Research at the Extremes

Presentations at Keystone 2005 focused on both very early and prolonged infection to try to discern what an effective vaccine might need to do

By Philip Cohen, PhD

This year's Keystone joint symposia on "HIV Vaccines: Current Challenges and Future Prospects" and "HIV Pathogenesis" were held in Banff amid the soaring Canadian Rockies. The seven-day symposia are perhaps the highlight of the HIV conference calendar, a forum for the world's top researchers to air new data and put forward new hypotheses, as well as get together to share ideas and spark collaborations.

Much of the talk at the social events this year centered on the recently announced funding streams and organizational mechanisms—the National Institutes of Health's Center for HIV/AIDS Vaccine Immunology (CHAVI) initiative generated plenty of discussion and opinion, as did the Bill and Melinda Gates Foundation's call for new research proposals. A more somber topic of discussion was the passing of the remarkable vaccinologist Maurice Hilleman, and a number of scientists took time in their presentations to pay him personal tribute (seeObituary).

The presentations at the Keystone symposia covered much of the diversity of topics within HIV research. A recurrent theme of many talks, though, took HIV or SIV infection at either extreme as their model system in which to look for immunological clues towards an effective vaccine—immediate early infection or established infection in humans and monkeys able to control their virus.

Getting acquainted with the host factors
A series of talks at Keystone focused on host intracellular factors that HIV engages in order to facilitate its replication or that the virus must overcome to avoid destruction. Either way, the hope is that therapies which target these factors may some day offer novel ways to attack the virus.All viruses must subvert human proteins and cellular machinery to their own ends, but for HIV the list of these proteins continues to grow. Mario Stevenson of the University of Massachusetts Medical School tracked down two more HIV collaborators while investigating the role of the nuclear envelope in infection. HIV's ability to infect some non-dividing cells has been attributed to its ability to transverse the nuclear envelope. In contrast, murine leukemia virus (MLV) can only infect dividing cells that go through mitosis, during which the nuclear envelope dissolves. But, unexpectedly, when Stevenson's group froze cells in mitosis using drugs like taxol, they discovered that MLV could integrate DNA into these cells but HIV could not, suggesting that HIV is actually dependent on nuclear factors to replicate.

Using RNA interference, Stevenson's team knocked down expression of some 20 known nuclear envelope-associated proteins and discovered two that were necessary for the viral integration: BAF and Emerin. Emerin is a protein inserted into the inner portion of the nuclear membrane, while BAF is its DNA-binding partner. Together the pair serve as a scaffold for chromosomal DNA. When either factor was knocked down, HIV DNA did not integrate but instead accumulated as a circular molecule of one or two copies of its genome—a dead end for the virus. Stevenson proposed that HIV uses BAF and Emerin to navigate quickly to chromosomes. "The virus is much smaller than the cell and needs sign posts," he says. The virus may end up circularized because chromosome repair enzymes mistake the linear virus for a broken chromosome and "heal" it by joining the ends.

While HIV benefits from its association with BAF/Emerin, it has a different and complex relationship with Murr1. Two years ago, Gary Nabel of theVaccine Research Center at the NIH and his colleagues found that Murr1—a protein formerly known for its influence on copper metabolism—contributes to the restriction of HIV replication in resting T cells (Nature 426, 853, 2003). Part of this effect was attributed to Murr1's ability to inhibit the activity of NFкB, a transcription factor that HIV requires to amplify its replication. Murr1 influences NFкB indirectly: it stops E3 ubiquitin ligase from tagging a protein called IкB for destruction by the proteasome. IкB is a natural inhibitor of NFкB, so the increase in IкB concentration drives down the transcription factor's activity and cripples the replication of HIV.

Since Murr1 has little homology to other known proteins, Nabel's team first assumed its relationship to HIV was probably unique. But at Keystone he reported that searching databases for human proteins with a similar predicted structure revealed that Murr1 is probably just one of 10 genes in a new family (dubbed COMMD for Copper Metabolism Murr1 Domain), which seem to pair off to produce dimers with different biological properties. So while the Murr1 (now called COMMD1) homodimers can inhibit HIV replication, COMMD6 disrupts this inhibition by forming an inactive heterodimer with COMMD1.

