Scientist Stories: Jeffrey Ravetch, Fc Receptor Engineering to Enhance Antibody Therapeutics

Okay, it's just 10 o'clock. We will get started. Hello and thank you to everyone around the world for joining us for this week's Global COVID Lab meeting. My name is Kristin Aboud. I am the science editor at the Human Vaccines Project, and I will be your moderator today. We are very happy to be back after a short summer break. With this talk in our lab meeting series. Unlike us, COVID did not take a summer holiday. However, the World Health Organization announced yesterday that the world has never been in a better position to end the COVID-19 pandemic. Last week, deaths due to the disease reached their lowest level since March 2020, a significant and for once optimistic milestone. In the fight against COVID. The WHO says now is the time to continue vaccinations against and testing for SARS-CoV-2 while at the same time strengthening systems to fight future variants as well as other viruses that are all but guaranteed to emerge in the future. Before we get to today's presentation, I just wanted to announce again an immunology prize. Through a partnership with AAAS and Science Magazine, Michelson Philanthropies is awarding a prize for immunology to early career researchers under the age of 35. A grand prize and two runner-up awards will be granted for transformative research that can accelerate vaccine and immunotherapy discovery. Applications are still being accepted until October 1st. Please see the link in the chat for more information or to apply. With that, it is my pleasure to now introduce today's speaker, Dr. Jeffrey Ravitch, the Teresa and Eugene Lang Professor at Rockefeller University and head of the Leonard Wagner Laboratory of Molecular Genetics and Immunology. Dr. Ravitch completed his MD-PhD at Rockefeller University and Cornell Medical School., and then pursued postdoctoral studies at the NIH, where he characterized the genes for human antibodies. His laboratory is focused on the Fc domain of antibodies and the receptors it engages, determining the mechanisms by which this domain enables antibodies to mediate their diverse biological activities. His work established the novel structural basis for Fc domain functional diversity and the preeminence of FCR pathways in host defense, inflammation, and tolerance, describing novel inhibitory signaling pathways to account for the paradoxical role of antibodies as both promoting and suppressing inflammation. These findings have been widely extended into clinical applications for the treatment of neoplastic, inflammatory, and infectious diseases. Dr. Ravitch has received numerous awards, is a member of the National Academy of Sciences, the Institute of Medicine, a fellow of the American Academy of Arts and Sciences, and a fellow of the American Association for the Advancement of Science. During today's presentation, please send all questions for our speaker using the Q&A function in Zoom. I will ask a broad selection of your questions after the presentation,, and we will have some time for discussion. With that, it is my great pleasure to introduce Dr. Jeffrey Rabrich.

Thank you very much, Kristin. It's a pleasure to be asked to present our work on Fc engineering in SARS-CoV-2 therapeutic antibodies. What I'll do this morning is give you an overview of the Fc receptor engagement pathways, how they've been applied to therapeutics for infectious diseases, and then focus exclusively on the SARS-CoV-2 antibodies and how engineering can enhance their activity. So with that, let me share my screen and begin today's presentation quickly. So as Christian said, the— See how I can— how can I advance my slides?

So should be able to use the arrows, or no, not—

nope, that does not seem to be working. Okay.

Any other ideas? I think you might have to just check your preferences in your PowerPoint.

I see. Well, that's gonna take a moment.

Just a moment. Yeah.

Bear with us. If I can do that.

I think if you just select how you advance, let me just double-check what it is actually called. Hang on one second.

Let's see if this works. Oh, there we go. Okay, I found, I found a button.

Thank you so much.

