Scientist Stories: Jon Yewdell, Fc Mediated Enhancement of Anti-Viral Antibodies

protein translations, defining mechanisms of influenza and coronavirus evolution, understanding of immune dominance of B-cell antibody responses, and identifying and characterizing novel viral proteins. So I know John for probably the longest time from all the immunologists because when I started as a physicist, I read his papers continuously because they were actually the easiest to understand, and he has a quantitative mind that actually was more accessible to me than many others. And today we're going to hear about— I don't see it here—

wait a second, there we go—

FC-mediated enhancement of antiviral activity.

Buzz Lightyear comes to the rescue. Awesome.

So that actually takes it back to your MD-PhD in Pittsburgh where you worked with Walter Gerhard.

Philadelphia. Pittsburgh, Philadelphia, Philadelphia, whatever.

Yeah, with Walter Gerhard on monoclonal antibodies. Essentially, he hasn't really progressed.

No, I have not. We're going to time machine back to 1977. That's when I started working on influenza virus with Walter Gerhard. Okay, but before we get to, uh, data, I want to talk about 3 short reviews, 2 of which occupied a good chunk of my time during COVID sitting at home. One is on antigenic drift. So antigenic drift, we all knew, was famous from flu. Now it's again famous because of SARS-CoV-2. And, you know, just how does antigenic drift work? This is pretty good, actually. This is— immunity has these primers for people who aren't necessarily virologists, just to bring you up to speed. So I'm pretty proud of this one. So I think this is good about getting you into antigenic drift and providing some of the complexities about it. So there's genetic drift. Viruses, like everything else, has certain mutation rates. Viruses basically, among all viruses, and Shane is a world expert on viral mutation rates, viruses basically have a polymerase to create at least a single nucleotide substitution in each virion. So viruses by definition are working as swarms. No two viruses are basically the same in a population. You have genetic drift. Because in going from person A to person B, not too many virions can be transmitted. There aren't a lot of hard numbers for these in human populations. Adam Loring has a good one for flu, and it's something like 3 virions on average is what's transmitted from person A to person B. So we have this genetic bottlenecking. So whenever we're passaging a virus, we're automatically going to select for mutations. Because as I told you, each virion has at least one mutation to start with, okay? So that's just neutral, right? This is just what's going to happen. So the virus is always going to drift because of the bottlenecking. That's inevitable, right? So there will always be sequence differences in viruses. And so, you know, people freak out when COVID— oh, it's mutating. Well, every virus mutates, and we're all infected with viruses, and the viruses within us themselves are mutating, and HIV in a single patient in 1 year will undergo as much antigenic— or excuse me, genetic changes as influenza will in the entire human population. So this is viruses, one of the cool parts of viruses and the talk is over, and now it's back, is that they're always changing. So antigenic drift is when we have selection pressure on a virus typically by a neutralizing antibody. And if it's a monoclonal antibody, it's going to be quite simple. We're going to get a single amino acid substitution is usually sufficient to reduce the affinity of the antibody sufficiently that the virus will escape. That's how we get antigenic drift. People are really worried about antigenic shift with flu, which will be the topic of the talk because, you know, flu, it's endemic in lots of animals. Billions of birds are infected with flu all the time. It's a different virus in birds. It's a gut virus. You have a bird virus that goes into a pig or sometimes straight to a person. It has a different hemagglutinin, right? We have— the hemagglutinin is a receptor protein for flu. We'll be talking about this, how a neutralizing antibody works. If you change the hemagglutinin from one subtype to another, there are 18 subtypes we know about in nature. 3 so far we know really can infect human and make a human virus, H1, H2, H3. If we get a new one like an H5 or we get a new version of an old one like a new H1 like happened in 2009, we get something called an antigenic shift. And our immunity then is much reduced. But everyone's worried about that. That's the next big 1918 flu pandemic, right? On the other hand, drift is a constant. And drift is occurring actually remarkably slowly. Flu is famous for drift. But on average, the two proteins that are subject to drift that are selected by antibodies, which are the hemagglutinin and the neuraminidase, the two surface proteins of a flu virion, they're only changing at an average rate of 1 to 1.5 residues per year in the human population. It's really not very impressive. And yet, this is sufficient to make flu a major public health issue. So people are super worried about shift, but 1918, A lot of people died, something like 2% mortality in a much smaller world, 50 to 100 million deaths. That was a big deal, you know, worse than COVID. There have been a number of shifts after, right? We had a shift in 1957, we had a shift in 1968, we had a shift in 2009. The '57 shift was pretty bad, but all told, if we add up the deaths from shifts versus the yearly death dying from drift, it's 100 times more people are dying from drift than from shift. So drift is really underappreciated and is important. This is crazy. So SARS-CoV-2 came along and experts on antigenic drift like me thought, well, it's not going to be a problem with coronavirus because coronaviruses don't drift. That's what we all thought. And there were 4 seasonal coronaviruses. There had been sequencing done and somehow The field missed this. I mean, they do drift. Uh, during COVID Jesse Bloom had a really nice paper showing that, um, one of the seasonal, uh, coronaviruses is drifting at least as fast as flu is. So coronaviruses do drift. And then SARS-CoV-2 comes along, and Omicron has 32 amino acid substitutions that are probably affecting antigenicity. Holy shit. What we think we know about coronaviruses is that their polymerase is something like 100-fold more accurate than the flu polymerase because it has a domain that's correcting mistakes that are made, an error correction domain. So we all thought drift is not going to be an issue. Omicron alone is worth 30 years of flu. So what? Uh, I don't know. I think this is really, really interesting. And you say, well, okay, Omicron probably came out of an immunosuppressed person with HIV in, in maybe South Africa. Okay, flu can infect immunosuppressed people in South Africa. Uh, RSV can. Human pneumo— uh, metapneumovirus can. Every human virus can. What is different about coronavirus? I don't know. I think this is really, really interesting, right. We don't understand this. Okay, next topic, antigenic drift inevitably leads to original antigenic sin, right. You've all— you've heard of this term. People here work on it, I