The function of all these proteins is still unclear but based on preliminary data Nabel proposed that these proteins might be more generally "gatekeepers to the proteasome," guiding proteins to or away from this disposal pathway. "HIV-1 may need to find its way around these elements to survive its journey through the cell," he says. One intriguing connection is that HIV inactivates the human protein APOBEC3G by misdirecting it to the proteasome. APOBEC3G is an antiviral protein that is currently a hot topic of HIV research (see Guardian of the genome). Nabel's team has shown that COMMD1 can bind APOBEC3G, but appears not to play a role in APOBEC3G's degradation. Now Nabel's team is collaborating with Michael Malim's laboratory at King's College London, where APOBEC3G was discovered, to see if any of the other COMMD members play a role in this pathway.

Greg Towers of University College London revealed unexpected connections between two other known antiviral proteins, the murine Fv1 and TRIM5α. Fv1 was first describe in the late 1970s and restricts the replication of MLV. TRIM5α was recently uncovered as a human protein that blocks MLV in human cells but also appears to be a major contributor to species-specific barriers to retroviral replication—including that of HIV—in many primates. TRIM5α and Fv1 are unrelated by sequence and seem to differ in their ability to restrict virus in a fundamental way: TRIM5α, as a rule, erects a replication block prior to reverse transcription while Fv1 blocks after DNA synthesis. However the same amino acid residue in viral capsid, 110, controls sensitivity to both factors. Now Towers' team has demonstrated some mechanistic links between the two factors. When Fv1 and human TRIM5α are coexpressed in the same cell there is less viral suppression before reverse transcription than when TRIM5α is expressed alone. This would be the case, says Towers, if Fv1 competes with TRIM5α for the virus. "They may both act at the same early point of the lifecycle, but end up killing the virus in different ways," he says. In addition Towers' team has found that the TRIM5α protein from squirrel monkey actually blocks retroviral replication after DNA synthesis, showing the exact point at which TRIM5α acts may not be as fixed as originally thought.

Early events

The course of HIV infection and its progression to AIDS may take decades but the early events in infection have always been of interest to researchers since the goal of any preventive vaccine is to stop the incoming virus from establishing a firm foothold. A quartet of recent publications has highlighted just how narrow this early window of opportunity might be, and they set the stage for much of the Keystone meeting. The critical battle between HIV and the immune system is now envisioned as taking place within the first few weeks of infection, during which virus irreversibly destroys the vast majority of CD4+ memory T cells in mucosal tissues, in particular in the gut-associated lymphoid tissue or GALT (see Research Briefs).

One key question is the nature of the virus that establishes early infection. Eric Hunter of Emory University in Atlanta focused on the virus that is transmitted in a cohort of heterosexual couples initially discordant in their HIV infection status. Previous work by his team involving 8 transmission pairs (Science, 303, 2019, 2004) suggested that sexual transmission of HIV is essentially a severe bottleneck where new infection is established by a small population of founding viruses with particular characteristics. They also found that the resulting population of sexually-transmitted virus had shorter V1-V4 region in their envelope, fewer glycans and were unusually sensitive to neutralization by antibodies from the chronically-infected partner. Hunter and his colleagues propose that these properties reflect different selective pressures on the virus during the long course of an infection and during transmission. During chronic infection the virus adapts its envelope to evade neutralizing antibodies. In contrast, during transmission they speculate the virus is forced to adopt a more compact structure, in effect throwing off its protective shield against antibodies.

It was possible, however, that the bottleneck was not driven by transmission but by the recipient's early antibody response. To address this issue, Hunter described 5 new transmission cases in which the researchers sampled virus in the first few weeks of infection, before antibodies were detectable. Consistent with his hypothesis, these early viruses were still subject to a severe population bottleneck.

