In the search for correlates of protection, AIDS vaccine researchers are once again starting to look beyond the classic neutralizing antibody responses
By Andreas von Bubnoff
The types of immune responses that are generally considered important for vaccine-induced protection against HIV are T cells and neutralizing antibody responses. But when it comes to antibodies, there are other mechanisms that may play a role in vaccine protection. Antibodies that coat an HIV-infected cell can recruit innate immune cells to either kill the HIV-infected cell or otherwise inhibit viral replication. The killing of the HIV-infected cell is referred to as antibody-dependent cellular cytotoxicity (ADCC). The inhibition of viral replication as a result of the death of the HIV-infected cell or other mechanisms is referred to as antibody-dependent cell-mediated virus inhibition (ADCVI).
Recently, there has been speculation that ADCC could in part explain the results of RV144, an efficacy trial in Thailand of a prime-boost vaccine regimen that reduced the risk of infection by about 31% (1). In addition, ADCC appears to be higher in elite controllers—HIV-infected individuals whose plasma HIV RNA levels remain very low for a prolonged period of time without antiretroviral therapy. Studies in recent years also suggest that ADCC/ADCVI might play a role in the protection of rhesus macaques from challenge with simian immunodeficiency virus (SIV) or SHIV, an SIV/HIV hybrid.
“One of the ideas is that neutralizing antibodies maybe on their own are not sufficient,” says Galit Alter, an assistant professor of medicine at the Ragon Institute in Boston who studies ADCC/ADCVI. “But if they could recruit innate immune function, now all of a sudden maybe you are able to really confer protection against infection.”
After initial interest in the late 1980s and early 1990s, AIDS vaccine researchers have generally neglected studying ADCC/ADCVI until recently, Alter says. Instead, they focused their attention largely on eliciting T-cell immunity and neutralizing-antibody responses. But studies conducted about 20 years ago showed that ADCC-related mechanisms were a possible correlate of protection against HIV infection. In 1987, Kent Weinhold, a professor of surgery at Duke University, discovered ADCC in HIV-infected patients (2). He found that CD4+ T-cell lines with gp120 bound to their CD4 receptors were sensitive to lysis by peripheral blood mononuclear cells (PBMCs) from HIV-infected individuals, but not from uninfected people. He later found that PBMCs from HIV-infected individuals contained natural killer (NK) cells with anti-gp120 antibodies bound to their Fc receptor, and it was these antibodies in HIV-infected individuals that recruited ADCC effector cells to the gp120 on the CD4+ T-cell lines. “We were quite excited about the potential [of ADCC directing antibodies] as a correlate of immune protection,” Weinhold says.
An important aspect of this mechanism, according to Weinhold, is that it does not require major histocompatibility complex (MHC) class I receptors. Therefore, innate immune cells, in concert with anti-HIV antibodies, can mount an ADCC response against infected target cells that are from a different host, he says. This is different from T cells, which only get activated once their T-cell receptor recognizes that an infected target cell comes from the host. This may make it possible for ADCC to defend against early transmission with HIV-infected cells from an infected partner.
But several years later, Weinhold was involved in an experiment that turned him away from studying ADCC further. In the experiment, chimpanzees infused with a high titer of antibodies from HIV-infected people were not protected against an HIV challenge. “That was what turned us away from a focus on ADCC,” Weinhold says. “We thought [it] was an intriguing phenomenon and still do, but the inability to demonstrate any kind of protective effect kind of took us away from ADCC.” What’s more, he says, no one had really demonstrated that ADCC-directing antibodies were a correlate of protection in any animal model of human disease. “There wasn’t a precedent,” Weinhold says. “[It] was hard to write grants on any of this stuff because it was speculation. We didn’t have good evidence that it could serve as a correlate.”
At the same time, the knowledge that T cells and neutralizing antibodies are the correlate of immunity for other vaccine models has led many researchers to largely neglect ADCC over the past 20 years, Alter says. “Everybody jumped on the bandwagon and left ADCC in the background,” she says. “It’s the irony of the cyclical patterns of fashion in HIV research. Something becomes fashionable [and] everyone starts working on it. It’s just so funny how people follow the crowd.” It also didn’t help, she says, that it was a difficult functional response to look at (See Detecting Antibody Activity, below).
|Detecting Antibody Activity
Antibody-dependent cellular cytotoxicity (ADCC) assays:
ADCC assays have three components: Target cells that express HIV antigens (CD4+ T cell lines or primary CD4+ T cells), effector cells such as natural killer (NK) cells, and a source of antibody such as serum, plasma, or monoclonal antibodies. ADCC assays measure the death of target cells by NK cells in the presence of antibody. This is usually done by measuring the release of a dye or another compound that the cells release once they die.
