeminent immuno engineers in the country. I believe he went on to do his to sell his company, Distributed Bio. And after Distributed Bio, he started his new company called Centivax. We're honored to have him. Today is March 9th and we're having an atmospheric river again. And on the same day we had Dr. Robert Lustig with another atmospheric river. So this is just going to be an amazing presentation by Jay.

All right, well, listen, I want to thank you, Ron, for having me over and thank you all for giving me a bit of your afternoon and to invite me over to talk about one of my favorite subjects.

And that is the question of why.

Is it that we don't all enjoy universal immunity? Why is it that we get infected by viruses that mutate? If parts of the viruses don't mutate every year, why aren't we all targeting those? Why do we have to update our vaccines? Why is it that after a couple of flu exposures we don't have universal protection? This is a great question to have, because if we understand why we miss, then we can engineer tools that can do better vaccines, that can provide broad spectrum protection against whole classes of viruses.

So what I'm going to do today.

Is talk about a bit about the history of the field, the basic emergence of understanding in vaccine science around 1993, from just rending our hair and saying there's no way we can ever solve this problem to the discovery of broadly neutralizing antibodies. These special antibodies that bind these conserve sites that got the whole field excited because those proved that it was technically possible to create universal vaccines or broad spectrum vaccines. They also annoyed the entire community for the next 20 years on trying to understand, well, if some people make those, why don't we all make them?

And that exercise in greed, why can't.

We all have those great antibodies is what I'll be talking about. And some of the technological advances in achieving that reliably and reproducibly in a vaccine.

So I want to begin by.

Just making the point.

Vaccines are great.

Vaccines are one of the greatest medical advances since sanitation and fire.

And I think I Just need to.

Emphasize that over and over again because the modern world forgets we have people that are concerned. And vaccines are not perfect.

But it's important to realize the magnificent.

Transformation in human health that has happened since this technology has emerged recently in the last couple hundred years.

What I'm showing you here is the first known case of smallpox. That was Ramses Van. We had smallpox on the face of.

A pharaoh 3,000 years ago.

And then some jokers with pocket protectors.

And many other people working with them eradicated the virus. They ended a pathogen that harassed us since the beginning of time, and they did it in the 70s. That's something that is remarkable about vaccines because unlike most other types of medicine.

They have the capacity to end the.

Disease that they treat.

It also is an area of frustration for me because it feels a little.

Bit like our trips to the moon, where we accomplished something so great and then we stopped. And because we live in this golden age of biotechnology, I felt that it is incumbent upon us to push further, to ask the questions, why can't we do more of this work? Why can't we advance our existing vaccines and kill more of our most ancient enemies?

And let me be clear.

Infectious disease is humanity's greatest enemy.

Every one of us experiences this enemy.

Within months of birth. It kills more in every generation than every war combined. And we have technologies that can end these diseases forever. So why aren't we doing it?

I also just want to emphasize it wasn't just smallpox that we in the.

First world take for granted the enormous gift that vaccines have given us in terms of what our modern lives are like. I grew up in a Zutuwil Mayan village in Guatemala. There is no vaccine hesitancy. At least there wasn't until the recent pandemic and the Internet caused the cloud of it. Zutu wiles are super pro vaccine because they have experienced firsthand the consequences that the first world has forgotten, which is what life is like when you do not have these kinds of easy and cheaply available defenses against pathogens that can steal children from us and waste a lot of energy on recovering death and injury from these diseases.

You can see here just a subset.

Of the remarkable transformation of what standard life is like. Imagine all the time that people can spend building things and making money and time with their families instead of fixing broken cells. That's the call to arms of the value of vaccine technology. Not to say that they're perfect, but they are remarkable and transformative.

Here's the problem.

Vaccines work great against slow mutating pathogens. Pathogens where a single target or the whole surface of the protein or the pathogen is enough that you can give it. And the immune system produces antibodies against it and T cell responses, and that gives you the anticipatory protection so that when you're exposed to the real agent, you can defend yourself. Unfortunately, vaccines don't work so well against rapidly mutating pathogens.

And this is showing a history of how well our flu shots work.

So flu mutates pretty quickly. We have to make a new vaccine for flu. We've been making vaccines for flu since the 1930s. It is by far the best studied class of vaccines ever, because we make one in the northern and the southern hemisphere every single year. And we've been doing it for decades.

And yet they suck.

The vaccine efficacy ranges typically from 30 to 60% efficacy. You know, it's a. That's less than a towing cost. That's going to protect you. And some years it's much worse than that. We had a 10% year during the swine flu epidemic. We've had some really bad H3N2s.

And the problem for that is that we don't. What you're injected with is not what.

You'Re infected with, that there's mutations that have accumulated and that government bodies, when they're planning to manufacture the vaccine, they need time. And so they have to make an educated guess about what the strains might look like and what they have now on hand. That's close enough that it'll do a good enough job on a population.

But that's not where we should be. Where we should be is in the 80 to 90% effectiveness range.

It's really hard to get 100% because you're going to run into people with immunosuppression and various other abnormalities. We rarely see 100% effectiveness with vaccines. There's a couple that get pretty close.

But we should be doing way better.

Than 30 to 60%. And the problem here is really comes down to mutation.

This is an illustration of the degree of mutation that accumulates just in a few flu seasons. And it's quite substantial. It's enough that we have to update. There are sometimes cheats that are made to try to keep components in from.

Two, even three years before. But we pay for it with less effectiveness.

And flu is not alone. We're all familiar with the coronavirus, and.

I think we've all lived through the experience of a vaccine that at first looked good. And then when we saw these variants of concern popping up, they progressively deteriorated the quality of protection. That's the Delta strain, the Omicron strain, Ba4, Ba5, the list goes on.

