Antibody Fever

Characterization of the slew of recently discovered broadly neutralizing antibodies was one of the advances highlighted at the recent HIV Vaccines conference

By Andreas von Bubnoff

For the first time, the annual Keystone Symposium on HIV Vaccines, which took place from March 21-26 in Banff, Canada, was held in conjunction with a symposium on Viral Immunity rather than with the symposium on HIV Biology and Pathogenesis, which was held from January 12-17 in Santa Fe, New Mexico. Wayne Koff, one of the organizers of the HIV Vaccines meeting and chief scientific officer at IAVI, told the delegates on the first night of the joint symposia that this was done to enable cross fertilization of the two fields. “It’s rare that many of us have the opportunity of getting out and seeing what is really occurring in the field outside of HIV,” Koff said.

Combining the two meetings was a good idea, said Silke Paust, a postdoctoral fellow at the Ragon Institute and Harvard Medical School. But Paust thought that there could have been more joint sessions. “I think that putting both together was a good idea, just from the mixing of the people that you got that way,” she said. “I would like to have them together in the future, but maybe [with] more joint sessions, because sometimes I really wanted to be at both places at the same time.”

Even within the HIV Vaccines sessions, there was a lot to digest. Much of the discussion focused on broadly neutralizing antibodies, both the characterization of those recently isolated, as well as a better understanding of how they develop in HIV-infected people. Other topics included antibody-dependent cellular cytotoxicity (ADCC), updates on HIV vaccine trials, the development of new vaccine strategies, further insights into early HIV transmission, and the role of CD4+ T cells in protection.

Characterizing new antibodies

Recently, several new HIV-specific broadly neutralizing antibodies have been identified, including VRC01, which was discovered by researchers at the Vaccine Research Center (VRC) at the US National Institute of Allergy and Infectious Diseases (NIAID), and PG9/16, discovered by IAVI researchers in collaboration with researchers from The Scripps Research Institute (see Raft of Results Energizes Researchers, IAVI Report, Sep.-Oct. 2009). Now, one question scientists are attempting to answer is just how these new antibodies neutralize so well.

Some clues come from their structures. Peter D. Kwong, chief of the structural biology section at the VRC, presented data on the structure of the antigen-binding fragment of VRC01 bound to an HIV clade E gp120 monomer. He showed that VRC01, which recognizes part of the CD4 binding site of gp120, neutralizes so well partly because it mimics CD4 very well. Parts of the VRC01 heavy chain align very well with parts of CD4, and VRC01 binds gp120 at an angle that is only a few degrees different from the angle of CD4 binding to gp120. VRC01 also contacts gp120 in regions that do not change conformation, while it has a gap to accommodate the variable part of gp120. This way, it can accommodate variations in gp120 while retaining its ability to bind.

Another question researchers are grappling with is how an antibody like VRC01 can be elicited by a vaccine. As discussed at the meeting, antibodies like VRC01 have a high degree of affinity maturation. Affinity maturation is a process that starts once a B cell is activated by an immunogen. This process creates mutations in the variable regions of an antibody. Versions of the antibody with higher affinity for the immunogen are then selected, and the population of B cells expressing these altered antibodies expands.

Kwong said that, compared with most antibodies, VRC01 has an exceptionally high degree of affinity maturation. At least 30% of the amino acids in its variable region (more than 60 amino acids) are changed, compared with 5-10% (10-20 amino acids) in most other antibodies. The changed residues include more than half of the ones that come in contact with gp120.

Affinity maturation could in part explain why it seems to take years until broadly neutralizing activity develops in HIV-infected people. For vaccine development, this could mean that to induce antibodies like VRC01, one might have to guide the immune system along the affinity maturation process by sequentially vaccinating a person with several immunogens, each of which binds to intermediate stages of the antibody as it undergoes the affinity maturation process, Bart Haynes, director of the Duke Human Vaccine Institute at Duke University, suggested in his talk.

Kwong is confident that, in principle, this should be possible because affinity maturation is a process that is quite well understood. “This is the first time where we have got an antibody [where] you actually know the mechanism [of how it might be elicited in high titer],” he said, adding that with hundreds of papers published, antibody affinity maturation is a well investigated area, and much of the process is well defined. “That’s why I am actually excited about this one.”