think, someone, no, maybe not. OAS, right, original antigenic sin. This phrase was brilliantly coined by a guy named Thomas Francis who you should know about. History of virology. This is the guy who probably should have gotten the credit for the polio vaccine. You have the Salk Institute. Tommy Francis was the brains with Jonas Salk's work. Salk was a postdoc in Francis's lab. Francis was a brilliant viral immunologist, brilliant. And he coined this term. And you could see when Francis was working, these figures are out of this brilliant review he wrote in 1960. People knew nothing about what a virus was. So on the top, this is his model of what flu was. It's just this blob, right? You students, you have to realize that, you know, not so long ago we knew absolutely nothing. Absolutely nothing. Now we have these beautiful crystal structures and models. It's just, it's an amazing world. They had no idea what an antigen was. At the bottom here, this is, it does kind of look like a virion. But that's strictly coincidental. They had no idea what the hemagglutinin was, what a protein antigen was, no idea what an epitope was, an antibody structure hadn't been solved. People were going to win a Nobel Prize, Edelman and Porter, for basically running SDS-PAGE gels and showing that there were two chains, right, and that the ends of the chains were where the antibody binding was. So the amount of progress there's been just in my lifetime, just in your lifetime, It's been absolutely astonishing. And I'll just read out of this paper because Francis was so accurate. The antibody-forming mechanisms that have been highly conditioned by the first stimulus so that later infections with strains of the same type successively enhance the original antibody to maintain it at the highest level at all times in that age group. Looking at people of different age cohorts. The imprint— this is the rage word now in viral immunology, imprinting— established by the original virus governs the antibody response thereafter. This we have called the doctrine of original antigenic sin. So brilliant. Tommy Francis was amazing. I urge you to go back and read old papers in your field. Flu started in 1931 with the isolation of swine flu virus by Shope. And these old papers, they're great. And sometimes you get really good ideas from reading the old papers. These guys were really good at at discovering phenomena, which is now a dirty word. Phenomenology, this paper is phenomenology, you get that, right, when the paper is rejected. And I'm like, yeah, that's what I'm trying to do is discover phenomena, right? So these guys were really good at discovering phenomena. The papers I think were much better written. They were much more direct. There was not so much jargon, much less bullshit. It's really fun to read the old papers. And if you're lucky enough to be in a virus field or immunology field where there's an old literature, I encourage you to go back and read these papers. You will enjoy them. You will learn a lot from them. You will get ideas from them. Okay, so original antigenic sin, brilliant name. I think very, very unfortunate because you have sin in it, it implies this is bad, right? And original, right, implies maybe it's something to do with viruses. So this is a review I wrote with Jeff Santos. OAS is not original with flu. First paper describing it, 1906, infected with bacteria A, boosted with bacteria B, you make a better response to bacteria A. So happens with bacteria, obviously happens with viruses. There's a paper I love from Frank Dixon who basically started Scripps Institute, a great virologist, came from Pittsburgh to the beautiful San Diego area that you all enjoy. Enjoy, as do I when I visit, although the two cloudy days were not a bonus feature of this trip. I was hoping for pure sunshine. Frank had paper with serum albumin, immunized rabbits with horse serum albumin, boost with rabbit serum albumin, you make a better response to horse serum albumin, OAS. And what it's based on, we know this, it's based on memory B cells being easier to trigger than naïve B cells. Okay, this is not inevitable. The immune system evolved to do this, obviously, right? You could make them harder to trigger, you can make them the same trigger, but you make them easier to trigger. So there's obviously a method to this non-madness, right? It's not— I'm arguing it's not a sin. Every immune response, we have original antigenic sin. Everyone that I could find, there's actually a pretty long literature on this. You won't necessarily find it by searching for antigenic sen in the title though, right? It's just basically what happens when we boost an animal or a person with a different antigen. So again, the immune system probably knows what it's doing. There's a really, really nice paper from a former postdoc in my lab named Scott Hensley who's got a brilliant career. He's at Penn. What Scott showed was that he came up— he's got an original antigenic sen system in mice. We prime with one virus, one flu virus, we boost with another one. What Scott can show with monoclonal antibodies is you take a monoclonal that doesn't bind, right, doesn't bind well enough in an ELISA to bind very well, and you can show it actually protects mice, okay. So you may not think these antibodies are useful, but I would argue the immune system probably knows better than we do. So what I would say is be very careful in thinking that OAS is a problem. I think this is really a feature, this is not a bug, really not a bug. And be very careful in what you think is a protective antibody response unless you actually test that, which is sometimes not so easy. Okay, so we make antibodies to SARS-CoV-2, we think they're useless, how do we know? Without an animal model, that's really going to be hard to tell. Okay, last thing, uh, Ploss-Pathogen's paper, I wrote one when there was all this discussion about herd immunity and you had this crazy guy, this— what is he, like a neuroradiologist from Stanford? With all these opinions on herd immunity and trying to, you know, influencing Trump, who was convenient to be influenced by, that, well, we don't need to vaccinate anybody, we're just going to have herd immunity and we're going to be better off, right? Wrong. So Got me thinking about something I should have been thinking about my whole career as a viral immunologist. What do we really know about the duration of immunity, right? And when we are infected with a virus, how long are we protected? So something— Jeff Tauberberger is my neighbor in our building at NIH. Jeff had this really nice paper where he showed that with influenza virus, he's got an influenza challenge model. Not many people have these. Virus challenge models in humans. They're not easy to get approved, they're not easy to do, but Jeff has done some really beautiful work here and he shows you can take people, infect them with flu, have them come back a year later, the same exact batch of virus, and reinfect them. Sometimes with the exact same symptoms. Okay, so that's flu, this is not COVID. And you can measure the antibodies in these people and from what's in the literature, which is that you have an antibody response in a certain test with a titer