Page History

Joe Sodroski: [00:00:00] Thanks to the organizers. It's a real pleasure and honor to be here. I'll start just by reiterating that the coevolution of retroviruses and their hosts is complex and very intimate. You've, of course, heard that many immunodeficiency viruses have adapted through specific primate hosts, and in some [00:00:30] cases, they cause disease. Being asked to reminisce about my experiences in this field, I realized that I had forgotten most of them or maybe repressed many of them. But I thought that, early on, one of the questions as a trained physician that really fascinated me was this question. What is the basis for this virus being cytopathic because I had studied retroviruses previous [00:01:00] to this, like human T-cell leukemia viruses that weren't cytopathic and the in vitro cytotoxic effects of HIV were really impressive to me.

We and Jeff Lifson also showed that really the HIV envelope glycoproteins really could recapitulate many of the cytotoxic effects of HIV in tissue culture, syncytium formation, and single-cell lysis. Then Norman Letvin (1949–2012) and I went on to develop SHIV models [00:01:30] that could cause CD4 lymphocyte depletion in vivo, an AIDS-like disease. We use those models to show that the membrane fusing capacity of the envelope glycoproteins, which are important for cytotoxic effects in tissue culture, actually can modulate the ability of viruses to cause CD4 lymphocyte depletion in rhesus monkeys, even when the levels of viremia are normalized.

This [00:02:00] is a theme that went through a lot of my research and really led me to think about envelope glycoproteins in detail. Along the way, I have to say I got distracted by a lot of other questions, some of which you've heard about already. One of the things that distracted us initially when we were thinking about animal models was this observation that we and Paul Bieniasz and others made that these immunodeficiency viruses encounter very specific blocks to infection [00:02:30] of certain primate lineages. SIV, for example, naturally infects Old World monkeys, and can infect the cells of apes and humans in tissue culture, but after the entry process, it encounters a block in New World monkey cells.

Likewise, HIV-1 encounters a post-entry block in the cells of the Old World monkeys. [00:03:00] These blocks basically target the viral capsid. If you take Old World monkey cells that are blocked to HIV-1, but can be infected by SIV, and you make chimeric viruses, these chimeric SHIV viruses that have SIV capsids but HIV envelopes will infect those cells and then the viruses that have their SIV except for HIV-1 capsids are blocked. [00:03:30] The HIV-1 capsid is actually targeted by these restriction factors. I convinced a very talented graduate student in my lab Matt Stremlau that this would make a really good thesis project.

Fortunately for me, Matt believed that story and went ahead and took on a very risky and challenging thesis, which was basically to screen monkey libraries [00:04:00] for factors that would block HIV-1 at this reverse transcription process. Matt was fortunate enough to pull out monkey TRIM5α, which basically recapitulated these post-entry blocks to HIV-1. We now know that TRIM5α is a member of these large family of proteins called the tripartite motif family of proteins that have [00:04:30] RING domains that can ubiquitylate proteins, B-box and coiled coil domains that allow these proteins to oligomerize. They also have a C-terminal domain called a B30.2 or a SPRY domain that's involved in specifically recognizing their targets.

Particular retrovirus capsids are recognized by this B30.2 or SPRY domain. These domains basically [00:05:00] form a lectin, almost like an immunoglobulin fold. Then they have these species-specific variable loops that project from them and determine the specificity for the viral capsid. They're almost like intracellular antibodies that basically recognize capsids when they come in. The capsid surface is a complex array made up of a single protein that forms hexamers. Then those hexamers oligomerize [00:05:30] to form the large capsid array. This is what the TRIM5α proteins have to see. They do so by themselves oligomerizing.

First of all, TRIM5α forms an antiparallel heterodimer with these B30.2 domains poised to interact with the capsid surface. The coiled coil and the B-boxes basically set up these [00:06:00] proteins to self-oligomerize. Basically, what happens is these proteins form a large array that binds to the capsid surface and causes premature uncoating so that the capsid falls apart, blocking reverse transcription. [unintelligible 00:06:20] have shown very nicely that these arrays of TRIM5α proteins are also hexameric in nature. [00:06:30] They form a different kind of hexamer that in some way causes the capsid to prematurely uncoat.

That's one of my distractions and I just want to turn back to virus entry of which I've been most interested in over the years because there's another selective process that occurs for these two lineages of immunodeficiency viruses, the SIVs and HIV-1s, in adapting to replication in monkeys [00:07:00] and humans respectively. I'll turn to the entry process and just give you a quick summary of what we think of the conformational changes that happen in the envelope glycoproteins during the virus entry process. The envelope glycoproteins sit in a hyperstable and metastable state that we now call state one.