And with each one of those additional.

Sets of mutations, the guidance was very frustrating to communicate where instead of saying, yeah, this is going to stop from getting sick, it was like, well, this is going to reduce your chance of severe hospitalized outcomes, but you're probably still going to get sick. And that was because of the deterioration. And the best method that they had to deal with this was to, okay, take more shots. Even if most of your antibodies don't work anymore, we'll just boost you like crazy.

So.

So the remaining antibodies that cross will be at a higher titer and we have to be able to do better than that.

This problem has been applied to hiv.

We do not yet have a working HIV vaccine because of the mutation rate of the virus.

And we know that this is the.

Problem because if we have a matched virus to a matched vaccine, then you can protect a monkey that way. The problem is the mutations, the heterologous.

Collapse, and there's a whole bunch of other pathogens that for all the praises.

I want to sing about how great vaccines are, they have not addressed these particularly well because of this problem of mutation. And so that's the area we're going to focus on today.

I mentioned in 1993, then a number.

Of studies in 2000, 2004, 2007, there.

Were these remarkable observations that, hey, in some people we are finding these antibodies. This is the hemagglutin and stem right here. And then right on blue are some antibodies. And these antibodies seem like they're not.

Just binding one strain, but they're binding a whole bunch of strains. They're hitting a conserved site that exists.

And if you can get an antibody.

Like that, it will broadly neutralize many different variants.

And that got everyone super excited because they said, well, if we can get vaccines to make those kinds of antibodies as a response, then your vaccine isn't.

Just by a single strain. It's broad spectrum across a whole class.

It would be revolutionary to create a.

Second generation of vaccine technology that could do this for all the, all the pathogens that are not currently well treated.

The challenge was, why aren't those happening all the time? To help you understand what I'm going.

To talk about next, I wanted to bring up something that happened during World War II because it's going to be illustrative of the mutational challenge of a mutating virus.

During World War II. I don't think this is apocryphal I.

Think this actually happened, but maybe it's just a fairy tale.

The Brits were wanting to go put.

Some more armor on their planes.

And so they said, I know, let's.

See what we do.

Let's go analyze all our planes and we'll mark all the spots where we see bullet holes on them and we'll make a map and that'll tell us where to put protective layering. And so they looked at all their planes from all the sorties they put them and they got something that looks like this. And someone's like, all right, well pip, pip, what we're going to do here is it looks like we need to put some more protection here and out at the wingtips and we're good to go. And then someone's like, no, you idiot. Those are the spots where the plane could get hit and it could still fly back. Where you need to put protection is these regions where if the plane got hit, you never see a bullet hole there, because if there was a bullet.

Hole there, the plane crashed and you never, it never returned.

And this exact principle applies in evolutionary space that there are certain sites which do not have mutations because if you.

Mutate it there, the plane can't come back. It destroys the via of the organism.

And that does two very cool things. One is that there are these conserved sites.

The second is that it tells you that conserved site is very important, it's sensitive to change.

And so that's a great place to.

Send an antibody because it's conserved and it's important.

I'm going to highlight that exact point by showing you this exact kind of.

Map on the surface of hemagglutinin.

So this is now the viral spike protein. In yellow are the super conserved sites, in green and blue are the bullet holes.

I should have colored it differently.

And so what you see is there's some semi conserved regions here, there's a lot of dark blue, and then there's this patch that's super conserved, but there's still like sites around it that can mutate.

But there's this hot spot in there.

And that actually is exactly where the.

Broadly neutralizing antibody binds. There's another, less broad one that can.

Bind up here and a couple other sites. And so those are those white spots.

On the plane that cannot afford a bullet hole.

And if you're trying to shoot the plane down, that's where you want to.

Shoot and that's where we want to send our antibodies.

So people tried a couple different techniques once they saw this, they're like sweet.

Let'S make a vaccine to focus the broadly neutralizing antibodies. No problem.

It turns out it is very difficult.

To herd the immune system to do what you want. And it was particularly difficult because that period preceded the computational immunology revolution where we had access to high throughput data to interrogate these systems.

So they were trying to engineer a.

Car without lifting up the hood, which made it challenging.

People tried generating consensus sequences of hemagglutinin. It was an interesting idea. The idea was let's make it sort of the average so the most number.

Of sites would look the most similar.

There were some interesting results, but it.

Required a lot of cross boosting. It ultimately was not successful. None of these were successful. We do not yet a broadly neutralizing vaccine available.

There was some elegant work saying, well, let's just chop off the more diverse head part and we'll deliver the stem or we'll create a recombinant version of the antigen. People tried that. It had somewhat promising results and then.

It kind of collapsed when it came down to neutralization and protection.

And part of the challenge there was the problem that the immune system overfits whatever it sees.

The other problem was regulatory, that our regulatory process revolved around detecting antibodies against the head and they were chopping it off, which wasn't helping. That was not the insurmountable problem. The other problem was it didn't work.

People tried saying, well, let's cross boost. Let's give them a 1970s vaccine, then a 1980s and a 1990s, and that'll.

Sort of create a Venn diagram that selects for the things in common and it'll boost this ultra breadth.

And that didn't work either. Indeed, how could it?

Right? Because if that worked, then everybody by the time they're in their mid-20s should have universal protection because that's what we experience as adults and children coming online immunologically.

I'm going to show you later an.

Exception to that point.

People tried heterologous VLP particles where they put like eight different versions of HA on the surface of a vlp. That actually does kind of work that's.

The most similar to our technology.

There are some limitations that this works.

