The Antibody Race
This year’s Keystone symposium on HIV vaccines, held jointly with B-cell immunologists, focused unapologetically on antibodies against HIV
By Andreas von Bubnoff
Organizers of the annual Keystone symposium on HIV vaccines like to mix things up a bit, combining their gatherings with those of other research tribes. They do this ostensibly to encourage the exchange of ideas—though to what extent that actually occurs is anybody’s guess. But this year, it wasn’t much of a stretch: The co-hosts of their latest meeting, which was held in Keystone, Colorado, from Feb. 10-15, were folks concerned primarily with B-cell development and function.
For observers of the field, this should come as no surprise, given the many antibody-related advances the field has seen in recent years. “We now have an increasingly detailed map of the vulnerable sites on HIV Envelope due to an array of newly discovered broadly neutralizing antibodies and the identification of their targets,” explained Georgia Tomaras of the Duke University Medical Center, who co-organized the HIV vaccines track of the conference. Capitalizing on that information, she said, is a priority of the field—sufficient reason, at any rate, to have the B-cell folks over for the conference.
And, indeed, antibody aficionados were probably not among the disappointed in Colorado that week.
Toward passive protection
Now that dozens of broadly neutralizing antibodies (bNAbs) have been isolated from HIV-infected people, the next logical question is whether their presence in uninfected people prevents infection. One approach to finding out is to devise vaccine candidates that might coax a naïve immune system to make anti-HIV bNAbs of its own. Another is passive immunization. The advantage of the latter approach is that researchers won’t have to cross their fingers and wait for the immune systems of vaccine recipients to go through the long process of affinity maturation that leads to the cherished bNAb.
Besides, a successful passive immunization trial would provide proof of principle for an antibody-based vaccine and perhaps some guidance on how much neutralization activity might be needed for protection from HIV, said Barney Graham of the National Institute of Allergy and Infectious Diseases (NIAID). “Many of our licensed vaccines [such as measles vaccine] were preceded by demonstrations that passive immunization is effective in preventing or diminishing disease,” he said.
Graham, who gave an overview of passive immunization projects at the NIAID’s Vaccine Research Center (VRC), said that VRC researchers have recently shown that passive immunization with VRC01, one of the broadest and most potent HIV-specific bNAbs, can protect rhesus macaques from challenge with SIV/HIV hybrid (SHIV) viruses. The VRC researchers injected human VRC01 intravenously or subcutaneously into animals. They then challenged the animals two days later, vaginally or rectally, with SHIV SF162P3—which is relatively difficult to neutralize—or the more easily neutralized SHIV BAL.
The result: At a concentration of about 40 micrograms per milliliter of blood, VRC01 protected all animals from challenge with SHIV SF162P3. Concentrations of only five micrograms per ml, meanwhile, sufficed to protect the animals from SHIV BAL. The higher concentrations, Graham said, are routinely achieved in transfusions of a prophylactic antibody named Palivizumab, which is given to infants to prevent infection with respiratory syncytial virus.
Graham said VRC01 is currently the only bNAb manufactured at a GMP grade, and the VRC plans to apply for FDA approval to begin a Phase I passive immunization trial using human VRC01 in adults at the NIH clinic in Bethesda in the fall of this year. “VRC is potent enough and manufacturable enough and deliverable enough to answer the question whether an antibody with a certain level of neutralizing activity can protect” people from HIV, Graham said.
One limitation of the VRC’s macaque experiments using human VRC01 is that human antibodies only have a half-life of 4-5 days in the animals, since the animals launch an immune response against the human antibodies. That’s why the VRC researchers waited only two days after infusing the antibody before they challenged the macaques with virus.
To reduce antihuman immune responses, VRC researcher Kevin Saunders grafted the human part of the antigen-binding region of VRC01 onto a monkey antibody. This increased the half-life of the molecule to about nine days. A one-time infusion with the “simianized” VRC01 protected the animals from infection for two weeks. While the half-life of VRC01 in humans isn’t known, the half-lives of antibodies of this type are typically about 21-24 days in the human body. Graham said this suggests that VRC01 will probably have to be given once a month to volunteers in the proposed trials.
