Over 30 years after the emergence of HIV-1, there is no effective vaccine, and AIDS remains an important threat to global public health. Following infection by HIV-1, the host immune response is unable to clear the virus due to a variety of factors, including rapid viral mutation and the establishment of latent reservoirs. The only target of neutralizing antibodies is the trimeric envelope spike complex, but HIV-1 can usually evade anti-spike antibodies due to rapid mutation of its two spike glycoproteins. We use structural biology, biochemistry, and biophysical methods to characterize antibody recognition of the HIV-1 Env spike, with the goal of using structural knowledge to design immunogens that can elicit broadly neutralizing antibodies. We also use structure-based protein design methods to engineer antibodies that can resist some of the common routes of HIV-1 mutation, with the hope that the designed antibodies could be used in passive immunotherapy methods for HIV-1 treatment and/or prevention.
Pamela J. Bjorkman, PhD, David Baltimore Professor of Biology & Bioengineering, Caltech
We’d like to welcome you to the Sanguine Speaker Series webinar. Today’s topic is Neutralizing Antibodies Against Pandemic Viruses. Our speaker is Pamela J. Bjorkman, PhD. Pamela is the David Baltimore professor of biology and biological engineering at the California Institute of Technology. I will now hand it over to Pamela.
Pamela J. Bjorkman:
Okay. Thanks very much. It’s a pleasure to be here. You’ll notice that I changed my title because normally we’d been concentrating on characterizing antibodies against HIV, but now about half and half my lab is now working on SARS-CoV-2 because of the pandemic. So I’m going to combine discussion of antibodies against HIV and antibodies against coronaviruses. So I called it neutralizing antibodies against pandemic viruses to reflect the shift in the emphasis. Okay. I wanted to say a few things that I’m sure you already know about antibodies. These are proteins in your blood that can also migrate into tissues. And the point of them is that you can make all kinds of different antibodies as a mammal, they’re specific for antigens from different pathogens, let’s say. And an antibody for the terminology here, an antibody can neutralize the virus when it stops the virus from infecting the target cell, and then that means it can’t make any viral progeny.
I wanted to emphasize here that I’ll be talking about the Fabs, which mean the fragment antigen binding. There are two Fabs and those are the part that bind to the pathogen, and those are the parts that are different from one antibody to another. At the bottom, I’m showing different colors of antibodies to show that as a human, you can make probably greater than 10 to the 16th types of different antibodies, depending on what you have been exposed to. So any new pathogen that comes, you can eventually make antibodies to it. And then the final reminder would be that the clonal selection theory is actually the way that B-cells and T-cells are selected by a pathogen. In the case of B-cells, they have membrane-bound receptors, which are just B-cell receptors. Those are membrane bound antibodies.
I’m symbolizing the different types of membrane-bound antibodies or receptors with the different colors here. Let’s say in this instance, the green one actually has the antigen binds to the Fabs of that one, that induces proliferation of the B-cells that actually make that type of antibody. And then you get a lot more of them. And the expanded clone then makes a soluble form of the B-cell receptor, which is actually the immunoglobulin [inaudible 00:02:58]. That will have gone depending on the exact way the antibody is made. If it’s made in a germinal center or not, that antibody will have undergone somatic hypermutation to increase its affinity. So I want to point that out because there are differences in the amounts of somatic hypermutation for antibodies against HIV versus coronaviruses.
The resulting antibody, this is the soluble B-cell receptor and it’s specific for the original antigen. Okay. So just as a reminder of what HIV does, it’s like every virus that there is. To summarize very simply, a virus needs to get into a host cell, that would be through a host cell receptor, make copies of itself and release new viruses from the cell that can go on and infect other hosts cells and also other people. I’m going to be concentrating mostly on the entry step, and that requires binding to a host cell receptor. I also wanted to point out that… Okay, so in the entry step in the part that’s in red, you see the envelopes fight protein, which I’ll call either on or spike trimer. It is a trimer of three identical subunits.
Each has a GP 120 and a GP 41. The GP 120 is responsible for binding to the host cell receptor, which is CD4. And then after that, it triggers conformational changes that allow it to bind to a co-receptor, which is a GPCR such as CCR5. That allows fusion to take place through further conformational changes. The point of this binding is to allow fusion of the membrane of the age of HIV with the host cell membrane, that allows it to get its genetic material into the cell, which allows it to go on to step two, which is make viral copies. Now, I want to point out that during the process of making viral copies, in the case of HIV, this is a retrovirus, so you make a copy of the CDNA, complimentary DNA copy of the RNA. HIV is an RNA virus. So it reverse transcribes its RNA genome into DNA. And during that process, it’s incredibly error-prone. The HIV reverse transcriptase is not what you would use in a lab for PCR, because it makes so many mistakes.
