Scientist Stories: Scott Hensley, Multivalent Vaccines

Okay, let's go ahead and get started. Welcome to the influenza webinar series hosted by the Human Vaccines Project. The Human Vaccines Project is a nonprofit public-private partnership with the goal of decoding the human immune system to accelerate the development of vaccines vaccines and immunotherapies. Before we begin, we want to remind you that registration is currently open for our upcoming scientific summit on universal influenza vaccines. The meeting coincides with the centennial anniversary of the 1918 influenza epidemic and will feature prominent researchers and thought leaders to discuss cutting-edge influenza research. Spaces are filling up quickly, so we encourage you to register soon. The full meeting agenda and registration link can be found at www.humanvaccinesproject.org/talks. In addition, we are still accepting pre-applications for the 2019 Michelson Prizes. This year, the $150,000 prizes have expanded to include 3 focus areas related to vaccine discovery: human immunology, computational biology and protein engineering, and neglected parasitic diseases. A global search is underway for young investigators who are applying disruptive research concepts and inventive processes to advance vaccine and immunotherapy research for major global diseases. The October 29th deadline is rapidly approaching. We encourage you to visit www.humanvaccinesproject.org/michaelsonprizes to learn more, and please be sure to pass along the information to other interested researchers. And now for the webinar. In order to make real progress toward the development of a universal influenza vaccine, collaboration among researchers around the world is essential. We hope this monthly webinar will foster the scientific foundation needed to advance influenza research. The series of webinars will be broadcast on the third Tuesday of each month and posted online. For a copy of today's webinar, please visit the Human Vaccines Project website at www.humanvaccinesproject.org/talks. Following the presentation, we will have time for questions from the audience. Please type your questions into the questions box into the GoTo interface. I will read these questions following the presentation. And now it is my honor to introduce Dr. Scott Hensley. Dr. Hensley is an associate professor in the Department of Microbiology at the University of Pennsylvania in Philadelphia. His group studies mechanisms that promote antigenic drift of influenza viruses and factors that influence influenza virus responsiveness. With the overarching goal of improving influenza vaccine development. Discoveries out of the Hensley Lab have influenced decisions related to seasonal influenza vaccine compositions. Today, Dr. Hensley will tell us about immune history and immune influenza virus susceptibility.

All right, well, thank you for that introduction. Let's see if I can, uh, Advance the slides here. Oh, there we go. Okay, well, thanks again for that introduction, and thank you for having me on this webinar series. So most of you are familiar with flu, but I thought I'd spend a few slides here in the beginning to introduce the virus and some of the challenges that we're faced with, with making a vaccine against against these viruses. So influenza virus has two glycoproteins, hemagglutinin, shown here in green, and neuraminase in yellow. Most neutralizing antibodies target the hemagglutinin molecule, and they prevent the virus from replicating by blocking a virus attachment to host cells. And some of these antibodies will actually bind and prevent the fusion of the virus membrane and the cellular membrane. There are some antibodies that target NA that can reduce disease severity, and it's thought that most of these antibodies bind to NA and prevent virus release. But today I'm going to be focusing mostly on HA, or the hemagglutinin molecule. And you can see here on the right, hemagglutinin can be broadly split into two domains: the head domain as well as the stalk domain. Okay, so let's just move on. So there are many different flu viruses, as many of you know, many different subtypes of viruses. These are all in animal reservoirs, and occasionally they jump into humans and cause pandemics. Shown here are the different subtypes that have caused pandemics over the last 100 years. We all know about the H1N1 virus of 1918. And that was then replaced by the H2N2 strain that first emerged in 1957. And then in 1968, we had the H3N2 virus come into the human population. And we all remember 2009 when the H1N1, so-called swine flu, took hold. Now, there's a couple points I want to make about this slide. Number one, when these pandemic viruses jump into the human population, it's not like they just go away, okay? Cause a pandemic, they infect a lot of individuals, there becomes a lot of herd immunity against these viruses, but they don't disappear. And the reason they don't disappear is because they gradually acquire mutations in the hemagglutinin and neuraminase proteins, which allow these viruses to then reinfect individuals that were previously exposed. Okay, so we have these viruses, they cause a pandemic, they don't go away, they just constantly change and reinfect us year after year. Now in 2018, what are we dealing with? We're dealing with this H1N1 virus that is really our seasonal H1N1 virus now, and we also have H3N2 viruses, which again have been circulating in humans for 50 years. Now we also have influenza B viruses, which I won't talk about too much today, but our vaccine that we receive each fall is comprised of antigens from one H1N1 strain, one H3N2 strain, and one or two influenza B strains. Now, this is a nice phylogenetic tree that, that demonstrates the point that these viruses are constantly changing. Each circle there is a viral isolate, and you can see on the x-axis, this is time, And so what are the points here? The points are that this is a phylogenetic tree showing that there has been a lot of genetic change in H3 hemagglutinin since 2006. And each one of these X's, well, this is a vaccine update. So the virus is constantly changing, as you can see here as we go up this phylogenetic tree. And what we do as a community is we, we are constantly updating our flu vaccine strain shown in each one of these X's, with the goal of matching our vaccine strains to the strains that are circulating. Of course, this is a challenging process. This is a slide taken from Trevor Bedford that demonstrates some of the challenges with, with kind of playing catch-up to make these seasonal vaccines each and every year. So in the northern hemisphere, we're getting ready for, for our flu season,, and we will be isolating flu viruses. Typically, the peak of the flu season in the Northern Hemisphere is in December, January, more January and February, but that can vary from season to season. And here's sort of the remarkable thing of this is that we actually will choose the vaccine strain for the following year here in February. Okay, so viruses are isolated during the winter, We