Keystone Symposium

Updates on Trials, New Candidates and Immune Basis of Protection  

By Mark Boaz 

Continuing our report of April’s Keystone Symposium on HIV Vaccine Development, one of the field’s premiere research conferences, we cover an array of talks ranging from animal models to ongoing clinical trials. Although there were no major surprises or new findings announced, the meeting gave a good overview of the broad efforts underway on many different fronts.

Update on human clinical trials Updates from ongoing vaccine trials and from monkey studies of candidates due to enter Phase I studies soon, featured heavily on the conference agenda. Since last year’s Keystone conference, four completely new vaccines have entered Phase I trials and several other Phase I and I/II studies were launched. 
(See database).

DNA candidates from the Vaccine Research Center

Barney Graham, who directs clinical studies at the NIH Vaccine Research Center (VRC; Bethesda), discussed early results from Phase I trials of the Center’s new DNA vaccines. These constructs are the first step in a multi-clade, prime-boost strategy that will use adenovirus-based vaccines as a boost and incorporate env from three different subtypes.

The first VRC trial, started in late 2001, tested DNA containing HIV gag and pol from subtype B at three doses (0.5 mg, 1.5 mg or 4 mg). Immune responses were determined by measuring interferon- (IFN)-gamma-producing cells using intracytoplasmic cytokine staining (ICS) assays. Each group had seven volunteers, two of whom received placebo. Graham presented unblinded data on responses after the second of three vaccinations.

CD4 T-cell responses were present in 4/7 individuals in the lowest-dose group, mainly to Gag but also to Pol, and Gag-specific CD8 cells were found in two of these volunteers. The 1.5 mg dose led to CD4 and CD8 responses in 2/7 volunteers. No data is available yet for the 4 mg group.

Graham also presented early results from the more recently launched trial of multi-clade DNA containing gagpol, and nef from subtype B and env from subtypes A, B and C. Groups again consisted of seven volunteers with two receiving placebo. After two of three scheduled vaccinations, 3/7 volunteers in the 2 mg dose group had HIV-specific CD8 cells, and two of these three volunteers also had CD4

T-cell responses.

An important question for vaccines with multiple antigens is whether responses are reduced compared to single antigen vaccines. Another is whether using 3 different Env antigens induces a broader response than a single Env protein. Graham said that macaque data showed no reduction in Gag responses when Env antigens are incorporated, and that data on the breadth of Env responses may be available by the September AIDS Vaccine 2003 meeting in New York.

High-dose canarypox study

Juliana McElrath (University of Washington, Seattle) presented data from HVTN 039, which tested a well-studied canarypox-based vaccine (ALVAC vCP1452) given at the most commonly-used dose and at one higher dose (107.26 and 108 TCID50). McElrath said that the immune responses to both doses, assessed using the Elispot assay to detect IFN-gamma-producing cells, were indistinguishable but low (between 40 and 70 spot-forming cells per million white blood cells), leading McElrath to conclude that use of the higher dose was not warranted. Some cross clade responses were seen to Gag, with three of the seven people who responded to subtype B Gag also responding to subtype C.

Preclinical studies in monkeys Bob Johnston from the University of North Carolina gave an update on the development of HIV vaccines using vectors of Venezuelan Equine Encephalitis (VEE) virus. One potential advantage of this approach is that VEE targets immature dendritic cells, which play a key role in presenting antigens to the immune system. Together with the biotech company AlphaVax, Johnston’s team has engineered a replication-incompetent vaccine vector encoding HIV clade C gag, which recently received approval from South Africa’s regulatory authorities for a Phase I study. The trial will take place in Johannesburg and four US sites of the US HIV Vaccine Trials Network (see Vaccine Briefs, p. 20).

Johnston described results showing partial protection against disease in two studies of vaccinated macaques. In one, 2/6 macaques vaccinated with VEE plus env (gp160 and gp140) completely controlled an intrarectal challenge with SIVE660, one of the pathogenic SIV strains; the remaining four animals showed partial control. In contrast, only 1/6 unvaccinated monkeys partially controlled the challenge, while the rest showed no control. Both CD8 and neutralizing antibody responses were associated with protection.

In the second study, eight macaques were vaccinated with VEE carrying env (gp160), gag and pol, and then challenged intravenously with SHIV89.6P. Vaccinated monkeys showed some protection from disease compared to controls, with peak viral load lowered about 10-fold and a 100-fold reduction in setpoint. The correlates of protection were less clear in this study, since pre-challenge CD8 T-cell responses were only detectable in 2/8 monkeys and no neutralizing antibodies were found—although Johnston said this may reflect differences between the different challenge viruses used in these studies.