Well on a naive animal. And then it gets more complicated once you have immune memory and imprinting. And I'll talk about that a bit as well.

There are some T cell approaches where there are these conserved T cell epitopes, little strips of peptide that you can feed in.

It turns out it helps, but not by itself. So you need antibodies for dealing with flu. The T cells are sufficient or supportive. If you give the T cell vaccine along with a normal flu shot, it'll boost it. But if you try to give it by itself, that's human, very expensive, clinical study failed.

Then there's the more cowbell approach, which.

Is just, let's apply more adjuvant, let's do more boosts. And it helps you increase the titer. If you have a few antibodies in there that are pretty broad. If you increase the, those will also increase, but it doesn't last a long time. That's why we had those guidances for all those repeated boosts of the Wuhan strain during the pandemic. And it's why we haven't solved this problem by just increasing adjuvant.

There's also a limit to how much.

Adjuvant you can give people before it starts irritating them. There's a safety and tolerability issue that starts coming into play.

So what I did when I got interested in this work was I was.

Engineering antibody therapeutics and I was doing quantitative analysis using deep sequencing of antibody repertoires. I was using high throughput synthesis methods to build better antibody libraries based on the sequencing where I was trying to understand my defects of why in a petri dish, why in a test tube, why was I missing in a library of antibodies that I had total control over.

And in the process of doing that.

I developed some quantitative basis for my guess on why we were having trouble in vivo missing certain epitopes we wanted to hit as well. I was coming at it from a purely engineering perspective, but I realized there was this in vivo application in vaccine science.

And I'm going to tell you the.

Story about how I attempted to apply that and how we're seeing some very promising results as a consequence.

So what I was trying to ask is like, okay, if I want to get an antibody here, why is that hard sometimes?

And that's really a question for any protein target. But it's a very foundational and very important question. Furthermore, what's wrong with the model? Why were all those other universal vaccine efforts failing against influenza?

So the model I'm going to show you, in my opinion, the problem was.

The model was a radical oversimplification because people know better, scientists are smart, but.

They also, when they're talking to each.

Other, they oversimplify because it's annoying to use language to be very precise, and there's a bad habit that the more you say something, you know to be an oversimplification the more your brain only remembers that thing. And I think that happened to immunology and vaccinology.

And this was the convenient lie that.

People talked themselves into.

And that was, well, the head is immunodominant and non broad, the stem is conserved. And if you believe that, then of course you should say, well, we should chop off the head and make stem.

Only vaccines or chimeras or approaches like that.

And part of the reason they had.

This separation, this massive geographical separation, was.

That you could order this reagent separate from this reagent.

You could express the two pieces independently.

Those methods all failed. And it gets back to this thing.

I showed you before.

Yes, there's a conserved site in here.

But there's also a mutational variant.

Yes, this is more variant, but there's.

Also some conserved sites. It was the failure of respecting that nuance that gave rise to frustrating results.

Because if you immunize with this and.

You think, okay, stim.

Is conserved, you've just lied to yourself because most of this is not. And all an antibody needs to do.

Is have one of these flexible positions in the binding site and that's enough to shed that antibody. And so those techniques did not work.

So what I wanted to do is get to the just kind of crack.

Open the hood of the immune system and get at least a decent understanding of what was happening. What were the quantitative diversity dynamics of a vaccine response, which is to put.

Simply, how many antibodies do I get.

After I give someone a flu shot? This is a basic question, but a really important one to be able to address these systems.

And so this is a summary of about 10 years of my work and that of others trying to quantify how.

Much diversity there is in an antibirus body response.

To get a handle on why we miss first, we now know from deep sequencing exercises that there's about 100 million.

Unique B cell receptors in the naive and memory compartments of a human that.

Was way lower than people thought people were estimating.

It was like e to the 11 because you have that many B cells.

It turns out when your body makes B cells. Oh cool, we made a good rearrangement.

Let's make 1,000 copies of it. Let's not spend all the time making every B cell perfectly unique.

And I discovered this when I was getting to the bottom of why antibody.

Libraries were much less diverse than people thought they were, because they were taking blood, making this hopelessly optimistic prediction of how diverse blood was and being like.

Well, if we have 20 people, that should be perfectly sufficient for a Super diverse library.

And then I'd steep sequence it and be like, guys, this library is like a thousand thousandth of the size you thought it was. And this was the underlying problem.

It played into vaccine science because then, therefore there were much less antibodies than people thought.

Unique antibodies after vaccination.

We know there's about 1000 unique responding.

B cell lineages after vaccination.

That's it. They have lots of copies of those.

Cells, but there's really only 1000 unique flavors of B cell receptor. After a flu shot.

And then using liquid chromatography and mass.

Spectrometry, the same type of biology that you do here at Emory, a great.

Group from Giorgio's group down in the University of Austin assessed the serological diversity.

From subjects who were vaccinated.

And they said, you know, only about a 10% of those receptors end up.

Becoming the plasma cells that protect you.

It was a really elegant assessment and.

A shocking reduction of the diversity that people thought that actually your serum response.

After a shot is about 100 unique antibodies. Now that's a great number.

It's a useful number because it lets.

You ask the question, okay, well, how many epitopes, how many binding sites are.

There on the protein?

If it's less than 100, you should be hitting them all every time. Why are we missing? And so I asked that question. It turns out there are way more than 100. I used a protein protein docking mechanism where I took the Amazon cloud and.

About 600 solved crystal structures of human antibody. Every human antibody I could find. I docked it hundreds of millions of.

Times against the surface, saturated, like carpet, bombed the surface of hemagglutinin and came up with a list of a few million different docking geometries. And of those, because we know an antibody doesn't care about all the positions in contact, it's actually a much more.