Another reason he is optimistic is that some of the recently identified broadly neutralizing antibodies are actually observed in high titers in the infected individuals they were isolated from, suggesting that in principle, a vaccine should be able to induce such antibodies. “That’s what’s so exciting,” Kwong said. “Humans can make antibodies that have the same phenotype as the PG antibodies and VRC01.”

Data as to how long it takes for broadly neutralizing activity to develop in HIV-infected people were presented by Lynn Morris, head of the AIDS Virus Research Unit at the National Institute for Communicable Diseases in Johannesburg. Morris said that broadly neutralizing antibodies may not be as rare as was once thought. While the definition of breadth varies, about one quarter to a third of HIV-infected people appear to have some level of cross-reactive antibodies after a few years of infection, according to Morris. She presented data from a longitudinal analysis of the CAPRISA cohort of HIV-infected women, mostly sex workers in the province KwaZulu-Natal in South Africa, most of whom are infected with HIV subtype C.

By three years after infection, several of the women had developed broadly neutralizing activity in their sera. In most cases, this activity developed gradually. Over time, the serum could neutralize an increasing number of HIV strains in a diverse panel. “I am pretty sure we are the first to show that over time, the development of breadth occurs incrementally,” Morris said. In one case, however, the serum became broadly neutralizing almost all at once—about 70 weeks after infection, the individual’s serum neutralized just one virus strain, but 10 weeks later, it could neutralize 16 different primary viruses. “That was very suggestive of a single antibody developing in this person,” Morris said.

The study also showed that the sera taken prior to the development of broadly neutralizing antibodies could neutralize virus that was taken from the same person a few months earlier, but could not neutralize concomitant virus. However, once the serum became broadly neutralizing, it could often neutralize concomitant virus as well, suggesting that escape from broadly neutralizing antibodies may be more difficult.

Typically, just one antibody specificity seemed to account for most of the broadly neutralizing activity of the sera. The researchers showed this by using known targets of broadly neutralizing antibodies to deplete the serum of antibody types binding to them and then checking if the sera could still neutralize a diverse panel of HIV strains. For example, in the person in which the serum acquired broad neutralization activity all at once, two thirds of this neutralization was due to antibodies that bind the membrane proximal external region (MPER) of gp41 envelope (1). In another case, in which the neutralization breadth developed gradually, the activity was mostly due to antibody that could be depleted by gp120. This was true at different time points after infection, suggesting that the broadly neutralizing activity was due to the same type of antibody that developed over the years.

Together, these findings suggest that one reason it takes years for the sera in HIV-infected people to develop broadly neutralizing activity is affinity maturation of one type of antibody. This is also consistent with the high degree of affinity maturation observed in broadly neutralizing antibodies like VRC01. “Initially when we saw those data we thought that neutralization breadth was due to the accumulation of lots of different antibodies,” Morris said. “But actually it probably isn’t—it’s probably a single antibody that’s affinity maturing.”

Dennis Burton, a professor at The Scripps Research Institute, presented a similar analysis of the types of antibodies or antibody specificities that account for the broadly neutralizing activity in 19 HIV-infected individuals whose sera had among the broadest and most potent neutralizing activity from IAVI’s Protocol G cohort. The cohort comprises about 1,800 HIV-infected people and includes the individual that was the source of the PG9/16 antibodies. Similar to Morris, Laura Walker, a graduate student in Burton’s group, used known targets or properties of broadly neutralizing antibodies to remove the corresponding antibody types from the sera and then determined if the sera could still neutralize a diverse panel of HIV. She found that the neutralization activities were typically due to one or two broadly neutralizing antibody specificities. Major specificities included antibodies binding the CD4 or the CCR5 binding site of gp120, and specificities similar to the PG9/16 antibodies. One donor had a binding specificity similar to the broadly neutralizing antibody 2G12, which binds to glycans on gp120. Four donor sera had specificities that bound to the same glycan, but not directly, suggesting that their antibody specificity was directed to a previously unknown target.