Adeno-Associated Virus (AAV)

Phil Johnson (Children’s Research Institute, Cincinnati) reviewed some new data on an HIV vaccine using vectors of adeno-associated virus (AAV), focusing on recent safety studies. A version of this vaccine containing gag from clade C is slated for human trials starting in late 2003.

Johnson presented data from animal toxicology and biodistribution studies, which are required for regulatory approval of clinical trials. No local or systemic reactions to the vaccine were seen in a 150- rabbit toxicology study, and analysis showed only trace amounts of vaccine present in about 15% of the tissues sampled at 180 days—mostly in the muscle at the injection site or in other highly perfused tissues. Importantly, no trace of vaccine was found in the gonads, indicating that vector is not transmitted into the genome of offspring. Overall, this safety profile is similar to that seen with DNA vaccines. Two immunized macaques sacrificed at 5 months also showed very low persistence of vector.

These studies also confirmed earlier findings that the AAV vector does not integrate into cellular DNA (J.Virol 2003:77;3495). Integration of the unmodified AAV has been seen in cultured cells and initially raised safety concerns about this approach.

Viral escape: Can it generate “immune-resistant” strains?

David Watkins described a recent monkey study which looked at some possible consequences of viral escape in a vaccinated population.

In infected macaques and in some vaccination studies, SIV and SHIV undergo mutations which enable them to evade the CD8 responses that normally control viremia. Escape also occurs during HIV infection, raising the possibility that T-cell-based vaccines—which allow infection but prevent replication—may lead to the selection of mutated viral strains lacking crucial CD8 epitopes, and which might then spread through populations unchecked by immune control

To test this scenario, graduate student Tom Friedrich created a strain of SIVmac239 with mutations in each of the three most important epitopes (in the Gag, Tat and Nef proteins) targeted by CD8 T-cells in monkeys of a specific genetic background (MamuA*01 and B17). Four rhesus macaque monkeys with this background, and 2 without (so they target different epitopes) were then infected with the mutant escape strain, which the researchers called 3xSIV.

Watkins said that, as expected, the MamuA*01/B17- positive animals generally failed to control replication of the escaped strain and showed no CD8 responses to the mutated epitopes. But surprisingly, in both of the monkeys negative for MamuA*01 and B17—where there is no immune pressure against these epitopes—the Gag and Nef epitopes reverted back towards their original sequence, suggesting that the escape mutations reduce viral fitness. Watkins called this surprising finding a “saving grace” for CTL-based HIV vaccines, since it suggests that even when escape occurs, the escaped strains are unlikely to sweep through human populations, which are genetically diverse; rather, important CD8 epitopes will probably be regenerated, thanks to the higher fitness they confer.

Correlates of protection

Identifying the specific immune responses which control HIV infection has long been a central issue for HIV vaccines, since this knowledge would greatly simplify the tasks of designing and testing candidates. In Banff, several speakers reported on studies that tackled this problem from different angles.

Jeff Lifson (National Cancer Institute) gave an interesting talk describing examples of control of SIV infection, whether achieved through early antiretroviral treatment (ART), vaccination, or spontaneously, and the associated immune responses. One SIV-naïve macaque inoculated with the virulent SIVmac239 controlled viral replication following the initial peak to less than 20 copies/ml of plasma. CD8 responses seemed important in this control as these increased progressively over time and the viral load rebounded upon depletion of CD8+ cells. Of note, Elispot responses in this animal during control of SIVmac239 infection were lower than levels seen in other animals that were unable to control this same virus in other vaccine studies. Neutralizing antibodies were low through the period of initial control, but were present at high levels as the animal re-established viral control after the depletion of CD8+ lymphocyes. Conversely, a vaccinated animal controlling virus to <20 copies had detectable low-level Elispot responses, yet upon CD8 depletion, no rebound of virus was seen. A greater role may have been played in this animal by antibodies, which were detectable all the way through.

In a different setting, a macaque which controlled SIVmacE660 infection following limited early ART also had modest levels of Elispot responses, but was able to control a heterologous challenge with SIVmac239. Lifson concluded that qualitative characteristics rather than the level of the immune response were likely important in these animals, since low-level responses were associated with control of viremia. Early control probably also limited virus diversification, thereby facilitating sustained immune control.