Now, this is actually, it was selected for in the case of HIV because this results in many viral strains even in a single infected individual, and this is basically why it’s so hard for your immune system to get rid of HIV, why it’s so hard to make a vaccine and so on. Now, it turns out that if you could, people do make neutralizing antibodies against HIV when they’re infected. So neutralizing antibodies would interfere with the attachment to the host cell receptor or maybe the fusion process. And then what happens is none of the subsequent steps take place at all. That’s a great way to get rid of the viral infection if you can make enough neutralizing antibodies. Now, the problem is for HIV, that most neutralizing antibodies neutralize only a subset of HIV strains.
In this animation here, this is the point of the original title, which was the molecular arms race. Because of this rapid mutation, the virus wins out over mostly antibodies that a host cell host, sorry, a person can produce. So let’s say someone was infected with this virus here. Even though there are many strains inside a single infected host, it appears that just one gets across, that’s called the transmitted founder virus. Let’s call it this yellow one. When that divides to produce, sorry, it doesn’t divide. But when it makes copies of itself to produce new progeny, they’ll be slightly different from each other as illustrated by these color changes. Now a lot of these won’t be fit at all. So those viruses don’t infect cells, but some of them are still fit.
A number of these will be recognized by neutralizing antibodies for those will go away, but a small subset of these will remain and they’ll produce more progeny, which are again, slightly different from each other. Many of these are unfit or they’re attacked by antibodies, but they can go on to produce more progeny, some of which are unfit or attacked by antibodies. And then you can see what’s happening here, is that basically inside the person who’s infected, you get a viral swarm. That’s what is depicted here in the different colors of HIV. So it’s been estimated that there are more different strains of HIV inside a single infected person than there are strains of influenza virus in the world. You can see that this is a real problem for both your immune system and for making a vaccine.
Now, what’s been done over the past 10 or more years by a number of different laboratories, is to characterize the antibodies from HIV infected patients. This was developed originally in Michel Nussenzweig’s Lab, our collaborator at Rockefeller University. They did single B-cell cloning methods that are described in this reference right here. What they were able to do is get heavy and light chain genes for individual antibodies and then produce those antibodies. And when they looked at them, you could express them recombinantly and look at their properties in vitro. Most of the antibodies turned out to be strain specific. But in rare patients from who these samples are isolated, they would find broadly neutralizing antibodies.
Not all the antibodies in a single person would be broadly neutralizing, but what would happen is when you looked at these broadly neutralizing antibodies, they would neutralize, say 50% or sometimes up to 95% of strains in vitro in laboratory assays. Now, unfortunately in the person from whom this antibody was isolated, all the viruses in that person are resistant to this antibody. But of course, it would be if we could figure out how to raise those antibodies, those would be protective in a vaccine. So a lot of the efforts in my lab are geared towards figuring out how to raise these antibodies so that when someone is challenged during an infection, the infection couldn’t take hold. Okay, so we call these bNAbs for broadly neutralizing antibodies.
Okay. That’s all I’m going to say about HIV, but I’m happy to address questions. Well, I have a few more slides on HIV, but I want to point out first that in general, antibodies against viruses tend to have really long complementarity determining region threes. So what you’re seeing on the left is the surface of an antigen combining site on an antibody. This is the part of the FAB that actually recognizes the antigen. And it has six hypervariable loops that are called complementarity determining regions, three of them from the light chain, three of them from the heavy chain. The CDR threes are in the center. And then in all the antibodies, the CDR ones and twos are on the sides. So in the center, you have the CDR three loops, and there’s various reasons genetically that those have more diversity when you’ve compare antibody sequences.
What’s been found in surveys, especially of HIV antibodies, is that you often see protruding CDRH threes in neutralizing antibodies, especially against HIV. So I’m just showing you this movie now that this is a structure now of two antibodies against HIV. There’s one in a teal color and one in blue, they have particular names that are over there, and you’re seeing the HIV spike trimer and the purple and cyan colors are different types of N-link glycans. Remember I said that it was useful for the antibody to have a long CDRH three. So you can see why because the HIV protein is extremely highly, the envelope protein is extremely highly, heavily glycosylated. So 50% of its molecular mass are glycans.