Neutralization is not the only mechanism thought to explain protection afforded by broadly neutralizing antibodies. A 2007 study by Ann Hessell, a staff scientist in Burton’s laboratory at The Scripps Research Institute, and colleagues found that eliminating the ability of the broadly neutralizing antibody b12 to bind to Fc receptors, which is necessary for ADCC, makes this antibody less protective in challenge studies in rhesus macaques (2; seeAntibodies: Beyond Neutralization, IAVI Report, Jan-Feb. 2010). At Keystone, Hessell showed that a b12 antibody that lacked fucose residues in the Fc receptor binding region had about a 10- to a 100-fold better ability to bind to the Fc receptor IIIA. In vitro, this translated into a 10- to a 100-fold better ability of this modified b12 antibody to mediate ADCC and antibody-dependent cell-mediated virus inhibition (ADCVI). Experiments in rhesus macaques are planned to see if this also translates into better protection from simian immunodeficiency virus (SIV)/HIV hybrid challenge in vivo. “We believe that the enhancement of ADCC will lead to protection by a substantially decreased dose of [modified] antibody to achieve the same effect,” Hessell said.

Moving beyond RV144

Researchers are still trying to find an explanation for the modest success of the vaccination regimen tested in the RV144 trial in Thailand, an efficacy trial involving more than 16,000 Thai volunteers that tested a canarypox vector-based candidate ALVAC-HIV in a prime-boost combination with AIDSVAX B/E (see box below). Nelson Michael, director of the US Military HIV Research Program, presented an update of studies that are designed to investigate the mechanism of protection. Michael said that six months after the final vaccination, uninfected vaccinees showed no CD8+ T-cell responses, while about one third had CD4+ T-cell responses to Env. Vaccinees that became HIV infected showed CD8 responses to Gag and Env epitopes that were different from the Gag and Env CD8 responses observed in unvaccinated individuals who became infected. “There really is little overlap between the ELISPOT epitope responses to Gag and Envelope in people [with] breakthrough [infections] that received vaccine versus those that received placebo,” Michael said. “To us, it’s the first evidence that there is an immunologic correlate at the T-cell level to what we saw clinically.”

Michael also addressed future plans and said that there are discussions to test a similar vaccine regimen to that evaluated in RV144, but in high-risk populations rather than the general population that was recruited for RV144. The vaccine regimen appeared to be worse at protecting volunteers who had indicated that they had any high-risk activity at any time during the trial compared with people who reported no risk behaviors (see Prevent and Conquer, IAVI Report, Jan.-Feb. 2010). But Michael said that does not mean that the vaccine shouldn’t be tested in high-risk groups, because it is not clear if the high-risk behavior took place early after vaccination, when the protective effect seems to have been highest, or later when the protective effect was much smaller. “It’s hopelessly confounded with the fact that the duration of this [protective] effect was also transient,” Michael said.

He said that tests of a similar vaccine regimen in men who have sex with men (MSM) in Thailand or in high-risk heterosexuals in southern Africa are being discussed. Researchers are considering using NYVAC, a poxvirus-based vector that has been developed by the company Sanofi Pasteur, as a prime. While similar to ALVAC, NYVAC might be easier to produce. “There is an impression that NYVAC might be a better vector,” Michael said. As a consequence, investigators are discussing whether to do two efficacy studies in high-incidence populations of heterosexuals in southern Africa, one with NYVAC, the other with ALVAC, or to select a single vector now which would then be tested in both high-risk populations in southern Africa and Thailand. At the same time, there are discussions to do additional trials in Thailand in higher-risk individuals, possibly MSM, with additional gp120 booster shots to see if the transient protective effect seen in RV144 is reproducible and if it could be extended by adding additional boosts.

