Human Immunodeficiency Virus Type 1 Vaccine Development: Recent Advances in the Cytotoxic T-Lymphocyte Platform “Spotty Business”^▿

CTL-BASED VACCINES: STEPPED ON?

The discontinuation of the Merck recombinant adenovirus type 5 (rAd5) phase II proof-of-concept STEP study is undoubtedly a significant setback for the field of HIV vaccine development. In the study, a trivalent, three-vector rAd5 (Gag, Pol, Nef) vaccine was tested in two cohorts of 1,500 individuals at high risk for HIV infection. The vaccine was scheduled for three immunizations of 1.5 × 10¹⁰ viral genomes of each rAd5 vector encoding Gag, Pol, or Nef administered on day 1, month 1, and month 6. There are concerns that preexisting vector serology can mute the effectiveness of a CTL vaccine approach. Accordingly, the first cohort was made up of 1,500 individuals with low levels of preexisting antibodies against Ad5, with titers of ≤1:200, while the second cohort of 1,500 included participants with anti-Ad5 titers of greater than 1:200. However, the STEP study was prematurely stopped by the data- and safety-monitoring board because it was clear that even in the cohort with low titers of preexisting antibodies against Ad5, there was no evidence of efficacy. Table 1 summarizes the results from the cohort with titers of ≤1:200, comparing the rates of infection and viral loads of the vaccinated versus placebo groups. This was the best-response group in earlier clinical studies, eliciting CTL responses in over 60% of vaccinees. As shown in Table 1, in the subset of individuals that received at least two immunizations, there was a trend toward more HIV infections in the rAd5-vaccinated (19/672) group than in those in the placebo (11/691) group. The difference in viral loads 12 to 18 weeks postacquisition of the infection between the vaccinated and placebo groups (40,000 versus 37,000 copies per ml, respectively) were not statistically significant. Since this rAd5 trivalent vaccine did not prevent infection or decrease viral replication in vaccinated individuals, the implications for CTL-based vaccines remain unclear, except that future studies should test hypotheses and concepts that are not identical to the Merck approach. Furthermore, this study draws a line in the sand in that the following CTL-based approach should induce higher levels of CTL responses and induce a different immune phenotype. The inclusion of different antigens or prime-boost strategies in future studies with defined immune endpoints will be important. We will consider many of these in the following sections.

THE IMPORTANCE OF MUCOSAL RESPONSES

The hallmark of HIV-1 infection is the rapid depletion of CD4⁺ T cells from the peripheral blood during acute infection. More recently, the ablation of CD4⁺ CCR5⁺ T cells in the gut was noted as a dramatic consequence of both SIV (87, 99, 156) and HIV (27, 54, 102, 103) infections. The majority of HIV-1 infections occur across the mucosal barrier, and the mucosal tissues are an active site for viral replication of SIV (87, 155) and HIV (134, 169). The induction of both antigen-specific immunoglobulin A (IgA) antibodies and CTLs at these critical sites may provide a first line of defense with immediate effector activity (Fig. 1). Antigen-specific CTLs in the mucosa could potentially eliminate infected cells within the mucosa and subsequently prevent the systemic spread of the virus. It has been shown in murine (16, 17) and macaque (18) models that mucosal, not systemic, CTLs are necessary to protect against mucosally transmitted virus. Clearly, the generation of CTLs that can target mucosal sites may be an important facet of a successful HIV-1 vaccine. Overall, there is a great need for analyses of mucosal immune responses in preclinical and clinical trials, as the currently available data are minimal.

VIRAL VECTORS

The use of viral vectors as vaccines for HIV-1 has been reviewed elsewhere (135). Viral vectors have been generally studied for gene therapy applications but have also been adapted for vaccine development based upon the intrinsic immunity elicited by their infection of the host. Generally, recombinant viral vectors are constructed by removing gene sequences encoding crucial viral proteins, rendering the resulting vector replication defective, and using the space generated by the removal of these genes for vaccine antigens of interest. As expression systems, recombinant viral vectors can be engineered to transduce a wide variety of cell types, including antigen-presenting cells (APCs), and, subsequently, to express HIV-1-encoded antigens in the cytoplasm of the target cell. Intracellular expression allows for the efficient priming of CD8⁺ T cells by the presentation of antigenic peptides through major histocompatibility complex class I (MHC-I). In addition, transduced cells serve as antigen factories, which can leak antigens for MHC-II presentation. Through cross-presentation, nontransduced APCs can take up and present extracellular antigens through MHC-I. APCs can “bite” off pieces of transduced cells (58) as well as phagocytosing apoptotic cells that contain transduced antigen (4, 15, 68, 148). It is also possible that pieces of, or whole, antigenic proteins can be transferred from transduced cells to APCs (125) and presented to CTLs through cross-presentation. In addition, viral vectors can activate innate immune responses through the pathogen-associated molecular patterns of Toll-like receptors on APCs (reviewed in reference 170).