choose the vaccine strain in the winter, and then the reason we have to choose so early, because look at all these things. We got to make high-growth reassortments. We got to make the vaccine. We got to distribute it for it to be available in October of the following year. So the vaccines that we're getting, you know, over the past few weeks and the population will continue to get leading up to this year's flu season in the Northern Hemisphere, these vaccines were basically, the strains to be included in these vaccines were decided upon in February earlier this year. So it's a pretty long process, and sometimes we choose strains that are antigenically matched. Sometimes the viruses actually change so that these vaccine antigens are not well-matched to the strains that actually circulate. Okay, so here are my two kiddos, and this slide is actually a few years outdated, but what I'm going to show you today is that immunological encounters early in childhood impact how you respond to viral strains later in life. And I love this slide because my kids look so mischievous in this slide, right? My daughter Hope has paint all over her hands, Her brother at this stage is, I don't know, he's probably 2 at this point. He has paint on his hands and he's holding the paint. So, you know, when we're children, I mean, you can see it here in this slide, you know, these early things that happen really impact, you know, all aspects of our life. You know, we read to our kids every night. Why do we do that? Because, you know, it's been shown that if kids get books read to them early in life, that they are some sort of, some kind of like hardwiring that they can actually see letters and read later on in life. And the same thing is true with their immune systems. Kids are infected, as shown here in this next slide, by the time that they're 3 or 4 years old with flu. Many kids, at least in our country, are vaccinated now, but this is looking at seroprevalence against H1 or H3 over time since birth. And so you can see that kids are born with antibodies against these strains, You know what these are? These are maternal antibodies, right, from mom. They decline over time, and then look what happens. By year 1 of age, yeah, about 40% of people are seropositive. By year 2, it increases. By the time we're 3 or 4 years of age, we've seen flu, we've been infected with flu. And like I said, in our country, you know, 6-month-olds are getting vaccinated. And what I'm going to show you today is that these very early exposures, again, influence how we respond to flu viruses 50 years later, okay, which is remarkable. Now, what happens during these initial infections? These initial infections prime B-cell responses, okay, so this is sort of an old-school immunology slide that I like to show during my talks, and so on the y-axis are number of antigen-specific B-cells, so in this case you can think of these as number of flu-specific B-cells, You can see that before antigen exposure, there's not a whole lot of these, and that's the beauty of our adaptive immune system. We have many different B cells that can recognize anything. You can throw anything at our immune system and we will mount an antibody response against it. Now, since we have this capability of mounting antibodies against essentially anything, then we have— before we actually see the antigen, we have very few of these against each specific type of antigen that we can encounter. So there are very few flu-specific naive B cells before antigen encounter, but look what happens upon, say, an infection early in childhood. Well, these guys explode. The numbers go up. You get something called class switching affinity maturation. There's actually mutations that go into these B cell receptor genes which make the antibodies actually bind to the antigen with a higher affinity. And here's the important part. You can see over time, which is shown on the x-axis, these, the B cell numbers decrease, but what you're left is with this memory B cell pool. There's two things that are important about this memory B cell pool. Number one, there are more of these guys than before the antigen exposure, so there's, there's an elevated number of these memory B cells. But perhaps more importantly, these guys are, are just ready to go. Memory B cells have a low threshold for activation, and when they see antigen again, they can dominate the response. They can actually outcompete new naive B cells, and I'll show you examples of that today. All right, let's go back to this slide. I told you that we have been— we've had different flu pandemics over the last 100 years. I kind of muddled through this slide, I think, the first time I went through it, But now I want to come back to it because this is important. I told you that we're all infected by the time we're 3 or 4 years of age, and that when we're infected, you get this explosion of memory B cells. Well, everyone here on the web can basically look at your year of birth, and you can almost guess what virus you were imprinted with or infected with early in childhood, because what we're learning now is that the specific type of virus that you first saw in childhood dictates how you respond respond to flu strains today. And so if you were infected with H3N2 early in childhood, you probably have a different memory B-cell composition compared to if you were infected with H1 viruses as a child. And of course, it gets more complicated than that. It's not just H3 versus H1, it's the specific type of H1 or the specific type of H3, because remember, these guys are changing all the time. So I just want to summarize some, some data that we have generated over the last, I guess, 8 years now on H1N1 viruses. I told you the pandemic H1N1 strain of 2009 caused the pandemic, and then it continued to circulate, and it continues to circulate today. What we found very early on in our experiments with this virus is that antibodies elicited in a naive animal have a different specificity compared to antibodies that are elicited in different age humans. So let's look at this example here. There, this is— pretend this is HA with 3 antigenic sites. You have a red antigenic site, a green antigenic site, and a blue antigenic site. In this example, antibody— if you infect a ferret with this virus, you have antibodies, these red antibodies, recognizing this red antigenic site. If you infect, in this case, a middle-aged individual with a virus, you don't have these red antibodies, but instead you have these green antibodies recognizing the triangle. If you take someone that's a bit older, you get not red antibodies, not green antibodies, but instead blue antibodies against this site here. Now what's, what's going on? I think I've spoiled my punchline. This is because of prior immunity. Now what do we do when we do animal research? We make sure that we use animals that have never seen flu before, So if you have a primary response, in this case in a ferret, you get antibodies that tend to react to these highly exposed epitopes on top of the HA. But if you take