Stepping along

Juliana McElrath, a professor of medicine at the University of Washington, presented results from continuing studies addressing why the adenovirus serotype 5 (Ad5)-based vaccine candidate MRKAd5 tested in the Phase IIb STEP trial failed. One explanation is that the immune response in the vaccinees was not very broad. After mapping HIV-specific T-cell responses to Gag, Pol, and Nef (the antigens used in the MRKAd5 vaccine) in 73 STEP trial vaccinees, these T-cell responses were found to be directed to a median number of one epitope per vaccinee. Last fall at the AIDS Vaccine 2009 conference in Paris, researchers had reported that the median number of epitopes was two (see Raft of Results Energizes Researchers, IAVI Report, Sep.-Oct. 2009). McElrath also reported results from an analysis of the viral loads in some STEP trial volunteers early after HIV infection, before they became seropositive, which suggested that the vaccine had an early inhibitory effect on virus replication, but that the effect then disappeared.

Researchers are also still investigating possible explanations for the trend toward an increased risk of HIV acquisition in some STEP trial vaccinees. One potential mechanism that has been discussed is that the vaccine might have increased the number of HIV target cells. But McElrath said that that does not appear to be the case. In a case-control study of 254 STEP trial volunteers at two time points, week eight (or four weeks after the second immunization) and week 30 (or four weeks after the third immunization), activated CD4+ T cells were not a significant predictor of the risk of HIV acquisition.

As for the other risk factors, initially, researchers observed that uncircumcised vaccine recipients with preexisting antibody immunity to the Ad5 vector used in MRKAd5 had an increased risk of HIV infection compared to placebo recipients with the same characteristics. In follow-up analyses, the increased risk among uncircumcised men who received the vaccine remains, although it is waning over time, while the effect of preexisting Ad5 immunity is not detectable at later time points (see Raft of Results Energizes Researchers, IAVI Report, Sep.-Oct. 2009). Studies in humans are planned to look at the effects of circumcision with vaccination. These will look at cell populations in the foreskin in the context of Ad5 serostatus and co-infection with HSV-2.

Dan Barouch, an associate professor of medicine at Beth Israel Deaconess Medical Center and Harvard Medical School, also addressed whether it is possible that in the STEP trial, preexisting immunity to the Ad5 vector may have created additional HIV target cells following vaccination that then could have migrated to the colorectal mucosa or the foreskin. He presented data from experiments conducted in nonhuman primates (NHPs) that suggest this is not the case. To induce Ad5 immune responses, the researchers infected rhesus macaques intranasally with Ad5 that can replicate in monkeys. They then vaccinated them intramuscularly with a replication incompetent Ad5-Gag/Pol/Nef vaccine that was similar to the one used in the STEP trial.

The Ad5 seropositive macaques did not have more Ad5-specific CD4+ T cells in blood or in the colorectal mucosa when compared with Ad5 seronegative animals. “These new data suggest in a rhesus monkey model that there is no greater increase in mucosal trafficking of either total or Ad5-specific CD4+ T cells in Ad5 seropositive versus seronegative monkeys following Ad5 vaccination,” Barouch said. In the blood, these Ad5-specific CD4+ T cells showed transient increased proliferation and activation for one week after vaccination. This resolved to baseline levels by week two and therefore does not likely explain the more than 52 weeks of potentially enhanced HIV acquisition in the STEP trial, he added. “Overall, these data would suggest that Ad5-specific cellular immunity does not appear to explain the potential for enhanced HIV acquisition that was seen in the STEP study,” Barouch concluded.

The use of alternative Ad vectors such as Ad26 and 35 has been suggested for the development of future candidate HIV vaccines, because of their biological differences to Ad5 as well as lower levels of preexisting immunity. Barouch presented epidemiological data from Africa suggesting that Ad26 and 35 antibody titers are substantially lower than for Ad5, and seroprevalence to Ad26 and Ad35 is also less common than to Ad5. The seroprevalence for these two vectors was less than 1% in 149 healthy two- to nine-month old South African infants. Of 346 South African school children age 6-18, 74% were seronegative for Ad26, and 86% for Ad35, while only 31% were seronegative for Ad5. And of 199 adults age 18-50 from several African countries, 58% were seronegative for Ad26, and 78% for Ad35, while only 16% were seronegative for Ad5. “In total we believe that these seroprevalence data support proposals for further evaluation of these vectors in Phase I trials in sub-Saharan Africa,” Barouch said, adding that in preparation for clinical studies, Ad26 and 35 vectors expressing mosaic Gag/Pol/Env antigens are currently being manufactured by the company Crucell. Mosaic antigens are designed to achieve optimal coverage of the many different versions of HIV proteins that are circulating. “We hope to have these ready sometime next year,” Barouch said.