No worries. Right, okay. So just to remind the audience that the idea of immunotherapy using antibodies has been around for a very long time. It goes back to the work of von Behring and Kitasato in the late 19th century where antitoxins were developed by immunizing animals like horses and cows to develop serum that could be used for diseases like diphtheria. And just showing on this slide is the effect of diphtheria antitoxin treatment, which when introduced as a therapeutic, reduced the mortality by over 70% since introduction, and then the vaccination, of course, reduced diphtheria as a significant disease. Today, we see that there's a large number of therapeutic antibodies that are in widespread use in clinical medicine, starting in the 1980s with the introduction of the first sets of antibodies, and then progressing fastening over the next 30 years, we see that there have been an explosion of antibodies. They're now over 130 approved antibodies in the United States for the treatment of a variety of diseases. The vast majority are for the treatment of cancer, but as you can see, infectious diseases are rapidly being introduced and catching up to the neoplastic antibodies. So in the monoclonal antibody immunotherapy category, we see antibodies that have been approved pre-COVID for diseases like anthrax, RSV, Ebola, as well as many antibodies that are in Phase 2 and Phase 3 studies for a variety of viral and bacterial diseases. Since the advent of the COVID pandemic, a large number of antibodies specifically targeting the COVID SARS-CoV-2 virus have received emergency use authorization shown here. Another number are still being considered in various phase studies. Let's focus on the monoclonals against SARS-CoV-2 that have been given emergency use authorization that have in fact been used for both prevention and treatment of SARS-CoV-2. The efficacy for these antibodies in mild to moderate cases is greater than 70%. That's a very impressive number. You can see that from some of the early publications that came out. Looking at the cocktail developed by Regeneron as well as the Lilly cocktail, and you can see the comparison in the prevention studies or in the therapy studies looking at the effect of treatment with these monoclonal antibody cocktails. However, what we also realize is that while antibodies have proven to be effective in prevention and to mild to moderate disease, they show limited activity in severe COVID-19 cases. And we can see those same antibodies that were initially promising in the early stage of disease were terminated for use in later stage disease because of, uh, the lack of efficacy. So it really raises the question, how do the antibodies work? How do they work to protect or reduce severity of disease, and why are they limited in more severe disease? And to understand that, we have to understand the antibody molecule itself, which as everyone knows is comprised of two distinct functional domains: the Fab domains, which govern antibody specificity. In the case of a viral disease like SARS-CoV-2 and COVID-19, they can block virus entry and fusion as well as release of viral particles from infected cells. While the Fc domain is specialized to interact with specialized receptors, Fc receptors, for example, on leukocytes, and they mediate the effector functions. So the questions that my laboratory have tried to address over the last 40 years is what are the relative contributions of these domains to the efficacy of an antibody, and how can we then use that information to better design antibody therapeutics for activity in clinical studies? So starting back in 2014 or so, my laboratory looked at a variety of infectious disease indications, including anthrax, Ebola, HIV, flu, and most recently SARS-CoV-2, to determine whether the Fc domain played a role in the in vivo protective activity of antibodies in these variety of diseases. And the types of data that we generated were based on in vivo animal models for, for example, influenza. And here what we look at as an example is a lethal influenza infection in a mouse, which is now protected from that lethal infection by a monoclonal antibody against the HA determinant, a neutralizing antibody against HA. However, if the antibody is modified in the Fc domain so it can no longer engage Fc receptors,, we see that same antibody is no longer effective at the doses that we see efficacy for in the intact antibody, demonstrating a clear role for the antibody Fc effector domain. The same rule applies for anthrax antibodies against lethal toxin of anthrax. In case of Ebola, there's a more modest decrease in efficacy when the Fc is compromised. And as I'll talk about in some detail, SARS-CoV-2 follows this rule as well. As does HIV. So, the main conclusions from these earlier studies pre-SARS-CoV-2 was that in vitro neutralizing activity cannot predict in vivo potency or clinical efficacy. Which is a somewhat dramatic statement, but I think I have the data to support that contention. That even the most potent neutralizing antibodies won't achieve maximal therapeutic activity if they are unable to engage Fc receptors and mediate Fc effector activities. And finally, the Fc effector function of antibodies is a critical component for their antiviral activity. So what do we mean by that? Let's look at the in vitro neutralization versus in vivo potency for antibodies against influenza. And we use the influenza system because it provides us with a readily accessible in vivo lethal model of infection in wild-type mice that we can then manipulate in some detail. So here's a— sorry, here's a standard experiment where we look at an antibody which is a very potent neutralizing antibody against the HA of influenza, an IC50 of 7 micrograms per mL, and we see that In a lethal infection model of flu, the PR8 virus, the antibodies protect at 4 mgs per kg quite effectively. However, taking an anti-HA monoclonal antibody that essentially is a non-neutralizer where we cannot determine an IC50 even above 200 micrograms per mL, we still see protection in vivo at 4 mgs per kg. So it raises the question of what's the difference between neutralizing and non-neutralizing antibodies in terms of their in vivo efficacy? And that was the question that we began to ponder as we started these studies. The same was true for antibodies against Ebola that we published back in 2019 when the Ebola pandemic was an issue for many countries around the world. And here too, We took a series of monoclonal antibodies that recognized different domains of the spike protein of Ebola, and we found that epitopes that showed essentially very little neutralizing activity still showed good in vivo potency, and antibodies that showed strong in vitro neutralizing activity in some cases showed relatively modest in vivo potency. So there was no clear rule to allow us to establish that in vivo activity correlated directly with in vitro neutralizing activity, which should not come as a surprise to anyone because the in vitro neutralizing assays really only interrogate binding and internalization of the virus to a susceptible target cell. They don't