In collaboration with Bette Korber's team at Los Alamos National Laboratory in New Mexico, the Emory group has analysed more closely clade C viral features that correlate with neutralization sensitivity, using their database of donor and recipient viruses. Again, the researchers found evidence for a correlation between shorter V1-V4 regions of gp120 and better neutralization. But in this new analysis amino acids in a region adjacent to the V3 loop, rather than in V3 itself, appeared to influence neutralization sensitivity, suggesting a novel mechanism for antibody escape in clade C viruses.

Ashley Haase of the University of Minnesota Medical School, Minneapolis, discussed data from his study on a rhesus macaque model characterizing the first critical days of SIV infection, which was published shortly after the meeting (Nature, 434, 1148, 2005). Haase's group and another team have independently reported on the massive depletion of CD4+ memory T cells in the first few weeks of infection (see Research Briefs).

Haase noted that while the virus was making short work of CD4+ memory T cells in the GALT, the response of the immune system to this onslaught was "too little and too late." In gut tissues that were awash with viral particles at 10 days, the researchers never detected significant numbers of CD8+ T cells. A robust CD8+ T cell response in vaginal mucosa only came up after the virus had done its damage, raising the possibility that prophylactic immunity induced by a vaccine could beat the virus. "Immediately following infection, the virus is vulnerable," says Haase. "You have a small founder population of infected cells fighting to keep the infection going."

Two other talks focused on the potential importance of cell-cell contacts in spreading HIV. Dendritic cells (DCs) are among the first cells the virus encounters in mucosal tissue. DCs normally act as antigen presenting cells, processing viral proteins and presenting them to T cells to fire up an immune response. But DCs can also enhance HIV infection in two ways: they can host HIV replication and transfer viruses from their surface directly to T cells without replication.

Yvette van Kooyk of the Vrije University Medical Center Amsterdam discussed the complex interplay of HIV with receptors on the DC surface that govern the path the virus takes inside the cell. HIV envelope glycoprotein gp120 binds the C-type lectin receptor (CLR) DC-SIGN leading to rapid internalization of the whole viral particle and its subsequent transmission to CD4+ T cells. So the DC-SIGN pathway is possibly one way in which HIV manages to avoid being processed to antigen or inducing DC maturation and inflammatory responses. But van Kooyk's team showed that when DC-SIGN was blocked with antibodies, presentation of gp120 on MHC class II and activation of gp120-specific CD4+ T cells was reduced. This suggests that after HIV engages DC-SIGN, the virus may simultaneously proceed on two intracellular routes that work together to boost infection: antigen presentation to enhance proliferative responses of CD4+ T cells and transmission of whole virus to these cells.

Clare Jolly, a postdoctoral researcher in Quentin Sattentau's laboratory at the University of Oxford, described her work on the virological synapse (VS), a tight union between the membrane of an infected and uninfected T cell. It's been proposed that HIV uses the VS to rapidly colonize closely-packed T cell populations, as is the case in lymph nodes, but recently-transmitted virus may also encounter clusters of T cells in the genital tract, especially in the inflammatory context of a secondary sexually-transmitted disease. Jolly used confocal and electron microscopy to visualize the synapses and fluorescent antibodies to probe their protein content in primary CD4+ T cells. Her data support a model where proteins are actively recruited to the VS since inhibitors of cytoskeleton or microtubule function stopped the accumulation of viral and cellular proteins there. Jolly found that cholesterol-rich lipid rafts—which HIV favors as a budding site—were required for VS formation and were clustered preferentially at the synapse. Jolly pointed out that the tight VS may shield emerging viruses from antibodies or recognition by immune cells, making these structures of interest to vaccine designers. "The question is whether neutralizing antibodies gain access to the synapse and can they block cell to cell transmission?" she says. Her team is now characterizing the ability of antibodies, peptides and small molecules to block HIV budding through the VS.

Long-term lessons

Even twenty years into studying HIV, researchers continue to learn about the complex interaction between the virus, its host and its environment. By exploring those interactions, researchers are gaining insights into why some rare people are able to control infection indefinitely, what factors determine and predict the immune response to particular viruses, and how antiretrovirals (ARVs) shape viral evolution and host immunity.