While there is concern that vaccines might induce CD4+ target cells, Hendrik Streeck, a junior faculty member at the Ragon Institute, presented evidence that suggests that vaccine-induced CD4+ T-cell responses might actually be important for protection. Streeck showed that secretion of the cytokine interleukin (IL)-21 is predominantly seen in HIV Gag-specific CD4+ T cells from HIV-1 controllers, but almost not at all in progressors. He also found that IL-21 could increase the ability of CD8+ T cells from progressors to produce perforin and granzyme B, resulting in their better ability to kill HIV-infected cells. This resulted in up to a 1,000-fold upregulation of the ability of these CD8+ T cells to inhibit HIV replication in a viral inhibition assay. “CD4 cells have been completely underestimated,” Streeck said, adding that the observation of fewer CD4+ T cells in acute HIV infection doesn’t necessarily mean that they are permanently depleted or not important for protection.


Using Endogenous Retroviruses to Fend Off HIV

One major challenge for HIV vaccine development is that a vaccine needs to protect against the huge number of HIV strains in circulation. Brad Jones, a graduate student at the University of Toronto, reported on experiments that could lead to an HIV vaccine strategy that might be able to circumvent this problem because it doesn’t involve targeting HIV itself. Instead, it involves human endogenous retroviruses (HERVs), which are remnants of previous retroviral infections. Their sequences are littered throughout the human genome and make up about 8% of our DNA.

Previous research has shown that HIV-infected people have more HERV sequences in their blood plasma than uninfected people (3). This suggests that HIV infection might lead to the activation of HERVs, expression of which could be used as a marker for HIV-infected cells.

For their studies, Jones and colleagues decided to focus on HERV-K, because it is the evolutionarily youngest and most intact type of HERVs and is the most similar to HIV. This might explain previous observations that elements of HIV and HERV-K can interact to facilitate HERV-K expression.

To see if infection of CD4+ T cells could indeed reactivate HERV-K expression, Jones and colleagues isolated HERV-K-specific CD4+ T cells from an HIV-infected elite controller. They found that infecting these cells with HIV in vitro led to reactivation of HERV-K protein expression. They also showed that HERV-K-specific CD8+ T cells from the same elite controller could kill these HERV-K-specific HIV-infected CD4+ T cells no matter what HIV strain they were infected with. 

This suggests that a vaccine could be developed that induces HERV-K-specific CD8+ T cells that should then be able to kill HIV-infected CD4+ T cells because all of them should express HERV-K, no matter what HIV strain a person is infected with. “I believe this is the first time it has ever been shown that a non-HIV-specific T cell can specifically kill HIV-infected cells,” Jones said. “It is a proof of principle for the strategy of targeting a surrogate marker of infected cells rather than the HIV sequence itself.” —AvB


Looking for better breadth

Given the limited breadth of immune responses induced by MRKAd5, researchers are trying to develop new vaccine candidates that are capable of inducing a broader response. In NHPs, mosaic vaccines have recently been shown to induce broader and deeper cellular responses than conventional vaccines with inserts encoding consensus sequence or natural sequence antigens, with depth referring to the simultaneous induction of different responses to the same epitopes (4,5; see Capsules from Keystone, IAVI Report, Mar.-Apr. 2009).

These mosaic vaccines were primarily developed to induce T-cell responses, but Barouch said that the mosaic vaccines developed in his lab also induce “remarkably potent antibody responses,” which are at least non-inferior to the antibody responses generated by other leading modalities such as consensus or natural sequence envelope. “It is a really important aspect that people have overlooked,” he said. Bette Korber, a laboratory fellow at the Los Alamos National Laboratory and external professor at the Santa Fe Institute, said that she and her colleagues have developed new strategies to design mosaic vaccines tailored for antibody responses, based on variation in regions of Env that are close together in three-dimensional space.