Okay. Now you may have noticed during that movie that there was a hole in the glycan, and this particular strain of HIV from which we got this on [inaudible 00:12:29] for our structural studies happens to have a hole in this red region. So if you do immunizations in an animal to try and make a HIV vaccine, what will happen is you’ll raise antibodies against this red region because that part doesn’t have… It’s a glycan hole. This is a part of the antigen that has exposed protein surface area that’s not conserved from one strain to another, so you’ll get a neutralizing antibody, but it will be strain-specific and that’s just not useful.
This is a big problem for HIV, somewhat of a problem for influenza, not as much of a problem for coronaviruses, I have to say. I should back up by saying that, of course, the N-link glycans are added to viral proteins by host cell machinery. So in general, they’re not immunogenic because they’re not recognized usually by the host system as being foreign. So they seem to be a self type of modification that is not recognized. The antibodies are trying to reach through to the protein surface, which now differs from host cell proteins.
Okay. That’s all I’m going to say about HIV, but you’ll see some obvious contrasts. I’m going to move on to coronaviruses. So just like HIV and many other viruses, coronaviruses are enveloped viruses, meaning they have a lipid membrane that’s derived from when they bud out of a membrane on the host cell. That encloses the genetic material. And for coronaviruses, that genetic material is also RNA, just like in the case of HIV. But there’s a difference. Coronaviruses are, of course, not retroviruses. They do not make a CDNA copy of their RNA genome that integrates into the host cell. And this is good because in the case of HIV, the fact that it is a retro virus, it will integrate its genome into the host cell. Even if that host cell is not producing viruses at any point, it can start to do that. There are latent reservoirs in the case of HIV, which is another reason that it’s really hard to get rid of it. So Coronavirus, that is not an issue because they copy their RNA into another copy of RNA.
Now, what I’m going to be emphasizing today are the spike glycoproteins, which are called S. So on some of my slides I’ll just call them S, and this is the counterpart of the HIV envelope. It’s also a trimer. And the reason Coronavirus has got their name is that when you look at them by electronic microscopy that was done like 40 years ago or so when they were discovered, it looked like all these spikes were large, and it looked like solar corona. So they got their name from that, which means crown. Okay. So SARS-CoV-2, of course, is just one of a number of coronaviruses that have been characterized so far. Now, a lot of these, the human coronaviruses that I’m sure you’ve all heard of are SARS-CoV-2, the current pandemic, and then there were outbreaks in the world of the original SARS in 2003, I believe, and MERS which was in 2012. And these fortunately, these outbreaks died out, but they did cross over into humans.
The really scary thing about these viruses is that they arise in animals. These are so-called zoonotic viruses that pass into humans sometimes through a secondary host. So in the case of SARS, that’s thought to be [inaudible 00:16:33]. In the case of MERS, it’s dromedary camels. The word is still out on what if there is a secondary host for SARS-CoV-2? No one’s quite sure. It was suggested it was a pangolin, but it’s very unclear. But what we’ve had in the world circulating for a long time are common cold coronaviruses, which infect humans, which are in blue here. So these particular viruses, there are four known strains of these, these arose sometimes in bats and sometimes in rodents. Sometimes they had a secondary host, sometimes they didn’t. These are clearly circulating, but they don’t cause a problem. All of us are probably constantly exposed to these. We don’t mount very good memory responses against them, but if we get infected, we don’t get very sick. So they cause very mild symptoms.
Okay. Now I’m going to go into some of the biology that’s necessary to understand about SARS-CoV-2. And this is in common with some the other coronaviruses as well. So for example, the SARS-CoV-2 and also SARS, but not MERS, enters a host cell through binding to its receptor, which is ACE2. That’s a protein on the surface of lung cells and the surface of cells and other organs as well, that is there to regulate blood pressure, but SARS-CoV-2 and also SARS and at least one of the common cold coronaviruses have co-opted it to use as their means of entry into host cells.
Okay. So the virus goes in and it can either go in at the cell surface or it can get into a [inaudible 00:18:32] out of low pH compartments. It releases its RNA and in a very complicated process that we’re not studying, but it’s very interesting. The viral RNA is translated into viral proteins. These are then assembled inside the cell and then released. And of course, then the new progeny can infect other cells in the same person or go on and infect the other hosts. So just like in the case of HIV, if the host makes neutralizing antibodies, that would prevent SARS-CoV-2 infection of cells. Once you have that happening, none of the other steps can happen. That’s what we’re studying right now for SARS-CoV-2.