One longstanding area of interest are long term nonprogressors (LTNPs), patients who have never received ARV therapy yet maintain normal CD4+ T cell counts and nearly undetectable levels of virus for years. "We have this excellent example in humans of immunologic control of HIV and it's incumbent upon us to develop a deeper understanding of it," says Mark Connors of National Institute of Allergy and Infectious Diseases.

Connors' team has found that in many ways the HIV-specific CD8+ T cell response of their LTNP and progressor cohorts are indistinguishable—similar quantitatively and in clonality, surface protein markers, and microarray expression patterns—and there was no correlation between mutations allowing the virus to evade T cell responses and the ability of the immune system to control viral replication. But they did find one striking difference in CD8+ T cells taken from LTNPs: these cells were much more likely to proliferate after restimulation with HIV-infected cells or HIV-epitopes in vitro. This proliferation was accompanied by the production of perforin and granzyme B, indicating cytolytic function. In contrast, restimulation of CD8+ T cells from progressors triggered the production of cytokines and degranulation but did not induce proliferation.

What this means for the function of these cells in vivo is still unclear, but Connors discussed three possible explanations for this proliferation defect in CD8+ T cells from progressors: either their cells have already been pushed to the point of replicative senescence, they are being restricted in their replication (anergy), or their developmental program has halted at an immature state. If the cells are senescent, Connors thinks there are few options for therapy to rejuvenate them. However if the cells are anergic or immature, then therapies may be devised to jump start them. "We're going to be looking at this in much greater detail and testing the likely actors, including factors involved with intracellular signaling and protein translocation," says Connors. His team has already tried treating CD8+ T cells from progressors with IL-2, a cytokine that is able to stimulate some types of anergic cells, so far without success.

While many aspects of how an immune response shapes an HIV infection remain a mystery, Bruce Walker of Harvard Medical School in Boston pointed out that some examples of virus-host interactions are highly predictable and could possibly be exploited for vaccine design.

In a study of 375 Zulus infected with HIV clade C, Walker's team found a strong association between the HLA MHC type and IFN-γ ELISPOT assay readout. "Our technicians got to the point where they could even predict a person's HLA based on the IFN-γ readouts they were seeing," said Walker. The researchers found that 111 of 410 overlapping peptides, representing epitopes from a reference strain, showed strong allele-specific HLA associations. In addition, his team has reported that HLA-B alleles have a powerful influence on viral load (Nature, 432, 769, 2005), providing strong evidence that the cellular immune response does indeed influence virus replication in humans.

Walker also presented a remarkable example of how host and viral genetics determine the course of infection, involving a rare incidence of identical male twins infected by HIV through intravenous drug use. Sequence analysis demonstrated that both twins were infected by the same virus, suggesting they were probably infected on the same day. Walker and his colleagues found that the subsequent course of infection appeared to proceed in parallel in the two brothers. The time course of their viral load and CD4+ T cell counts were almost indistinguishable and antibody sera from one twin were able to neutralize virus from the other. A very similar pattern of response to different HIV epitopes was also found in IFN-γ assays of both twins. In addition, the same viral escape mutants from the initial immunodominant epitope arose around 29 months after infection in both men, a clear indication that the virus and host genetic interplay rather than stochastic events determine the course of HIV infection.

ARV therapy can also play a role in shaping the course of infection. In the best case scenario, drugs beat down viral replication to undetectable levels in some HIV-infected people, halting disease progression indefinitely. But in an intriguing talk, Steven Deeks of the University of California, San Francisco, argued that even in cases where drug therapy "fails"—that is, drug-resistant virus emerges—there still may be therapeutic benefit.

Deeks and his colleagues have defined an HIV-infected group they refer to as partial controllers on ARV therapy (PCATs), which comprise 35% of people who fail drug therapy. These individuals show a persistent 10-fold or greater reduction of viral load from their pretreatment viremia (consistently <10,000 vRNA copies/ml) despite the emergence of highly drug-resistant virus.

