AIDS Vaccine 04

International conference in Lausanne provided an update on recent scientific progress

By Richard Jefferys*

Overlooking the glittering waters of Lake Geneva and surrounded by the stunning snow-capped peaks of the Alps, the third international conference on AIDS vaccines took place from 30 August to 2 September, 2004 in Lausanne, Switzerland. The mountainous terrain offered an appropriate metaphor for the long climb that lies ahead for researchers attempting to develop an effective AIDS vaccine. In keeping with this theme, many discussions at the conference revolved around the idea that—like mountaineers scaling a precarious rock face—scientists may need to link more closely if they are to reach their lofty goal.

In terms of AIDS vaccine science, the meeting featured several reports highlighting incremental but significant progress in developing strategies for inducing broadly neutralizing antibodies against HIV, suggesting that this perennial obstacle in vaccine development might eventually be overcome. Less encouragingly, the utility of modified vaccinia Ankara (MVA) vectors and current DNA vaccine constructs came into question with the unveiling of disappointing immunogenicity data from trials involving a DNA prime/MVA boost approach developed by scientists from Oxford University in collaboration with IAVI and the Kenyan AIDS Vaccine Initiative (KAVI). Given that many AIDS vaccine candidates are currently based on these platforms, this unwelcome news added to the general sense that the vaccine pipeline might not be as robust as it had once appeared and efforts to identify additional novel vector strategies need to be stepped up. A related issue is the problem of circumventing pre-existing immunity to certain vectors, such as adenovirus, which also received considerable attention in Lausanne. Among scientists focused on human immunology, the potential limitations of the most commonly used assay for evaluating vaccine-induced T cell responses—the interferon (IFN)-g ELISPOT—were widely discussed, and a broad consensus reached that assays evaluating additional T cell functions should be standardized and validated for use in future vaccine trials.

Progress in neutralizing antibody research

Still the most significant problem facing AIDS vaccine researchers is HIV's notorious ability to avoid neutralization by antibodies, a critical component of the protection afforded by licensed vaccines for other diseases. However, a small number of monoclonal anti¬bodies (MAbs) have been isolated over the years that do show some neutralization of primary HIV isolates from multiple different clades (see J. Virol. 78, 13232, 2004 for a recent report on the activity of these MAbs). In Lausanne, several presentations focused on efforts to better understand the mechanisms by which the most effective MAbs work, in order to develop vaccines that can induce neutralizing antibodies capable of preventing HIV infection.

Figure 1. The crystal structure of a broadly neutralizing anti-HIV antibody in complex with its epitope on Env. The monoclonal antibody, 2F5 (blue), is shown in complex with its epitope (red) in the membrane proximal region of the HIV-1 envelope ectodomain. The indole side chain (red spheres) of the final residue of the epitope is coplanar with a hyrdrophobic region (blue spheres) on 2F5. This region probably interacts with the viral membrane (depicted as opaque plane at bottom of picture) to substantially enhance antibody affinity for HIV.Image courtesy of P Kwong and R Wyatt and the American Society for Microbiology. [Large image].

A considerable body of research has focused on a neutralizing antibody epitope called ELDKWA in gp41, the transmembrane component of the Env glycoprotein. This epitope is targeted by one of the more broadly neutralizing MAbs identified to date, christened 2F5. However, attempts to use the ELDKWA epitope to elicit similar neutralizing antibodies have so far met with failure, perhaps because the epitope needs to be presented in its correct three-dimensional conformation in which it occurs naturally as part of gp41 (see below, Richard Wyatt's presentation). A presentation by Gail Ferstandig Arnold from Rutgers University described a novel approach to try to present the epitope in its correct conformation by randomizing the amino acids surrounding it. Arnold and colleagues have inserted the ELDKWA epitope into rhinovirus as a vaccine vector after introducing amino acid substitutions on either side where the epitope joins the rhinovirus sequence. These amino acid substitutions were introduced in the hope of hitting on a combination that forms the right framework around the ELDKWA epitope so that it is presented in its authentic conformation. A combinatorial library of variants was engineered and then screened using an elegant competitive immunoselection strategy. This involved immobilizing MAb 2F5 in an assay dish, adding the naked ELDKWA peptide to bind to the antibody, and then washing the recombinant rhinovirus vectors over it; the rhinovirus recombinants presenting the ELDKWA epitope in the most natural conformation would competitively bind to MAb 2F5, displacing the naked ELDKWA peptide, and these bound recombinants could then be collected and enriched by tissue culture passage.