Okay. So just one more thing to say is that the Coronavirus spike protein trimer, that structure was determined really quickly right after the pandemic. I believe two papers came out in March of 2020, [cryo-EM 00:19:38] structures of SARS-CoV-2 spike trimer from Jason McClellan’s Lab and David Veesler’s Lab. This is a depiction of what that looks like on the surface. It’s a schematic, but the structure part is accurate. It’s a surface from their coordinates. But prior to that, a lot was known about coronavirus spike proteins from cryo-EM structures of other coronaviruses before people knew about SARS-Cov-2. One of the things that is really interesting is that, well, first of all, the spike is going to be the major, if not the only target of neutralizing antibodies on coronaviruses. But what’s interesting about this spike is that actually unlike HIV, it undergoes this very strange process where it has three copies of something called the receptor binding domain or RBD.
Those RBDs can only bind to the ACE2 receptor when the RBD is in a so-called up conformation. And when that happens, you can get these conformational changes that are necessary for membrane fusion, and then delivery of the host genome into the cell. So I just wanted that as background, which is what we’re going to be looking at, are structures of spike protein. You can look for the RBDs as being either in the up or the down conformation. So the down conformation do not allow fusion because it can’t bind to ACE2. So here’s what we started doing in collaboration with Michel Nussenzweig’s Lab. We started in February, and that was when Michel’s lab got donors who had been diagnosed with COVID-19 and they isolated plasma. They got plasmas from them. And from B-cells that they had gotten, they then isolated antibodies.
They got their genes from single B-cell cloning and they expressed them. So this [inaudible 00:21:53] paper here from the Nussenzweig Lab describes the cloning of monoclonal antibodies from about 140 or 150 donors in the New York area, early in the pandemic, I should say. So this is now the reason, obviously these are polyclonal. That’s why different colors of antibodies are shown in those schematic donors. Then what my lab went on to do is actually isolate the IgGs from those plasmas, which were sent to us from the Nussenzweig Lab. Then we compared recognition of different coronaviruses. And what we did was we had SARS-CoV-2, we had MERS, and we had three of the common cold coronaviruses. I’ll just give you a summary of that without showing the data. It’s in the paper that is referenced at the bottom of the slide.
We then went on to use a mapping technique to determine where the antibodies mainly bound for different plasmas. And then finally, we did higher resolution structures of individual monoclonal antibodies bound to the spike trimer. I’m not going to go into detail now about these [inaudible 00:23:19] ELISA and neutralization experiments. They’re all in that first paper. Basically, the plasmas were just different. So people had different, their antibodies were different in terms of their binding strengths, their degree of neutralization and their patterns of Coronavirus as protein recognition. What we found was people’s plasma IgGs would bind to the spike proteins from SARS, conventional SARS, and also MERS. And that likely represents cross reactive recognition since plasma donors from New York, or of course, unlikely to have been infected with either SARS or MERS.
They also bound the receptor binding domains of the common cold coronaviruses, but we did competition experiments to show that this was due to the fact that there were common exposures. So within the plasmas of these people who had had COVID-19, they still had antibodies to common cold coronaviruses. So now what we were trying to do with the plasmas originally was, can we determine the structural correlates of antibody mediated neutralization of Coronavirus infection? So the question we were first asking is, are different epitopes targeted in recovered individuals in their anti-polyclonal mixtures, or is there a single predominant epitope? So in order to address this, we use a technique called negative stain polyclonal epitope mapping, or negative stain EMPEM. This is a technique developed in Andrew Ward’s Lab at Scripps, mostly he’s done this for HIV and flu and other viruses. We just developed that using what he had worked out.
Let me show you the typical pipeline for a negative stain EM experiment to do polyclonal mapping. And this is based on Christopher Barnes, the postdoc in my lab, who led this work. This was based on, he worked it out in sera from HIV immunized rabbits. So you take the serum and you purify IgGs from that. You cleave it to make the antigen binding fragments, and then you determine how well they bind by ELISA. And then you incubate the Fabs that you’ve got and you run size exclusion chromatography, because you have to do this before the EM, or you’ll have an absolute mess. And you have to figure out how much to add because you won’t see anything unless you add enough. So the tighter binding the Fabs are, the better this will work.
You can see here a shift to the left on the third panel, on the right you can see that the complex in red, it bound very well. So we got Fabs binding in this case. Then you do the electron microscopy and you get three-dimensional reconstructions. You can see on the lower left, you can see an HIV spike trimer with three Fabs bound mainly at the top, which is a particular epitope we were interested in in these immunized rabbits. So note there are three there because HIV, the envelope is usually a symmetric trimer.