necessarily look at the full range of activity that are mediated by a variety of immune cellular populations. So let's look at the evidence for the Fc effector function in these models. So here once again is the flu model that we talked about. And here we see an antibody called 2G02 recognizing the HA spike, which is showing protection at 6 mgs per kg in this mouse model. If we modify the antibody's Fc, the mouse 2As are very potent. Activating subclass for effector function. If we modify that Fc to mutate a single amino acid position at amino acid 265, that antibody now can no longer engage Fc receptors. And at 6 μg/kg, the protective dose, we see no protection over PBS in these animals, demonstrating that activity of Fc domain was in fact required to the in vivo potency of this antibody in this protection model. Which brings us to what the Fc is actually doing. And from this schematic, you can see that the Fc mediates a large number of in vivo biological functions, everything from the activation of leukocyte populations like ADCC and phagocytosis, to the elicitation of cytokines polarizing macrophage to inflammatory populations to the adaptive responses that begin with dendritic cell maturation, antigen processing, which lead to T-cell responses, and of course B-cell activation as well, which can also feed into affinity maturation and antibody production. And this type of diversity of Fc effector activities raises the obvious question of how the so-called constant domain, this monomorphic domain of the IgG, is able to mediate such a wide variety of biological activities? And the simple answer is the Fc domain is by no means monomorphic. There is in fact diversity in the Fc domain based on its structural heterogeneity that I'll be talking about, and that leads to selective Fc receptor engagement to give rise to functional diversity. So what do we mean by that? Oops, sorry. We'll go back. Right. So Fc domain heterogeneity is generated by several determinants. One, of course, are the subclasses. For all species, including human, we have multiple IgG subclasses that were actually first defined back in the 1960s. These subclasses will have preferential engagement of different receptors. But far and away the largest driver diversification of the Fc comes from the glycan, which is invariably found on the CH2 domain of IgGs. This glycan is a complex bi-antennary glycan, which can occur in over 30 different configurations, which is typical for glycosylation. And these different glycan configurations together with the subclasses can thereby generate a combinatorial number of over 1,000 different variants of an Fc for any one Fab. So the Fc is by no means invariant. It is in fact heterogeneous based on subclass and glycan composition. And the way you can understand this is looking at a simple, simplified idea of how this works. This is an experiment that we performed quite a few years ago now when Falk Nemmerhan was working in the laboratory as a fellow. We looked at the mouse IgG subclasses. As you know, mice have 4 IgG subclasses, 3 of which engage Fc gamma receptors. The mouse IgG1 subclass, the mouse IgG2a subclass. What we did was first measure the various subclasses' ability to bind to the various mouse IgG Fc receptors. There are 4 mouse IgG Fc receptors listed here. 1, 2B, 3, and 4. We found that the mouse IgG1 subclass preferentially engaged the so-called 2B receptor, which has inhibitory signaling, and very weakly engaged an activating receptor, R3 or R4. This allowed us to calculate a ratio of activation to inhibitory binding affinities, which showed that it had in fact a very low AI ratio. In contrast, the mouse 2A subclass, which binds preferentially to activating FC receptors like the R4 receptor and weakly to the inhibitory receptor, has an AI ratio of about 70. When you compare 2A to G1, you find almost a 3-log difference in the AI ratios. Now, how does that translate into in vivo activity? Well, we could take an antibody against a melanosome protein like GP75, so-called TA99-FAB, and use different Fc's for that antibody, a mouse G1 or a mouse 2A or a mouse 2B, and then use these antibodies in an in vivo model of cytotoxicity, which in this case is a metastatic melanoma model where we inject melanoma cells IV into a mouse and they rapidly metastasize to the lung. The lungs turn black If you treat the mice with a G1-based version of this antibody, you see no therapeutic activity. But at the same dosing, a mouse 2A subclass shows essentially complete therapeutic activity, showing that Fc subclass truly matters. And that's because of this engagement of selective Fc receptors. How about the glycan? What does the glycan do? And from this little cartoon, you can see the glycan occupies a very critical space in the IgG molecule on the CH2 domain where it is required for maintaining an open conformation. And it's in this cleft formed by the A and B chains that the Fc receptors are actually binding. So as you modify the glycan, you can modify the binding interactions to the various Fc receptors, and that can be seen quite simply when you look at the composition of the glycan and the composition of the various subclasses. So what are the receptors that these Fc's bind to? Well, there are two families of Fc receptors that we define over the years: the so-called Type I receptors, which are members of the Ig superfamily. We also call them the canonical Fc receptors. They're the first ones that we define, and there are a family of cell surface glycoproteins that have extracellular domains that engage with the Fc of IgG antibodies with low affinity, with the exception of the R1, which is a high affinity receptor. And they have different signaling motifs in their cytoplasmic domains, either an activating ITAM motif or an inhibitory ITIM motif. And the rule simply is that these low affinity receptors bind through avidity interactions to surface receptors on various leukocyte populations like macrophages or dendritic cells or NK cells. And by engaging activating receptors, they can trigger cross-linking and cellular activation that's gated by the presence of the inhibitory receptor. So the rule simply is there is co-expression of activating inhibitory receptors to set a threshold for binding affinity on a certain cell. The Type 2 receptors are quite different. They're members of the C-type lectin family, typified by DC-SIGN and CD23, and they're involved in modulating the immune complex engagement of cells like B cells or dendritic cells for their ability to suppress or modify normal inflammatory responses. So the way the glycan contributes can be seen very nicely in this experiment where we look at the IgG molecule with and without sialic acid. And what you see is that in the absence of sialic acid, the antibodies will engage activating receptors, but in the presence of sialation of the FC, that same antibody is now anti-inflammatory and suppresses inflammation by virtue of these Type II Fc receptor binding. In contrast, the presence or absence of glucose will lead to enhancement of the inflammatory response by increasing the binding affinity to an