AIDScience Vol. 2, No. 9, 10 May 2002
Warming trends at Keystone vaccine conference
Talks highlight human trials, T-cell markers and antibody approaches
by Emily Bass and Richard Jefferys
From the IAVI Report, the newsletter of the International AIDS Vaccine Initiative. Address correspondence to: Reprinted with permission from the IAVI Report.

nseasonably high temperatures and record-low amounts of snow set the backdrop for this yearís Keystone Symposium on "HIV Protection and Control by Vaccination" in the Colorado Rockies. Inside, the thaw was more welcome, as early clinical trials data started to flow through the vaccine development pipeline. In the corridors, participants noted the welcome move towards convergence of virology and immunology, with virologists looking increasingly at approaches for immune control of HIV and immunologists envisioning therapeutic uses for vaccines alongside the goal of preventing disease and ultimately, infection.


Two keynote speakers addressed some of the challenges faced by HIV vaccine developers specifically tackling either the cellular or humoral arm of the immune system.

Cellular immunologist Rafi Ahmed (Emory University) focused on a subset of T-cells responsible for immunological memory, which enables the immune system to "remember" microbes or vaccine antigens encountered long ago and to respond quickly if and when (re-)infection occurs. After reminding the audience that he works not on HIV but on viral infection in mice, Ahmed plunged into his teamís extensive studies on the generation, maintenance and function of CD8 memory cells and on what these findings imply for designing more effective immunization regimens.

Using sophisticated microarray technology, he showed that the process of generating antigen- specific memory cells from "naÔve" precursors involves changes in the expression of thousands of genes, extending over several weeks—suggesting that prime- boost immunization schedules should plan enough time between prime and boost for this full maturation of memory cells. Zeroing in on one specific molecular change, Ahmed identified a cell surface protein called CD62L (or L-selectin) that could be a marker for resting ("central") memory cells. CD8 T-cells appear to lose and reacquire CD62L as they transition from naÔve to central memory T-cells, and reacquisition is correlated with the ability to mount a robust proliferative response to viral infection.

Ahmed then examined how antigen dose affected both the initial number of antigen-specific CD8 T-cells ("burst size") and the eventual number of CD62L-expressing central memory cells. The larger the initial burst, he said, the larger the eventual memory cell pool—but the longer it took them to emerge. Drawing a message for vaccine developers, this suggests that the interval between prime and boost immunizations should increase relative to the strength of the prime.

Ahmed also reviewed data showing that antigen is not required to maintain memory CD8 T-cells and described his groupís search for factors other than antigen that might play a role. One of his lead candidates is the cytokine IL-15, based partly on studies of "knockout mice" that lack the IL-15 gene; these animals showed a gradual attrition of their memory CD8 T-cell pool, apparently stemming from a drop in proliferation rate. Later in the meeting, Ahmedís Emory colleague FranÁoise Villinger presented a poster supporting the notion that IL-15 might be a useful adjunct to vaccination: 4 macaques immunized against influenza and then given a short course of IL-15 had 4-5 fold more flu-specific memory T-cell than animals receiving vaccine only, and the elevated levels have so far persisted out to six months follow-up. In contrast, IL-2 given to 4 animals led to only transient increase in memory T- cell numbers.

Next, Joseph Sodroski (Dana Farber Cancer Institute, Boston) turned to the problem of inducing broad neutralizing antibodies (NAbs) to primary HIV strains, which no vaccine has yet achieved, reviewing the obstacles and outlining one new approach he is pursuing. He and collaborator Rich Wyatt (Vaccine Research Center [VRC], Bethesda) found that the binding of HIVís gp120 envelope protein to CD4, one of two cell surface receptors for HIV, requires an atypically large expenditure of "ordering energy" to rigidify gp120ís flexible structure. Likewise, antibody binding to neutralizing epitopes in gp120, usually concealed within the moleculeís interior, is also energetically unfavorable. To overcome this obstacle, Sodroski and Wyatt are exploring "molecular plugs" for a 10-angstrom-deep cavity in the gp120 structure that is a pivot point of floppiness. Here, the goal is to stabilize gp120 in a conformation that exposes usually-concealed NAb epitopes, thereby allowing easier NAb induction.