Arnold selected 21 ELDKWA-encoding rhinovirus recombinants for immunogenicity studies in guinea pigs. Sera from animals immunized with either the rhinovirus recombinants alone or the recombinant plus a booster using ELDKWA-based peptides were tested in Virologic's luciferase-based HIV neutralization assay. Neutralizing activity against some primary HIV isolates was demonstrated, including viruses from clades A, B, D, and E (now known as CRF01_AE), the first report of antisera elicited by any ELDKWA-based immunogen that can neutralize HIV primary isolates in vitro.

Arnold's research team are now planning immunogenicity studies in non-human primates and efforts are underway to determine the three-dimensional structure of the most promising constructs. Arnold suggested that rhinovirus, which causes the common cold, may have a number of advantages as a vector, including its mild pathogenicity, serological diversity (there are >100 serotypes), robust immunogenicity (both systemically and mucosally) and the well-characterized nature of its structure and immunogenic sites. Arnold also noted that rhinovirus acts as a multivalent carrier for encoded epitopes, expressing around 60 copies per virion, and, because rhinovirus cannot grow in guinea pigs, Arnold is optimistic that the approach will be more immunogenic in a permissive species.

During the conference late-breaker session, Richard Wyatt from the Vaccine Research Center (VRC) at the US National Institutes of Health (NIH) debuted the three dimensional structure of MAb 2F5 (see Figure 1). Wyatt and colleagues, including Peter Kwong, have succeeded in producing a three-dimensional crystal structure of 2F5 in complex with its complete epitope in the membrane-proximal region of gp41 (which includes the ELDKWA sequence). This structure has revealed that 2F5 recognizes its epitope in a fully extended conformation rather than the helical structure that it adopts in the fusion-competent conformation (which, in its entirety, is referred to as the 6helix bundle), suggesting that the mode of action of 2F5 might be to bind to the pre-fusogenic state of gp41, "hang on to" this extended conformation and thereby interfere with the fusion process. The structural analysis also revealed that the 2F5-bound surface of the epitope is charged and that the non-bound surface is occluded, possibly by the viral membrane or some other hydrophobic elements of the Env spike. This interpretation is consistent with binding data that was presented demonstrating that 2F5 binds to its epitope with a relatively higher affinity in the presence of lipid. The group hopes that this structural and biochemical analysis will provide important new clues to enable them to present the gp41 epitope in a stabilized conformation, occlude the distal surface, and provide lipid to generate an immunogen that will better elicit 2F5-like neutralizing antibodies.

In a summary talk on envelope-based antigen design, VRC's director Gary Nabel cited the ongoing work of Wyatt and Kwong and also made a number of general points about this area of research. Firstly, Nabel stressed that investigators need to compare liganded and unliganded conformations of the envelope antigens under study in order to look for differences in their ability to induce neutralizing antibodies. Nabel also noted that the conformation of potential immunogens can be stabilized (e.g., by disulfide linkings) and argued that researchers should confirm that their candidate envelope constructs retain their ability to bind to CD4. Another potential approach to enhancing envelope immunogenicity mentioned by Nabel involves testing the effects of mutations in the stem of the V3 loop, part of gp120 involved in CD4 binding.

Nabel closed his talk by revealing that VRC has constructed chimeric immunogens based on transposing the V3 loop from clade C HIV into a backbone based on a combination of clade A and B viruses. Preliminary results suggest that this construct may have an improved ability to induce neutralizing antibodies.