activating receptor called R4— sorry, R3A in the human, R4 in the mouse. So, to summarize, there's functional diversity of the Fc receptors. They are able to mediate activating responses like cell activation, phagocytosis, and cytokine chemokine expression, or through the inhibitory receptor that can override the activating receptor signaling and limit the pro-inflammatory responses, setting thresholds for cellular activation by an IgG molecule. So how this applies to an antiviral antibody is typified on this slide. So there are a variety of mechanisms that Fc's will contribute to antiviral protection. So for example, a viral particle shown here in this little ball with spikes on its surface, when bound by an antibody, can engage Fc receptors on neutrophils or on macrophages to lead to the clearance of these infectious particles. And this mechanism of clearance is important in opsonization and clearance. It can also be involved in the elimination of infected cells. So if a cell has been infected by a virus and it expresses on its surface various viral glycoproteins. Antibodies that recognize those glycoproteins will bind to that cell and recruit NK cells, for example, or macrophages to eliminate these virally infected cells. And a third mechanism, which is an adaptive long-term mechanism, involves the ability of these immune complexes formed by antibodies and viral contents to engage activating receptors on dendritic cells, leading to their maturation and thereby upregulation of co-stimulatory receptors that convert naive T cells into activated T cells, giving rise to CD8 cytotoxic memory responses. So there's a variety of pathways, all of which I can demonstrate in our in vivo studies. Let's look at SARS-CoV-2 now. So in order to understand how antibodies through their Fc domains might contribute to in vivo potency of anti-SARS antibodies, anti-SARS-CoV-2 antibodies, you need an in vivo model. And fortunately, Ralph Baric and his group with Tim Sheehan had developed mouse-adapted viruses that would infect murine strains. One of these was called the MA virus. And in an experiment that we did in collaboration with his laboratory, as well as Michelle Nissenzweig, who developed some of these monoclonal antibodies, we asked a simple question: did the number of viral particles that we found in an infected mouse with these MAE viruses show any dependence on their FC when treated with neutralizing antibodies? So here is the wild-type version of this neutralizing antibody, C104 or C110. And we see that in the case of C104, it reduces dramatically the number of viral particles. But when we mutate the Fc of C104 to knock out Fc receptor engagement, we lose at least a log of antiviral potency. And the same was true for the C110. So, the Fc variants that had diminished Fc receptor binding showed reduced antiviral activity, despite retaining in vitro potency in neutralization. Other models that were developed included the hamster model, which allowed us to look at actual isolates from human samples. And using the Wuhan or the Washington strain, the hamster model, one could identify a potent in vivo— sorry, a potent in vivo model for SARS-CoV-2 infection. Problem was we knew very little about hamster Fc receptors and how one could modify an antibody to differentially engage hamster receptors. So after characterizing the hamster receptors, demonstrating that an Fc variant of a human IgG1 antibody would knock out binding to hamster receptors, we were then able to do an experiment using a monoclonal antibody developed by Vir, S309, that was either in the wild-type human G1 configuration or in this Fc variant that lost its ability to engage hamster Fc receptors. And in this model of hamster infection, we could see once again that the viral titer, which is significantly reduced by the wild-type antibody, was essentially unchanged when the Fc variant was used, demonstrating once again that there was an Fc component to the ability of these antibodies to reduce the viral load in these animals. Since those studies, a large number of studies have been published by our group and other groups demonstrating that, in fact, this is a general finding for a large number of anti-SARS-CoV-2 antibodies. That show an Fc dependence for their in vivo protective activity in a variety of models. And I'll run through some of those in a moment. But our real goal, of course, is to understand how an antibody, a human antibody used for the treatment of human COVID-19, is able to be modified to improve its activity. And to do that, we need to rely on models that recapitulate what goes on in the human. Now, this just shows that there's a diversity of structures to various species that are commonly used in a lot of these viral studies for their Fc receptors. And notably, the macaque, for example, has an Fc structure which is more similar to the mouse or the rat than it is to the human. Because in the human, what we found many years ago, what that was a recombination event had occurred in the process of differentiation into the hominid species to give rise to this amplification and duplication of the human Fc receptor locus. So we have both new genes that were— that emerged in the human Fc receptor locus, as well as unique binding affinity profiles for human IgG. So in order to be able to look at a human antibody that would recapitulate what might be happening in a human disease, we needed an animal model that would recapitulate the human Fc receptors, not only in terms of the number of receptors and their binding affinity, but their cellular distribution as well, because the human receptors are expressing a profile that is unique to the human, not found in other species like the macaque. Or the rodent. So to do that, quite a few years ago, our laboratory generated animals that were initially deleted for all of their mouse Fc receptors. And then they were— they were then modified by putting in the human BACs for the human Fc receptor genes so that we express all the human Fc receptor genes on a background that was negative, deleted for all the mouse FC receptors, and demonstrated that these animals, in fact, had both the appropriate genes as well as their cellular distribution to recapitulate what we see in the human. These, these mice have been quite widely used in both the academic and biotech communities, and there are over 50 biotech and clinical academic laboratories where these mice have been used successfully. So now we can switch over to looking at what happens when we use these mice. And once again, we relied upon Ralph Baric's group, who developed another variant of their mouse-adapted virus, so-called MA10 variant, where we could look at the development of a rapid disease where you see severe disease 1 day post-infection, and look at the cocktail that had been given emergency use authorization, the Regeneron cocktail of two antibodies, where we compare once again the antibodies that are wild-type human IgG1 compared to those that are silencing their Fc by two amino acid substitutions. And as you can see, there is a significant difference in their ability to confer survival at the same