someone like Yang Li, who was my first grad student, I guess he joined the lab 8 years ago, he's been gone for quite some time at this point, but if you take someone like Yang Li, well, he was imprinted or he was infected early on with a virus like this. This is a 1983 strain, but let's look at the similarities between the '83 strain and the 2009 strain. The circle is black here, it's red here. Hopefully my pointer is, is tracking okay. You get this guy, it's black here, blue here. The only conserved epitope is this green epitope. So the model is, is that Yang probably had an antibody response against this 1983 virus that might have been skewed towards one epitope or the other. And even if this was a subdominant response against this triangle, well, when he then encountered the 2009 HA, that antibodies against that triangle were the only conserved epitope. And so the model is, is even if you have a subdominant response against one of these epitopes during a primary response, if you're then exposed later in life with a virus that's quite different but has that epitope intact, that you'll have this preferential recall of memory B cells that go on to produce antibodies against that conserved epitope. Now, what's going on with Dr. Fauci? Well, he's not the same age as Yang, and so he was probably exposed to a virus like this. Same idea: circle black, circle red, triangle black, triangle green. The only conserved epitope is this blue epitope. So depending on which virus you saw in childhood, you're going to have sort of a different memory B cell repertoire. You're going to have different memory B cells that can be recruited, and now you can be infected with the same virus as your father or your son, and you're going to have a different antibody response because you have different, uh, you have, you have different memory B cells that will recognize the different conserved epitopes depending on your year of birth. Now, I will say that in this case, Yang and Dr. Fauci, these antibodies were a good thing, okay? Recalling antibodies, as long as they're neutralizing, recalling memory B cells that produce antibodies, as long as they're neutralizing, that's a good thing. Where it can become problematic is if you have such a skewed antibody response that mutations can then arise in those epitopes, which then can block the antibodies from binding, and I'll show you an example of that. I better move along here. I'm carrying on too long, I think. So here's, well, here's what it looks like structurally, okay? And you see the citations here. So 1983, this is what HA looked like. 2009, this is what it looked like. And red shows residues that are different. Try to find an epitope that's conserved, because I just told you if you're primed in 1983 and now infected again in 2009, you should have a skewed response against conserved epitopes. And sure enough, well, I see one right there in green. And sure enough, if you look at people that were born in the late '70s, early '80s, they tend to have a skewed antibody response against that, what I have highlighted in green. And that's because, again, that's one of the few epitopes that are conserved between these late '70s, early '80s virus as well as the 2009 virus. Again, these antibodies were awesome. They helped people in 2009, but here's the problem. Get ready for this animation. It's going to blow you away. Boom! When a mutation then arose in this virus, which, which took place actually in 2013, now these antibodies became useless. Okay, so having preferential recruitment of this antibody response, this very narrow epitope, it was a good thing. It protected people in 2009, but it sort of came in at a cost. Everyone kind of put their eggs in that basket, the people that were like middle-aged individuals born in the '70s and '80s, and now Later on, what did H1N1 do? It acquired mutations right in this epitope, and then these individuals became susceptible again. This is, this is a nice schematic from Jesse Bloom's lab showing what this actually looked like in the example I'm telling you here. So in, in 2009, there was a lysine at this residue 166, which is smack in the middle of that green area that I showed you in the slide before.. And now during the '13-'14 season, look what happened. A K166Q change occurred where the viruses all that were circulating that year basically had this mutation. This mutation became fixed, and concurrent with that, middle-aged individuals became susceptible. And again, that's because all their antibodies were focused on this epitope involving 166, where there had to be a lysine in that position for their antibodies to be effective. This is a collaboration with Arnold Monto's group that demonstrates how important this is. So in this, in this, uh, in these experiments, Arnold Monto sent us serum, um, and from, from a cohort that he had established. What Arnold's group did was he bled almost 400 individuals prior to this '13-'14 season. 20 of these people just, just got infected with H1N1 over the course of the '13-'14 season. So this was a great cohort for us to look at because we asked a simple question: Did these 20 people have preseason antibody titers against the vaccine strain but not the circulating strain? Did these people have antibodies that could recognize the 2009 pandemic, but they were so focused in a way that they couldn't recognize this drifted 13-14 virus? And so we, we did hemagglutination inhibition assays, an assay I like quite a bit, and we did it with a wild-type 2009 virus or a virus that was engineered to have this single mutation, K.1.66Q. And we put people into 3 different groups: those individuals that did not have antibodies against either virus, those that had antibodies against both viruses, and here's the important group, those that had antibodies against the wild-type virus, but they were so focused that they couldn't recognize the 1314 virus with this single mutation. Who were the people that were infected? Well, this is PCR-confirmed infection. You imagine, and this is what you would expect, people that didn't have antibodies they were infected. There was 202 of those individuals and 15 were infected. The people that had antibodies against both strains, they weren't infected, they're protected. And here is the important group: 41 individuals in this cohort had antibodies that, that could give an HI, hemagglutination inhibition, to the wild-type strain but not the mutant strain. And guess what? If you were in this group, you had just the same chance of getting infected as if you had no antibodies at all. This, I love this study. I think this is still one of my favorite studies because number one, it only took us 2 weeks to do the study. It took Arnold and his colleagues, I'm sure, a lot longer to collect the samples. But number two, Arnold sent these samples blinded to us. We did these assays without knowing who was infected, and you can see that susceptibility is really tied to antigen or antibody specificity. And this is a great example of how immune history really can affect your