Tomas Hanke, a reader in immunology at the University of Oxford, described another approach to induce broad cellular immune responses. It involves delivering a single DNA construct that contains the 14 most conserved regions of HIV, which are taken from four major HIV clades. Hanke and colleagues injected this insert intramuscularly into rhesus macaques in different forms. Three DNA primes were followed by a boost of the insert in a human Ad5 vector and then by another boost of the insert in a modified vaccinia virus Ankara (MVA) vector. This “DDDAM” heterologous prime-boost regimen was then followed by two injections of the same conserved sequences in the form of 46 separate peptides of 25-28 amino acids length. To avoid immunodominance, the peptides were injected at six separate sites.

Data on the immune responses to this “DDDAMSS” regimen in three rhesus macaques showed that the injection of the peptide pools further increased the magnitude of the cellular immune response induced by the “DDDAM” regimen by 30%, and also made the T-cell responses broader. Hanke said that while the immune responses induced by the DDDAMSS regimen are not as broad as the ones recently reported for mosaic vaccines, the responses induced by the regimen are of a higher magnitude because they are focused on the conserved epitopes. “[The DDDAMSS regimen focuses] all the might of the T-cell immune responses on the fewer, but invariant epitopes,” Hanke said. Early clinical trials will start later this year with a modified “DDDCM” regimen that uses chimpanzee instead of human adenovirus, he added.

Further insights into transmission

In the last few years, researchers have learned that in most heterosexual transmissions just one transmitted virus variant is responsible for productive infection in the recipient, suggesting that there is a genetic bottleneck which limits the degree of variation as the virus is transmitted (see HIV Transmission: The Genetic Bottleneck, IAVI Report, Nov.-Dec. 2008). However, it is still unclear where this bottleneck is located, and which factors determine which virus ultimately gets transmitted.

At the meeting, Cynthia Derdeyn, an associate professor of pathology and laboratory medicine, and Eric Hunter, a professor of pathology and laboratory medicine at Emory University, presented data from their labs that suggest that the genetic bottleneck is not in the donor. Debrah Boeras, a postdoctoral fellow in Hunter’s lab, used single genome amplification to analyze the viruses present shortly after transmission from vaginal swabs from six female donors and semen from three male donors. A comparison with the sequences of the transmitted viruses showed that while the donor fluids contained many different virus variants, it was not the most common variant that ended up getting transmitted. “What’s new about our data is the idea that in the genital compartment of the donor partner you have a major dominant variant circulating, but that doesn’t seem to be the one that’s transmitted,” Derdeyn said. “It suggests that it’s not stochastic, it’s not simply that the most frequent variant is the one that’s transmitted. It’s something else about the variants that get transmitted. The bottleneck is not in the genital compartment of the donor.”

In addition, Derdeyn and Hunter found no difference between viruses taken from the blood of donors and recipients around the time of transmission in terms of which cell types they preferentially infect. They all depended on high levels of CD4 and CCR5 receptor expression to infect target cells, and all infected CD4+ T cells well, but not macrophages. “It seems like macrophages are probably not the initial target cells,” Derdeyn said.

The observation that there are many different viruses in the genital fluids of donors is consistent with evidence presented by Suzanne English, a graduate student at the University of Oxford, which suggests that different viruses get transmitted to different recipients even when the donor and route of transmission are the same. English and colleagues analyzed HIV sequences from two MSM that were both rectally HIV infected from the same donor in the same night, just minutes apart from each other. The researchers used sequence analysis to confirm that the transmitted founder viruses in the two recipients came from the same donor, but they found that the sequences were too different to have come from the same transmitted virus. Instead, the two were so different that they are estimated to have been evolving prior to transmission for at least several years. One thing is for sure, the researchers were lucky they had the opportunity to study a case like this, English said. “It’s very, very difficult to find two patients who were infected by a single donor by the same route in the same night who were then both sampled on the same day 63 days later.”

1. J. Virol. 83, 11265, 2009
2. Nature 449, 101, 2007
3. PLoS Pathog. 3, e165, 2007
4. Nat. Med. 16, 319, 2010
5. Nat. Med. 16, 324, 2010