dose responses. And that led us to understand that what we're actually looking at is about a 5-fold difference in their ability to mediate protection in vivo, meaning it would take 5 times as much antibody if we were using an Fc null than using a wild-type antibody. So that just gives us the starting parameters. The question now is, can we go beyond that? Can we actually go beyond the human IgG1 Fc and enhance the binding to selective Fc receptors. And for this— I'm sorry about this— and for this, we rely upon the fact that we have characterized a variety of IgG1 human Fc variants, shown here by their schematic: 1 amino acid, 2 amino acid, or 3 amino acid substitutions that confer selective binding affinity on the human Fc receptors. What's important to note is the inhibitory receptor here, 2B, and the activating receptor here, 3A and 2A. As you can see, some of these variants enhance activating receptor 2A, some enhance the activating receptor 3A, but don't modify the inhibitory receptor. The variant that we rely upon mostly that we develop has both enhancement of activating receptors, 2A and 3A, and reduction to the inhibitory receptor. This antibody will have the maximal in vitro and in vivo cytotoxic potential by virtue of these variations. What happens if we now use these antibodies in vivo? What you can see is the efficacy of the same cocktail is now enhanced quite dramatically. Here's the experiment that you can see. The same antibody cocktails now administered at 1 mg per kg. Which is a minimal amount of antibody for this protection effect in this mouse model using our Fc receptor humanized mice. If you enhance only 2A or 3A, we see some improvement but not as dramatic as activating or enhancing all activating receptors while reducing inhibitory receptors. And that once again leads to about a 5-fold enhancement of its potency meaning that we can get activity at 1 mg per kg when the parental IgG1 required a 5 mg per kg. And that gives us some sense of the contribution of the effector domains of the Fc in the in vivo potency of these antibodies. So in the— we looked at it both in prophylaxis as well as in treatment. In the prophylaxis model now, Looking at a 0.5 mg per kg activating compared to wild-type version, you see improvement and enhancement. Once again, somewhere between a 3 to 5-fold enhancement in potency, as well as in a therapeutic model. So to summarize, we confirm there's a key role for the Fc-Fcr interactions in driving therapeutic activity of anti-SARS-CoV-2 antibodies. We can engineer the Fc for selective activating Fc receptor binding to improve its in vivo antiviral activity, both in prophylaxis and therapy. And then a full therapeutic effect can be achieved at a dose 5 times lower than the dose that's required for a wild-type unmodified IgG1. And most importantly, as I'm sure many of you are wondering, Does this enhancement of in vivo activity by enhancing the pro-inflammatory activity of antibodies lead to increased pathology, the so-called ADE? And we find no evidence that antibodies that are improved for their enhanced cytotoxicity against SARS-CoV-2 show any increased pathology or mortality. Okay. So, I want to now just kind of end with this issue of how we can go beyond the ability to mediate enhanced clearance of viral particles, enhanced killing of virally infected cells to drive a long-term T-cell memory response that will give us an adaptive long-term immunity. So, those studies once again were based on our earlier observations in the flu model where Dave de Lillo, when he was in the laboratory, observed that we could in fact demonstrate once again antibodies against flu conferred protection in an Fc receptor-dependent fashion, and that Fc receptor dependence could then be dissected into the specific FCRs that are involved in their activity. And once again, using our Fc receptor humanized mice and a variety of Fc variants for human IgG1 Fc with selective Fc receptor binding, we could in fact now demonstrate which domains of the antibody are involved in specific in vivo protective activities. And What was clear, and in a sense somewhat surprising, was that the optimal ability to generate a protective response came by antibodies that were enhanced for the so-called 2A receptor. And I'll remind you, the 2A receptor is found on dendritic cells. It's also found on monocytes. And that if we enhance the 2A receptor, we demonstrated once again a significant improvement in the in vivo protection against a lethal infection by the flu virus. The mechanism behind that we were able to dissect depending upon the dendritic cell expression of 2A as opposed to the monocyte expression of 2A. And that DCs in the human and in our humanized mice uniquely express a single activating receptor 2A. And the inhibitory receptor 2B, and they're negative for the 3A receptor. So that selective engagement of the 2A receptor would have a different effect than engagement of the inhibitory receptor. So what normally happens in an immature dendritic cell is that an immune complex will generate a tolerogenic signal by binding to inhibitory receptors that will dominate on these immature dendritic cells. However, we can shift that by forcing engagement to activating 2A receptors by engineering the Fc to that domain. When we do that, we now see upregulation of co-stimulatory molecules that will lead to the activation of the naive T cells to activated T cells, giving rise to their expansion and generation of memory responses. And you can see that in vivo by looking at mice that are treated with a 2A-enhanced variant that exhibits increased frequency of mature DCs. And you can see both cDC1 and cDC2s when treated with a 2A-enhanced antibody show upregulation of CD80 and CD86. And you see that in both populations of DCs. You also see activation of CD4/CD8 T cells looking at the 2A enhanced version compared to the FC-null versions shown here. All right, so treatment of 2A enhanced variants led to the increased frequency of activated CD4 and CD8 T cells in the lungs of the flu-infected animals. So, the critical experiment now is, do we see evidence of a CD8 dependence on our in vivo enhancement of protection? So, we once again repeated the experiment now with an Fc that's modified to enhance 2A binding, as is this variant or this variant, and compared the protection in mice that were treated with an isotype or a depleting antibody for CD8. And what you can see is that the protection that we see in the wild-type mice is lost when the mice are depleted of their CD8 cells. So that demonstrates that after the infection, the CD8 cells that are generated can be depleted, and they will lead to a reduction in the long-term survival these animals. So with that, I'd like to conclude and thank the various members of my laboratory, as well as our collaborators who have been instrumental in allowing us to carry out these studies in a variety of somewhat challenging in vivo models. And of course, our support from both the NIAID and NCI, the university, Rockefeller University, the Gates Foundation, and from DARPA. And with that, I'll be happy to take any questions.