This suggested that the selection of some types of HIV drug resistance may result in the evolution of less fit viruses which are more likely to achieve a balance with the immune system. Indeed, the researchers found that the immune response of PCATs in many ways resembles that of LTNPs. CD4+ T cells in PCATs and LTNPs produced high levels of both IL-2 and IFN-γ in response to Gag, while this was not true for patients whose viral load was reduced below detection by ARVs or in patients where ARVs had no lasting benefit. Deeks' team also showed that non-specific activation of CD8+ memory T cells was significantly lower and similar in PCATs and LTNPs compared to progressors. In LTNPs and PCATs, Deeks concluded, the host generates a strong, multifunctional HIV-specific CD4+ memory T cell response in the context of relatively low inflammatory response.

Deeks then presented preliminary data on drug resistance to the fusion inhibitor T20, which suppresses viral load to undetectable levels in many patients, even those with virus resistant to many other ARVs. His team found that when T20 drug resistance emerges it does so very quickly, in as little as 2 weeks, and viral load rebounds back toward baseline. Even so, T20 treatment still resulted in a higher CD4+ T cell count that persisted for more than 2 months and data suggested this improvement was driven by a fitness defect in the virus caused by the drug resistance, an idea that may have implications in vaccine design (see Perspective). Deeks concluded that the study of the interplay between drug-resistant virus and the immune system holds important clues for clinicians, immunologists and vaccinologists. "Attenuating the virus with drugs may be the closest thing we have to an attenuated vaccine for humans," he says.

Monkey business

Primate SIV models are also being used to study immunological control of viral replication or to reveal aspects of HIV/SIV pathology. David Watkins of the University of Wisconsin began his talk presenting what may be the Indian rhesus macaque equivalents of LTNPs.

When these animals are challenged with the highly pathogenic SIVmac239, the infection typically settles to a baseline viral load of over 1x106 vRNA copies/ml and the animals progress to and die from AIDS. But Watkins' team has found some rare spontaneous "controllers" that maintain viral load below 50,000 vRNA copies/ml and "elite controllers" below 1,000 vRNA copies/ml. These controller populations often express the Mamu-B*17 MHC class I molecule and tend to mount broad, low frequency SIV-specific CD4+ and CD8+ T cell responses, but show no sign of neutralizing antibodies in conventional PBMC neutralization assays.

While these controllers may be valuable for elucidating immune correlates of protection, Watkins noted it argued for excluding Mamu-B*17+ animals from vaccine studies. He also presented encouraging data from such a study of Mamu A*01+ macaques using a DNA prime, adenovirus serotype 5 (Ad5) boost vaccine regimen with both vectors expressing Gag, Tat, Rev and Nef immunogens, but no Env component. After repeated low-dose SIVmac239 challenge at 6 months, the regimen suppressed peak viremia more than 10-fold, suppressed chronic viral load set point by 1.5 logs, and was able to slow the loss of memory T cell subsets that are now thought to play a critical role in establishing the course of HIV disease.

Watkins noted the need to test the vaccine regimen in genetic backgrounds other than Mamu-A*01, and while his DNA/Ad5 vaccination protocol proved effective, he noted that the experiment did not reflect a realistic course of HIV infection since the vaccine and challenge viruses were exactly matched and the challenge came just 6 month after vaccination. His team plans to address these issues in future experiments.

Jeffrey Lifson of SAIC Frederick, Inc. at the US National Cancer Institute presented preliminary data on correlates of viral control in Indian rhesus macaques, characterizing virus-specific responses of circulating immune cells using polychromatic flow cytometry with intracellular cytokine staining. The animals had controlled replication of pathogenic SIV isolates (SIVmac239 or SIVsmE660) either spontaneously or after short term ARV treatment during primary infection, or after various vaccination regimens.

The results suggest that more than one pattern of immune response may be associated with effective virus control. One macaque that had controlled infection with SIVmac239 for 5 years with minimal viral load (~ 100 vRNA copies/ml) demonstrated robust, broad CD4+ and CD8+ T cell responses to SIV antigens. At the other end of the spectrum was an animal that was "vaccinated" by transient early ARV treatment during the first weeks of infection with SIVsmE660. Except for a few occasional blips, this monkey held viral load to below 100 vRNA copies/ml despite high dose intravenous rechallenges with the virulent viruses SIVsmE660 and SIVmac239. Yet this animal's T cell responses in peripheral blood were barely detectable. "This animal is a poster child for the argument that in focusing on higher frequency immune responses measured in peripheral blood, we are almost certainly missing an important part of the story," says Lifson. He plans to follow up the work by looking at qualitative features of responding cells and antiviral responses in lymphoid tissues.