Merckís adenovirus-based vaccine
Emilio Emini, director of Merckís vaccine division, opened the session on clinical trials with updated data on its adenovirus-based vaccine, a proof-of-concept construct containing HIV-gag in an Ad5 vector. Reporting on the highest-dose group (1011 viral particles) in a Phase I dose-response study presented at the recent Retrovirus conference (see IAVI Report, Jan/Feb 2002, "Retrovirus 2002: Merck Debuts Phase I Data"), Emini said that after the first two injections 7/9 volunteers showed CD8 T-cell responses. (Responses were measured by ELISPOT assays using interferon (IFN)-gamma, which is secreted by activated T- cells.) This was the highest proportion of responders seen with Ad5-gag in any dosage group so far, although coincidentally this group had fewer volunteers with pre-existing adenovirus immunity (defined by Merck as NAb titers above 1:200) than the next- lowest dose group (1010). This random distribution of people above Merckís threshold makes it difficult to draw conclusions for now about dose response in the context of pre-existing immunity.

Emini also presented data using much lower doses of Ad5-gag, in this case as a boost for a DNA-gag vaccine. Macaques boosted with 107 Ad5-gag particles after a DNA-gag prime responded as strongly or better than those given 1011 particles without a prime, depending on NAb levels. In addition to these trials, Merck has started a prime-boost study of DNA-gag plus Ad5-gag, along with several trials in HIV-infected people, looking towards possible therapeutic use. For preventive vaccines, plans are underway to begin a much larger Phase I study.

Oxford-IAVI prime-boost trials
Oxford Universityís Andrew McMichael followed with preliminary results from trials of DNA and MVA-based vaccines (DNA-HIVA and MVA.HIV) in the UK, sponsored by Britainís Medical Research Council and IAVI. He began by reviewing the separate DNA and MVA studies, which detected T-cell responses to HIV in the majority of volunteers (based on ELISPOT assays using IFN- gamma). Responses to the DNA-HIVA vaccine (100 or 500 Ķg given at days 0 and 21) were mostly low, and were dose- independent; a few came up only months after the final immunization. MVA.HIV vaccinees showed peak immune responses 21 days after the second of two immunizations with 5 x 107 plaque-forming units. Responses to DNA-HIVA were a mixture of CD8 and CD4 T-cells, while MVA.HIV responses were predominantly CD8 T-cells.

In the prime-boost study, McMichael reported on 6 volunteers from the DNA-HIVA trial received an MVA.HIV boost one year after their final DNA injection. All six showed responses after the boost. "The prime-boost might give a better mix of responses than either one alone," said McMichael. The Oxford team is now conducting intracellular cytokine (ICC) assays for a closer qualitative analysis of the responding cells.

Similar DNA and MVA trials are being conducted in Nairobi, and the UK team has just begun a Phase IIa prime-boost study (see article, "Expanded DNA-MVA Prime-Boost Trial Begins in UK").

Table 1. HVTN 203 immunogenicity results (day 182)
Test antigen/volunteer groupResponsea
vaccine8/122 (7%)
vaccine16/122 (13%)
Env or Gag
vaccine20/124 (16%)
aNumber and percent of volunteers (relative to total number tested) showing HIV-specific CD8 T-cell responses as measured by standard cytotoxic lymphocyte assays (51Cr-release) on fresh PBMCs, 182 days after the last immunization.