As target cells, researchers often use CD4+ T-cell lines that are resistant to lysis by NK cells in the absence of antibody. The target cells are either infected with HIV or coated with gp120 or Env glycoprotein, and effector cells are added. One source for NK effector cells are peripheral blood mononuclear cells (PBMCs) from uninfected donors. These can be enriched with NK cells, so that the killing can then be assumed to be mostly due to NK cells.
Recently, Michael Alpert, from the lab of David Evans at Harvard Medical School, reported at the 27th Annual Symposium on Nonhuman Primate Models for AIDS that he is working on a standardized ADCC assay that provides less variable results because it uses cell lines for both the target and effector cells (see Monkey Models: Far from Extinct, IAVI Report, Nov.-Dec. 2009).
Antibody dependent cell-mediated virus inhibition (ADCVI) assays:
The ADCVI assay also has three components: Infected target cells (primary CD4+ T cells or CD4+ T-cell lines), effector cells such as NK or other innate immune cells, and an antibody source. Unlike the ADCC assay, the ADCVI assay does not measure the death of infected target cells, but rather the degree to which NK cells or other innate immune cells used in the assay inhibit virus yield from the infected target cells in the presence of antibody.
Since this inhibition of virus replication can come from different kinds of innate immune cells, the effector cells used in ADCVI assays can be NK cells, unfractionated PBMCs, or other kinds of innate immune cells, such as monocytes or macrophages. The target cells in this assay can be primary CD4+ T cells or a CD4+ T-cell line that is resistant to lysis by NK cells in the absence of antibody.
The assay is usually done by infecting target cells with HIV for 48 hours. Then cell-free virus is washed off and the HIV-specific antibody and the effector cells of choice are added. About a week later, the virus is measured in the supernatant, usually by measuring the amount of HIV core protein p24. This is then compared with a setup where everything is the same except that the antibody is not against HIV. —AvB
Results from other fields
Meanwhile, developments in the fields of cancer research and autoimmunity also suggested that antigen binding and neutralizing activity of an antibody via its Fab region (the tips of the Y-shaped antibody) is not the only thing that accounts for its biological activity. The Fc region (the base portion of the Y-shaped antibody) is also important, says Jeff Ravetch, a professor at Rockefeller University who has been studying Fc receptors for the past two decades. Ravetch found that in mouse models of autoimmune disease, development of the disease phenotype required the activation of a certain type of Fc receptor. He also showed that in mice, the activity of monoclonal therapeutic antibodies developed to treat certain cancers required Fc receptor activation and therefore ADCC. “ADCC was required for the in vivo activity of these antibodies,” Ravetch says.
Later, several clinical studies found that people with alleles of Fc receptors that bind better to the Fc regions of antitumor antibodies responded better to the antibody treatment. “So our prediction is that if you take an Fc domain and modify it so it binds more robustly to activating receptors and enhances ADCC, it will be a more effective therapeutic,” Ravetch says.
Drug companies have already taken notice. “The therapeutic antibody field has flipped over—now every therapeutic antibody is ADCC," Ravetch says. "All the therapeutic drug companies are generating Fcs for enhanced ADCC activity, and clinical trials [have] now progressed to Phase III for Fc-engineered antibodies to enhance ADCC in vivo.”
HIV vaccine researchers have more recently found additional hints to suggest that ADCC, ADCVI, or related mechanisms might be at play in protection from HIV infection. Don Forthal, an associate professor of medicine at the University of California in Irvine, found that polymorphisms in Fc receptors appear to track with HIV disease progression, possibly by changing the binding affinity of antibodies to Fc receptors. He found that people show faster disease progression if they have a variant of the Fc receptor gamma R2A that is less efficient at binding antibody (3). As a result, it is possible that more immune complexes stay in circulation, which could contribute to immune activation and accelerate the person’s progression toward disease. “That got people excited,” says Alter.
In addition, studies have shown that ADCC might play a role in protection from challenge of rhesus macaques with SIV and SHIV. Marjorie Robert-Guroff, head of the immune biology of retroviral infection section at the US National Cancer Institute, showed in several studies starting in 2005 that ADCC might account for at least some of the protection against SHIV and SIV challenge in macaques.
|Antibody Activities in Mucosal Tissues
In antibody-dependent cellular cytotoxicity (ADCC) or antibody-dependent cell-mediated viral inhibition (ADCVI), the Fc regions of antibodies bound to HIV-infected cells bind to Fc receptors of innate immune cells such as natural killer (NK) cells, monocytes, or macrophages. Depending on the type of innate immune cell, and the Fc receptor involved, this binding either activates or inhibits the innate immune cell. Some innate immune cells such as NK cells then kill the target cell that has the antibody bound to it, as happens in the case of ADCC. This killing of target cells can be measured as ADCC in in vitro assays. In the case of ADCVI, the innate immune cell, upon Fc receptor binding, can secrete chemokines that inhibit viral replication, says Don Forthal, an associate professor of medicine at the University of California at Irvine, who developed an assay that measures ADCVI of HIV-infected cells (4). Figure adapted by permission from Macmillan Publishers Ltd: Nature 449, 29-30, 2007.