Complicated list of billions of unique epitopes.

And these estimates had to be correct. They lined up with, first, the level.

Of epitope diversity and crystal structures that were coming in, that almost every one was unique.

And second was some nice studies where.

People were doing immunoprecipitation. They were taking people's serum and running it through heterologous immunoprecipitation to pull out antibodies that were broad. And they found that the ratio of broad antibodies to non broad were lining up about with my estimates.

And so I have this huge diversity set of epitopes. And then I can go, okay, well, let's look at the last hundred years of sequence evolution.

Of the virus.

And let's just ask what proportion of.

Them are the broadly neutralizing sites?

And the answer is about 80% of those sites are easily mutated.

In the next one to three years they have one of those green or blue spots, at least one in the binding site.

About 20% or 19% are pretty conserved as to H1 but not H3.

Pretty conservative to H3 but not H1. These are pretty good, but they're only 20% of your titer.

So you're going to lose the remaining.

80% on small changes in the virus in the next couple years. That's why you need a booster shot.

And then the really sexy broadly neutralizing epitopes that people publish on, the reason.

It takes them a while to go.

Find them is that they're super rare.

They'Re less than 1%. The really good ones are less than 1 in 10,000.

And that's why some people produce them, a lot of people don't.

But even if you produce them, it's 1% of your immune response, so it's not immunodominant. Maybe we all know that one annoying hippie friend who's like, oh, I never get sick, I'm always good.

And that might be true, he might be very lucky.

And that was the high titer response.

But that is an abnormal case. Most of us do not produce it. You do not need to invoke any bizarre theories. It's just a numbers game that you only have 100 antibodies or 100 darts.

Hitting unique epitopes and you're throwing them.

At a vast dartboard with a very.

Tiny little red bullseye. And so we're missing by probability.

And the utility of this model was that first, it gives us useful prediction.

Power in our laboratory, but second, it explained why some of the other methods people are attempting could not work.

If I'm right, then some of these.

Methods can ever help to succe.

You can't mutate one site at a.

Time to make a perfect immunogen because it's going to be whack a mole. Every time you make one change, there's going to be 100 million other wrong answers you're going to pick next.

It also suggested an approach that we.

Applied that could fix this.

And the approach was, okay, well, I need to find a way to dilute.

Out the strain specific epitopes. How do I do that?

What I'm going to describe to you is going to focus on influenza. What we have is a broad spectrum.

Vaccine platform which is generally applicable to Mutating pathogens.

We actively work on influenza.

That's the thing that we're getting ready to go into humans with. We're gearing up to initiate the manufacturing process.

The coronavirus and hiv, those are earlier.

But we're testing the same technology platform.

In both of those.

They're in vivo, they're showing promising results.

And we've even gone all the way out to antivenom. I wanted to take something that wasn't a virus, was as far away as.

Possible, but I wanted to show we could use our same technology to produce broad spectrum immunity in the most distant.

Thing, for intellectual property reasons.

But also there's a humanitarian benefit to doing so. And I'll show you a little bit of that data as well.

So here is the concept, and at heart it's pretty simple. I said for any vaccine, we give.

A dose and you get an immune response.

All right, well, what if you lower the dose? If you keep lowering the dose, at.

A certain point, there's not enough protein to produce an immune response. What's happening at a molecular level is.

Your B cells are not internalizing enough.

Of the target, the antigen, they're not.

Displaying it on mhc, so they don't.

Get enough T follicular help. The lymph nodes are basically inactive because there's not enough drug for them to activate against it.

Sweet. So I go get 10 versions of hemagglutinin from the last hundred years, 19.

18 up to modern times.

I set each one right below that threshold of activation. There's not enough of the vaccine for each strain. Strain specific B cells, they don't have enough dose to activate. But the B cells that recognize a.

Conserved site across all of these can.

Gobble up to 10 times the dose. Through this mechanism, we give a bigger dose to B cells. The more broad the epitope they recognize.

The bigger the dose of the vaccine they receive. This selectively rewards B cells that recognize.

Conserved epitopes and they can activate, proliferate.

Somatically, hypermutate class, switch and convert to.

Plasma cells without the competition of the.

Overwhelming number of less useful B cells.

And that's really part of this problem.

Is there's this crowding problem that if most of the B cells are the wrong answer, but your body's restricted to.

The top hundred, you better make sure.

That the really good top hundred make it to the front of the feeding trough, otherwise they're going to be starving in the back, which is what happens all the time.

So we started out, this is the first primitive teensy but critical experiment and what I did here, hopefully you can all see it reasonably well, is I had sets of pigs, I gave seven pigs nothing. I gave them squalene and then I tested their serum. After multiple boosts against H1N1s, H3N2s, and.

A few other H5N5 H7N7, and they didn't really respond.

I then gave some pigs a set of seven, a single H3N2 at 50 nanograms, that's a hundredth of a normal.

Dose that you would normally give a vaccine to a pig, and nothing happens.

As expected, to that antigen or to anything else. Same thing happened here to an H1N1 2007. And yet when I give these two, as well as 25 other HA's that.

Are all different from each other, all.

At that 20, 50 nanogram level, sorry, this magic happens where suddenly they respond to both of those things and everything.

Else and stuff that was not in the set.

So there's this level where you go below the threshold. It's suddenly they respond to the conserved sites because there's only 50 nanograms of.

Each one of the components, but there's 27 times that for the conserved sites, if they're able to recognize it.