Ronald Veazey of Tulane University in New Orleans used data from a number of primate species to argue that pathogenesis of HIV/SIV infection is a function of target cell availability, turnover and destruction of CD4+ T cells in the gut—what he called a unifying hypothesis. His team has been building on their work suggesting that the infection and subsequent direct viral destruction of gut CD4+ T cells is a crucial part of SIV pathogenesis. This observation has proven to be also true in humans and has recently been independently supported by other research groups (see Research Briefs). His new data show that the sudden loss of these cells can't be accounted for by three other mechanisms: CD4+ T cell redistribution, cytotoxic T lymphocyte (CTL) killing, or lowered CD4+ T cell production.

Veazey's team looked for redistribution of CD4+ T cells during SIV infection in rhesus macaques and found no evidence for migration of these cells to other compartments, including blood, spleen, liver or bone marrow. "If they are redistributing, it must be to hair or toenails because we've looked at every other tissue," he joked. To address whether CD8+ cells could be targeting infected CD4+ T cells, the researchers treated animals with anti-CD8 antibodies before infection. CD4+ T cell destruction was even greater in the CD8+ cell-depleted animals. Finally, they measured CD4+ T cell replication by pulsing recently infected animals with BrdU label for 24 hours. They found steady production of CD4+ T cells, especially in the bone marrow.

He then talked about sooty mangabeys and African green monkeys, natural hosts of SIV that maintain high plasma viral loads but don't develop disease. His work suggests these animals can withstand infection because their gut CD4+ memory T cell population lacks the CCR5 coreceptor for the virus and, as a result, aren't targets for SIV and aren't destroyed.

Veazey ended his talk with a cautionary tale for vaccine researchers based on a vaccine study using a SHIVsf162p3 vector. His team found this virus, which Chinese macaques easily clear in about 90 days, appeared to give robust protection against the otherwise deadly SIVmac251. Two out of 5 animals had a 100-fold viral load reduction over an unvaccinated control animal, and the remaining trio had no detectable plasma viremia. Nonetheless, when Veazey examined the gut-associated CD4+ T cells in these 3 "protected" animals, he found the cells were completely destroyed. "This argues we need to really monitor mucosal immune response in vaccine studies or we might be missing something very important," he says.

Clinical update

Barney Graham of the National Institutes of Health Vaccine Research Center (VRC) began his presentation with an analogy to the setting of the meeting, saying that we were stuck in the foothills and "we still have to get over the mountain in AIDS vaccine research." He gave an interim report on three ongoing multi-center safety, immunogenicity and dose-escalation clinical trials of DNA and Ad5 AIDS vaccine candidates, preliminary trials with an eye to future prime/boost trial regimens.

Trial VRC 004 is testing a 4 plasmid DNA vaccination protocol (subtype A, B, C env genes plus a subtype B gag/pol/nef fusion construct) administered three times at four-week intervals to 40 volunteers at 2, 4, and 8mg doses. Encouragingly almost all vaccinees had CD4+ T cell responses against Env (and 25% against Gag) detectable by intracellular cytokine staining (ICS) that persisted for a year. After 6 weeks (between the second and third vaccine dose) these were of broad phenotype—either IL-2+ only, IFNγ+ only, or IL-2+/ IFNγ+. CD8+ T cell responses were detectable by ICS against Env in 40% and Gag in 25% of subjects who received the two higher doses of DNA vaccine, the majority of these cells being IFNγ+ only, with some IL-2+ only and very few expressing both. Graham said that these cellular responses evolved over time so that a large proportion of cells switched to the IL-2+ only phenotype by the one year mark. At 6 weeks, 40% and 80% of vaccinees given the 4 and 8mg doses respectively also showed Env-specific antibody detected by immunoprecipitation Western blot.