of greater than 1:40, these people should have been protected, but they weren't. Okay, so this is happening with flu. And then it's been known for a long time that other respiratory viruses, other challenge studies dating back to even with coronaviruses in the 1970s in England, immunity doesn't last long. If the criteria is just reinfection there's very poor protection over the— over a single year even, okay. And the same thing with gut viruses. We're not protected from being reinfected with noroviruses, with rotaviruses, right. We're just not. That's what the clinical evidence is, right. And the common feature, of course, is that these are all mucosal viruses that cause local disease. Most of the human viruses that we worry about are, you know, particularly the— the ones that can cause pandemics. Okay, but you say, well, yeah, but there obviously there are some viruses that work. The pediatric vaccines work really, really well. You know, for example, poliovirus. This vaccine provides lifelong immunity, right? We all know that. But poliovirus infection doesn't prevent mucosal reinfection, right? This has been known for a long time. What it does is reduce viral replication and transmission. And it prevents serious disease, i.e., paralysis in the case of polio. And arguing that this is the key, that effective immunity to vaccines and infection is based largely on preventing disease, right? And in the case of polio, disease requires dissemination via the blood or the lymph, body fluids. Okay, another vaccine that was incredibly effective, smallpox. Vaccine, right? This eradicated smallpox. It's the only virus, only human virus that's been completely eradicated by vaccination, right? So this is the ultimate herd immunity. Smallpox is a mucosal virus, right? It spread via the respiratory route, but the disease and the transmission that are required for dissemination requires a blood-borne phase. Right? And in fact, in the historic literature of smallpox, vaccination does not prevent reinfection. People, you couldn't detect this clinically. How you could detect it was someone would come in, they'd be, you'd test their blood for smallpox antibodies, you'd see a rise with no disease. This was common. Okay? So even with smallpox, the same deal. Okay? Rinderpest, the other man-exterminated virus, although it's an animal virus. Also depends on viremic spread for transmission and disease. So that got me thinking, and I went through the textbooks, and you thought someone would have already noticed this, had written it down. And I think people did know this, but it wasn't written down anywhere that I could find. You take all the viruses that we, we know about that are, you know, the most common human pathogens, including coronavirus, and then we look at whether, uh, how durable, uh, immunity is And what we find is that for reinfection, nothing is durable. What's durable is preventing disease, and in every case, disease is then based on blood or lymph dissemination. So the idea here is that antibodies work really well, right, in the blood, in the tissues, in the lymph. They don't work well on surfaces. They don't work well on the mucosa. Okay, so why is the durability of mucosal immunity so limited? I think this is a really interesting question, right? What about all those secreted antibodies? 90% of the antibodies we make are delivered to mucosa, right? 70% of total antibody produced, even though we rarely measure it, is IgA. Okay, at the same time, selective IgA deficiency, the most common inherited Immunodeficiency in European-origin populations, 1 in 400, it's a pretty big number, has modest effects on viral disease. Okay, and I think what we have to start thinking about is why the secretory antibody system evolved, what the main target is, and maybe it's not viruses, right? Maybe it's bugs. Why is it limited? Well, you know, this is a tautology, but Always have to think about evolution, you know. It's as good as it needs to be to maintain host fitness. And it may be less effective in species like us that evolved in widely scattered small populations. I think this really points to the importance of comparative immunology, which is very hard to get funded, right. But some amazingly cool findings come out of it that turn out to be very useful, like studying lampreys and finding a whole new system immune recognition. And I think it'd be interesting to know, and maybe people already know this, I don't know, I haven't found any vet immunologists who have been able to answer the question, do herd animals have durable immunity to mucosal vaccines? Take an animal that evolved in a herd of a million individuals that are always living together. Do they have better herd immunity than we do? Maybe they do. Interesting. Second, I mean, do we know that viral infections are only deleterious? I mean, that's our bias, right? Viral immunologists, we want to prevent viral infections. But, you know, something that's always puzzled me, interferon works really, really well. Why isn't it always on? Why do we wait? It's like, you know, all right, give us, infect us a little and then we'll deal with you. Are there beneficial effects to being infected? Maybe because we evolved with infections, we need them. It's kind of a circular argument, I know, but maybe. Or maybe it's very useful to maintain gene transfer. A good chunk of our genome are retroviruses, retroviral elements, right? Maybe if animals have too good an antiviral, they don't get that, they lose the battle of evolution. It's worth thinking about, right? And the key theme of my talk today is going to be evolution. Charles Darwin is going to show up. Here he is, first time, right? The immune system did not evolve to prevent diseases that don't impact host reproductive success. I know that's obvious, but sometimes we forget that. The immune system doesn't care that we feel shitty for 2 weeks unless it means we're going to have less children. Right? Really. So always think about immunology. So critical questions for COVID. How much do we care about protecting the upper airway against infection with coronaviruses, flu, and other viruses? Do we really need you know, to fix the vaccines? Is it such a problem? Okay, and then if we do, how do we increase the durability of secretory IgA responses, right? If we really want to do that, how do we do that? And then again, you know, be careful what you wish for. But at the same time, you have to do clinical trials and things have to have some risk. But I'll just say there are probably some risks there. Okay, so now the actual science. So the work I'm going to show you is a one-man band, actually. It's a guy who came to my lab from Slovakia as a postdoc and was a fantastic postdoc. And we have these great positions at NIH, which I think everyone should have. I know we don't have the funding for them, but they're staff scientists. So basically it's your number one helper in your lab. They help run the lab. They're fantastic scientists. They have their own projects. Ivan, we get one at NIH. Ivan is my staff scientist and fantastic guy, fantastic guy. So Ivan's interested in antibodies against viral surface proteins, hemagglutinin and neuraminidase. This is a, you know, a