Now Christopher did this for the plasmas we got from the New York patients who’d had COVID-19, and this slide illustrates what happened. He got distinct predominant epitopes that were targeted by these convalescent plasma antibodies. The control is just the negative stain three-dimensional reconstruction of the spike trimer on its own. Then from a plasma from someone who’s called COV21, you can see in orange, you can see a predominant epitope. It’s actually on the RBD. There’s only one copy of it though, not three. And then for COV57, there was a predominant epitope in green, that has all down RBDs that spike. And that epitope is outside of the RBD. Most neutralizing antibodies, in fact, most of the antibodies are against the RBD, but this one was outside of the region. Then if you look at the reconstruction for antibodies from COV107, you don’t see anything. And the reason is in general, these Fabs from these convalescent plasmas bound very weakly compared to the immunized HIV animals that we’ve looked at by polyclonal [inaudible 00:28:08].
That was one thing that was interesting right off the bat, was that they don’t bind very well, the Fabs. So the binding is not great compared to what you get from immunizations with HIV immunogens. Then we were interested, we had these two plasmas from which Christopher was able to get this technique to work, and I’m just showing you where there are, of course, mutations in Coronavirus spike proteins. If you look in GIS A, G-I-S A, they keep sequencing Coronavirus genomes, they tell you where the mutations are. So we mapped them in the low resolution structure here. Basically from these plasmas, the identified mutations in SARS-Cov-2 spike were unlikely to affect these particular epitopes, which I think is good news. In particular, you may have heard of the D614 gene mutation, which has pretty much taken over in most parts of the world. That’s down there near the bottom. That is really far away from any of these antibody binding sites. There’s some evidence that it affects the ability of the spike to actually enter cells, but antibodies don’t usually bind to that place.
Now I’m going to go on and talk about doing single particle cryo-EM structures. This is the way you’d get to higher resolution so you can look at the particular interactions. All of these antibodies we looked at are against the receptor binding domain. So we’re going to be looking at anti RBD monoclonal antibodies. So the first one structure that Christopher got, again, these were cloned by Davide Robbiani and Michel Nussenzweig Lab. These are human monoclonal antibodies from COVID-19 patients. The first one Christopher got, a structure of, was called C105. And in cryo-EM, as you may know, you can get multiple states in a single structure. What Christopher found was what we called state one, where you had two Fabs bound. We’re only showing you the variable regions of the Fabs here for clarity. Those are in green, and the RBDs are always going to be in red now.
The two Fabs that were bound were both bound up RBDs, and the other RBD was there, but it was in a down position. And then there was another state where there were three Fabs bound and they were only bound to up RBDs. So we characterize that particular epitope for C105. So on the left, you see just the RBD in the gray space filling representation. And then you see the variable heavy and variable light chains of C105, they’re bound to it. Now the RBD has been rotated. So you can see the actual epitope and mapped upon the RBD, are the complementarity determining regions for that antibody. And then to the right of that, you can see in space filling representation, where those complementarity determining regions actually, where they have the most contact with the RBD.
Here is a comparison now to the surface recognized on the RBD by ACE2. So you might ask why does C105 neutralize? The answer’s really simple. It just blocks binding to ACE2. But in analyzing this further, it was interesting to us that actually most of the contacts were from CDRH1, CDRH2 and CDRL1, not from the CDR threes. In fact, if you look at the third over from the left, you’ll see a big gap in the center of the epitope, where CDRH three doesn’t really do much. I think this means this has been shown in a lot of structures of these antibodies from a number of labs in this point. So I think what it means is that it should be possible to use protein engineering, machine learning and so on, techniques to improve binding or selection in vitro selection to improve binding of C105 and other neutralizing antibodies. I think you’d probably get more potent antibodies that you could use at low concentrations therapeutically, for example.
Okay. So this epitope, it turns out, represents a binding class that is encoded by the germline VGM segment called VH3-53. It can also be encoded on something that’s really similar to that and called VH3-66. This is a recurring class that Michel Nussenzweig’s Lab found in many of their donors, not just one, and then all kinds of other papers from other labs finding this as well. So what we did, we meaning Anthony West and my lab, is he did some bioinformatic analysis and showed statistically that this was, it was something that was happening recurring in the different donors. Furthermore, this type of antibody had to have a short CDRH3, otherwise, if it was long, it can’t find to this particular epitope.
Now I’m going to go on to present some of Christopher’s new results. The interesting thing is he was able to very quickly solve eight new structures of antibodies bound to SARS-Cov-2 spike. Those are presented in a second paper. He actually led a group in my lab that was able to work together to get all these structures, a number of graduate students and so on. So everyone was working together cooperatively to get this done quite quickly. I’m only going to show you four of the new structures but they’re all in this paper that is now out. So the first, and these are examples, so C144, for example, binds only to down RBDs. And we found that C144 Fabs are now shades of blue and those bind three, C144 Fabs bind to three down RBDs. For C002, one of the states showed two down RBDs and one up RBD and all three had Fabs bound. So C002 combine to either up or down RBDs. That’s also true for C121 and C135.