[00:07:30] They're driven initially by receptor binding, CD4 binding to an intermediate state, state two, and then ultimately to state three as more CD4 molecules attached to the trimer and form this pre-hairpin intermediate conformation. This is competent for binding to the second receptor, CCR5, which drive the envelope to the post-fusion six-helix bundle in gp41, bringing the viral [00:08:00] in the target cell membranes together. Now, most of what I'll talk about today involves this first transition, which is very tightly regulated in viruses between state one and this new intermediate state that Walter Mothes initially found by single molecule fret, and which we've now shown to be a functional intermediate on the virus entry pathway.

Well, it turns out that the SIV and HIV-1 lineages have adapted to different target [00:08:30] cell levels of CD4 expression in monkey and human target cells. The levels of CD4 for many of the target cells for SIV, probably in macrophages, as well as some of the lymphocytes seem to be low. SIV, therefore, has had to adapt to make sure that it triggers the state one to state two transition more readily. It has to be more prone to respond to these low levels of CD4 than as HIV-1 [00:09:00]. Now, how does SIV accomplish this? Well, this slide shows you the gp120 glycoprotein binding to the CD4 from crystal structures that we solved 18 years ago with Wayne Hendrickson.

Basically, I want to show you the hotspot for gp120 CD4 interaction shown in this box here. If we cut away to that, there's a critical residue on CD4, phenylalanine 43 that inserts [00:09:30] into a highly conserved pocket on the gp120 surface. In back of that pocket, though, is a, what's called the Phe 43 cavity which goes deep into the interior of HIV-1 gp120 and is essentially empty in HIV-1. In all the SIV lineages, that Phe 43 cavity is filled by bulky typically hydrophobic residues like tryptophan 375 [00:10:00] and the filling of that cavity basically makes the SIV envelope glycoproteins more prone to respond to low levels of CD4. They're essentially pre-triggered to go into the CD4-bound conformation. That is a somewhat dangerous predicament for a virus because if you open this State 2 before you get to the target cell, this state is highly susceptible to antibody [00:10:30] neutralization. If you look at all of the SIVs they've evolved a very large variable gp120 loops, which we think basically prevents this premature opening. They're particularly triggered to spring when they see natural CD4, but they don't spring spontaneously to make them susceptible to antibodies.

That's how the SIV lineage has evolved. HIV-1 is actually [00:11:00] different than we think. Mostly HIV-1 really likes lymphocytes with lots of CD4 on them and it doesn't have to set these triggers at quite a low level as we see for SIV. If we look at the phenylalanine 43 cavity and HIV-1s, it's essentially open and available. That creates particularly, I think, attractive targets for small molecule drugs [00:11:30] which we with Wayne Hendrickson and Amos Smith at Penn have been designing rationally over the years.

These small molecules, CD4-mimetic compounds, basically fit into this highly conserved cavity on the gp120 surface, and the consequences of their binding are quite interesting. First of all, they block CD4 binding by directly competing for CD4 binding. They can also [00:12:00] trigger this conformational change from State 1 to State 2 which basically is a short-lived state that is irreversibly inactivated. These compounds have direct antiviral effects on their own.

What I want to talk about today is the fact that they can open up the envelope and basically increase its susceptibility to antibody neutralization and to antibody-dependent cytotoxic responses. I won't talk about ADCC at all [00:12:30] but basically infected cells that have envelope-like proteins can be triggered to become highly susceptible to ADCC mediated by antibodies that are present in all infected individuals.

Today I want to talk about antibody neutralization. I'll introduce this afternoon session where you'll hear a lot about attempts in vaccine research to make broadly neutralizing antibodies, the bNAbs. Most of those bNAbs [00:13:00] recognize State 1, the mature unliganded state of envelope although we know now that a few of them recognize the intermediate State 2. The problem with bNAbs, as you'll hear, is that they're very difficult to elicit. They're elicited in some HIV infected individuals and so far have been very difficult to elicit by any vaccine immunogens.

I want to turn to another set of antibodies directed against the envelope glycoproteins, the easy-to-elicit [00:13:30] antibodies. These are antibodies that are made at high titers during natural infection, and they're elicited by almost all of the vaccine candidates that we've looked at to date. Many of these antibodies recognize highly conserved structures on this State 2 intermediate state but because this state is difficult to get to for antibodies in the virus target cell synapse, those antibodies sterically cannot [00:14:00] access those envelope. These antibodies are typically non-neutralizing or weakly neutralizing. We call them weNAbs.