He then presented data from trial VRC 007 involving 15 volunteers each given 4mg of six DNA plasmids—subtype A, B, C env genes and subtype B gag, pol, and nef genes, each on separate plasmids—three times at four-week intervals. At 6 weeks, CD4+ T cell responses were again seen in all volunteers against Env, but there were also more vaccinees with responses against Gag than was evident with the gag/pol/nef fusion construct used in trial VRC 004. About half the vaccinees had CD8+ T cell responses against Env and some also against Gag and Nef. Eight of the volunteers also had detectable Env-specific antibodies.

Trial VRC 006 used a cocktail of four Ad5 recombinants (expressing either subtype A, B, or C Env or a Gag/Pol fusion) to vaccinate 10 volunteers each three times with either 109, 1010, or 1011 particles at four-week intervals. At the highest dose about half the volunteers had mild systemic symptoms (fevers, chills). CD4+ T cell responses to all of the immunogen proteins were seen in nearly all vaccinees and the majority also had CD8+ T cell responses, some even in the face of pre-existing anti-Ad5 antibody responses at titers as high as 1:8000. Antibody responses were dose-dependent, with nearly all vaccinees given the 1011 dose showing a response but only 20-30% of vaccinees given the lower doses.

"The most exciting thing is that the data from this combination of trials has led to a set of protocols that will be submitted to the FDA for approval for use in Phase II DNA prime/adenovirus boost trials" says Graham. These trials will be an "exciting collaboration" between the HIV Vaccine Trials Network (HVTN), the US Military Research Program and IAVI at sites in eastern and southern Africa, Haiti, Puerto Rico, Jamaica, Brazil and the US. A priority trial that is already recruiting is the six plasmid prime combined with the four Ad5 recombinant boost.

John Shiver of Merck Research Laboratories gave an update on various aspects of their AIDS vaccine program. Merck are finding the ELISPOT assay more sensitive than ICS—they can detect more than twice as many CD8+ T cell responders with ELISPOT, in contrast to researchers at the VRC. Shiver described a study designed to see if higher doses of Ad5-Gag (from subtype B) can overcome the blunting effects of pre-existing anti-vector antibody immunity on induced CD8+ T cell responses. Eight weeks after vaccination with Ad5-Gag at one of four doses (108, 109, 1010, or 1011 particles) the expected lowered frequency of responders was evident when anti-Ad5 antibody titers had exceeded 1:200 before vaccination. The highest 1011 dose was not able to improve either the frequency or magnitude of response.

He then described a trial with a trivalent Ad5 recombinant expressing Gag, Pol, and Nef (all subtype B) administered to volunteers at the same four doses. After 8 weeks there was a dose-dependent frequency of CD8+ T cell responders—for instance, 24%, 38%, 76%, and 68% of vaccinees given 108, 109, 1010, or 1011 respectively had responses against more than one protein. However there was little difference between the 1010 and 1011 doses, suggesting that the highest dose may not be worth pursuing given the mild systemic symptoms seen in trial VRC 006 at this dose.

Merck is also looking to see how effective their candidates might be against other subtypes of HIV. CD8+ T cells from vaccinees given the subtype B trivalent Ad5-Gag/Pol/Nef were tested for responses against subtype B, A, and C derived peptides. Gag-specific responses were not drastically lower—61% responded to subtype B Gag peptides, whereas 43% and 45% responded to subtype A and C peptides respectively. But Nef-specific responses were very different—47% to subtype B but only 19% and 7% to subtypes A and C respectively.

Shiver also described a clinical trial of Ad5-Gag prime, ALVAC canarypox vector (expressing an almost identical Gag) boost, a vaccine modality that has previously been shown to induce potentially synergistic immune responses in monkeys. In humans, however, there appeared to be no such effect and the frequency of responders (and magnitude of response) remained the same.

Merck, in collaboration with the HVTN, is currently conducting a proof of concept clinical trial with their trivalent Ad5-Gag/Pol/Nef candidate to asses the efficacy of the vaccine-elicited immune response.