Thank you so much, Dr. Ravitch, for that really stimulating and outstanding presentation. We do have several questions in coming. Let me start maybe with— we have a question on the role of non-neutralizing antibodies and that it's clear from your study in the mouse model that the non-neutralizing antibody is protective. However, immunization with HIV proteins, even a very high titer of non-neutralizing antibodies are elicited and not protective. So wondering if you could maybe explain some possible reasons for that. And then also on top of that, wondering why, if non-neutralizing antibodies can be protective, why do our immune systems make neutralizing antibodies and in most cases broadly neutralizing antibodies?

Right.

Right.

So I think it comes down to, once again, in a polyclonal antibody response, as we see in the vaccination system, the immune system is able to generate both neutralizing and non-neutralizing antibodies with FCs that are potently cytotoxic as well as FCs that are more modestly engaging. And that allows for a diversity of potential mechanisms to offer protection. The studies that I've been focusing on are using monoclonal antibodies. And when you use a monoclonal antibody, you collapse all that diversity into a single specificity. And when you do that, you then rely heavily upon other pathways. So neutralizing antibodies as monoclonals will certainly confer an activity that is based on neutralization, but with the correct Fc can also enhance that potency. So the earliest studies from Von Varencki and Kitasato, of course, used polyclonal serum. And some of the early SARS-CoV-2 studies attempted to use patient neutralizing serum as well, but it was found not to be very effective. And I think that's the lesson that we learned, that the immune system over its hundreds of millions of years of evolution has not relied upon a single pathway, but has diversified mechanisms to allow for alternative approaches. No one approach is going to be adequate. When we adopt as a therapeutic, we make our best guess. And what I'm saying is that our best guess is often a bit narrow in terms of what the functions might be.