And so it creates this reward that spans out to a whole bunch of HA's that weren't even in the set of 27. So that was the core mechanism. What's happening here, if you look at the structure, is that as we dilute the components, so the B cell has to hit, say, two to get a threshold dose, or it has to hit three or four or five. You're selectively enriching B cells that recognize this one stem region and this one, all the way down to just being.

The stem region and not even the head.

We actually dose sort of right around in here.

We hit this conserved site and we.

Hit this, but we skip most of.

Those other sites that are prone towards mutation.

Then we wanted to go test this. I had worked with pigs because I.

Think they're a better model than mice for influenza.

They actually get the flu.

There's $180 million per year market for vaccines for pigs. And I think just as a philosophical, organizational principle, you should work on a system that rewards you for being right.

You had to make a vaccine actually.

Work in pigs, whereas you can trick yourself and fool yourself and create orthogonal systems in mice that were concerning to me.

So I chose pigs and ferrets. I was able to work with the Gates Foundation.

We won a grand challenge in the Pandemic Threat Award to be able to support us doing live challenge experiments at Auburn University and University of Georgia.

And what we did is we pretended it was 2007. So we used components from 1918 up to 2007, but nothing after that.

And the reason. And we compared that to a 2007 flu shot.

And the reason we did that is that then we could go test these animals. We could spray them in the face.

With future virus to see if they're protected from getting sick.

Sick. We could test their serum 13 years.

Out into the future to see if it had neutralizing capacity compared to a 2007 flu shot that we knew would not be able to do that.

And so those are the experiments. We did some clever things, as much as you can do on a computer, to try to predict what the valency and the dosing was. And we did a lot of animal.

Work using pig as the primary discovery model to figure out what's the minimum.

And effective valency, which is predicted here and borne out where we think that.

And what's the dosing level that we think would be appropriate. And then we did a lot of pig experiments.

I rolled up my sleeves and shoveled.

Through pig manure with the rest of us to be able to get to.

The data until the Gates money came in.

Then I get to sit on my butt and I get to get great collaborators to do that work for me, which I really appreciate.

So this is the first set of data. I'm going to show you binding.

I'm going to show you something called a hemagglutination inhibition assay, which is the key assay the FDA wants to see. And I'm going to show you neutralization assays. I'm also going to show you spraying the animals in the face with the virus to see if they get sick or not, which is what you really care about. And we're even going to look at the lungs and the trachea to see if virus has been inhibited from growing.

So this is a real simple assay. It's just saying pigs that had just the adjuvant or just a bivalent, these.

Two, or our vaccine, how broadly were they producing an immune response after three shots?

And so pigs with the adjuvant doesn't do anything.

The adjuvant doesn't produce protection.

A bivalent does what you'd expect. It responds. If we inject 1997 H3N2, it'll cover that. It'll hit some 1995, it'll hit some.

2002, but it deteriorates in both directions. That's why we update vaccines.

Same idea. A 2007 flu shot works against 2006.

Okay.

There's some stuff earlier, absolutely nothing useful.

Against 2009, because that's when the pandemic swine flu hit. Like, this is the whole problem with.

Vaccines versus our little piggies responded to everything. A century of influenza, including all five.

Pandemic strains of the last hundred years, I guess a little more than 100 years now.

They responded to a whole bunch of stuff, not even in the vaccine. The things that were the lowest responses were these exotic avian strains, like from H7N7.

Some of the H5N1s, some of the other ones we responded pretty well to. So that was encouraging to us, but.

It was just binding. We next had groups of pigs where we separated out, okay, we have this mixture. What happens if we lower the dose of the mixture? Because what you'd expect to find is.

As you lower the dose, you'd actually increase the breadth. There's this paradoxical inverse dose response effect.

And indeed we saw that when we went into animals at 500 nanograms per antigen to 100 to 50 compared to.

A bivalent at the traditional 5,000 nanograms.

Per antigen, there was an increase of a breadth shown in dark blue as.

We lowered the dose. At a certain point, there's a balancing act because as you lower it too much, you lower your titer, so you have super good breath, but not enough tighter. And so we play a game of tackling the middle of that problem.

But next we wanted to go to function. And so we chose that hai assay, which is what the FDA cares about.

It's the standard correlative protection. If you have a 1 to 40 dilution, hai, that means you're correlated to being protected. That's what the current vaccines aim for.

To accomplish a control flu shot in the pigs and the ferrets. You can see against a wide panel of strains spanning many years, they don't achieve that 1 to 40 protection for.

The majority of the strains. And that's the whole problem with flu shots.

That's why you update them in the ferrets. Similar thing. They got a. It was a 2007 shot is what both of these guys got. So it protected against 2007.

Absolutely nothing against the rest of H1N1s.

H3 and 2s did a little better. It was sort of up and down a bit where they did pretty well against 2017.

Otherwise pretty rough.

The Habs, they did pretty well, I think better against one of the two types of HAB, but in general a good response.

HABs are more conserved.

So when we put our vaccine in, we see this radical shift in this.

Massive increase in breadth that occurred against.

Every one of the components that's matched adjuvant. And then we found this additional adjuvant.

That increases the titer. It basically cranks up the response so we get essentially universal response. As to the pigs 13 years out into the future, even though through a.

Pandemic shift event similar in the ferrets, we see this massive increase. This adjuvant is my favorite right now. Our breath is really good here, but it doesn't have that crazy titer that we've achieved here.

That's part of the reason why we're so excited looking into LNPs. There's a number of reasons why the RNA technology is exciting to us, but.

One of them is it has this.

Built in immunogenicity property which is attractive.

I'm going to go pretty light on these slides.

I just want to touch upon them briefly.

This slide shows in green, that's our favored formulation compared to blue, which is an almost 2007 flu shot.

And red is our vaccine on the squalene background.