NIH-sponsored trials
Rounding out the human trials session, Barney Graham (VRC, Bethesda) began with data from HVTN 203, the Phase II trial that was pivotal in NIHís decision not to move into efficacy trials with this prime-boost strategy, which combines the canarypox-based ALVAC vCP1452 vaccine with a gp120 boost. (In Thailand, Phase III plans for a prime-boost trial of another ALVAC vaccine plus gp120 are continuing, although this trial was not discussed by any Keystone presenters.) Six months after the last immunization, only about 16% of the vaccinated volunteers showed T-cell responses against HIV Env or Gag, as measured by cell-killing assays (see Table 1). Graham did not present ELISPOT data, but said that interim results appeared "no better" in terms of immunogenicity.

He then described future trial plans for the VRC, which is working on a DNA-adenovirus strategy similar to Merckís. Within the next few months, they will launch a Phase I trial of a DNA-gag prime delivered via Biojector, a needle-free delivery system, and a boost of recombinant adenovirus (rAD) containing env from clades A, B and C, delivered both intramuscularly (IM) and intranasally (IN). Building on the VRCís DNA-gag-pol vaccine now in Phase I studies, new DNA and adenovirus constructs will include HIV- gag, pol nef and env from clades A, B & C. Graham said that the VRC is pursuing a "multivalent, multiclade" approach.

Wyeth Ayerstís HIV-VSV vaccines
Last year, John Rose from Yale University published encouraging protection data on an SIV vaccine made in a vesicular stomatitis virus (VSV) vector (Cell 2001;106:539-549). At Keystone, Michael Egan from Wyeth Ayerst (the company that has now licensed the VSV platform) presented a new study comparing mucosal vaccination (IN) to the earlier IM immunization protocol (see details in Table 2) in ten macaques, which were then challenged intravaginally with the pathogenic HIV/SIV hybrid SHIV89.6P.

Table 2. Effect of HIV-VSV immunization on CD4 countsa
Study groupCD4 cell count (% of baseline)
at 150 days in each animal
IM group
vaccinated (n = 3)86, 76, 47
control (n = 2)21, other animal died < day 150
IN group
vaccinated102, 92, 83
control (n = 2)104, 28
aAnimals were immunized at 0, 8 and 16 weeks with VSV vectors encoding SIV Gag and HIV Env genes and challenged intravaginally with HIV89.6P at week 21. Data show CD4 counts from individual animals, 150 days after challenge.


Overall, they found that animals immunized IN had higher levels of pre-challenge cellular immune responses to SIV (measured by tetramer staining and ELISPOT) compared with the IM group, as well as somewhat better post-challenge control of peak viral load and preservation of CD4 counts. Antibody titers were similar in both vaccinated groups, and (unusually) all animals—except one control which died by 150 days post-challenge—showed similarly low viral set points. Based on these data, Egan concluded that IN administration is the preferred delivery method for this vaccine. Wyeth Ayerst is continuing to develop the VSV platform for potential human use.

In search of AAV
Last year at Keystone, Phil Johnson (Childrenís Research Institute, Columbus) showed promising macaque data on an HIV vaccine made in Adeno-Associated Virus (AAV), a harmless virus widespread in human populations (see IAVI Report Feb/Mar 2001, ). The potential advantage of AAV vectors is that they persistently express foreign genes—at least 17 months has been seen in macaques—and might therefore work as single-dose vaccines. However, a key safety issue is whether the vector integrates into the host cell genome, which could have unknown consequences.

This year, Johnson presented data showing that the AAV-SIV vaccine does not appear to integrate into the DNA of immunized mice. Using a battery of different tests, his team found no integrated AAV in muscle cells at the injection site, indicating that over 99.5% of the vector molecules that persist are unintegrated, said Johnson, and "in the same ballpark as plasmid DNA vaccines."