model of a flu virion on the left. There's about 500 HA spikes on a virion. We depict it as a sphere, but flu actually in humans is filamentous. People are just now coming to grips with what effects dealing with a filament versus a sphere would have. They're not zero. The virus evolves immediately to make a sphere in vitro because it grows faster. Put it back in a person, it's a sphere again. It's a filament again. So virus really wants to be filamentous for whatever reason, which we don't understand. Spherical particle, 500 HA trimers, about 50 NA tetramers. So 10 to 1, HA to NA. For reasons we don't understand, the NAs are actually probably in little islands of NA, like 3 or 4. So they're not just scattered randomly. There's some geometry here that the virus thinks is important. We don't know how that happens. We don't know what it actually means functionally, but I'm sure it means something. Okay, so the antibodies most of us have studied for since the 1930s when flu was first isolated from people are antibodies that block the interaction of the hemagglutinin, which literally hemagglutinates red blood cells where it gets its name. This is the viral receptor. It binds this incredibly ubiquitous molecule called sialic acid. The linkages matter, but lots of cells have lots of linkages. Different species have different linkages. The hemagglutinin binds sialic acid. The antibodies to hemagglutinin in the globular domain block this interaction, and these are the most efficient neutralizing antibodies. And these are the ones we all think were the most— were the best antibodies to make. They certainly work the best. Okay? So that's one way to block the virus. There's a new set of antibodies that we'll be talking about in this part that are against the conserved part of the HA called the stem, right? This is what the— the globular domain is sitting on. There were clues that these antibodies were around from the early '90s. People got really interested when they started cloning human antibodies from B cells, and they found that these were not rare antibodies. It turns out they're very common antibodies, and they are binding conserved parts of the HA, and they will neutralize the virus, right? And I'll talk— the first part of the talk is how they're working. One, two ways we know they're working. First, we know that they can bind the stem of the HA and they can prevent the conformational changes required for the virus to fuse with the host membrane. Flu is internalized into early endosomes. The endosomes, just as they function, they lower the pH. The virus evolved to take advantage of this. The HA undergoes a conformational change. It liberates the fusion peptide. Which then can integrate the viral membrane with the host membrane, deliver the viral cores to the cytosol, and initiate the infectious cycle. Okay, so anti-stem antibodies can block this. People get very excited from that. Something Yvonne showed, which is quite cool, is that— let me go back, sorry, got ahead of myself— is that antibodies to the neuraminidase but also anti-stem antibodies can block the neuraminidase. Okay, and then we had— Ivan had a Journal of Experimental Medicine paper, very clever, where we show if we increase the height of the NA, we can actually prevent the anti-HA stalk antibodies from blocking because the active site of the neuraminase is too far away now. And we can show that these mice are not as well protected by the anti-stem antibodies. So this is a whole new mechanism. That Yvonne found for how the anti-stem antibodies are working, okay. So not simple, complicated, and I just, I'm not going to talk at all about the neuraminidase. We haven't done a lot of work on that, but the neuraminidase, I did mention the neuraminidase is drifting at the same rate of the HA, which means according to flu, these antibodies are as important as HA antibodies. And these are antibodies that everybody has ignored until recently. Do we immunize against neuraminidase? Not intentionally. But the way that the least expensive vaccines are made in eggs, what you do is you take virus, you treat them with a mild detergent, you pellet the cores out, and what's left is the hemagglutinin and the neuraminidase. And drug companies, the vaccine companies were never required to measure how much neuraminidase is in the vaccines. It's there. We've shown if you infect mice with flu, by weight, the neuraminidase is the most immunogenic part of the virion. Okay, so do you make antibody responses to the neuraminidase in the vaccine? You do. Why don't the companies measure how much neuraminidase is there? Because they're lazy. The neuraminidase assay, which I'll, I won't show you, but I'll show you we use it, An 8th grader can do this assay. You take something from one vial, put it in another one, it turns color, and you can measure this in a spectrophotometer. I could do this assay. That's how easy it is, right? My lab skills are much degraded as I'm aging here. So all they would have to do is take a batch of— or we could do it— is just take batches of vaccine and measure how much neuraminidase is there. And if it's native, it'll be immunogenic, it will be enzymatically active. We could know this. We could correlate one vaccine versus another. Maybe it's better because it has neuraminidase. I bet it is. But this is things that humanity does not know yet that people will now find out in the next 5 years, I would imagine, right? Because it's there and antibodies do work. John, these NA islands that you mentioned, that exists in both the spherical and the filamentous? Probably, yeah. There aren't a lot of EM cryo— there's much more cryo now. There aren't so many published ones that actually show the islands. But I think people think they're real, you know. And what they're doing, there's obviously something to it. And this is one of the reasons that NA antibodies don't affect the HA, there's much less and they're all concentrated. And there's plenty of HA everywhere else, right. So yeah. We just hired someone, Tijana Ivanovic, who's moving to our department at NIH from Brandeis who who's got this fantastic Nature Microbiology paper on what the spherical particles are doing. And she's also worked out flow virometry, which we're really— we've been playing around with, but we're going to— you need— she's really good at that. So flow virometry, you can take viruses and antibodies and measure the interaction of antibodies in a flow cytometer. This is an amazing advance for virology then, amazing advance. And there are papers on this. If you search for this, as of a year ago, there were like 50 studies that come up with flow virometry. So people are using it, but this is really crucial. You get a lot of information measuring antibody interaction with other forms of antigens other than on a virion, but at the end of the day, you have to know what's on the virion. And speaking of the literature, my scientific grandfather was a guy named Fasigus de Saint-Growth. Rob Webster, who's 90-ish now, was his student. Rob, as a graduate student from the 1960s, is on these papers with Fasigus where this was not easy, but they did it. They were measuring interactions of antibodies with virions with gel