Vaccine vectors: Problems and promise

Immunogenicity data was presented from several Phase I and Phase I/II trials of a DNA prime/MVA boost approach developed by Andrew McMichael and Tom Hanke from Oxford University. Development of this vaccine was supported by IAVI and the trials were conducted in the UK, Kenya (in collaboration with KAVI) and Uganda. The results proved disappointing; only 10-25% of volunteers developed cytotoxic T-lymphocyte responses to the HIV Gag protein contained in the vaccine (which encodes the clade A gag gene fused to a string of 25 partially overlapping CTL epitopes from gag, pol, nef and env; see Vaccine Briefs). As a consequence of these data, IAVI announced that ongoing studies will be completed and additional immune responses evaluated but "unless there are new immune response data that are dramatically different, IAVI will not develop the candidates further, and will focus on its other research and development projects."

In another presentation McMichael hypothesized that the problem may have related to the poor immunogenicity of the DNA construct, and that MVA appears to be poor at priming T-cell responses but could still have a role as a boost. In support of his argument he offered a glimpse at unpublished data from a trial using the MVA construct as a therapeutic vaccine in individuals on HAART; in this setting detectable CTL responses did appear to be induced in a higher percentage of participants, which McMichael suggested was likely the result of boosting pre-existing responses that were primed by natural infection. An issue often raised about MVA as a vector is the large size of the genome and the possibility that immune responses may be directed more toward the multitude of MVA proteins rather than the HIV antigens it encodes. A study presented recently by Bavarian Nordic, who make an MVA vector encoding the Nef protein, found that while 8/8 HIV-negative trial participants developed T cell responses to MVA, only 3 showed evidence of a response to Nef (and these responses were borderline, ranging from 9-49 spot-forming cells [SFC] in IFN-g ELISPOT). Analysis of MVA-specific T cell responses in the IAVI trials is ongoing.

Although it remains unclear to what extent the poor immunogenicity of the Oxford MVA vector is specific to the particular construct rather than the platform as a whole, researchers are looking at ways to enhance the utility and immunogenicity of MVA vectors generally. David Garber and colleagues from the Emory Vaccine Center at Emory University, Atlanta have constructed an MVA vector that has the uracil-DNA glycosylase gene deleted; this deletion prevents progression from the early to late phases of the poxvirus replication cycle and thereby restricts the expression of antigenically complex MVA proteins that may interfere with the generation of immune responses against the inserted HIV (or other) antigens. The deletion is also associated with increased apoptosis of infected cells, potentially increasing priming of T cell responses via uptake of the apoptotic cells by antigen-presenting cells.

Garber described macaque experiments in which groups of four animals each were immunized with the modified MVA encoding the HIV Gag protein (MVADudg-gag), the parental MVA or a vector based on measles virus (both also encoding Gag). Six weeks after a single immunization Gag-specific T-cell responses were analyzed using IFN-g ELISPOT. The three macaques immunized with MVADudg-gag displayed an average of 376 SFC/million PBMC (peripheral blood mononuclear cells) compared to 26 SFC in animals immunized with the parental MVA vector (a statistically significant difference, p=0.034). While the results are preliminary, Garber suggested that MVADudg is likely to be a better priming vaccine than currently used MVA vectors. Confirmatory studies in a larger group of macaques are ongoing and Garber also plans to compare the efficacy of the modified and parental MVA constructs as a boost subsequent to vaccination with a measles virus vector.

Two presentations reported on the potential of recombinant measles virus (rMeV) as an AIDS vaccine vector. Hussein Naim from the University of Zurich cited a number of potential advantages to this platform, including that it is easy to manufacture and in widespread use as a vaccine already, long-lived protective immunity is induced after single dose, MeV replicates in the same compartments as HIV (macrophages, dendritic cells and T cells), and a relatively large 5kb of DNA can be inserted.