I should say, for the other one I showed you for C105, that sterically cannot bind to down RBDs, it can only bind to up RBDs. So there’s all kinds of different types of antibodies, which I’ll go into in a moment about classifying them. So let me tell you a bit more about C144, that’s the one that binds to the down RBDs. So there’s the picture of its structure again, and this one’s really interesting. It actually bridges between adjacent RBDs and that locks the spike into a closed conformation, such that the RBDs can’t possibly come up. That’s its neutralization mechanism.
It could in theory, bind to an up RBD because its epitope is not occluded, but we’ve never seen that happening. But if it did, it would also block the ACE2 binding site on the RBD. But if you look at how it actually bridges between the RBDs, what you see is you’ve got, if you look at the heavy chain of C144, which is in the darker blue color, it binds to what we call, let’s say the primary RBD, and then the light chain binds to the adjacent RBD. And the amazing thing is that the CDRH3 of the heavy chain reaches across to the secondary RBD to insert two hydrophobic residues into a hydrophobic pocket on that RBD. And then this is three-fold symmetric. That’s repeated three times. So you’ve just locked this thing down. And on the right, you’re seeing the details of that. There are two adjacent hydrophobic residues, Phenylalanine 100D followed by Triphenylene 100E. And those interact with hydrophobic residues on the RBD Phenylalanine 342, phenylalanine 374 and so on. So this was a very interesting neutralization mechanism.
I just want to point out that we can use this type of structural data to assess whether the SARS-Cov-2 mutations confer resistance to antibodies. I already showed you an example of that in the negative stain mapping, but we can also do it for the monoclonals. But first I want to point out that really SARS-Cov-2 doesn’t mutate very much compared to other viruses. This is partly because it has a [inaudible 00:37:39] when it copies its RNA. But compared with influenza, let’s say, where we have circulating strains that vary every year, that’s why you have to get a flu shot more than once because they keep varying which type of influenza strains are in the shot. But compared with that, you can see the SARS-Cov-2 exhibits relatively few mutations. It’s almost like at the level of measles, which we have a very effective vaccine against, where you get it once in your life and you’re immune. Also, if you get the measle, if you are naturally infected, you’re immune for the rest of your life.
Now, it’s not real clear that if you get SARS-Cov-2, you’ve probably heard that some people can get reinfected. That has more to do with, not to do with the mutation necessarily, but due to memory responses, whether or not your antibodies stay around. But we started mapping where sequence variants that we found in the databases of all the SARS-Cov-2 sequences of the spike. You can see these here mapped on the two RBDs, where C144 binds on the left. Some of these would be effected by mutations that are known to occur. And then you can see on the other antibodies here. The interesting thing about this of course, is that we put all these mutations here, but they’re only found in less than 1% of the circulating SARS-Cov-2 isolate. So in a natural infection, this might not matter that much. But if you’re using monoclonal antibodies for treatment of an infection, you could select these mutations. In fact, in the selection assays here, Paul Bieniasz’s Lab [inaudible 00:39:26] were able to select these mutations in in vitro selection assays.
I’m almost done now, but we had this long table now in this paper here, where we did structural classifications of neutralizing antibodies against the spike and we classified them into four groups, class one through class four. Class four, we don’t have any structures of these, but other labs have done structures. These are not in general, very potent if they neutralize at all, some of them are non-neutralizing. So if you want to develop therapies, you’d want to use a class one, class two or class three antibody. Class one, an example is C105 that I showed you in the first Cryo-EM slides. That’s the one that has a short CDRH3. It comes from the VH3-53 class, and it binds to the RBD only when the RBD is up. That’s one type of antibody.
Class two also block O and it directly blocks ACE2. Class two also directly blocks ACE2 binding, but it can bind to its different epitope when the RBD is either up or down. Class three doesn’t directly overlap. Its epitope doesn’t directly overlap with the ACE2 binding site, but it neutralizes in a different mechanism and it can bind to the RBD epitopes when it’s either up or down. This is just a depiction of where these different types of antibodies bind. They don’t completely overlap with each other, but they’re in general, this is where they would bind. We’re showing you a single representative of each class on an isolated RBD right here. But I want to point out that this is somewhat confusing because if you want to ask, do different antibodies compete with each other, it’s more complicated than depicted on that slide. Let’s give an example.