filtration methods. These papers are absolutely brilliant. The entire literature of how antibodies interact in real natural circumstances with virions is basically restricted to what these guys did with Fasigus. Basicus did, right? And we have to get back to that. You can't know how antibodies are working with virus bound to polyvinyl or whatever the plates are, right? You can't know. You get ideas, but in the end you have to know in the context of a real virion. Okay, so this is what I was showing you. This work, you know, the antigenic sites was actually what my PhD thesis was. We made— I just happened— I got thrown out of One lab, a guy was moving out to Scripps at Penn. His wife got divorced, and one of the cascading side effects was I became a flu immunologist because he said, yeah, John, I don't— we don't want to take too many graduate students. What should I do? Well, Norman, go talk to Walter. Walter was Walter Gerhard. I walked in the lab, 1977. He had just made the first monoclonal antibodies to viruses. Success in science, a lot of it is luck, right? And for me, this was just pure luck. I would have had some boring-ass project in Norman's lab with some bacteria, right? And which I didn't really— I love viruses. I was— I liked viruses in undergrad. So for me, this just worked out incredibly well. We selected antigenic escape mutants. This was my PhD. Walter made antibodies. I made the escape mutants. I did a lot of virology. This was great. We all completely miss the stem antibody. So stem is on the bottom there, globular domain's on top, right? People are super excited about the stem antibodies because of universal flu vaccination because it is so conserved. So there's a lot of interest in this. Okay, so here's me and Charles. This is like Jesus College at Cambridge. This is the beautiful garden with the statue. I'm sorry if you're English, I'll offend you here. Well, old chap. You do know that evolution didn't select antibodies based on their fitness in in vitro assays, right? It didn't. Okay, so something that we all ignore but Yvonne didn't is something called— it's not CQ1, it's C1q, sorry for that typo. This is an abundant complement component that binds avidly to multivalent antigen complex Fc domains. So the concentration of C1q in the blood is 320 nanomolar. The affinity for individual antibodies is 100 micromolar, so that ain't going to work. But put them in the right array like on a virion, now we're talking. Now we can have an affinity of 10 nanomolar. It's never coming off, okay. And C1q is huge, right. This is kind of like to scale, right. It's over 400 kilodaltons, right. So you got to think that's going to affect how antibodies work, right. It's got to, and it does. First guy to do it, this is Walter, about when I was in his lab. Old picture from a party. Walter had this really nice paper in Journal of Virology 20 years ago showing, looking at the effect of C1q on the antibodies to the globular domain. And showing you got this major enhancement of antibody neutralization activity, blocking attachment, particularly as the antibody moved down, moved down the spike in terms of the epitope where it was. The ones right near the RBS receptor binding site, not so much. But move down the spike, something called the CB site if you're a flu virologist, then those are working much, much better. Okay, so Yvonne read these papers and he wanted to say, well, what happens with the stem, right? So here's the assay. You take a highly sialated protein that's cheap from Sigma called fetuin. You bind this to plastic. The virion will bind that and you can detect the virion binding just by its neuraminidase activity. Flu has a self-reporting protein. And again, you just throw in the substrate, it fluoresces. It's cheap, easy, incredibly quantitative. It's a linear measure for how much neuraminidase is binding to something, right? So in the— go back. Stem antibody doesn't block attachment. That's what's published. That's true. Stem antibody plus C1q blocks attachment pretty well, actually, right? And we can completely block attachment, 100%, right? So that's attachment to a plate. This is the epitope that this 310— this is a human antibody. Immunology has changed now. You can just, you know, read a paper, people publish a sequence of the antibody, you just get them to make a cDNA, you express the antibodies, and in T93 T cells, you got antibodies. So no more hybridomas. Sad but great. Okay, and this is just what Walter had shown. There's a minor effect with an antibody right at the RBS, a greater effect if— as the antibody moves down, but nowhere near the effect as with the stem antibody, which goes from zero to good, right? There's just no attachment blockade with the stem antibody even at 100%. Binding, 100% occupancy. C1q, then it goes down to zero. Okay, so it works. Okay, does it neutralize? Yes, it does. So it blocks attachment and it doesn't— stem antibodies neutralize anyway, but the way they neutralize is, right, inefficient. It's either by blocking release, so they neutralize much better in a multi-step assay. Or they can block endosomal fusion, but that's not terribly efficient. But in the presence of C1q, there's a huge increase in the ability of all the stem antibodies that Yvonne looked at. He looked at 5 specific for clade 1, 5 for clade 2. Clade 2 are the H3 viruses. We have 2 circulating viruses in humans now, H1, H3. This works for every antibody that Yvonne tested. These are all different human monoclonal antibodies published by other labs, okay. So it enhances virus neutralization. And just kind of a control experiment, just, you know, due diligence here, make a mutation that should block the interaction from what's published of the FC region with C1q, this K to A substitution at position 322, and the effect goes away. Okay? So everything is kosher. Okay, now this is a complicated assay and you're virologists so I wouldn't try this with most audiences but let me just go through this and if you don't understand it, don't worry, I haven't explained it well. No, I'm not going to do it. If you're interested, I'll tell you later. This is really tricky assay. This is like high-level. Flu, flu antibodyology. This is cleverly done by Ivan, but just trust me, it also blocks fusion, enhances fusion, and then it also enhances the neuraminidase effect as well. So everything the stem antibodies do, they do much better in the presence of C1q. Okay, this is quite cool. Okay, and this now is the only thing we have relating it to COVID. What Ivan did was very clever here, I think. He asked, okay, what does he want to do with anti-spike antibodies but with a twist? This is a VSV pseudovirus system, okay, where we can control the density of the spike on the pseudovirions by taking off the ER-Golgi complex retention sequence. So we take it off, we get a denser spike on the surface. And what we find, first of all, and remember with on the, remember I didn't tell you on the, on the SARS-CoV-2 virion, the spikes are really far away from each other. And you can imagine, and I'm not gonna share any experiments with SARS-CoV-2, so I can't give you an answer here, but this is like a model for what could happen. So it's again, very