But a major stumbling block, in all but infants, is pre-existing immunity to measles virus. However, previous studies in macaques (J. Virol. 78, 146, 2004) and Naim's work in mice suggest that rMeV vectors may remain immunogenic despite the presence of high levels of MeV-specific antibodies. Naim reported the construction of an rMeV vector that can encode multiple genes at three insert sites, and pointed out that expression levels of the encoded gene are site-dependent. In mice, rMeV vectors encoding either Env alone or Env plus Gag elicited both antibody and T cell responses. Naim noted that responses to Env were enhanced in mice that received the constructs containing both Env and Gag and he hypothesized that this might be due to the formation of virus-like particles. In concluding his presentation, Naim emphasized his feeling that rMeV is a promising platform for developing an inexpensive AIDS vaccine that could be utilized by both adults and children.

Frederic Tangy from the Institut Pasteur followed Naim's talk with a description of his group's efforts to develop a MeV-based AIDS vaccine. He stressed that low cost, the fact that it could easily be produced at large-scale in many countries and the existence of distribution systems for extant MeV vaccines are all important parts of the rationale for pursuing this approach. Tangy's construct is based on the Schwarz strain of MeV that is included in licensed vaccines, including MMR. For macaque challenge experiments, a MeV vector encoding the Gag, Env, Tat, and Nef proteins from SHIV89.6P was constructed. The construct induced T-cell and antibody responses to both MeV and HIV proteins in macaques and led to control of a SHIV89.6P viral challenge in 6/8 animals. While the relevance of this particular challenge system to human HIV infection has been a subject of debate, Tangy concluded that as a proof-of-concept it supports the further development of MeV vaccine vectors. However, in questions after the presentation Stanley Plotkin (consultant to Aventis-Pasteur) also raised the important point of persistence of MeV, a phenomenon that, along with the widespread pre-existing immunity, might well dampen enthusiasm towards this as a potential vector.

Another vector widely discussed at the meeting was adenovirus. These viruses naturally cause severe colds in humans and circulate in several distinct serotypes, the most common being type 5 (Ad5), which Merck has developed into a vector for their lead AIDS vaccine candidate. Merck's initial studies involved an Ad5 vector encoding Gag alone (Ad5-gag), but at the conference Robin Isaacs (Director of Clinical Vaccine Research at Merck) presented the first preliminary data on the new "trivalent" construct that encodes Gag, Pol and Nef (see Table 1). Based on these results, Merck has decided that the trivalent Ad5 vector will be tested in the Phase II trial.

Table 1. Interim results post-prime (Week 8) with trivalent MRK-Ad5 in participants with anti-Ad5 titers <1:200
VP/dose	Gag	Pol	Nef	≥ 2 antigens
3 x 109	48% (10/21) 171 SFC	38% (8/21) 270 SFC	48% (10/21) 103 SFC	48% (10/21)
3 x 109	59% (13/22) 166 SFC	45% (10/22) 301 SFC	50% (11/22) 134 SFC	55% (12/22)
1 x 1011	71% (12/17) 289 SFC	53% (9/17) 406 SFC	47% (8/17) 261 SFC	53% (9/17)
Vaccinations (VP = viral particles) at weeks 0, 4, (priming series) & 26 (boost). Responder defined as ELISPOT≥55 spot-forming cells per million PBMC and ≥4-fold over background; summaries are based on 15-mer peptides. Results shown as the percentage of T-cell responders (absolute numbers), and the geometric mean spot-forming cells (SFC) in IFN-g ELISPOT. These data were presented at AIDS Vaccine 04 and have not yet been peer-review published.

 

A major problem facing Ad5 as a vaccine vector is pre-existing anti-Ad5 immunity. Previous studies in the US have indicated that about a third of the population has anti-Ad5 antibody titers greater than 1:200, levels which have been shown to compromise substantially the immunogenicity of recombinant Ad5 vector-based vaccines for HIV. In Lausanne, Dan Barouch and colleagues from Beth Israel Deaconess Medical Center and Harvard Medical School in collaboration with Crucell Holland BV reported that in South Africa, Zambia, and Botswana, more than 90% of tested individuals had pre-existing anti-Ad5 antibodies, with median titers over 1:1000. In comparison, pre-existing antibody responses to the less common adenovirus serotypes Ad11 and Ad35 were seen in only 20-35% of individuals, in whom titers were generally less than 1:100, suggesting that these serotypes may be worth pursuing as vaccine vectors for developing countries.