Johnson concluded by outlining a novel strategy for combining HIV-specific cellular immune responses and neutralizing antibodies (NAb) in macaques—the combination widely viewed as a "next-generation" HIV vaccine. Since no vaccine so far has induced broad Nabs against primary strains of virus, Johnson and collaborators Dennis Burton and Paul Parren (Scripps Research Institute, La Jolla) instead took an existing one—a monoclonal antibody called b12, derived from an HIV-infected person—and inserted the DNA encoding it into AAV vectors. Immunodeficient mice injected with these constructs have so far shown sustained b12 expression for over six months. Johnson is now looking to test this b12-AAV vector together with his existing HIV-AAV vaccine, which induces cellular responses, and develop the combination as a human vaccine.

T-cell responses, viral escape and avidity
David Watkins (Wisconsin Primate Center) presented data on HIV escape in three macaques that became infected despite vaccination. The animals (all MamuA*01) had received a series of immunizations with DNA and MVA constructs encoding the epitopes Gag CM9, Tat SL8 and whole Tat, Rev and Nef proteins. The choice of immunogens was based on a long-term non- progressor monkey that had CD8 cells directed against these epitopes (plus CD4 T-cells targeting Rev, Nef and Gag); moreover, the CD8 cells bound these epitopes with high avidity, which Watkins has correlated with the emergence of SIV escape mutants in infected animals. He therefore reasoned that inducing these responses in animals prior to challenge—in effect giving them a head start over SIV—might induce effective immunity. Twenty weeks after the last immunization, animals were given a high-dose i.v. challenge with SIVmac239. But, despite an initial reduction in peak viral load, all animals had high levels of viremia by week 18, along with SIV escape mutations in the CM9 and SL8 epitopes.

Watkins also showed that this vaccine failure occurred despite indications that the animals had HIV-specific mucosal responses. About 12% of T-cells in the gut and 8% in blood expressed a cell surface molecule, alpha4-beta7, associated with cells trafficking to the mucosae. James Allison (University of California, Berkeley) offered another perspective on avidity and epitope selection. Allison found that high-avidity binding is associated with the up-regulation of CTLA-4, a cell surface molecule that stops T-cell replication. Allison speculated that this may help increase the diversity of immune responses by shutting down the proliferation of T-cells binding to high-avidity epitopes, thus leaving room for sub-dominant responses to emerge.


Continuing the theme of identifying immune parameters of viral control, Brigitte Autran (HŰpital Pitiť-SalpÍtriŤre) described recent results from her groupís extensive studies on long-term non-progressors (LNTP) and outlined how these findings are being used in developing therapeutic vaccines. In terms of cellular immunity, control of viral load in LNTP correlated with high levels of HIV- specific CD4 cells (those producing IFN-gamma), and, to a lesser extent, with the intensity and diversity of CD8 T-cell responses. Non-progression also correlated with levels of IgG2 antibodies directed against Gag and gp41 (AIDS Res. Hum. Retroviruses: 2001;17:1435). Whatever role these antibodies play (they do not appear to be neutralizing), Autran proposes that they may give data on T-cell functionality not captured by IFN-gamma alone. Overall, the combination of IgG2 and HIV-specific CD4 responses were the best predictors of LTNP status.

Autranís talk finished by reviewing efforts to enhance CD4 responses in infected people on HAART, using canarypox-based HIV vaccines (ALVAC vCP1433 or vCP1452) with or without a whole-killed HIV vaccine (Remuneģ, originally developed by Jonas Salk). After one or two injections, 65-75% of the 200 volunteers had detectable T-helper responses (50-800 spot-forming cells in chronically infected people given ALVAC, 50-1500 SFC in acutely infected people given ALVAC or both vaccines). Autran, who plans similar trials with newer vaccines, added wryly that therapeutic vaccination is "Elysienne [like heaven]—we always talk about it, but never see it."

In the same session, Bruce Walker presented a prospective, blinded study of Zambian serodiscordant couples. Walker and a team led by Marilyn Addo analyzed samples from 36 individuals, 25 of whom were defined as high-risk due to being in serodiscordant couples, and found no evidence of anti-HIV responses at the first time point or one year later. Walker concluded with the bold suggestion that "There are no CTL responses in HEPS [highly exposed, persistently seronegative people]—thatís our feeling based on the data we have." Of note, the study criterion for high exposure was 3-4 high-risk sexual events per month—much less than the 6-7 exposures per day or week seen in other HEPS cohorts, such as the Nairobi sex worker group, a difference that could explain the absence of responses in this study.