susceptibility. So if we go back to this slide, Yang and Dr. Fauci— I should call him Dr. Li because he's a PhD now— Dr. Li and Dr. Fauci, they have different antibody repertoires, and those antibodies are different than naive ferrets, and they're very protected against the 2009 strain. But look at this, when a mutation comes in, as this asterisk indicates, Then you have differences in susceptibility in distinct aged populations. In this example, Dr. Fauci is still protected. He has blue antibodies, but now these middle-aged individuals, Yang, Li, and company, they are susceptible because those green antibodies no longer work. I'm going to shift gears and talk about H3N2 viruses, and then I'm going to try to race through this so I can talk about some new unpublished data. So the H3N2 vaccine has been not that effective over the last few years, and I think most of us are aware of that. But something that we find very interesting is it has been effective in one age group, and that is the very young. So this is looking at preliminary vaccine effectiveness from last season, and you can see that young individuals here defined as, excuse me, people that, I don't know, are 8 years old, 9 years old, or younger had about 50% vaccine effectiveness against the H3N2 strain. And we get excited whenever we see a cliff like this, because if you are 9 to 18, there's no vaccine effectiveness and very poor vaccine effectiveness in older populations. So what could be going on here? Is childhood imprinting involved? Is this because somehow individuals that had seen H3N2 strains in the past are responding inappropriately to the new vaccine. That would be the hypothesis. And so I want to tell you about this, this study from Seth Zost. We'll call him Dr. Zost since he just got his PhD last week, and he's going to actually move down to Nashville to be a postdoc with Dr. Crowe, which should be, which should be pretty amazing. So what did Seth do in this study? Well, what he focused on was this mutation at residue 160. I don't want you to get too bogged down in these residue numbers, but this one's an important one. You can see that H3N2 viruses had a lysine at residue 160, and during this '14-'15 season, this guy went away because essentially all the viruses acquired this K160T mutation. Now, why is this so important? Well, it introduced a glycosylation site shown here in blue, and it wasn't just any glycosylation site. This glycosylation site was right on top of HA. It was in what we call antigenic site B, which is a really important antigenic site in the case of H3HAs. Seth did the biochemistry. He introduced a single mutation through reverse genetics and made viruses that had either the T or K at residue 160. And he found that this K to T change really does result in a glycan addition. As you can see, the HA from this virus migrates slower compared to the HA that has the K. And if you use PNGase treatment, they migrate at the same, at the same level as you would expect. Okay, so in 2014-15, we had a bad flu year. The vaccine didn't work very well, and it didn't work very well because of this addition of this new glycan that was smack on top of HA. You couldn't get a bigger antigenic change than that. This is a slide that we can go over fast, but I'll tell you, I really like collaborating with Dr. Patrick Wilson from University of Chicago. He has a lot of monoclonal antibodies that he has shared with us, and I think this is a good example of how important this antigenic change was. This is a single monoclonal antibody. You can see it binds to a 2009 HA, and look what happens. It doesn't bind at all to the 2014 HA. Well, it binds very poorly, And now if we revert that T back to a K, this, this antibody has perfectly fine binding. Okay, this glycan, this single glycan, had a profound antigenic change. We tested many different antibodies with Dr. Wilson's group, and you can see many had this profile. They all bound, in this case, to the 2009 HA. These antibodies were derived in 2010. They did not— most of them did not bind to the 14-15 strain. But if we reverted that glycan back to what it looked like prior to 14-15, you can see a handful, more than a handful, a lot of those actually down. Okay, this was a major antigenic change. Now, what if— so, okay, well good, let's just update the vaccine to have one of these strains. These viruses are called 3C2A viruses, and that's exactly what happened. We updated the vaccine strain with this 2014 Hong Kong strain. It looked like a perfect match. But here's the problem. When we grow these viruses in eggs, and in particular these H3N2 strains, they grow very poorly. And so even though we, you know, we, I'm saying we as the community, chose the right vaccine strain to update the H3 strain, it really changed as it became a high-growth reassortment. And why am I so fixated on this? Because guess what? This guy had this glycan. It had T160HA. And upon egg passage, it spit it out. It reverted back to what the virus looked like in essentially 2009. So the glycosylation was in all of the circulating strains. It has been for the last few years for H3s, yet our vaccine strain doesn't have this site B glycosylation. This is a problem. Oh, so this is the slide that everyone likes. So let me get— I think I have the light on. You can really You zoom into the visual here, if you have the webcam on, you can see I have a lot going on in the sides but not so much on top. Well, this is a good analogy. You can think of me, my head, as the vaccine strain, and the circulating strains have this big fluffy glycosylation site right on top. And so you would expect that if this is the vaccine antigen, you might have some antibodies on the top of HA, that failed to react to the circulating strains, and that's essentially what we found. So here's Seth's work in ferrets. So he infected— ferrets were infected in this case with the vaccine strain, and then neutralization assays were done either using the vaccine strain, the one that had K, or the circulating strain, the one that had T. And you can appreciate that these antibodies are highly reactive to the vaccine strain, but they fail to neutralize the circulating strain very well. And then this is something interesting that we continue to follow up on, that if you, if you do the reverse, if you infect ferrets now with the glycosylation strain that has this glycosylation site at site B, you get antibodies that recognize the glycosylation, the glycosylated H3, as well as the egg-adapted H3. And that's consistent what we know about glycans. They basically shield the antigenic site. So the antibody repertoire elicited against this glycosylated HA is likely directed against epitopes lower on the HA. All right, what about humans? I gave you the example with H1N1 showing you that H— that humans and ferrets can have different antibody repertoires. Well, most of our vaccines are produced in eggs, but not all of our vaccines are produced in eggs. We have some of our vaccines, a very small percentages are produced in cell culture or through recombinant