To develop DNA vaccine candidates for HIV, we developed a series of synthetic genes designed to express HIV Env mutants in human cells. In the absence of HIV regulatory proteins, these codon-altered envelope protein genes were readily expressed in human cells. Like other DNA vaccines, immunization with these vectors elicited strong CTL responses in mice, and antibody responses were not robust in mice immunized with wild-type Env expression vectors. Mutations in highly conserved N-linked glycosylation sites did not significantly alter humoral or cellular immune response to native Env. In contrast, a mutant Env with deletions in the cleavage site, fusion domain, and a region between the heptad repeats elicited a more-potent humoral immune response and retained its ability to stimulate Env-specific CTL.
The immune response to HIV infection in long-term nonprogressors (9
) and HIV-exposed sex workers suggests that specific viral immunity may limit infection and the symptoms of disease, but no single characteristic correlates with protective immunity. Depletion of CTLs in chronically infected macaques enhances viremia (19
). In humans, higher CTL responses correlate with lower viral load and stabilization of clinical symptoms (30
). In animal models, passive transfer of neutralizing antibodies can also contribute to protection against virus challenge (8
). Neutralizing antibody responses can also be developed in HIV-infected individuals (8
) and are associated with lower viral loads in long-term nonprogressors (25
). Though broadly neutralizing antibody responses are uncommon, they are directed largely against the Env protein of the virus (40
). In early human vaccine trials, gp120 protein immunogens have yielded disappointing results: vaccine-induced antibodies have not been broadly neutralizing and have sometimes enhanced infection in vitro (5
). Monomeric gp120 loses oligomer-dependent epitopes and does not include sequences in the ectodomain of the gp41 that become exposed during virus entry (6
). It is assumed that broadly neutralizing antibodies must bind well to native gp120/gp41 on the surface of the virus (8
Recent reports suggest that gp160 forms trimers in vivo and the domain required for trimer formation resides in the ectodomain of the gp41 (55
). Such trimeric forms of HIV envelope protein are likely to present different epitopes to the immune system compared to monomeric gp120. In addition to the linear epitopes in the envelope, this trimeric structure is likely to display conformational epitopes important for B-cell triggering of a relevant antibody response. In this regard, gp140ΔCFI, which induced the greatest antibody response, is released in a soluble trimeric form (Fig. and ). This finding was confirmed both by sedimentation gradient and gel filtration analysis (Fig. ). Biochemical purification is likely to yield dimeric and monomeric forms, and another potential advantage of gene-based vaccines is the ability to make physiologically relevant complexes in vivo. In contrast to gp140ΔCFI, wild-type Env did not elicit high-titer antibody responses. The toxicity of Env in mammalian cells has been seen previously and could limit both the amount and duration of envelope protein expression in vivo, thus affecting immunogenicity. The envelope is also heavily glycosylated, and removal of partial or complete gp120 glycosylation sites has resulted in higher titers of strain-specific neutralizing antibody responses to mutant simian immunodeficiency viruses in monkeys, presumably through enhanced exposure of critical epitopes (4
). Though it seemed reasonable that deglycosylation would reveal epitopes otherwise masked in the native protein, we did not observe enhanced immune reactivity by DNA vaccination using different glycosylation site mutants, both in gp160 and gp150. This difference with the previous study is likely due to the fact that DNA vaccination rather than viral infection was utilized for immunization. Though glycosylation mutants are unlikely to prove helpful with the former method of immunization, it remains possible that modification of glycosylation sites may be effective with other vectors or adjuvants.
HIV-1 Env is proteolytically cleaved by a cellular proprotein convertase into gp120 and gp41. The gp41 subunit is composed of cytoplasmic, transmembrane, and ectodomain segments. The role of the ectodomain of the envelope in membrane fusion, particularly its hydrophobic glycine-rich fusion peptide, is well established. Two regions with heptad coiled-coil repeats in the ectodomain of gp41 are involved in viral fusion (13
). Upon fusion, these two alpha-helices, connected via a disulfide-stabilized loop (14
), presumably undergo a transient conformational change to a fusion active state. These changes allow the formation of a six-member helical hairpin intermediate structure that presumably exposes the fusion peptide at the NH2
terminus of gp41, allowing fusion to the target cell membrane (2
). The ΔCFI mutation was intended to eliminate cleavage of gp140, remove the unstable hydrophobic region, and stabilize oligomer formation. Though detailed structural data are not yet available on this protein, the immunologic and biochemical analyses indicated that these mutations retain native antigenic determinants, defined by known monoclonal antibodies, and oligomeric properties, similar to native viral envelope (Fig. ). The gp140ΔCFI mutant also elicits both humoral and cellular immune responses, probably by virtue of protein stabilization and secretion. For example, the neutralizing epitope (ELDKWAS) in the ectodomain of gp41 (31
) is present in the series of deletions and truncations of the envelope,
and gp140ΔCFI is reactive with the 2F5 neutralizing monoclonal antibody that binds to this epitope (Fig. ). It is also relevant that protein misfolding is unlikely to account for enhanced antibody responses, as several vaccine candidates that induced such responses showed expected cell surface expression and/or secretion (Fig. , 6, and 7 [gp140ΔCFI and gp145ΔCFI]), and several immunogens that were not appropriately exposed had no effect (data not shown).
Importantly, these immunogens also induced CTL responses to Env. It is evident that DNA immunization often elicits predominant humoral or cellular immune responses that are determined by features of the immunogen that are not understood. For example, we have found that responses to Env are primarily cellular while those to Nef display enhanced humoral immunity. Though gp128ΔCFI induced slightly more potent CTL activity, gp140ΔCFI was better able to elicit an antibody response, retained the 2F5 epitope, and was readily able to elicit such responses, both to peptide-pulsed cells and stably transduced target cells. Thus, the enhanced humoral immune response introduced by this vaccine candidate did not appear to diminish the CTL response. Taken together, these results suggest that gp140ΔCFI serves as an improved immunogen that can more effectively elicit a neutralizing antibody response against the envelope by DNA vaccination while preserving its ability to induce a CTL response.