While VRC01 is potent enough to get meaningful results in human trials, VRC researchers have isolated another bNAb called VRC07 from the same donor who provided VRC01. This antibody is about 2.5 times more potent and neutralizes 93% of circulating HIV variants, as opposed to VRC01’s breadth of 91%. They have also made dozens of modifications to several bNAbs that make them up to 15 times more potent, or increase their half-lives, Graham said.
But these modifications come at a price: The “vast majority,” said Graham, render the antibodies autoreactive—meaning that they are likely to bind to the body’s own tissues. Because the immune system normally eliminates B cells that make such antibodies, naturally occurring bNAbs are less likely to be autoreactive.
There are exceptions, however: in vitro studies suggest, for example, that two bNAbs— b12 and 4E10—might be autoreactive. But in vitro experiments may not always reflect the in vivo situation, said David Nemazee of The Scripps Research Institute, adding that the real test of whether the autoreactivity of an antibody matters in vivo is to study whether these antibodies are actually likely to be eliminated by the immune system.
To test this, Nemazee and colleagues made transgenic mice whose B cells either produce b12, which targets the CD4 binding site of Env, or 4E10, which targets the part of Env on the HIV surface that’s closest to the viral membrane, known as the membrane proximal external region (MPER). They found that most of the 4E10-producing B cells were indeed eliminated in the mice. However, this was not the case for b12, suggesting that in vitro tests of autoreactivity don’t always reflect the situation in vivo. Two years ago, Barton Haynes and colleagues also used transgenic mice to show that the MPER-specific bNAb 2F5 is eliminated in mice. Together, these findings suggest that the MPER part of Env might have a tendency to induce autoreactive bNAbs (see Research Briefs, IAVI Report, Sep.-Oct. 2011).
Inducing the elusive bNAb
The alternative to directly infusing bNAbs into people is, of course, to develop a vaccine that coaxes the immune system to make them itself. That’s no easy task. bNAbs are only found in a fraction of HIV-infected people, and they take years to develop because they need to mature and mutate away from their germline precursors through a process called affinity maturation. Unfortunately, the native HIV Envelope protein binds these germline precursors only weakly or not at all, while it binds strongly to the final, mature bNAbs. A vaccine that contains the naturally occurring Envelope, or parts of it, as an immunogen would not be expected to reliably bind and activate B cells in uninfected people. It would therefore fail to kick-start the affinity maturation process that gives rise to bNAbs.
To address this problem, researchers are developing versions of HIV Envelope, or artificial immunogens that are similar to Envelope, that can bind the germline precursors of bNAbs and so initiate the affinity maturation process. Later immunizations could then use immunogens that resemble more natural versions of Env that can bind the more mature bNAbs and so guide the affinity maturation towards the most potent versions of bNAbs.
At the meeting, Bill Schief of The Scripps Research Institute reported that his team has constructed artificial immunogens bearing versions of HIV Env sites recognized by HIV-specific bNAbs. To present these “engineered outer domains” (eODs) to the immune system, the researchers made a virus-like nanoparticle that can present 60 eODs on its surface. Researchers believe that the repetitive display of an antigen on the surface of a virus-like particle is an important feature of successful vaccines (see Taking the Gritty Approach, IAVI Report, Nov.-Dec. 2012).
Schief said his team has developed several different eODs. One of them, eOD17, resembles the CD4 binding site on HIV Env. It binds strongly to CD4-binding-site specific bNAbs and activates mature B cells that have a VRC01-like B-cell receptor. Because it binds much less strongly to non-neutralizing antibodies, it is also less likely to induce them. Other eODs have been engineered to resemble the CD4 binding site of five different HIV clades, so that they might induce neutralizing antibodies that are broader in their specificity.