If you had C144, which is class two, and C135, which is class three, their epitopes are different as you can see on this modeling on to an isolated RBD. So if you did a competition assay with RBDs, you’d get the answer that they don’t compete. But depending on what type of RBDs you have in the spike trimer, the RBDs are dynamic and they go up and down regardless of what is bound to them. Even if ACE2 [inaudible 00:42:06] bound, the RBDs are in an equilibrium, apparently between up and down conformations. You could get a situation where the same antibodies would actually compete with each other, depending on if they’re adjacent to each other. Both of them are bound to the same spike and the RBDs are in the particular positions where they would clash.
Here’s an opposite example. If you look at a class two antibody C119, and a class three antibody C135, these appear to clash on an isolated RBD, but depending again, on the way the RBDs are on the spike trimer, they don’t necessarily clash. So I just wanted to point out that for choosing cocktails, it’s a bit more complicated than just determining epitopes on isolated RBDs. You have to think about what’s going on on the spike trimer.
So with that, I’d like to close and take any questions you might have. I first want to acknowledge, well, the plasma donors and the fact that many researchers worldwide have switched to working on Coronavirus projects. There’s been a lot of structures out there. I didn’t have time to go into things that other labs have done, but it’s discussed in our recent paper where we tried to classify the antibodies into groups and the neutralizing antibody structures for the monoclonals. This is a collaboration with Michel Nussenzweig and his lab at Rockefeller at Caltech. The people that worked on this were led by Christopher Barnes, who’s the person in the red right here. And many people from my lab worked on this together with Christopher to get all these structures. I also want to thank the people in Michel’s lab and also Paul Bieniasz Lab at Rockefeller.
This is obviously a pre-pandemic picture. It’s my lab and the Protein Expression Center at Caltech, who helped us make, well, antibodies and viral antigens. We couldn’t do any of this work without a lot of help on the Protein Expression. So with that, I will close. I’ll take any questions. Okay. So I’ll read the questions and just stop me when the hour’s up. There’s a question from John Framer, how is the latent reservoir established and in which tissue and cell? How does the mutation change cell tropism? I think you’re talking about, yeah, you must be talking about HIV. So the latent reservoirs are every time you infect a cell, that has to be a CD4 positive cell. In the case of HIV, like any retrovirus, it takes the DNA copy of its RNA genome and integrates that into the nucleus of the infected cell. And then it may or may not be actively producing virus, but at any point that it gets activated, which is usually by, these are T-cells it infects because it’s CD4 positive T-cells for the most part.
If you get an infection or something like that, and the T-cells are activated, then that turns on the production of virus. How does the mutation change cell tropism? Well, you can eventually sometimes change to early in infection. The co-receptor is CCR5 G protein coupled receptor, and it can switch to CXCR4. You can have some macrophages infected, but really the majority of the cells that are infected throughout the infection are T-cells, their CD4 positive T-cells. When you lower the CD4 positive count way down as people do as they’ve progress to AIDS, your immune system is dysregulated and you get opportunistic infections and that’s what can kill people if they actually progress to AIDS.
Okay. So how has the molecular… This is from [inaudible 00:46:39]. These are all from, yes, I get it. John is sending them to me from other people. Okay. So how has the molecular arms race changed in the face of prep? So prep is, you give a single HIV drug prophylactically to people, which is really great. I didn’t go into this in detail, but as you may know, because of the high mutation rate of HIV, if you give a single antiretroviral drug, the virus will mutate to become resistant to it. So early on in the AIDS pandemic, they had reverse transcriptase inhibitors, such as AZT, and then other ones that were developed after that. And if you give just one drug, the virus just mutates and it’s resistant to it, and it’s fine.
After that, they started giving combinations of antiretroviral drugs. Like let’s say two reverse transcriptase inhibitors that bind to different places and then a protease inhibitor or an integrase inhibitor. So they did various things. Most people take a combination of antiretroviral drugs, usually three, four, or sometimes more. But you have to have at least three or the viruses can mutate fast enough to get around these drugs. And if you give at least three, they cannot mutate fast enough to get around three. So what prep is, is a reverse transcriptase inhibitor. I think it’s called Truvada. It’s just one of the antiretroviral drugs that’s given in combinations, but you give just one and then the person will not be infected by HIV because it’s enough to stop an infection, which as I said, happens with usually just one strain. So the prep works.