And you see that we perform equal or better on every category. And when we add in the new.

Adjuvant, then the titers go nuts. That also means we only need one shot and one boost and we actually.

Only need the boost for the pandemic strains.

Where you see a flat landing response, a complete failure to be able to cross when the virus changes enough and we're able to overcome it because we're focusing an immune response against these conserved sites.

Similar story, this is called a neutralization assay. It asks a different question, but it's.

Also doing the same thing.

What it asks is if you put the serum from an immunized animal in.

With the virus in a cell it would normally infect, does it stop it from infecting?

And we saw the same result. A bivalent shot protects against the year we immunized, against little to none against the rest. We see a century of protection going.

Back to 1934 and up to the most recent strains tested.

This was independently validated at the work.

At UGA and at Auburn University in ferrets and in pigs.

In live challenge experiments, we did what no other group.

Has been able to do, which is.

Take a 2007 earlier composition and protect.

An animal against 2009 pandemic shift event. That's like the Omicron of influenza.

We protect the animals from all the weight loss, which is when they get sick, they.

They start losing weight.

That's a readout. And then they have a bunch of.

Pathology, you know, little sneezes and sad things that ferrets do when they're sick.

We protect them from almost any symptoms whatsoever. We are also able. That's a 2009. We're also able to protect against 2007.

H3 and 2, which is 10 years into the future of the vaccine composition.

You know, we may need to Update.

Once every 25 years, so maybe a few times a century. But that is a radically superior value proposition than having to update the vaccine two times a year like we've been doing for decades.

When we look into the nasal turbinates, the trachea, the left lung, the right lung, we see for our favored composition.

100 to 1000 fold reduction in virus.

Undetectable viral load in most sites in most animals, although we would always detect a little bit of the virus at.

The site where we inoculate them. So it wasn't sanitizing the initial epithelial layer of exposure.

It was making the rest of the.

Body inhospitable and therefore protecting from symptoms.

This was so good that we believe.

That our vaccine also inhibits transmission. And we have current experiments set up to do that. We didn't do it last time.

And the way these experiments are set up is you expose the vaccinated pigs to the virus and then you go give them a roommate, a little buddy piggy who's not been exposed and not.

Been vaccinated, and you wait to see if the second pig gets sick. And so they're pretty straightforward experiments. And those are being set up now with the work we're doing.

Finally, the intention of our product was to create a seamless replacement for the current flu shot that works on H1, H3, and HAB, except that it works 80 to 90% rather than 80 to 90% of people rather than 30 to 60, and that it provides pandemic protection.

Against H1, H3 pandemics.

Those account for 80% of the pandemics.

Of the last hundred years. And they account for like 97% of the deaths. And that's because H1N1 from 1918 killed 51 million people. It was 3% of the human population.

Back then that said, you know, there's H5 going around right now in birds.

People are always worried about this. I get asked it by government regulatory agencies.

So I said, look, I didn't Put those in the vaccine, but we'll test it. The answer is yes, we see some outbinding again, the favorite adjuvant increases the titers. Those look pretty good to us.

But I'll just caveat by saying we haven't tested them on a neutralization assay. We have a partner set up to do this right now. This is just saying, yeah, we're seeing binding crossing out to the exotic avian subtypes, but we haven't confirmed it's neutralizing binding.

That's just a caveat I'm always very cautious about because I think that was.

A source of false starts that happen in the first 20 years of broad spectrum vaccine design is you'll see binding.

But it's not enough.

That antibody binds, it needs to bind tight enough and at the right site that it neutralizes. And so stay tuned for the update on that point.

And this is summarizing all that work basically by in pigs, in ferrets, now also in mice, by hemagglutination, inhibition by serum elisa, by neutralization, assay by live.

Challenge phenotype, by weight loss, by virology.

By every measure that we have, we.

Have created an unprecedented broad spectrum immune response against a traditionally challenging pathogen that we have struggled creating vaccines against since the 1930s nearly 100 years ago.

That is the basis of the technology. There are unique and clever things, or.

I don't know, we think they're clever.

That we have to do in some pathogens. So coronavirus, we don't have a century of data for.

And so we had to do some innovative techniques to come up with the variant cocktail to work with. And we have an engineering way to introduce mutations for hiv.

It is a less stable trimer than flu. It's like this delicate little petal that.

Just want the petals want to fall off the flower and you want to.

Keep it stable enough in a syringe so that it's able to deliver the.

Appropriate parts of the virus you want to target.

And so we have to do a.

Little engineering to stabilize it.

There are these tricks that we apply. My principle at the company is to use the minimum set of things necessary.

To accomplish the task. So absolute engineering parsimony.

I think that that's important because smart people will sit in a room and.

They'Ll come up with 20 different things that they can do. And secondly, next thing you know, you have a Rube Goldberg machine and something breaks and you don't know which piece it was. So no, don't do that.

But there are some steps you have.

To take in some of these to.

Make sure that we can do it right.

And luckily my group evolved out of one of the world's best antibody discovery groups. And so we're super comfortable doing the right engineering techniques, applying them to enhance the thermostability and the diversity characteristics of these novel folds.

I want to shift gears a little.

Bit to just talk about the consequences of this.

So you have so gearing up this year to initiate the manufacturing the GMP.

Process to be able to get us to go into human clinical trials with our first application, which is influenza, the.

Coronavirus and the HIV are currently in vivo.

They're going to require some more. They're about a year behind. So we're working on both of them.

Okay, let's say we get this out on market. What does that mean? So obviously it's good for patients.

I'd rather take.

I take the flu. Look, I take the flu shot, I'm an asthmatic, I will take the flu.