In 2007, Ann Hessell, a staff scientist in Dennis Burton’s group at The Scripps Research Institute, and colleagues mutated the Fc receptor binding region of the broadly neutralizing antibody b12, knocking out its ability to bind to Fc receptors. As a result, four out of nine macaques infused with the mutated antibody became infected after high-dose vaginal SHIV challenge, while most (eight out of nine) macaques infused with wild type b12 antibody were protected (5). “[The study] caused people to think that maybe we kind of gave up on ADCC or on these mechanisms a little bit too early,” Weinhold says.
The b12 mutant in the 2007 study eliminated interactions between b12 and all Fc receptor types, but Hessell and colleagues are making additional b12 mutants to see which of the effector cells recruited by certain types of Fc receptors are the most important. “We have now cloned a new panel of b12 Fc variants and are in the early stages of characterizing their phenotypes for effector function activities,” Hessell says. Alter, who collaborates with Burton in these studies, also says it will be important to determine which type of innate immune cell is most relevant for ADCC/ADCVI protection. “If you knock out the ability of an antibody to bind to different innate immune components, you can start to tease out the effect of Fc receptors on protection from disease acquisition, which is huge,” she adds. “We are trying to figure out, do you want to recruit NK cells, neutrophils, monocytes, dendritic cells? [And] which Fc receptors do you want to target?”
In addition, Hessell says that other broadly neutralizing antibodies like 2G12 might also protect at least in part by mechanisms other than neutralization, especially since a recent study found that 2G12 protects much better in vivo than would be expected from its neutralization ability in vitro (see 2G12 Revisited, Research Briefs, IAVI Report, July-August 2009). “There have got to be other things going on and of course we want to explore that,” says Hessell.
Sufficient for protection?
Hessell thinks it is unlikely that non-neutralizing antibodies alone could be sufficient for protection via processes like ADCC. She says it is better to call antibody functions such as ADCC “extra-neutralizing” rather than “non-neutralizing.” “Both neutralizing and possibly non-neutralizing antibodies may contribute to ADCC,” she says.
But others say ADCC/ADVCI could have a possible protective effect even in the absence of neutralizing antibodies. ADCVI probably requires a lower binding affinity between the Fab part of the antibody and its cognate antigen than what is required for neutralization, Forthal says. “For an antibody to mediate ADCVI, all it has to do is attach to an infected cell with perhaps not the greatest affinity, but just enough affinity to hang on long enough to have a natural killer cell come by and allow crosslinking of the natural killer cell’s Fc receptors,” he says, adding that the same would be true for other innate immune cells that also have Fc receptors and can mediate ADCVI such as monocytes, macrophages, dendritic cells, or neutrophils. “So I think the threshold for this activity with respect to antibody affinity is probably lower, and in that sense there may be epitopes which allow ADCVI to occur that don’t allow neutralization to occur.”
Forthal says ADCC/ADCVI-related mechanisms could also explain why the prime-boost regimen tested in RV144 showed some efficacy even though the candidates do not appear to induce neutralizing antibodies. The vaccine regimen tested in the RV144 trial didn’t result in neutralizing antibody responses to primary HIV isolates, according to Josephine Cox, a director of clinical immunology at IAVI who participated in that study (6). Cox was also involved in a 2005 study that showed that the vaccine regimen did induce elevated ADCC activity (7).
In addition, Forthal was part of a 2007 study that looked at ADCVI responses to antibodies from vaccinees of the Vax 004 Phase III trial of AIDSVAX B/B, which is similar to the gp120 AIDSVAX B/E boost used in RV144. While that vaccine candidate did not show any efficacy and did not induce much of a neutralizing antibody response against clinical HIV strains, the study found that the lower the ADCVI activity, the higher the rate of HIV infection was after vaccination (8). “I know there was no overall efficacy in that [Vax 004] trial, but it may be that certain subgroups who have high levels of [ADCVI] antibodies were protected,” Forthal says. “We don’t expect a lot of classical neutralization by the antibodies of clinical strains of virus after that [RV144] vaccination, [but] it may very well be that ADCVI, ADCC, or some other Fc-receptor mediated non-neutralizing antibody may play an important role in this case.” As a result, there are discussions about measuring ADCC or ADCVI responses in blood samples from RV144, Forthal says.