Schief’s team also developed an eOD named eOD-GT6 that can bind mature VRC01-like bNAbs as well as their (inferred) germline versions (Science 2013, doi: 10.1126/science.1234150). When presented as a 60-mer on a nanoparticle, the immunogen potently activated germline B cells in laboratory studies. Because the immunogen binds to both germline and mature versions of VRC01-like bNAbs, it could in theory be used as a prime to kick-start the affinity maturation process, and then possibly again as a boost to guide the process towards the more mature bNAbs (which it binds more tightly). Next, the Schief team will use their eOD-GT6 nanoparticle immunogen to immunize rhesus macaques.
Andrew McGuire of the non-profit research institute Seattle BioMed reported that he and his colleagues have developed versions of Env that can also bind putative germline ancestors of bNAbs (J. Exp. Med. 2013, doi: 10.1084/jem.20122824). He achieved this by changing just one amino acid in the protein to prevent the addition of a sugar that is ordinarily added to that site. McGuire said the resulting Env protein can bind germline versions of the bNAbs VRC01 and NIH 45-46, which were obtained from the same donor. The modified Env can also activate B cells that express the B-cell receptors of the germline precursors of those bNAbs. This suggests it might induce the production of bNAb precursors if used as an immunogen.
McGuire isn’t the first to show that removing sugar groups enables Env to bind germline versions of bNAbs. Two years ago, Haynes and colleagues showed that removing part of the sugar groups enabled Env to bind to the germline versions of the MPER-specific bNAbs 2F5 and 4E10 (see Research Briefs, IAVI Report, Sep.-Oct. 2011).
One challenge when using Env for immunizations is that the protein is highly unstable. It even vibrates, according to Quentin Sattentau from the University of Oxford. As a result, it’s difficult to induce protective immune responses using the protein—its constant instability distracts the immune system by offering up a parade of changing structural targets. “The B cells have trouble hitting the right epitopes because it’s like jelly,” Sattentau said.
Schief and his colleagues address this problem by creating stabilized artificial immunogens. Sattentau, on the other hand, stabilizes Env by treating it with glutaraldehyde (GLA), a chemical that crosslinks certain amino acids and so “fixes” proteins in a relatively static structure. When Sattentau and his colleagues treated soluble gp140 Env trimer with GLA, the fixed Env was less likely to fall apart than untreated Env. What’s more, the GLA-treated Env bound as well to bNAbs such as VRC01 as the untreated Env, indicating that the GLA treatment hadn’t changed the bNAb binding sites. In fact, the fixed Env actually bound weakly neutralizing antibodies less well than the unfixed Env, suggesting that it offers fewer distracting targets to the immune system than the “jelly-like” unfixed version.
In collaboration with IAVI, Sattentau and colleagues then used the fixed Envs to immunize rabbits. After four immunizations, there were more neutralizing antibodies in the animals immunized with the fixed Env than in those injected with the normal protein. Further, the modified Env elicited antibodies against the CD4 binding site, and to exposed parts that are normally quite flexible, such as the variable regions V1 and V2. In contrast, the unfixed Env elicited antibodies to Env regions that are normally hidden, consistent with the view that the unfixed Env is less stable. Next, Sattentau plans to immunize macaques with fixed Env trimers.
Researchers are also trying to better understand what factors are important for the induction and maturation of bNAbs. These certainly include T follicular helper (Tfh) cells, a type of CD4+ T helper cell that interacts with B cells in the germinal center of lymph nodes and is thought to be central to affinity maturation. Analysis of Tfh cells might therefore help in assessing how well a vaccine induces bNAbs. While it is not possible to routinely do biopsies to isolate Tfh cells from lymph nodes, researchers have recently identified markers that can be used to track them in blood.
Using such markers, Hendrik Streeck, who is now at the US Military HIV Research Program, isolated Tfh cells from the blood of HIV-infected people who make bNAbs and from those who do not. When he cultured the Tfh cells from the former with naïve B cells, he found that the Tfh cells were better at inducing antibody-producing plasma cells than their counterparts isolated from folks who do not make bNAbs.
Streeck also found that IL21, a cytokine made by Tfh cells, was essential to this capability. In other words, it appears that Tfh cells from people with bNAbs are better at inducing plasma cells, and that IL21 might be involved in this process.