Barouch also conducted immunogenicity studies in mice and showed that Ad35 and Ad11 encoding SIV Gag elicited 10-fold higher Gag-specific T cell responses than did the Ad5 vector in animals with anti-Ad5 antibodies at levels comparable to humans. He also evaluated several prime-boost vaccine regimens (including Ad5/Ad5, Ad35/Ad5, Ad11/Ad5, and Ad35/Ad11) and demonstrated that the Ad35/Ad11 regimen was more immunogenic than any combination that included Ad5 in the presence of pre-existing anti-Ad5 antibodies.

Merck has explored a different strategy for overcoming the problem of pre-existing immunity to Ad5. In macaque experiments, priming with Ad5-gag followed by boosting with Aventis-Pasteur's canarypox vector (ALVAC) also encoding Gag led to significantly improved immunogenicity. Unfortunately, Robin Isaacs reported at the conference that these macaque results were not mirrored in a Phase I trial in humans: individuals primed with Ad5 and boosted with ALVAC displayed levels of HIV-specific T cells that were statistically indistinguishable from those seen in participants primed with Ad5 and boosted with Ad5.

Adeno-associated virus

Adeno-Associated Virus (AAV) is a novel vector approach developed by Phil Johnson from the Children's Research Institute in Columbus, Ohio, with support originally from NIH and more recently IAVI. AAV is a parvovirus that is dependent on adenovirus for replication; the vector has been further modified so that it is completely replication-incompetent. An attractive feature of AAV is its prolonged persistence as an episome within cells, which may facilitate the induction of robust and long-lived immune responses to the immunogen after just a single dose. In Lausanne, Alan Schultz from IAVI offered the first look at immunogenicity data from a study in which rhesus macaques were vaccinated with an AAV serotype 2 vector encoding SIV Gag. Among 24 animals receiving the construct there was a clear dose-dependent T cell response to the Gag protein, reaching 500 SFC/million PBMC (by IFN-g ELISPOT) in macaques that received the highest dose. These robust Gag-specific T-cell responses were also accompanied by high and persistent titers of Gag-specific antibodies averaging 1:1600 (and as high as 1:8000 mean peak titer). A Phase I safety study of the AAV vector in humans is ongoing.

Targeting toll-like receptors

A current hot topic in immunology revolves around toll-like receptors (TLRs). This family of molecules (eleven have been reported so far, simply numbered TLR1 through TLR11) was discovered relatively recently and they appear to play a critical role in allowing immune system cells to sense "danger signals" within potential pathogens and then initiate an immune response. Basic research on TLRs suggests that they may have the potential to be targeted with adjuvant in order to enhance the immune response to vaccines. Bob Seder from the NIH presented results from a macaque study that tested the immunogenicity of a single vaccination with the HIV Gag protein emulsified in the adjuvant Montanide ISA 51 (an oil-based compound with similarities to the standard Freund's adjuvant) in the presence or absence of synthetic TLR7/8 agonists, TLR8 agonists, or a TLR9 ligand, CpG oligodeoxynucleotide (ODN). Two weeks post-immunization all animals that received the TLR agonists had significantly greater Gag-specific T cell responses (assessed by IFN-g ELISPOT) than the macaques that did not (five animals per group). Six weeks post-immunization, levels of both IFN-g- and interleukin (IL)-2¬producing HIV-specific T cells remained significantly higher in the animals immunized with CpG ODN or TLR7/8 agonists compared to those not given a TLR ligand. TLR8 is present on the myeloid dendritic cell subset while TLRs 7 and 9 are present on plasmacytoid dendritic cells (PDCs), and Seder suggested targeting PDCs may be important and that further studies of TLR-based adjuvant modalities are warranted.