In a talk on CTL-based vaccines, Andrew McMichael (Oxford University) emphasized the importance of looking at memory cells in vaccinated individuals, rather than at only the more short-term post-vaccination responses. He pointed out that CD8 memory responses in EBV-infected people do not strictly reflect the responses seen shortly after infection due to a change in immunodominance: whilst directed at a similar range of epitopes, cells that showed smaller responses after the initial infection come to dominate the memory response, whereas those that were initially dominant decrease proportionally (J Immunol 2002;168:3309). Since vaccinees are most likely to be exposed to virus when their immune responses are in the memory phase, he suggested it would be important to study these responses. McMichael described several different assays that can be used to measure memory responses, including Elispot and ICS to detect cytokines and cell markers, and killing assays using cultured cells. He said the ICS tetramer assay and ICS cell proliferation assay (the CFSE assay) would probably be the best ways to quantitate long-term memory.

Measurement of immune responses

Current measures used by researchers to determine the level of HIV specific CD4 and CD8 T-cell responses focus mainly on production of IFN-gamma, a cytokine with antiviral activity that is relatively easy to detect by Elispot or ICS assays. However, some studies presented at Keystone add to the growing evidence indicating that this measurement is not sufficient to identify all responding cells.

For example, data presented by Michael Betts (VRC, Bethesda, Maryland) showed that not all cells which kill virally-infected cells produce IFN-gamma. To do this, he used flow cytometry to examine CD8 T-cells which degranulate (by measuring the CD8 cell surface degranulation markers CD107a and b), the main mechanism by which they kill virally infected cells, in addition to IFN-gamma. These results suggest that both the quality and quantity of HIV-specific CD8 T-cell responses determined by IFN-gamma measurement may be underestimated.

The magnitude of CD4 T-cell immune responses also appears to be underestimated by the measurement of IFN-gamma alone. Stephen De la Rosa (VRC) illustrated this point with data on 9 individuals who received booster shots of either tetanus toxoid (TT) or Hepatitis B vaccine. For the HepB-specific CD4 T-cells, and to a lesser extent the TT-specific cells, IL-2 and MIP-1 beta production were the predominant cytokine responses, with IFN-gamma-producing cells accounting for as little as 25% of the responding CD4 cells.

Other data suggested that the methods used to stimulate HIV-specific cells so they can be measured are also part of the problem. Normally, researchers stimulate cells with consensus or reference virus strains (or peptides) that are representative of circulating viruses. However, a poster by Marylyn Addo and Marcus Altfeld from Bruce Walker’s group (Massachusetts General Hospital, Boston) suggests that this method may underestimate response levels, especially to the more variable HIV proteins. 

The researchers analyzed CD8 responses in 6 acutely infected patients using sequences based on consensus virus or patients own (autologous) virus to the p24, Tat and Vpr proteins. The breadth and sum of CD8 responses to Tat and Vpr were underestimated when consensus sequence was used rather than autologous sequence, with 8/24 peptides not being recognized with consensus sequence and an average magnitude of 179 SFC versus 514 SFC. In contrast, a slight underestimation of CD8 responses to p24 were seen when using consensus sequence compared to autologous sequences, with 4/18 peptide responses missed using consensus and a magnitude of 328 SFC versus 464 SFC. The responses to consensus and autologous sequences differed more to Tat and Vpr than p24 because the virus was more different in these regions—approximately 10% in Tat and Vpr compared to only 1.8% in p24. 

Altfeld said that underestimation of responses to the more variable HIV proteins may be a problem that plagues research if autologous sequence is not used for stimulation. (This was recently published in J Virol 2003:77;7330.) 

HIV transmission and envelope protein selection 

From the swarm of different virus strains present in infected individuals, only a small subset is transmitted sexually. This partly reflects the selection of viruses that use CCR5 as a co-receptor to enter cells, but other viral properties also appear to play a role. Since it is these newly-transmitted strains which vaccines must initially combat, a fuller characterization could be enormously useful for vaccine design. 

Cynthia Derdeyn (University of Alabama) gave a thought-provoking presentation on an analysis of Env proteins in eight transmission pairs, drawn from a serodiscordant couples cohort in Zambia. In addition to sequencing env genes from 8 newly infected people and their partners, the researchers also measured the sensitivity of transmitted virus to neutralization by antibodies present in the donor’s plasma. 

Derdeyn reported that the transmitted env sequences were all homogenous in the V1-V4 region, regardless of the complexity of viral strains present in the donor. Interestingly, they also showed a compact variable loop structure, suggesting that variants with larger V1-V2 and V4 loops were selected against during transmission, as were heavily glycosylated strains. Recipient strains also proved to be roughly seven times more sensitive to neutralization with donor plasma than the donor’s viral population at large, indicating that strains which escaped neutralization are not the source of transmitted virus. 

Further understanding of exactly what is transmitted should come from analysis of newly-infected individuals in the VaxGen trial, where over 300 full envelope genes have been sequenced.