Right.

Okay.

So we had a question on whether you could address the importance of the soluble pattern recognition receptors and the binding to the various types of FCR.

FC receptors? So, the TOL receptors, right? So, TOL-like receptors don't engage the canonical FC receptors or the type 2 receptors, but what TOL receptors— what TOL ligands offer, the TOL receptors provide, are pathways for immune stimulation. They're potent adjuvants. So, I showed you how dendritic cells can be activated by immune complexes. But only if they engage in activating FC receptor. A toll ligand, like an LPS, for example, will override all of that. It is a very blunt hammer by which you can activate globally dendritic cell activation. Tolls are very potent activators, and during an infectious challenge, of course, we have products, both RNA products, lipid products, that act as very potent adjuvants to activate the immune system quite broadly. Once again, we're dealing with a therapeutic approach. We try not to use that kind of activation, but rather try to be much more selective in how we activate to avoid the potential pathologies that can arise from hyperactivation of the immune response. And we certainly saw that during the SARS-CoV-2 pandemic, that hyperactivation of the immune response was one of the unfortunate sequelae of some patient conditions.

Yeah. So, we have a couple of questions on the role of complement and if there is a good way to dissect the functions of FCR and complement system.

Yeah. So, that was the focus of my lab for the first decade or so. That we started that, because the textbooks will all tell you that the way IgGs mediate their in vivo inflammatory responses is by engaging the classical pathway of complement activation. And we demonstrated very early on in models like the arthris reaction, which was thought to be a complement-dependent reaction, that in fact it was independent of complement activation, and that the mechanism by which antibodies triggered in vivo inflammatory responses was independent of the classical pathway of complement activation. That paradigm has held up over the last 30 years. Complement is not contributing significantly to the IgG-mediated mechanisms of cytotoxicity and inflammation. And what complement does, however, is provide potent mechanisms for IgMs, for example, or IgAs also provides a very potent mechanism by which immune complexes can be deposited on follicular dendritic cells and drive affinity maturation responses. So there are important selective roles for complement. However, its role in triggering antibody-mediated inflammation through the classical pathway is quite limited.

Okay, okay. So we have two questions on this. Stability of the modified Fc IgGs and whether they are applicable or feasible for monoclonal antibody therapy, for example. And also if you've had any experience where the yield is much lower compared to the wild-type antibody when you're producing them.

So in fact, the, uh, the antibody that I showed you with the 3 amino acid substitution that shows enhanced cytotoxicity and reduced and reduce inhibitory receptor binding has been shown to be stable in macaque models. We've published that and has now been developed by the VirGSK groups as a therapeutic, both in HPV as well as in SARS-CoV-2. It's not immunogenic, it has good stability, it's easily manufacturable. There is no limitations on that particular variant. Other companies like Zencor have developed, or Macrogenics, have developed Fc variants that have specific Fc receptor engaging properties. Once again, no evidence of instability in those systems, no evidence of immunogenicity, and good manufacturability. So there does not seem to be any problem in generating these Fc variants and having them be deployed as therapeutics. It's not— doesn't mean that everyone you make will be accessible, but one can in fact find ones that are quite, quite useful.

Great, great. And maybe a slightly related question is, can adjuvants change the Fc effector function?

So, so the Fc receptors themselves can be modulated in their expression. And we've known that for quite a while. Both the inhibitory and activating receptors, in fact, are regulated by the immune system, which makes perfect sense. If you're a dendritic cell circulating, you don't want to be activated by a stray immune complex. So dendritic cells preferentially express inhibitory FC receptors. Upon an inflammatory challenge, like a viral or bacterial infection, the toll ligands can in fact induce upregulation of activating receptors with reduction of inhibitory receptors. Now those cells are poised to engage immune complexes and trigger additional maturation and T-cell activation. But the primary role is to maintain tolerance and only to selectively engage these receptors. That's generally true. Pro-inflammatory cytokines, like IL-12, IL-6 will upregulate activating receptors. Anti-inflammatory cytokines, IL-10, for example, TGF-beta will in fact upregulate inhibitory FC receptors. It's far more complex depending upon which cell population at what time of an infection. But these are a carefully orchestrated and coordinated family of molecules that will be deployed during different stages of the immune reaction.

Okay. So we had a question from someone saying that some publications have shown different Fc effector functions based on specificity of the FAB to different domains of SARS-CoV-2. So can you comment on why and how the specificity can drive Fc effector function?