How has it changed? The idea might be that eventually, you would evolve strains that are resistant to that drug. Maybe you have to give people different types of drugs. I just don’t know enough about whether or not HIV strains have actually changed in the face of prep, but the real sad thing is that this could eliminate the HIV/AIDS pandemic in the world. Prep code is just not widely given. So it’s quite sad that due to social reasons or accessibility of the drugs or whatever, or people not even knowing that they’re positive. It’s just not given to everyone that it should be given to. I don’t know how that… I think it’s too early to know how that might’ve affected the so-called arms race.
Another thing is, how does HIV impact the aging processes? Well, that’s really interesting. I have no idea. I’m sure people are studying that. I just don’t know. How to make the immune system to identify latent virus? I don’t see how that could happen. There are various ideas for how you… How can you get rid of the entire latent virus? If you could somehow get all the HIV that’s integrated to come out, so the latent cells would start producing viruses. You could get rid of those viruses with antibodies or something like that. There are attempts to actually activate the immune system so that viruses come out of these latently infected cells and then kill them somehow. There are many ways to completely kill them, but so far no one’s managed to make a cure for HIV. So what you can do is you can treat HIV so that a person can have a fairly normal life on these antiretroviral drugs, but they’ll still be latently infected.
You may have read that recently, someone who was a long-term non-progressor actually is now negative for HIV message, HIV RNA that is. She went from having basically an infection that was under control for reasons that no one really understands, to not having any detectable virus. So I’m sure people are going to be studying her to figure out how that happened. Let’s see. Can you talk more about immunogenicity of engineered antibodies? Yeah. That’s an issue, In general, whenever you make a new antibody that hasn’t been seen before, that’s potentially a monogenic. So you get infected with some virus that your immune system hasn’t seen before and you make a new antibiotic with a new sequence. That’s immunogenic. So your body actually has mechanisms to make sure that new antibodies are not that immunogenic.
One of them is that apparently the FC region of an IgG antibody contains T-cell epitopes that are called Tregatopes or Tregatopes. It means there are peptides that are produced from an antibody, anything that has an FC on it that interaction with regulatory T-cells to turn down the immune response against new antibodies that are being produced. So one hope it that an engineered antibody that has an FC on it would not have all that many immunogenicity problems. So far the engineered antibodies for the most part, and we’ve been trying to do this for a long time, what happens to them is that you engineer them, they become more potent. We can make them more broad. We can do all kinds of things in vitro, but what happens is they become polyreactive, which means they bind to other things besides HIV, that our host cell proteins on host cells and so on and anything that has a lot of anti self-reactivity, even if it’s not dangerous to the person, just gets cleared immediately. So polyreactive antibodies just get cleared.
For us, we’ve had a lot of trouble engineering HIV antibodies so that they’re not polyreactive anymore. Immunogenicity is a problem in a sense, but if you wanted in the studies of antibodies given clinically to, so Michel Nussenzweig’s Lab gives antibodies to HIV infected people in a human clinical trial, and they haven’t seen that many anti-drug antibody responses. They’ve seen a few, but those are not engineered. Let’s see. If possible, can you discuss the molecular targets for diagnosis through the life cycle of the disease? I think that’s about SARS-CoV-2. Oh, okay. All right. So in the PCR test, what you’re doing is you’re amplifying up the viral RNA and almost always the primers for that amplification are to the N or the nuclear protein gene.
Usually in most of the tests, there are two primers to the N gene. Sometimes there’s a third primer to something called [ORF1 00:55:13]. So if you think about it, RNA comes in a coronavirus and a long, it’s just a single RNA. And at the five-prime end, it’s directly translatable because there’s a positive sense virus. At the five-prime end, or I think if one is over at the five-prime end, but then at the very end you have the end protein gene in the RNA. As it’s copying itself inside the infected cell, you get a lot more copies of the N gene. These are called sub-genomic fragments, a lot more, maybe 100 to 1,000 times more copies of N are in an infected cell than in the virus. So the virus has all the genes representative equally.
The infected cell has many more copies of N. So the PCR tests detect the N gene because you get greater sensitivity. So it depends on what you’re detecting. If you’re detecting infected cells, you’re going to pick up N really easily. If you’re detecting virus, you’re going to pick up N as easily or not as you would for any other gene in the virus. So they chose N deliberately, but of course, in the swabs that are done, you never know if you’re detecting virus or infected cells. You’re probably detecting both if the person is positive. Okay, that was the last question.
Thank you for joining Sanguine for our S3 webinar discussing Neutralizing Antibodies Against Pandemic Viruses. For a list of upcoming webinars or to request patient samples, visit sanguinebio.com. Thank you again and enjoy the rest of your day.