protein technologies using the baculovirus system. In 1617, the egg-adapted, this K160, was of course in the egg-adapted or egg-grown vaccine strain, but in 1617 it was also in the cell-based vaccine. That didn't change until last year, but the baculovirus was interesting in that it still had— it matched the wild-type sequence. It had the glycosylation site still intact, and so the vaccine, if it was prepared in during the '16-'17 season, either in eggs or in cell culture, it was expected to be mismatched essentially, whereas the vaccine, if it was prepared in the recombinant system, we expect that those vaccines would be a better match because the glycosylation site is intact. Granted, glycosylation machinery could be different in insect cells. So we— this is a study, a collaboration with the University of Rochester, where different individuals receive these 3 different vaccines, and I'll give you a little cheat sheet here. So Flublok is the one that has the glycosylation site intact. Flucelvax is the cell-based one that doesn't and Fluzone is the egg-based one that does not have the glycosylation site intact. And this is looking at antibody responses, fold changes against the glycosylated virus. And I think what you can appreciate is that the FluBlock one, the one that has the glycosylation intact, that there was more seroconversion, a better antibody response against the circulating strain when using the antigen that actually had the glycosylation site intact. Now when we look at antibody responses against the egg-adapted strain, there's no changes at all, and that is because most of us already had immunity against this strain because the strains with this epitope had circulated for quite some time. Okay, so if you look at the past couple seasons, you know, we can go back to this, you know, here are two different individuals with, in this case, antibodies, a pre-existing antibody repertoire, some that are blocked by the glycan. Some of those antibodies are not. Now imagine that individual 1 gets an egg-grown HA, whereas individual 2 gets the baculovirus HA. You can see the notable difference here that the baculovirus HA has the, the blue glycan, whereas the egg-grown HA does not. What happens if you get the egg-grown HA? Well, you're just going to perpetuate this, uh, the memory B cell response that has already been laid down that recognizes these, that is producing antibodies that recognize this exposed epitope on site B, and this is going to have very low neutralizing antibodies, low neutralizing antibody capability against these circulating strains. However, if this glycan is, if this epitope is covered up with the glycan and now your individual 2 gets this antigen, you can see you would expect now that you are going to preferentially expand B cells that are producing this antibody type that was subdominant prior to vaccination, and we expect these antibodies will be more neutralizing. Okay, but I want to revisit this idea of immune imprinting, and so again, coming back to what I've explained is why a lot of adults have, you know, a poor response, but what about those kids? Why do some of these young kids actually have decent vaccine effectiveness against these egg-adapted H3N2 strains. And, you know, so here it is again, just looking at the numbers. Again, just to remind you, kids that were 9 years of age or younger, and it would be really interesting to see this data not grouped like this, but rather just grouped by single year of birth. I would love if anyone that has these data, they could email me that, that would be great. But young kids have vaccine effectiveness, right? Well, here's why we think that's the case. We think it's the case because these people that are older, including anyone that was, you know, 10 years of age, well, they've been imprinted with the strain that essentially looked like this egg-adapted strain, because this egg-adapted strain did not make the virus look like it was from outer space. It made it look like a 2009 virus. That made it look like a 2010 virus. It made it look like a 2012 virus. And so if you look at this residue 160, okay, here's the T. I told you this T is the one that introduces this glycosylation site, okay? Most of the viruses today have this, and this, this was a recent thing. This happened in 2014-15. But if you were born, say you're really old and you were born in 2012, That's a joke for my daughter. If you were born in 2012, then you were probably imprinted or infected with the virus that had a lysine at this position. And it's expected that if you were first infected early in life or vaccinated, that you probably have a memory B cell repertoire that is favored against this egg-adapted strain, and that if you receive this egg-adapted strain, that those memory B cells would be the ones that would expand. And so we think that you have high vaccine effectiveness for, for this population. These are individuals that were either not infected yet, not exposed yet, or maybe infected with a 3C.2A virus that has circulated since 2014-2015. In that situation, you'll have memory B cells, we expect, against other epitopes because this glycosylation site covers site B. This is exactly what our lab is doing right now. We are modeling this in ferrets, and we're really excited about this possibility. I look forward to sharing this data soon. So this is a study we did in collaboration with Jonathan Grad and Sarah Kobie, and I think it supports it. You know, we published this and some people said, well, egg adaptations are not important anymore, what's going on? I think this completely supports the hypothesis that I put forth in that previous slide. And in this study, we looked at the 2012-13 season. In this case, There's an egg adaptation that we identified at residue 156. Man, this guy is right on top of site B. We would think it's important. And when we, we tested, when ferrets were infected with this virus, we found that this egg adaptation really does matter. But when we looked at humans, we found that most humans that received this egg-adapted strain had antibodies that can recognize the circulating strain as well as the egg-adapted strain. Well, why is that the case? It's because of immune history. If you look in this case, this egg adaptation didn't revert site B to look like some historical strain. It made it look like a virus that humans had never seen before. And if you look at residue 156 identity, you can see it was H in this case, but, and it made it look like a Q. But look, there's just a little sliver of the population that probably had seen the Q virus in the past. And so this egg adaptation probably did not have as great of an effect because the population didn't have memory B cells against the egg-adapted strain. Um, and that's written here. So the equation really is egg adaptation plus memory B cells that recognize that egg adaptation equals something bad. Okay, if you have memory B cells that can get recruited and you recruit them through an antigen that's mismatched, in case— in this case, an egg adaptation with