In another session, Sarah Fidler (Imperial College Faculty of Medicine, London) presented a UK-based study of 20 predominantly homosexual male couples showing that 13/14 had quantifiable HIV-specific CTL on at least one occasion. Here, highly-exposed individuals were defined as those who had 2 to 3 acts of unprotected, receptive anal or vaginal intercourse per week.


From Ahmed to Autran, presenters echoed the need to define the most important functional markers associated with protection or immune control of HIV. IFN-gamma production is now commonly used to detect the presence of immune-activated T-cells in vaccine trials, but several presentations suggested that other markers may also be informative. For example, Ahmed pointed out that CD8 T-cells continue to produce IFN-gamma long after they lose other key functions, such as ability to proliferate or to produce IL-2 and TNF-alpha—suggesting that IFN-gamma may detect not only fully active T-cells but also functionally "exhausted" ones. Ahmedís thinking is echoed in a recent paper from Norman Letvinís laboratory (Harvard University, Cambridge) which showed that protection from disease progression in SIV-infected macaques correlates with the presence SIV-specific CD8 cells producing both IL-2 and TNF-alpha (J. Immunol.2002;168:332-337).

Backing this up, a poster by Helen Horton (Fred Hutchinson Cancer Research Center, Seattle) compared results of classical cell killing (51Cr-release) assays and IFN-gamma production measured by ELISPOT, concluding that T-cells which produce IFN- gamma after antigen stimulation are not always able to kill target cells.

What other markers might play a role? Mark Boaz and colleagues from Kings College School of Medicine (London) characterized the HIV-specific CD4 T-cell responses in 16 long-term non-progressors (LTNPs) and compared them to 15 patients with progressive disease. LTNPs turned out to have significantly more cells expressing both IFN-gamma and IL-2. The two groups showed no difference in the proportion of T-cells expressing IFN-gamma only.

Table 3. Immune responsesa in blood and breastmilk
 BloodBreastmilkMilk+ / blood-
2 week env23/35b (66%)12/35 (34%)3/10 (30%)
2 week gag15/35 (52%)9/35 (31%)0/10 (0%)
4 week env15/35 (52%)9/35 (31%)3/12 (25%)
4 week gag11/21 (52%)3/21 (14%)2/14 (14%)
aImmune responses were measured on CD8 lymphocytes using and ELISPOT assay for IFN-gamma. A positive result = PHA stimulation > 100 SFU/106.
bNumber showing positive T-cell response/total number of volunteers.


Mucosal immunologists complemented the marker-hunters with more data on HIV immunity in compartments other than blood. Yan Ding from Juliana McElrathís group (University of Washington, Seattle) looked at CD8 T cells cloned from the blood, rectum and cervix of 12 (10M/2F) sexually-infected long-term LTNP and found that cells from these different compartments shared features such as epitope specificities and MHC restriction patterns. This suggests a common origin as well as trafficking between blood and mucosa, in contrast to several other studies that found distinct populations in the two compartments. To see whether her finding is specific to LTNP, Ding will do the same analysis in chronically infected individuals.

Another view came from Barbara Lohman (University of Nairobi) who showed that T-cell populations from blood and breastmilk of HIV-infected women differ in the proportion that respond to specific HIV antigens (see Table 3), suggesting some degree of separation between the compartments.

Emily Bass is Senior Writer at the IAVI Report. Richard Jefferys is the Basic Science Project Director at the Treatment Action Group, a New York-based organization advocating for HIV research. He was previously director of the Access Project at the AIDS Treatment Data Network, and most recently worked as a writer for the IAVI Report.

Copyright © 2001 by The American Association for the Advancement of Science