RV144: the gift that keeps on giving
The analysis of plasma samples collected in the RV144 HIV vaccine trial—the only one so far to have come up with even a trace of a positive result—suggests that IgG antibodies to the V1 and V2 loops of Env were associated with the modest protection afforded by the regimen evaluated.
To better understand what class of IgG antibodies those were, Tomaras and colleagues compared plasma samples from RV144 vaccine recipients with samples from an unsuccessful trial called VAX003. In VAX003, vaccine recipients didn’t receive the pox vector prime that was used in RV144, but did receive the HIV gp120 immunization that was used as a boost in that trial. Given that RV144 detected protection from HIV and VAX003 didn’t, Tomaras was looking for any immune responses that were better in RV144 than in VAX003. Such responses, she reasoned, would give clues about which immune responses can protect from HIV infection.
Tomaras and colleagues found that the RV144 samples had higher levels of an IgG subclass called IgG3 that was specific to V1/V2 of Env. This suggested that IgG3 antibodies might be responsible for the IgG-mediated protection observed in RV144. To test this hypothesis, Tomaras and colleagues reanalyzed plasma samples from 41 RV144 vaccine recipients who got infected and 205 vaccine recipients who did not. The samples in question were taken just before the first vaccination and two weeks after the final vaccination. Their analysis revealed that protected vaccine recipients indeed had higher V1/V2-specific IgG3 responses than unprotected vaccine recipients.
IgG3 antibodies are known for their ability to recruit antiviral effector cells by binding to the Fc receptor of such cells through their Fc portion. Therefore, the IgG3 involvement in protection suggests that such Fc receptor-mediated mechanisms might have played a role in protection in RV144.
One example for such a mechanism is antibody-dependent cellular cytotoxicity (ADCC), in which antibodies that are bound to an HIV-infected cell recruit innate immune cells (such as natural killer [NK] cells) that then kill the infected cell. This may explain another, rather surprising, finding from the analysis of RV144 samples—that blood levels of IgA antibodies specific to a part of Env called C1 were associated with greater risk for HIV infection in vaccine recipients: Tomaras reported that IgA antibodies might have blocked IgG-mediated ADCC. To show this, she and her colleagues isolated monoclonal C1-specific IgA and IgG antibodies from the same vaccine recipient and found that, at least in vitro, these C1-specific IgA antibodies kept the C1-specific IgG antibodies from binding to their target. As a result, the IgG antibodies couldn’t activate the NK cells required for ADCC.
The vaccine used in RV144 contained alum, an adjuvant that has been around for decades and is used even today in most licensed vaccines. It significantly boosts antibody responses, but is less effective in inducing cellular immune responses. To find better adjuvants for future HIV vaccine candidates, researchers are studying the mechanism by which alum and other adjuvants enhance the effects of vaccines.
But surprisingly little is known about how adjuvants, including the old standby alum, exert their effects. That includes their in vivo effects in the first 24-72 hours after immunization. To find out what happens in those early stages, Karin Loré from the Karolinska Institutet, who is currently a visiting scientist at the Vaccine Research Center (VRC), injected different muscles of rhesus macaques with HIV Env protein, combined with three different adjuvants: alum alone; alum combined with the toll like receptor agonist 7 (alum/TLR7); and the oil-based adjuvant MF59. Loré said she chose to study the mechanisms for these adjuvants because studies by VRC researcher Robert Seder in macaques suggest that both alum/TLR7 and MF59 elicit better B- and T-cell responses than alum alone. MF59 is an adjuvant that has been licensed in flu vaccines in Europe and other countries outside the US.
Peering into the injected muscles and lymph nodes that drain the injection sites 24 hours later, she found that all adjuvants caused a migration of more immune cells into the draining lymph nodes than did no adjuvant.
But there were differences: Alum/TLR7 induced type I interferon alpha, which inhibits viral replication and increases antigen presentation by dendritic cells (DCs) to CD4+ and CD8+ T cells. In contrast, MF59 had less of an effect on DCs, but caused a migration of neutrophils into the draining lymph nodes.