the current H3 strains, well, that's a bad thing. I think it's interesting because you could think now maybe we will strategically engineer egg adaptations to look like, you know, viruses that we've never seen before. That's, that's interesting, right? We can make them look like they're from outer space in site B. Of course, that would be an antigenic mismatch, but it would, it would preclude, or we hypothesize that that would not allow the recall of memory B cells against that foreign epitope. So I am running late. So here's some unpublished stuff. So can we make a better vaccine? You know, we're interested in the stock antibody responses to the stock. I think most people are aware of this. The stock antibodies work through different mechanisms. They prevent the fusion of the virus membrane with the cellular membrane. A lot of them work through cellular processes, ADCC, where they recognize HA that's expressed on the cell surface. You know, these antibodies are good. I'm going to skip through this slide. This is what a classic study from Patrick Wilson's lab, you know, first identifying these, and that a lot of these antibodies are elicited against the 2009 virus. Here's a head antibody. You can see it, it protects in a passive transfer experiment against pandemic H1N1, but it fails to protect against antigenically drifted H1 strains. Here's a stock. These stock antibodies are great because they're broadly protective. You can see they recognize not only, in this case, the 2009 H1N1 strain, but these drifted strains. But I will note that they're not quite as awesome, right? That you, at least in this example, and I think we need a systematic study to look at this with a lot of different monoclonals, but it seems that, you know, although they're broad, most of them, this is like very broadly speaking, most of them are broadly reactive, but not as potent. I think that's true, but I think we need a study that actually looks at that. So we wanted to ask, are these antibodies associated with protection in human? And again, this is another collaboration with Arnold Monto's group, uh, and what we looked at in this case, this is, uh, these are donors, uh, that were hospitalized against H1N1, okay? And these donors were hospitalized during, I believe, the '15-'16 season, and, uh, 62 of them were infected and 117 were in the hospital for other reasons, other respiratory illnesses. And so we asked simple questions: are head antibodies enriched in this group or in the, in the non-infected group, and stock antibodies and so forth? And so here are head antibodies, and you can see the head antibodies, well, there are more of— if you do a logistic regression there, these head antibodies are associated with protection. There are more, uh, higher levels of head antibodies in the uninfected group compared to the infected group. And if you do, if you look at stock antibodies, yeah, and you do the logistic regression, I'll show you the numbers on the next slide. These are also associated with protection, albeit to a lower extent. Now, these are the same data, stock antibodies on the x-axis, HI titer on the y-axis, and what you can really see is in the uninfected individuals, I should have a label here, the uninfected individuals are gray, the infected are red, You can see that the gray dots, if you go above a certain HI level, that the gray dots all reside here. Whereas if you look at the stock antibody response, well, they're sort of interspersed. Yet if you do the analyses, so if you look at are HI antibodies associated with protection, and you just take into account only HI antibodies, you can see that there's about a 24% reduction in susceptibility for every twofold rise in HI titer. If you look only at stock antibodies, there's about a 15% reduction in infection for every twofold rise in stock antibodies. But now, you know, it's really hard to sort of look at the independent effects of these guys, uh, with the population size that we have. If you take into account HI and stock antibodies, uh, the, the effect size is about the same. In the case of HI antibodies, it's about a 20% reduction for every twofold increase in HI titer. And, but if you look at the stock, the effect size is about the same, about a 10% reduction, but that does, that fails to reach significance. And again, this is, this is probably just due to our, to our sample size. So, you know, taken together, it really does look like head and stock antibodies are both associated with protection. Head antibodies probably a bit more, but stock antibodies do appear to be associated with protection, although our, our population size of only whatever it was, 150 is probably a little too low to look at the interdependence of these. So we've done passive transfer studies. So we've tried, we've transferred these human serum into mice that have been engineered to have human Fc receptors. If you transfer PBS, you can see the mice lose weight after challenge, in this case with H1N1. We challenged, in this case we've passively transferred serum from humans that were uninfected that had this HI high, hemagglutination inhibition high phenotype. You can see that these are very protective. And so here's where it gets a little interesting. Now we take the uninfected individuals that had HI low, but they had stock antibodies. We still get some protection, and that protection is actually better than when we do the passive transfers with serum from infected individuals that also were HI low. So we think that these stock antibodies are worth pursuing, and the question is, how do you induce high levels of them? And I'm going to just rush through this because I want to leave time for questions. Well, there's a lot of promising strategies, right? These chimeric HAs look great. The sequential vaccinations with these chimeric HAs that were developed at Mount Sinai. The headless HA, I think, is a really cool idea that the antigens developed by the VRC. We've been working with this modified RNA vaccine, and it's sort of a bold title there, right? The influenza vaccine of the future. These RNAs, they're lipid, and then they're RNA, in this case expressing hemagglutinin, and they've been engineered, they've been modified in a way that these RNAs are less immunogenic. And you might think to yourself, Scott, that's a terrible idea. Why would you ever modify the RNA in a way that it's not very immunogenic? Well, guess what? If you put an RNA into a cell, the innate immune system gets so revved up that you get no protein expression. And so if you modify these RNAs in a way that they sort of slip under the radar and cannot be sensed by the innate immune system. What you get is you get long-term antigen expression, and what we found in the case of RNAs that encode HA, a great immune response. This is a paper published earlier this year in JM showing the longevity. This is a single injection of an RNA expressing HA, and, and we did HI titers using sera collected from these mice over time. Here's a conventional vaccine, and I mean, you know, look at these titers, 