Cellular immunology

Several presentations reinforced a recent shift toward evaluating additional markers of antigen-specific T-cell function beyond just IFN-g. At the closing ceremony, Clive Gray (National Institute for Communicable Diseases, South Africa) reviewed these data and noted the consensus among researchers that intracellular cytokine staining (ICS) for IL-2 production should be included when evaluating T cell-based vaccines; efforts to standardize ICS assays for this purpose are underway. An example of the potential importance of IL-2-based ICS was provided by Helen Horton from the University of Washington. Horton showed that, in a Phase I study of GlaxoSmithKline's Nef-Tat-gp120 protein vaccine, the majority of vaccine-induced CD4+ T cells produced IL-2, not IFN-g, and would therefore have been missed with the standard ELISPOT assay. Clive Gray also cited additional techniques that may contribute to a more comprehensive evaluation of vaccine-induced T-cell responses, such as flow cytometry-based monitoring of antigen-specific proliferation utilizing staining with the dye CFSE (every time a T cell divides half the CFSE dye is lost, so cells with proliferative capacity can be quantified based on their loss of CFSE) and multi-parameter tests that assess the ability of T cells to produce multiple cytokines and chemokines simultaneously.

Conclusion

The final presentations in Lausanne reflected upon the impact of the European and US political landscapes in shaping AIDS vaccine research efforts. Michel Kazatchkine, head of the French Agence Nationale de Recherches sur le Sida (ANRS) discussed the current European Union situation, noting that while representatives of the EU are generally morally supportive of AIDS vaccine research, this is not mirrored by appropriate fiscal support. Currently European scientists lack any kind of coordinated lobbying mechanism to advocate for funding at the EU level and instead there is a patchwork of efforts highly dependent on the munificence of individual governments, coordinated wherever possible by the relatively youthful European Vaccine Effort against HIV/AIDS (EuroVac). Kazatchkine advocated strongly that this problem needs to be addressed if the EU is to develop a more credible, coordinated and well-financed AIDS vaccine research program.

Bart Haynes, chair of the AIDS Vaccine Research Working Group which advises the US government's Division of AIDS (DAIDS) on vaccine-related issues, noted that the situation in the US is somewhat better due to a significantly higher level of research funding support both from government and independent entities like the Bill & Melinda Gates Foundation. Haynes cited the Partnership for AIDS Vaccine Evaluation (PAVE), a DAIDS-sponsored initiative chaired by Peggy Johnston that is making progress in coordinating research across the differing government bodies (including the National Institutes of Health, the NIH Vaccine Research Center, DAIDS, HIV Vaccine Trials Network, Centers for Disease Control and Prevention, Walter Reed Army Institute of Research, Henry M. Jackson Foundation, Food & Drug Administration, Adult & Pediatric AIDS Clinical Trials Groups) and also external groups such as IAVI, Merck and the Canadian Network for Vaccines & Immunotherapeutics (CANVAC).

PAVE's ongoing work includes developing the laboratory support necessary for evaluating vaccine immunogenicity in efficacy trials (including standardization of immunological assays), building capacity for clinical trials by standardizing site development and investigator training tools, and harmonizing research protocols to better facilitate cross-trial comparisons and data sharing. Haynes articulated his hope that PAVE will enhance the ability of the US to participate productively in the global collaborative effort being proposed under the Global HIV Vaccine Enterprise, with the ultimate outcome being the speedier development of an effective AIDS vaccine.

In the final period of questions and comments from conference attendees, Larry Corey (head of the HIV Vaccine Trials Network) noted that the type of coordinated research being advocated under these various mechanisms will require something of a cultural shift among scientists used to a more independent approach—in Corey's pithy words: "everyone likes to collaborate but no one likes to be coordinated." Peggy Johnston, Director of Vaccine Research at DAIDS, emphasized the need to focus on training new investigators, stating: "this [HIV/AIDS] may well be a problem that will be handed to the next generation." Only time will tell whether the new commitment to global coordination discussed in Lausanne can prevent Johnston's dire scenario from coming to pass.

------------
*Richard Jefferys is Basic Science Project Director at the Treatment Action Group, a New York-based organization advocating for HIV research.