Shot, I will take the 50% chance of not getting sick. That's great.

I would much rather take a flu shot where I know I have like a 90% chance of not getting sick.

If I take the shot.

That's super useful. This is also good for the company.

That wields the technology.

Right.

So right now I'm showing you is this crazy process we go through every year where the cdc, the fda, the World Health Organization, you know, shoot the.

Racing gun in the air at the.

End of February and they go, okay, here's the strains. Everybody goes. And then all these groups that are.

Producing vaccines have to scramble.

They're doing some work in advance at risk, they're doing some guessing. There's compromises made about what strains they have to go. GMP a new drug in a six.

Month window, which is astonishing. We may normally take much longer to do that.

And then they have to start producing it. It's efficient batches and deliver it out to a market where a flu season is starting by September and it's a drug that only serves that season. And if you have any manufacturing delays.

Which happen, that's a huge risk for that organization.

Because the way this market works, the.

Drug is worth less, the vaccine is worth less. Every two weeks you delay on delivering it to market.

And the drug is worthless by the end of that season because that is the drug for that season. And the same process happens six months later in the southern hemisphere.

So you don't get to rest. You can't just use the northern hemisphere vaccine.

They have to make Adjustments, in contrast, what I've described to you is annoying.

The first time you build it because there's multiple components.

But once you do so, you do.

Not need to change those components.

And that means you have the luxury of behaving like a normal company in biotechnology, to be able to mass produce and stockpile the drug, to be able to have your sales teams out talking six months in advance, that you're hitting those markets and you're soaking up all the deals with security.

You can deliver the product on time, and when the market is at its.

Premium, you can serve the northern and the southern hemispheres. So this is a dramatic advance in.

The stability, notwithstanding the quality of the immune response and the preference organizations or people will have towards getting vaccinated in this manner.

This reduces a lot of the operational.

Risk of manufacture of vaccines.

By the way, this is the exact.

Scenario we're evolving into with the coronavirus where they're talking about doing these bivalent updates.

So that's that.

I wanted to check on time because.

There'S one final interesting story, but I.

Want to make sure we have time for it. Otherwise I could stop right there.

10 minutes.

All right. 10 minutes. Magic. All right.

So right in the beginning I said.

That there was like one exception to the general rule that if you try to give someone a flu shot, 1970s, the 1980s, the 1990s, they don't establish breadth.

And I found this exception and I.

Wanted to talk about it because it.

Fascinates me because I think if you can find what's called phenotypic extremes, unusual.

Immunological phenomenon, I think it gives you insights into all of immunology. And so I go looking for these weird stories and I'm really delighted to share this one with you.

What I was looking for were people that are extreme in their ability to.

Create broad spectrum immunity. And if so, what was different about them and their process and their history that gave rise to it? And what can we learn about that for vaccine science? Science.

And where I'm looking to this problem is an unusual place.

It's not a virus, it's actually anti venom.

So snakes are responsible for probably more.

Death and disability than you might think. It's about 140,000 people a year that die, another 400,000 lose a limb, mean they can't help themselves and their families.

Most of these people are in the.

Developing world where that's a significant hardship for them to be able to access anti venom, if it even exists or is available. It's certainly a hardship for their family. That's already maybe doing subsistence agriculture, and now they have to support someone with a disability.

The reason that this area gets neglected.

Partially is because most of the people who need this are poor. Let's be blunt. We focus more on cancers for rich people.

But the other problem is it's a.

Really hard biological problem.

There are over 600 venomous snakes of them. Each one is injecting you with somewhere.

Between five and up to 70, and in extreme case, different proteins.

So even if you could make the 3,000 to 42,000 different antibodies, you couldn't.

First off, that's an insane task. Like, the entire community hasn't done that. In terms of drugs produced, you can't.

Mix them all together at a dose.

Where each one would be at a useful dose and then deliver it to a person. And that's why no universal anti venom's been created.

Instead, there's over 40 products that treat one or a few of those snakes. And your current treatment strategy sucks.

Right now what happens is you get.

Bit unless you're lucky enough to be.

Where there's only really one dangerous snake in the area, in which case, if they have it, they just give you that anti venom, whether that was the snake that bit you or not.

They go, oh, I'm sorry, I'm sorry, were you bit by like a scary, poisonous, venomous snake? Could you go back and like, look around in the bushes and try to describe to me what the snake looked like that bit you? So we know if you're lucky enough that we happen to have, have an unexpired vial of the right venom to treat you, if it even exists. Like, that's not great.

That's not an excellent medical solution to a biological problem.

What you really want is a single product you can go just slam into.

Their leg or give them an IV and have them be able to be reliably treated.

And it's certainly an unreasonable ask of.

Someone who just got bit, who might be in shock or is stressed out.

And I don't know. My first thought would not be to go examine the phenotype.

If I got bit by a venomous.

Snake, it would be flea, you know? Yeah, so, so what I wanted to point out here, so I started looking.

Around being like, it would be really.

Cool if I could find someone who.

Had some existing antibody immunity because I'd built a shiny technology to pull out unique antibodies from individuals and I was looking for someone to apply it to.

And so I thought maybe I could.

Find someone who works with snakes and had been bit, maybe One to three times like a clumsy snake researcher is what I was basically looking for.

And so I was contacting vivariums and then to my amazement I meet this guy named Tim Friedy. Tim Friedy had spent 17 years and nine months giving himself escalating doses of.

16 different species of venom.

There are 856 self immunizations that he.

Exposed himself to over that nearly 18.

Year period that's recording them roughly grouped according to classes of snakes that he was exposed to every continent except Antarctica.

Which is to say every continent that has snakes.