unlikely that C1q is gonna be able to bridge. Okay. And VSV is probably the intermediate step. With antibodies against receptor binding domain in this assay, we see nothing. But with the non-RBD ones, the ones against the, the, the other domains on a spike, we do see an effect, but only when the— when the spike is denser on VSV. And I'm not really trying to make a point about SARS-CoV-2 here as much as I'm telling you that the geometry of the viral receptor protein that's the target is going to be really important to the effect of C1q. Because, again, a single interaction is not enough. It's got to be— probably have to have at least 2 and maybe 3 to really have a high affinity interaction, okay? So, this is an experiment documenting that. Okay, just a little sidetrack here before getting on the main story. This is the work of Zey Hu in my lab, and what Zey did was You know, that SARS-CoV-2 is a BSL-3 agent still is insane. You know, you're 1,000 times more likely to get it out of the lab than in the lab. And you're working on the ancestral strain, which is not even— which can't affect humanity anymore because everyone's immune to it. So why this is still a BSL-3 agent is— it's beyond stupid, right? And I've looked into trying to get this changed. Apparently this is— forget it. Everyone says It's impossible, we're never, it's not gonna happen. Okay, so we need to have something that's a proxy for SARS-CoV-2 that's better than VSV as I'll show you, right? And what they did was she made a recombinant human seasonal virus OC43 where we replaced the OC43 spike with the spike from SARS-CoV-2. And these are different neutralizing antibodies. If you compare, I know it's too much data, it's too small,— and you can see correlation on the bottom, but if you just listen to me. If you look at individual antibodies with VSV versus SARS-CoV-2, they're all over the place. Some are better, some are worse. They don't map terribly well. It's true with monoclonals, it's true with polyclonal serum from vaccinees. OC43 is almost a perfect recapitulation. So that's good, right? Makes sense because the geometry is going to be much more similar now, and I think that's really important. And how neutralizing antibodies work. This is actually a promising vaccine because it's not pathogenic in mice, and yet it protects them completely from SARS-CoV-2 infection. And the important bit here is the non-pathogenic bit. It's much less toxic for cell lines. So we had to jump through a whole bunch of hoops, as you can imagine, to get this designated as BSL-2, including going in front of the Dual Use Research Concern Committee at NIH, but We did all that, and this is— this has now been classified as a BSL-2 virus, okay? We just submitted this for publication. Once it's accepted, I'll give it to anybody in the world who has permission to use it, okay? And I think this is a much better virus to do for your neutralization assays, and obviously you don't need BSL-3 conditions. It's going to be cheaper, okay? I think this is something that could be very practically useful. Okay, so Charles is back. Can we get back to C1q? Yes, we can. From my point of view, Charles's, the critical point, again, old chap, he's very British, is how C1q affects viral evolution. And this is crazy, but we now select escape mutants in the presence of C1q and— it neutralizes much better, and paradoxically, it increases the escape repertoire. So you just use stem antibody alone, you get what you think you'd get. You get changes in the stem. When you use stem and C1q, you get changes all over the molecule, including changes in the head. Some of these change the receptor affinity. Of HA for its receptor, and we've shown— Scott Hensley had a 2009 Science paper where we showed this is what you select for in mice polyclonally. You select for absorptive mutants that bind better, so that's going on. But then we also see mutations that are changing the flexibility of the molecule. And I had a paper in 1993, completely ignored by the world, where I showed that we had this one monoclonal antibody which people have rediscovered in humans but I had in mice, which requires breathing of the heads for the antibody to bind. What I showed was if you select a variant with this, you get variants with this, you get changes in the stem including in the fusion peptide, which no one's ever explained. But I could show that that molecule was more rigid. And we have these tools to measure functionally measure the flexibility of the molecule, and we're selecting mutations that affect the flexibility. Not just in the direction I would have predicted. I— this is more— bring back the ape scratching the head. We don't know what's going on here. But this is really interesting. And as a practical point, from this and other papers we've published, you can't look at amino acid changes in a molecule and absolutely know what the selection pressure was. It is very, very easy to get fooled, right? There's a change right near the receptor binding site. I know what changed that, monoclonal antibody. No, antibody neuraminidase can select for that and vice versa. Crazy shit goes on with viruses. They are really, really clever, right? You think you know what they're thinking, you don't, right? And figuring out what's going on in a human being when you get an escape mutant, Good luck with that. You can't even do it in mice necessarily. Okay, so conclusion, we'll run through this because time is running short. And I'll just point out, C1q changes everything, right? So I just gave this talk at the VRC a couple weeks ago. Everyone's on their phones to the labs, you know, C1q, have we done C1q, right, in our nude assays? So do that. And you know Ted, right, Ted Pearson. So Ted had this really nice paper showing, maybe this only other paper with Walter really showing how C1q is working in this case with flaviviruses, really changing the properties of antibodies. So whatever virus you're working on, if you're working on neutralization, think about C1q. Okay, act 2. Universal flu vaccine, according to my institute, should be 75% effective. Protect against group 1 and 2, that's H1, H3, have durable protection, last at least a year, that's kind of a low bar, be suitable for all age groups. Okay, so what's a universal flu vaccine? Here's something coming out of left field, right? So we know that antibodies can't cross the plasma membrane, right? We know that. So antibodies against internal proteins they should only work if the internal protein is on the cell surface, which they are. I had a paper in 1981 showing that. We had a paper last year and another one about to be published with SARS-CoV-2. With SARS-CoV-2, there's more nucleocapsid on the cell surface of infected cells than there is spike, right? There's a lot of NP there, but there's also not much spike because of the ERGIC retention sequence. And we can show that cell surface nucleocapsid protein binds chemokines and is a target for ADCC, so there's that. But then here's the other thing. What if we can get an antibody against nuclear protein into the cell? Well, we all know antibodies don't get in cells. Well, that's not really true. For a long time, it's been known if you take antibodies from