What can we do about this situation? Well as I mentioned, the CD4-mimetic compounds can basically activate the State 2 on viruses that have not yet engaged the target cell. These viruses become very susceptible to neutralization by these, [00:14:30] otherwise, weakly neutralizing antibodies. This is shown in tissue culture on this slide. Basically, here we're taking monkeys that have been immunized with gp120 and gp140 immunogens. Six plasma from those monkeys were basically tested at different titers for their ability to neutralize this primary virus, HIV JRFL. You can see that basically they don't neutralize the virus [00:15:00]very well.

If the virus is exposed to a sub neutralizing concentration of a CD4-mimetic compound, BNM-III-170, you can see that those viruses all become exquisitely sensitive to neutralization by these plasma. This effect is specific because viruses pseudotyped with a mirroring leukemia virus envelope are not inhibited in the absence or presence [00:15:30] of the CD4-mimetic compound.

Can we take this tissue culture observation and ask whether these types of compounds can synergize with HIV envelope glycoprotein vaccine candidates to protect monkeys from intrarectal challenge with a transmitted founder SHIV virus? I guess this is my contribution to thinking about the future because I think these types of compounds might be used in prep type settings to interface [00:16:00] with immune responses that we can elicit by vaccines to protect against immunodeficiency virus infection.

Can we establish proof of principle in this kind of an animal model? Well, here's the study that we did where we basically took three groups of monkeys. Group 1 monkeys were immunized with a control immunogen, human serum albumin. The Group 2 and Group 3 monkeys were immunized with a clade C [00:16:30] gp120 glycoprotein. At the time of intrarectal challenge, basically, the Group 1 viruses were mixed with the CD4-mimetic compound, and then that virus compound mixture was immediately applied to the rectal mucosa. That same interaction between the CD4-mimetic compound and the challenge virus occurred for the Group 3 monkeys. The Group 2 monkeys simply, [00:17:00] the challenge virus was mixed with DMS the carrier and then applied intrarectally.

The challenge virus is a SHIV-C5 challenge. This is a heterologous challenge. It's a different clade C envelope than the immunogen. This is a clade C transmitted founder virus SHIV. The SHIV has tier 2 neutralization sensitivity, [00:17:30] and we used three high dose intrarectal challenges. These monkeys were challenged with 3.5 animal infectious doses per challenge. We expected that this dose about 91% of naive monkeys will become infected after a single inoculation. This is a very stringent challenge.

The next slide shows you the Kaplan-Meier survival curves for those monkeys showing you the percentage of monkeys that [00:18:00] remain uninfected after the three SHIV-C5 challenges. Let's start with the Group 2 monkeys that received the gp120 immunization alone without the CD4-mimetic compound. Seven out of eight monkeys became infected after the first challenge and the remaining uninfected monkey became infected after the second challenge. This is about what we would expect for naive monkeys. The vaccine didn't do very much against this very stringent [00:18:30] high dose intrarectal challenge.

The Group 1 monkeys that received the CD4-mimetic alone also became infected. Importantly, the Group 3 monkeys that received the CD4-mimetic compound and the gp120 vaccination were significantly protected from these high dose intrarectal challenges compared with either of the other two groups alone. Essentially, all eight of the monkeys resisted two high dose intrarectal challenges and 75% [00:19:00] of the animals resisted three high dose intrarectal challenges.

CD4-mimetic compounds can sensitize primary HIV to neutralization by readily elicited antibodies. BNM-III-170, one of the more recent CD4-mimetic compounds that we've rationally designed and a gp120-induced antibody response can synergize in protecting against high dose intrarectal challenges [00:19:30] with a heterologous transmitted founder clade C SHIV virus. Basically, CD4-mimetic compounds should increase the protective efficacy of any HIV-1 vaccine that is not itself 100% protective and that elicits antibodies capable of neutralizing State 2 HIV-1. That includes most of the envelope immunogens that we've looked at today.

We think that CD4-mimetic compounds in the future may be a useful

tool to synergize and to basically complement efforts to make protective vaccines. I want to thank all of the people involved in this work. Certainly many people from the University of Pennsylvania and Amos Smith's lab, Walter Mothes' lab at Yale, Wayne Hendrickson at Columbia, and Bart Haynes and his group at Duke as well as Sampa Santra at [00:20:30] Beth Israel Deaconess Hospital. I want to thank our funders and thank you for your attention.

[applause]

Paul Bieniasz (moderator): Okay. We have time for questions. Perhaps I'll start with one. What's the breadth of the CD4-mimetic compounds? Is it different? Is it antiviral versus a weNAbs sensitizer?

Joe Sodroski: The breadth of the CD4-mimetic compounds is dictated by the compounds themselves. They don't [00:21:00] develop more breadth as sensitizers than they do as antivirals. Their breadth is about 85% of HIV-1 strains. At this point, they will not neutralize, for example, the Ae recombinant viruses, which have a histidine in their Phe43 cavity, so the compounds can't get into those viruses. The AAE viruses are not going to be susceptible to this type of inhibition. All the other viruses are potentially sensitive and [00:21:30] as we make more and more potent compounds the breadth is getting better. We think we can improve them even further. Yes.