And yes, those are in hundreds of immunizations. So a radically unique immunological exposure profile from this person. He also took astonishingly meticulous records for.

The entire period, including a self reported discomfort score, a diameter of swelling, the milligrams, the exact species and subspecies like.

This was all recorded. And so at first I was just like oh God. But I got very interested because he.

Had done it so many times and.

You see, he kept all the species going. I felt that he may have overcome.

The failure mode of why vaccinating a 1970s to a 1980s to a 1990s flu vaccine does not produce breadth.

I felt that he may have cycled.

So many times, he may have beat the overfitting problem and actually produced broad spectrum immunity. And that if we were going to find a universal anti venom anywhere, we would find it in Tim Friedi's blood.

And so I contacted him and we spent three months studying very carefully considering.

A bioethical framework to be able to work with him. What we did is a strictly non interventional study where we said look, we just want 20 milliliters of your blood and then 28 days later, 28 more milliliters. We don't want to have any feedback regarding your immunization schedule. Here's some documents on snake venom being dangerous. And he's like.

We got the serum.

That'S all we needed for the approach. We got serum and we got B.

Cells from the serum. We ran Serum Elyses and confirmed that he, he did, at least from the ELISA says he did what he said he did. He had radically powerful serum responses to.

Four different genera of neurotoxins from snakes compared to healthy controls where there's absolutely.

No response in his B cells. We found compared to healthy controls, he had this heavy enrichment of somatic hypermutation.

Which is an evidence of immune mutation on B cells as they improve their receptors.

That was even more enriched when we looked at the universe of antivenom specific.

B Cell receptors that we got through panning full biotinylated antigens.

And then the highest ones are these.

Broadly neutralizing antibodies that were the most.

Mutated, which is what you'd expect.

Every time he immunized himself, the broadly neutralizing B cell would get stimulated. And so therefore it would be expected to be the most mutated.

So this was largely our experimental design. We grabbed some blood from tem, we've created a library and then we searched it against the key toxins that occur in all snakes.

And there really are five key toxins that we think are what the important piece is that snakes are using to kill you or make you lose a leg. The other toxins are homologs or you can walk them off and you don't care.

We would switch out between which species.

Homolog we were working with. So we'd pan first against the cobra.

Version of the the neurotoxin, then against.

The type N, then against the cobra, then against the crate.

And that would select for the antibodies.

That were broad and weed out the ones that hit just one strain.

And that was the process. We chose to go after the alpha neurotoxins because they have, it goes back.

To that plane again.

They have this conserved site on the neurotoxin because it goes and recognizes a.

Super conserved site on your nicotinic acetylcholine.

Receptors that are shared all the way out to fish. And that means a snake can bite a lizard or a bird, a frog.

Or a little mammal, and it can.

Have it for dinner. But it also means that its venom.

Has not been able to change that interface for over 100 million years. And that's where bodily neutralizing antibody, if it exists, it's going to be right there. So it's the same exact story as the flu work and the HIV work, the coronavirus.

We pull out an antibody, there's a family of them with one being super dominant.

And we began to characterize it.

First thing we observed was astonishing affinity of the single antibody against three different genera of snakes. Cape cobras, Black mambas, coastal taipans on Biakor. The affinities, the sequences are somewhere between.

50 to 64% identity of the amino acid level.

So these molecules, almost one out of.

Every two amino acids has changed.

And yet this thing is maintaining picomolar.

Affinity interactions to all three of them.

We built an assay using BLI to be able to characterize rapidly that antibody binding to a whole bunch of venom. We had a panel of 20 different.

Snakes from every continent, except for Antarctica.

Of all the ones that contain, we hit 18 of them. This one's known not to contain the long neurotoxin.

This one's suspected too.

So we either have 100% hit rate.

Or it's 18 of 19. We're characterizing that final point.

This is the range of the toxin.

Species we were looking at with this super high affinity.

We went to go demonstrate using immunoprecipitation pull down. And we pulled down, using that antibody.

From total venom, a band of exactly what the length you'd expect it to be.

Then we, again, using the technologies you.

Use here at Emory, we ran mass spec to characterize and the sequences were exactly the sequences we were expecting.

So we're able to, at a molecular.

Level, confirm the identity of the target of a broadly neutralizing antibody. And we could do that quickly and.

Efficiently in new snake venoms, which is.

Super attractive because it got us out of the cycle of having to figure out how to recombinantly produce each one of these toxins.

It also validated our platform. Peter Kwong's group at the NIH got.

Excited about this project and has been a phenomenal collaborator.

Him and his team, they generated crystal structures. So four crystals of our antibody in complex with the taipan, the cobra, the.

Mamba, and the krait homolog.

These things are like 50, 60% identical. It's binding in an identical mode in all four cases.

Four genera that are spanning three continents.

When you zoom in, you see the.

Mechanism of broad neutralization.

Here is the nicotinic acetylcholine receptor binding.

That conserve site I was talking about.

Then here's the antibody binding that exact same conserved site. It's even carrying about the same residues. There's same amount of buried surface area. And so Tim's friedi's blood, through 18 years of torturous retraining over and over again with these different snakes selected for.

An antibody and systematically enriched an antibody.

That perfectly molecularly mimicked its binding preference.

To that of the endogenous receptor.

And it put the snake venom between.

A rock and a hard place. A receptor that it cannot mutate its.

Venoms or it will not bind evolutionarily. And then the antibody now recognizes that.

Exact same same site.

This was encouraging to us because this reduced the, the idea of an antivenom to a little machine with a crank.

On it where we can go, okay, now we'll just apply our same process to the remaining toxins. And now we have universal antivenom.