people with rheumatoid arthritis, they have anti-DNA antibodies, and you label them and you throw them on cells, they show up in the nucleus. Now, People don't believe that, which is bullshit when someone tells you, I don't believe it. I mean, science. I mean, tell me which part you can't reproduce and we could talk. But you can't tell me you don't believe it. I'm telling you this is what happens. And you get this all the time from people. I don't believe it. You should be a priest, and then you can worry about your belief. As a scientist, belief is neither here nor there. We have evidence. Attack the evidence. OK? But this happens. And again, this is something so easy you can't get it wrong. You know if a cell is alive if you've done 2 days of microscopy. Here's a living cell. Here's the antibody. It's in the nucleus. So antibodies can cross the plasma membrane, OK? And why is that? There's a literature on this now sprung up in the last 5 years. If something is positively charged, it's got a shot of getting into a cell. Okay, there are these things called cell-penetrating peptides, right? And Ivan, when he came to my lab, was obsessed with cell-penetrating peptides. He thought this was the coolest thing ever, to the point where I would make a joke every lab meeting, and in the end he had the last laugh because not only is it true, it's really interesting. So what Ivan did was he took a monoclonal antibody that I made a long time ago against nuclear protein And he put like 6 to 8 arginines at the end of this protein, this monoclonal, okay. And he asked, okay, what does this thing do now, this cell-penetrating antibody? So nuclear protein, this is the antibody we use, it's called HB65. It's one of the first antibodies we made. It's ATCC, the HB65 is the Hybridoma Bank. It binds every flu strain. So it's for universal, if this would be a universal passive vaccine, this would be great. 'cause it should work against everything. Okay, and you put the sequence on the virus and you get a lot more of— this is not in the cytosol, mind you, this is in the cytoplasm— but you get a lot more sticking to the cell surface that's endocytosed. A lot more. I mean, this is probably 1,000 times more antibody that's taken up into the endolysosomal system. And the penetration efficiency is shit, but it's good enough. To work, as I'll show you, okay? I'm sure we can do better. This was just taking the simplest possible sequence and putting it on the end of the molecule. So I'm sure this is engineerable to be much better. Okay, so this is just a functional experiment. If we take this antibody, which is getting internalized much better, we can decrease nucleoprotein expression in infected cells. So the antibodies are 500 nanomolar overnight. That's a lot of antibody. You infect the cells with PR8 at a low multiplicity. We're measuring nuclear protein in a Western blot, and you can see that we have a 50% reduction in how much nuclear protein the cells are expressing. So it's doing something, right? And this is immunofluorescence then, and it's kind of hard to see, right? But what you can see is that there's antibody that's colocalizing with nuclear protein that's not in the nucleus where it should be at this time in the infection, it's in the cytoplasm. So that's cool. So some of the antibody is getting out of the endolysosomes into the cytoplasm where it's having this biological effect. Okay, and in a multi-cycle assay, right, so we're infecting the, we're putting the antibody on, we're infecting with the virus, the virus has to spread from cell to cell. To register infection. You can see that the nucleoprotein antibody alone is doing nothing. The cell-penetrating sequence, it's neutralizing the virus pretty damn effectively, right? I never thought this would work. Years ago in the early '90s, we actually electroporated the same antibody in the cell. We got a lot in. That worked. So I knew it could work, but I didn't think this would work. And obviously we're not going to electroporate antibodies clinically. That's probably a tough ask unless you're a fish or something, maybe that would work. But a human being, probably that's hard. Okay? So here's an assay. This is a mistake. It's 6 hours prior to infection, not post-infection. I couldn't fix it. It's an animation as you'll see. We add the antibody and then we're going to measure how much virus is released in the supernatant. And as— sorry— as predicted, we can reduce the amount of virus that's released from cells. This accounts for this neutralization in a multi-step assay. So the antibody is working as a cell-penetrating form. Does it work in vivo? It does. And what Ivan does, he gives us basically like a nasal spray. He doesn't do a deep intranasal, just a shallow one with a small amount. The mice are not anesthetized, so they're not breathing it in their lungs. It's just the upper airway. And then he infects— he infects with flu and we can get almost complete protection, right? So it's working prophylactically. It's a control with an anti-SARS-1 with the same sequence on the C-terminus of the heavy chain. And then it's also working therapeutically, not as well, but a bit, okay? So this may actually be, This may actually be a therapeutic enzyme. I'm going to finish in about 2 seconds here. How it's working, we don't know. There actually isn't a viral titer release in vivo, right? We don't see that. So this in vitro mechanism, if it's working, it's not obvious how it's exactly the same. And, you know, you do this arrays and you get 100 things and you get a couple of hits. And I don't think we're going to try to figure out what's going on here. This could be a rabbit hole you never escape from, particularly when the effects are subtle. Something's going on with cytokines that's changed. We don't know how it's working, right? We don't know how the cell biology is working. There's this really, really cool TRIM21, which is a ubiquitin ligase that's an FC receptor. There's a guy named Leo James in Cambridge who studies this, got some really nice papers. And the idea is with adeno is that adeno comes in, it penetrates cells with antibody bound to it. And this TRIMP21 prevents the adenovirus from expressing its genes. That's the model. Whether or not that's true in people or not, we don't know. But the odds are this is how it's working. We're just in the process now of overexpressing TRIMP21 in our cell lines we can use and knocking it out. So I can't tell you this is likely how it's working. We have the cell lines. We're about to do the experiments. Conclusions, though. We basically can show that we can take an antibody and target it to the cytosol. We weren't the first group to do this. There's a group in China that did it with hepatitis virus, but I think the possibilities for this are limitless. I mean, you could target oncogenes, you could do a whole bunch of things, right? I think what's interesting is that why doesn't the immune system do this? I mean, Yvonne could do it. In 2 years, the immune system had 400 million years, why doesn't it do this? All right, it can do it. I told you the anti-DNA antibodies get in. I told you everything's about evolution, so more head scratching. Okay, but that's it.