5,000. Unbelievable. So, you know, as you might expect, these vaccines protect. Mice don't lose weight. They survive. But this is something that just is coming out now. I guess it's published, where this is a collaboration with Drew Weissman's lab. All this is a collaboration with Drew Weissman's lab. And in this case, we looked at the induction of stock antibodies by these vaccines. So this is a little hard to get, get through, but let me walk you through it. So black is day 0 after a vaccination, orange is day 28, red is 238, sort of random numbering there. So this is PR8-HI titers, okay, and so if we inject, in this case, a single injection with PR8-HA, that is a modified RNA, and we complete HI, you can see there are incredible incredibly high HI titers, consistent with what I showed you in the previous slide. But the cool thing is, is not only are there HI titers, there's very high levels of stock titers too, and that's shown here. So in black is, uh, so this is looking at ELISA binding to HA, and in this case it's, uh, I forget this example, I think this is a chimeric HA. This is either a chimeric HA coated on the ELISA or a headless HA. The point is, is you get a lot of antibodies and at the same level of total antibodies at day 28 and 238, but now if you look at stock antibodies, here's day 28 in black, and by 238 you actually have more of these, okay, and that's probably because they're affinity maturing. You get long-lived stock antibody responses, and it's not as if this vaccine is just awesome at eliciting stock responses. What I think is it's awesome at everything. It just, the antibody response is just so high that you get a lot of HI antibodies, but now you're getting stock antibodies that are at physiologically relevant levels because you get such an induction of the total antibody response. We're still trying to figure this out. This is very encouraging because you could see here in this case the antigen was a California 2009 HA, and as you would expect, if you vaccinate mice with this or just the conventional vaccine shown in black, you get protection against California. But the interesting thing is now we could do sort of an infection with a different virus, in this case the PR8 H1N1 strain, which is different than this. And you can see in this case the conventional vaccine doesn't protect against this drifted strain, but these, the modified RNA expressing the California HA is protected. Okay, I think I've reached the end where we still have 10 minutes, so let me run through the main points and acknowledge all the people in my lab and our collaborators who have done the work that are presented today. Early childhood immunological imprints shape the specificity of antibody responses against new flu strains. This is clearly the case with H1N1, and we continue to identify interesting H1N1 specificities every day. And it appears to be the case with H3N2, and I really think this is tied towards not only the H1 or H3 response, but also against pandemic strains. And so we're all familiar with this classic paper, instant classic, from Katie Gostik and colleagues that shows that H1 or H3 imprinting is associated with protection against either H5 or H7, and we have some really cool studies looking at that as well. So, you know, it really matters what you see in childhood. Egg adaptations are our problem. Of course, we should continue to get vaccinated. That should go without saying, because even this egg adaptation is a big problem in that we have a mismatch in a key antigenic site, but What we're learning is that even when these— we have these vaccine mismatches, that although that may— these vaccines may fail to prevent infections in certain individuals, they are most certainly producing or eliciting some sort of protective immunity that will likely keep you out of the hospital, for example, and will reduce your disease severity. But it sure would be good to get away from that egg adaptation, I will say, and I think You know, what we have to do is just, you know, I would advocate for expanding the baclovirus-produced HA, expanding the capability of getting that thing around. So, okay, so what else do we got? I think the head and stock antibodies is very interesting. These cohort studies, they both look like they're associated with protection. Again, larger cohorts are probably needed to look at the interdependence. And I'm really excited about these modified RNA vaccines. You know, let's see when they get to humans, right? And I guess some of them are starting to get in humans now, but If the response that we see in humans is even close to what we see in mice or ferrets or rabbits, you know, the sky's the limit for these guys. I think they're very exciting. So I'd like to thank people in my lab. Dr. Zost, recently crowned PhD from my lab, did the H3N2 work. The H1N1 work was done by grad students Ben Chambers and Susie Linderman, as well as Seth, Ben, and Susie, and Yang Li. They're not on this acknowledgement slide because they've already moved on, but they probably should be. They, they were terrific grad students early on when I first started my lab. Shannon Christensen has spearheaded the stock project in our lab, and Kayla Parkhouse has really driven the modified RNA vaccine project. And I have my collaborators listed here. I try to acknowledge them as I went through. Drew Weissman is sort of the inventor of all this modified RNA stuff, and we are just sort of the flu people, I guess, that are, that are working with him, and it's been a really exciting collaboration. Sarah Kobe, we work with all the time. She's crucial in helping us analyze some of these larger datasets. I mentioned Arnold Monto's group as well, and Josh Petrie, as well as Emily Martin. Josh and Emily are just fantastic, and Arnold's fantastic. They have the HIVE cohort, they have the Haven cohort, and I think everyone should fund those cohorts because those samples are just so crucial. Not only to our lab, but to a lot of other labs across, across the world. Patrick Wilson has been a close collaborator making monoclonal antibodies that we've used. John Treanor orchestrated that, the Rochester, University of Rochester vaccine study. And Jesse Bloom is a close collaborator that helped us figure out exactly when that K.1.66Q mutation arose, and we're actively working with his group on a number of projects. I'm sorry I took so long. I think we have a few minutes for questions. Thank you for your attention.

Thanks, Scott, that was excellent. We are taking questions now, so just go ahead and type your questions into the question box in the GoTo interface. And we already have two. So the first one is, did the pre-1978 H3N2 strain contain the same glycosylation site and lost it, or was no glycosylation site present in the earlier isolates?

Let's see if I could— I go back in slides. The glycosylation site, so I'm not sure if I'm understanding the 1978. So there's— I was talking about the H1N1, the H1N1 story involved about that time period. The H3, this is a question about H3, right?

Is that right?