Paul: Michael.

Michael Emerman: Joe, I want to ask a historical question. Actually, I used the Nature paper where we found TRIM5α and teaching as an example of how to do a good screen. Question, at what point did you realize that the clone [00:22:00] you had matched the biological activity you were looking for?

Joe: Pretty early on. I guess, one thing that we learned in that screen is that it's worth getting someone to make a good cDNA library for you that knows how to do it. Matt made a few cDNA libraries. We screened them, and we didn't really come up with anything. Then we said, "Well, maybe we should get one made by a company." That $10,000 [00:22:30] that we spent to make that cDNA library was well worth it.

[chuckles] Fairly early on, we had this TRIM5α and where we could do all the specificity controls for SIV and HIV-1. I would say pretty early in the game we knew that we had something that seemed to recapitulate the biology.

Paul: Lee.

Lee: Hey? Hi Joe. I wonder if those CD4-mimetic compounds. [00:23:00] Do they affect the amateur C class two molecule binding to CD4? Therefore me a fact the energy presentation cell interaction with the CD4s?

Joe: No. The CD4-mimetic compounds interact specifically with gp120. They affect the ability of gp120 to bind to CD4, but they have no effect on CD4 or cellular functions as far as we can tell. They're relatively [00:23:30] nontoxic and have no effects on cellular functions that we can see.

Paul: Yes.

John Coffin: Have you selected resistance to those compounds?

Joe: We can make virus resistant to those compounds.

John: What are the properties of that virus?

Joe: The viruses can become resistant to CD4-mimetic compounds by basically stabilizing State 1. We know we can drive viruses to become [00:24:00] more State 1 preferring those viruses don't infect cells quite as well with low levels of CD4. They don't make the State 1 to State 2 transition as easily. Yes, you can make viruses resistant. That's why I think the best application of the CD4- mimetic compounds is going to be in prevention rather than using them for treatment.

There's a lot of interest in cure because they will make HIV-infected [00:24:30] cells highly susceptible to ADCC by antibodies that are very high titers and infected individuals. If you shock all the cells into producing envelope maybe you could do limited treatment with the CD4-mimetic compounds.

John: Would not prevention require continuous dosing of uninfected individuals at risk for infection?

Joe: You want to present this in a sustained fashion just as you would in any prep type of approach. [00:25:00] Sure.

Paul: Harriet.

Harriet Robinson: I just wanted to clarify that your mimetic was at the time of challenge that you were adding it.

Joe: Yes. The mimetic—

[silence]

Professor Sodroski: [00:25:30] Obviously, we have to do more in terms of primate toxicity for these who are testing those now. At this point, they don't appear to be toxic. We've used them in BLT mice to show that they basically protect mice from vaginal challenge. More has to be done there. We're at a very early stage. This is the proof of principle, I think.

Participant 1: The second and last question. [00:26:00] You weren't here earlier, which is common. One of the points, from our experience, I don't know if it's your experience, but we find antibodies that are protective in the primary experiments they don't last very long. When we try to make them last longer and we activate too many T-cells, we lose all efficacy. Have you thought about that as a problem? We see it with almost any gp120 antibody that has a useful [00:26:30] effect. To us, no matter what the immunogen is, we've got a big problem in the future even if we have the perfect immunogen to get the antibodies to last a bit longer.

In other words, if you go out further, you didn't go out very far. When you compare, let's say, HPV, which goes out for years, these things crash. If you look at the RV 144 trial, which I assume Bart will mention, the army trial, if you see they originally they had 62% protection. That was not bad, [00:27:00] but it crashes. If you follow the antibodies to gp120, they paralleled the crash and they have very short protection time. Then, the overall efficacy came down to mediocre, less than mediocre level and they didn't cause much excitement.

Professor Sodroski: Well, I think we have to study the duration of these antibodies. All of these experiments were done shortly after the boost. Within two to four weeks after the last boost, that's when we did the challenges. [00:27:30] We're currently letting the animals rest and seeing how long those antibodies stick around. I think that is a problem. It's something that's going to have to be addressed by all vaccine candidates. There is various delivery methods that we hope will make more sustained responses. But, yes, that's clearly an issue.

Participant 1: [inaudible 00:27:51] get the information [inaudible 00:27:56]

Professor Sodroski: Yes. Certainly, that will be [00:28:00] something that will be important for all vaccine.

Paul: Okay. Thanks Joe.

[00:28:07] [END OF AUDIO]