Thanks, John. Before I forget, we have a young scientist lunch right after this. I think we started a little bit late, so we have time for a quick question.

Yes, please. Yeah. Do you see escape mutations for anti-N in cell-penetrating antibodies? We haven't looked yet. Yeah, we haven't looked. That was the original reason I wanted to do it in 1991, and we never did that. We got it to work. A woman got— she was from Australia. Pressed and went home. And then it was rescued by Yvonne. Things like that happen all the time. You have this promising project and you have this one person who was quite good and all of a sudden it doesn't work or they don't want to do it anymore or they leave. You know, you never go back. It's not a very efficient process the way we do science, you know, it really isn't. A lot of hit or miss.

Another key question?

John, I don't remember, in the C1q, did you get rid of C1q in vivo? No, no, no, no, we didn't do that. There are knockout mice, we didn't buy them. We don't have a lot of mouse space now. NIH is poor. Well, my lab is not particularly rich. You guys have more resources than we do actually, the well-funded labs on the outside. You'd be surprised. We have some subsidies, but But not for that. But it's an obvious experiment we didn't do. Yeah? Is there a size limit to what the membrane-spanning peptides can get through? Or if you take a nanobody— Yeah, no. Yeah, good question. I don't know, Mitch. There should be. Yeah. And how it's working, I don't know. The flip side of that is how nuclear protein gets out of cells. And I don't know if you remember, but there was this old literature with SV40 T antigen before we knew anything about how MHC was working and people thought T antigen was a target for T cells. And there's this literature on how T antigen and a whole bunch of really nice biochemistry, but no one ever figured out the pathway for how these internal proteins get out. It's related probably to how IL-1 and other cytokines without signal peptides get out of cells. What it looks like is there's interaction with— I'm just blanking on the term— interaction with this ubiquitous molecule that's on the outside but also on the inside that somehow is carrying, in this case, again, positively charged substances out of the cell. So it's probably related to that. Yeah, but there's no knockouts, no genetic screens that have been published. And in the end, if it's not a protein, you can't do a genetic screen, right? And it may not be. But this is a really important pathway cell biologically that there has not been a lot of work on. And again, a whole bunch of cytokines get out in something like this, right? So it isn't some weird thing with viruses. And every— I haven't published so many. I think I've published 3 examples of viral proteins. This was the 4th. Every nuclear protein I ever looked at is on the cell surface. I worked on p53 for a while. I don't think that's published, but that's on the cell surface. I mean, there's a lot of proteins on the cell surface we don't expect there, right? A lot. And I wonder when they do proteomic screens whether they just throw those out because they aren't in the— they don't expect them to be there. But I bet they do. Yeah? [Speaker:STUDENT] So Skip Virgin many, many moons ago used to tell us that we're all chronically infected with 5 to 6 viruses. True. So what's up with virus B? But, for, well, these are herpesviruses. But, well, I mean, so, you know, you have to distinguish persistent viruses from acute viruses, right? And so herpesviruses have, they don't need to escape or they don't bother with it because they have other ways. Right? So I don't think there's much, I don't think there's much antigenic variation on their surface receptors. Of course, they have multiple receptors, which makes it hard to block anyway because they can get in a whole bunch of different mechanisms probably. But herpesviruses deal with the immune system in a different way. I mean, they're just obviously brilliant in dealing with particularly the cellular immune system. But, you know, that being said, GD is an FC receptor from Simplex. Right? A really good one. So, yeah, escape, the RNA viruses escape, all of them probably. And it's just a question of degree. A friend of mine at the FDA works on norovirus and again it's not widely known but noroviruses are extremely antigenically variable. They're probably evolving faster in people antigenically than flu is. And it's just a topic that hasn't had a lot of attention. And SARS-CoV-2, again, people— no one was really interested in antigenic escape until SARS-CoV-2 came along.

So I think we have to stop here. Thanks again, John. We started with your mutation analysis and saying how surprising Omicron is.

Oh really?

How much would you think is that due to the virus not being adapted to the host yet? How much of it is viral escape?