H3N2? H3N2. Yeah, so H— that glycosylation site that I told you about with H3N2, that is a new glycosylation site that, that first arose during the 14-15 season. I don't know if I can go back, but yeah, I'm not sure what the question is. So the glycosylation site first arose during the 14-15 season, and what we think— here, I think I could go back here. This is probably close to it. So okay, yeah, I think they're— oh, I see, this is what people are asking about. So there was a T here, but this was not a glycosylation site early on because glycosylation sites involve an N at position residue 1, an X could be anything except proline, and then an S or a T in the third position. So in this case, in 1968 through '78, there wasn't a glycosylation site because there wasn't an N at residue— at the first residue. Okay, so there wasn't the glycosylation motif. That's a good question. I— sorry, it took me a little while to figure that out. So yeah, these individuals were primed with a virus that basically had a naked site B. It didn't have that glycan intact, and that's again why we think that they probably respond not as well as individuals that sort of had an immunological blank slate going into last season.

Okay, the next question is, does imprinting also influence the response to the stock? And then a follow-up to that is, why can't you imprint later when the first exposure was group 1 and the later exposure was group 2?

Yeah, so very good questions, and we are actively looking at this right now. And, you know, you look at Katie Gostic's paper, uh, and it's certainly— and everyone should go look at that, Gostic et al., 2016, maybe, uh, Science, uh, and You know, that paper was really an important paper because what it showed was that if in early in life, if you received an H1 or an H2 virus, which are group 1 viruses, that those early infections were associated with protection against H5 later in life, which is a group 1, but susceptibility to H7, which is a group 2. And interestingly, the reverse was true. If you were infected early in childhood with an H3, which is a group 2, it appears that you have lifelong protection against H7, which is group 2, but not H5, which is group 1. A lot of, a lot of H's there. So if you're not as familiar with the terminology, just look at the paper. Um, but you know, so that is probably at the level of stall community, I would think. And we are— look, I will just tell you, we're looking at this now, uh, you know, probably a little too early to talk about it, but we do have data that suggests that that's at the level of stall community. Now why does, why does the second infection why don't you get a good stock response to, to, to the second, uh, uh, uh, antigen? It would be, you know, if you have like an H3 followed by an H1, why don't you get the good group 1? We don't completely know, but, uh, but, but, uh, we're looking at that. Let's, let's leave it at that.

Okay, we have 2 more questions that I want to try to get through. Um, the next one is, for the case of RNA vaccine, when did the challenge test take the place and how many days after vaccination?

We've done a lot of these studies now. We've challenged outward to a year after vaccination, so it's a very long-lived response. I'm not sure the exact experiment that I showed here. I would have to go back and look at the paper, but yeah, these vaccines are very incredibly long-lived and And you get heterologous, uh, I guess heterologous is probably isn't the right word, but protection against drifted strains, you know, upwards to a year after vaccination. This is a slide I love. Um, yeah, I mean, look at that, right? It doesn't really fall. So, okay, one more.

Yeah, the last question that we have is, so in terms of vaccine clinical trials, we've seen low VE and we've failed our trials due to the reasons you mentioned. Mismatch season, egg adaptation. What would you advise FDA in terms of vaccine approval?

Uh, well, I mean, I do want to make it clear that our vaccines, even though they have low VE, vaccine effectiveness, based on the criteria that we use, that they probably save lives and protect people by reducing this disease severity. Okay, so just get that, you know, as since this is a public on the web thing, I mean, that is important to know. So, but ideally, we'd have vaccines that would not only prevent us from dying, but also preventing us from getting sick. And so why is the VE so low, and what recommendations would I have moving forward? Well, you know, in the case of the H1N1 story, it's clear that some individuals have antibody repertoires that are different than animals, and that we all have different antibody repertoires based on our prior exposure, as well as other things, probably like, you know, what our genetics, our host genetics. And so I think it is important as we choose vaccine strains to pay a lot of attention to the human antibody response. And I think that that is a major focus in how these vaccine strains are selected today. And so that's a good thing that that is already a part of, an active part of how we choose these vaccine strains. This egg adaptation becomes a little bit trickier to talk about because at this stage, I think our capacity to produce a ton of vaccines in, say, a baclovirus system is very low. And I think that vaccine only, you know, takes— it only comprises of, I don't know, 5 or 10% of the US supply. So, you know, some of this is just practical. I think we have to really gear up our capability to make more of these kind of vaccines. But before we do that, what we really need is a legitimate vaccine effectiveness trial that directly compares the effectiveness of cell-based vaccines to egg-based vaccines, to tobacco virus-based vaccines. You know, who's gonna— you're gonna gear up the, you know, increase the supply of baculovirus recombinant protein-based vaccines based on a study of 30 people, as Seth Zost studied? I wouldn't. I think what we need to do before we make these huge policy changes, or before we really you know, before we have any sort of drastic changes in vaccines, is we need to have some sort of sponsored vaccine effectiveness trial to ask the question, are these egg-adapted strains really associated with lower vaccine effectiveness? You need hundreds of people, thousands of people maybe vaccinated against the egg-adapted strain, other groups vaccinated against these other vaccines, and then see who becomes infected. And And I would argue not until we have this kind of large vaccine effectiveness trial that directly compares these vaccines, not until that point will we have enough sort of ammunition to really persuade some of these companies to change the way that they make the vaccines. But moving forward, I think it's clear that it's time to really, you know, think about where we are and, And modern ways to make vaccines should probably be implemented for flu vaccines. And once we actually have direct vaccine effectiveness data, I think that argument would just be stronger. So thank you all for watching, and thanks.

Yeah, thank you, Scott, for the great webinar.