In additional experiments, Loré showed that neutrophils taken from the injected animals’ lymph nodes could act as antigen presenting cells: They presented Env antigens to activate memory CD4+ and CD8+ T cells, albeit at lower levels than DCs.
This, Loré said, suggests that MF59 might make up for its deficiencies in antigen presentation by DCs by inducing antigen-presenting neutrophil migration into the lymph nodes. “We found,” she said, “that although DCs are superior at presentation, as expected, neutrophils can do some of the job. The fact that [neutrophils] are so frequent [with MF59] makes you wonder if what they lack in potency they make up for in numbers.”
How this translates into vaccine efficacy remains to be seen. But, for now, Loré’s experiments suggest that alum alone may not be the best adjuvant to get responses that protect against HIV. —AvB
Antibodies and mother-to-child transmission
Even though antiretroviral therapy (ART) reduces the incidence of HIV transmission from mother to infants to 1-2%, more than 300,000 mothers transmit HIV to their infants each year. About half of these transmissions are believed to occur through breast milk.
But even without ART, 90% of children are protected from HIV transmission from their mothers. One reason for this, a recent study suggests, is that women who don’t transmit HIV have antibodies in their breast milk that can better induce ADCC (PLoS Pathog. 8, e1002739, 2012). “It’s not just how much antibody you have in breast milk that’s specific for HIV,” said Sallie Permar of Duke University Medical Center. “It’s really the function of the antibody that’s in breast milk [that determines] whether you will be better protected.”
To learn how to reduce mother-to-child transmission even further, Permar is studying why African green monkeys (AGMs), which are natural hosts of SIV infection, rarely transmit SIV to their babies through breast milk, while rhesus macaques, which aren’t natural hosts of SIV infection, transmit it in most cases. Permar and colleagues induced lactation in chronically SIVmac251-infected rhesus macaques and in chronically SIVagm-infected AGMs, and studied their breast milk. While viral load, CD8+ T-cell responses, and levels of Env binding antibodies were similar in the two species, only breast milk from the AGMs contained IgG antibodies that could neutralize the difficult-to-neutralize founder virus that initially infected the animal.
HIV-infected human mothers also have HIV-specific neutralizing antibodies (NAbs) in their breast milk, but these can only neutralize easy-to-neutralize HIV variants, Permar said. Permar therefore wants to develop a vaccine that can induce more potent NAbs similar to what she observed in the AGMs to further reduce mother-to-child transmission in humans. She and her colleagues have immunized lactating rhesus macaques with different types of vaccines to see if they can induce the kinds of virus-specific NAbs she has observed in AGMs. The results of the study, which have been submitted for publication, are promising, Permar said: Some of the vaccination strategies induced antibodies that can neutralize viruses that are relatively resistant to neutralization.
A solution that sticks
Much of the research reported at the meeting focused on neutralizing antibodies, but Olga Malykhina from Tom Hope’s lab at Northwestern University said that even antibodies that do not neutralize HIV might be able to prevent infection. She reported that many endogenous IgG antibodies bind to vaginal and cervicovaginal mucus, probably through their Fc receptor end. The mucus binds the antibodies and keeps them from moving.
If such mucus-binding antibodies bind HIV, they should be able to prevent sexually transmitted infection, Malykhina ventured. “As long as [the virus] is trapped, it could be expelled from the genital tract” along with the mucus, she said, adding that mucus naturally exudes out of the vagina. Such antibodies wouldn’t need to neutralize HIV for this to happen, she said, because as long as they bind to both mucus and HIV, it wouldn’t matter to what part of HIV they bind. Perhaps, she said, some of the protection observed in the RV144 trial was due to mucus-binding antibodies.
Now Malykhina and Hope want to better understand why some antibodies can bind mucus better than others. By applying this information, they say, it might be possible to develop a vaccine that preferentially induces certain types of HIV-binding IgG antibodies that bind mucus especially well.