Right. So I should point out that that's a very controversial area. The notion that FAB engagement modifies Fc structure. There is no structural data to support that. There are, as you mentioned, a few publications initially in the studies that were done for various fungal pathogens where Fc effector domain interactions seem to be dependent in some degree upon FAB engagement. I think the general rule is in all the variants that we've made and all the systems that we've looked at, we have not seen that type of engagement. I think it's a somewhat difficult technical area to work in because changes in avidity will certainly change the interactions with FC receptors because they are low affinity receptors. But the idea that specificity of the FAB will modulate the Fc is really not an area that I believe has, one, received the attention of structural biology, which I would love to see, and has not at this point been demonstrated to my mind in a conclusive fashion that it's a general phenomenon. That doesn't mean it may not be the case for certain selective monoclonal antibodies. But once again, it's not a general sense. And the simple structure, of course, of the antibody tells you that there's a hinge in between the Fab and the Fc. And the purpose of the hinge is primarily to segregate those two parts of the molecule so they act independently.

Got it. Okay. All right. Thank you very much. We also had a question asking if you've described human alleles of FC receptors that alter how FC binding to Fd receptors can alter susceptibility to infection?

Absolutely. Yes, I didn't get into the alleles of FC receptors. Essentially, all the receptors I described exist in allelic forms. They're not an enormous number of alleles. It's somewhat species-specific. Macaques have an enormous number of FC receptor alleles. Humans have a limited number of FC receptor alleles. But we do know that the alleles will contribute both to disease susceptibility for certain autoimmune diseases as well as to efficacy of therapeutic antibodies. And the best examples really came early on when rituximab, an anti-CD20 antibody that's used initially in non-Hodgkin's lymphoma and subsequently in a variety of autoimmune indications, was demonstrated to show a very clear bias in patient populations. If you had a higher binding allele of the so-called 3A FC receptor, you had a better clinical course when treated with rituximab than if you had a lower binding allele. And that notion has been recapitulated for other monoclonal antibodies in the neoplastic area. And I believe there are studies that have been carried out in various infectious diseases where there's suggestion of that as well. So clearly which allele you express can influence the efficacy of a wild-type IgG1 antibody. However, the goal of Fc engineering is to mitigate that. So the Fc variants that I described that enhance cytotoxicity, for example, are different to allele. They will show the same enhancement whether you have high or low binding alleles of 3A or 2A Fc receptors. And that is significant. Alleles of the inhibitory receptor 2B, we showed early on and others have certainly shown subsequently, are considered to be part of the susceptibility network of genes involved in various autoimmune diseases like lupus, for example.

So we have a question on whether the route of immunization or the type of vaccine, whether it be a protein, nucleic acid, or viral vector, Does that in any way modulate the FC function?

Right. So good question. I think the— that's an experiment that, you know, obviously many of us are now very interested in doing with the advent of the RNA vaccines, comparing RNA vaccination to protein vaccination, what role the adjuvants play in all those pathways. So those are ongoing studies. I would not be surprised if the composition of the vaccine, the type of adjuvant, And perhaps the administration may influence some of those studies. There's some really intriguing studies going on right now by Ali El-Obaidi and Rafi Ahmed looking at patients who are vaccinated with the RNA vaccines and then doing fine needle aspirations of the draining lymph nodes to determine exactly that question. What's happening in the draining lymph nodes? We never had the opportunity before to look at the initial responses in a human population to a novel vaccine to see how various receptors might be responding. So very important area, one that I think will inform us going forward on how to rationally design vaccines for specific types of responses.

Yeah, yeah, absolutely. A lot of interesting work to come. I think we'll try to squeeze in one more question, which was just on the difference between the in vitro and in vivo neutralizing ability. How much does the use of cell lines contribute to that difference?

It might. It's not my area of expertise. We use the established in vitro neutralization values based on the cell lines that have been developed, and then compare that to our in vivo potency in the animal models that we interrogate. Is it possible? I'm sure it is. But it's not something that we have investigated.

I confess ignorance on that point. Well, thank you for answering all those questions and I apologize to people who put in questions that we didn't have time to get to. Such an outstanding presentation and really thank you so much for coming and sharing this really interesting work.

My pleasure.

Thanks for inviting me. Yeah, we really appreciate your time. Also, thank you to all of our attendees who participated today and for submitting such great questions for our speaker. And I would just like to invite you to join us for the next Global COVID Lab meeting, which will be held on Thursday, September 29th. And our speaker that day will be Dr. Michael Birnbaum, Associate Professor of Biological Engineering at MIT. And he will present on library-based approaches to decode adaptive immunity. And if you're interested in more research on COVID-19, please do sign up for the HVP COVID report. Newsletter providing insights from experts around the globe and highlighting the latest scientific articles and data. And finally, please visit our website and follow us on LinkedIn where the full webinar series and all the recordings are available. And with that, I'll say thank you again for participating today, and we look forward to seeing you on the next Global COVID Lab meeting. Thank you so much.