Cox says she would also expect ADCC to be present in the mucosa. But measuring that in the RV144 trial will be difficult because very few mucosal samples were collected in the trial. “Unless we do further trials we are not going to be able to show that,” she says.
One challenge is that there is so far no standardized assay to measure ADCC or ADCVI, Weinhold says. Alter agrees. “We have no idea if all the ADCC assays are comparable or even what they are measuring,” she says. David Montefiori at Duke University Medical Center is organizing efforts to standardize the protocols and make the assays comparable. “We are organizing an effort to examine the readiness of various ADCC and ADCVI assays for case control samples from the RV144 trial,” Montefiori says. “Our goals are to evaluate intra-laboratory variability and to compare results between assays.” There are also efforts to develop new assays with better defined components (see Monkey Models: Far from Extinct, IAVI Report, Nov.-Dec. 2009). Ravetch says the RV144 trial volunteers should also be analyzed for the types of Fc receptor alleles they have, which could affect their ADCC or ADCVI response.
But not everyone believes there is enough evidence to suggest that ADCC or ADCVI may have a role in the small efficacy observed in RV144. “I think there’s no meaningful evidence that ADCC or ADCVI is either protective against HIV transmission or involved in controlling HIV infection in vivo,” says John Moore, a professor of microbiology and immunology at Weill Cornell Medical College, adding that he doesn’t trust the relevance of many of the assays that have been used to measure these responses. He says the only paper on this topic that is of any real interest is the 2007 study by Ann Hessell and colleagues on the effect of Fc-mutated monoclonal antibodies in passive protection studies (5). “But it’s a big leap of faith to extrapolate from that work to the RV144 trial and whatever effect those vaccine components may, or may not, have had on acquisition,” Moore says.
Researchers are also investigating whether ADCC-like mechanisms might account in part for the viral load control seen in elite controllers. One recent study found that ADCC was higher in elite controllers (9).
In a separate project, Alter is collaborating with William Hancock at Northeastern University to see if anti-HIV antibodies in elite controllers differ from antibodies in chronic progressors, for example in the structure of the glycans in their Fc receptor binding region, which modulate the binding affinity to Fc receptors of effector cells. Differences in the strength of Fc receptor binding to IgG antibodies have already been shown to explain part of the differences in the efficacy of therapeutic antibodies used to treat cancer, Ravetch says, and might explain the higher degree of ADCC seen in elite controllers. With Hancock and Genoveffa Franchini at the National Cancer Institute, Alter is also planning to look at whether ADCC could account for the protection some vaccine candidates have afforded in nonhuman primate studies.
Eventually, such studies might result in the identification of markers on antibodies that researchers might want to induce when they develop vaccine candidates. “There may be a chance that we can improve the protective nature of a vaccine if it can induce antibodies that can recruit this function,” Alter says.
For too long, the AIDS vaccine field has focused on the antigen binding region of antibodies and neglected the Fc region, which is involved in ADCC and ADCVI, Alter says. “I think that on antibodies we are missing the boat,” she says. “[People] are not even thinking about the larger section of the antibody which is the constant region.”
Ravetch also urges the AIDS vaccine field to take a closer look at Fc receptor-mediated mechanisms. “Stop being so narrow minded and thinking about everything the same way,” he says.
Not everyone is convinced. “Overall, I think ADCC/ADCVI is just another of the bandwagons that roll along every now and then in the HIV vaccine field,” says Moore. “Some people will be interested in joining it for a while, but not me.”
But others say that views in the field may already be changing. “I think everyone is sort of starting to believe [ADCC] is important—it’s just not sure how important it is because there is not a lot known on this subject,” Alter says. Weinhold agrees. “I think with Burton’s  paper and now the speculation that accompanies the 31% efficacy in the Thai trial, people are now putting [ADCC] back on the table,” he says. “If there is any inkling or hint of ADCC as a protective mechanism in the Thai trial, I think it will spur people to revisit this with some of the technologies that we have now that we didn’t have 20 years ago.”
1. N. Engl. J. Med. 361, 2209, 2009
2. Proc. Natl. Acad. Sci. 84, 4601, 1987
3. J. Immunol. 179, 7916, 2007
4. J. Virol. 75, 6953, 2001
5. Nature 449, 101, 2007
6. J. Infect. Dis., 190, 702, 2004
7. Vaccine 23, 2522, 2005
8. J. Immunol. 178, 6596, 2007
9. AIDS 23, 897, 2009