ONE STEP FORWARD: LESSONS LEARNED FROM NONHUMAN PRIMATE STUDIES

Studies comparing cellular responses induced by recombinant adenovirus, poxvirus, alphavirus, and plasmid DNA vaccines, all expressing SIV Gag in macaques, supported that rAd5 elicited the highest levels of antigen-specific CTLs, supporting its further clinical advance (136, 137). The results of studies with rAd5 in animal models highlight considerations for evaluation of candidate vaccines. The rAd5 vaccine never protected against infection: regardless of the monkey haplotype or challenge virus studied, all monkeys became infected following challenge, leading to the unrealistic expectation that vaccination with the rAd5 trivalent vaccine would protect against infection in humans. There was a clear hierarchy of control of viral replication by rAd5 vaccines in primate challenge models. In hindsight, the “easiest” challenge model, A01-positive (A01⁺) macaques challenged with simian-human immunodeficiency virus (SHIV) 89.6p, produced a clear 3-log reduction in viral loads compared to those of the control (137). Challenge with the more-virulent strain SIVmac239 resulted in a sustained 1.5-log decrease in the viral loads of rAd5-immunized A01⁺ monkeys (101). It is clear that Mamu A01⁺ rhesus macaques are useful models for observing antigen-specific responses by using tetramer staining, but it is questioned whether they are accurate models for predicting vaccine efficacy in humans. In preclinical studies with the rAd5 vaccine testing outbread macaques challenged with SHIV 89.6, a more-variable 2- to 3-log reduction was observed (89). However, when outbred macaques were challenged with SIVmac239 post-rAd5 immunization, no significant reduction in viral load was seen in vaccinated monkeys compared to that of the control (29). This hierarchy establishes that the use of outbred macaques and challenge with SIV should be considered the most-stringent vaccine model and be a focus for future studies evaluating HIV vaccine efficacy. However, based on a lack of evidence of protection competency in this model, it is clear that all primate studies are still at-risk studies without known correlates or clear immune targets and without a guarantee that protection in stringent SIV challenges guarantees similar protection in humans.

ONE STEP BACK: IMPLICATIONS FOR CTL-BASED VACCINES

Merck, the HIV Vaccines Trial Network (HVTN), and the Division of AIDS, NIH, should be commended for their valiant effort, creativity, organization, and management of this important efficacy trial. The bar for the STEP study was set high, as rAd5 was the most-promising CTL platform at the time, based upon clinical and preclinical data for rhesus macaques. In earlier clinical trials, significant response rates to all three HIV antigens (73% to Gag, 59% to Pol, and 68% to Nef) among individuals with low seropositivity to Ad5 who were vaccinated three times with the trivalent Ad5 were observed at week 30. The reported geomean of responders as measured by gamma interferon (IFN-γ) enzyme-linked immunospot (ELISPOT) assays were 205, 342, and 177 spot-forming units (SFU) to Gag, Pol, and Nef, respectively (37). Response rates were lower at the same timepoint for Ad5-seropositive individuals with high virus titers, with 69% responding to Gag (406 SFU), 38% to Pol (834 SFU), and 54% to Nef (428 SFU) (37). The majority of responders responded to only one eptitope in the Gag, Pol, or Nef peptide pools, with approximately 30% of responders recognizing three or more epitopes (37). Although the trial consisted primarily of men who have sex with men, it is interesting that one woman in the placebo group was infected. This critically important observation highlights the importance of the clinic as a teaching ground for HIV vaccine development. This observation has implications for HIV transmission between men who have sex with men versus heterosexual transmission that must be considered in future trial development. With this in mind, the analysis of the responder data from the STEP trial will be critical in determining a relationship among vaccine responders, the levels of the response, and the resulting incidence of infection. In addition, we anxiously await the data analysis from the high-titer vector-seropositive cohort as well as the Phambilli study to see if these results reflect an increased trend for infectious compared to low-titer Ad5-seropositive cohorts. The increased trend toward infection is not likely a general trend for all vaccines that elicit T cells. The Vaxgen protein trial, which mobilized B and T helper cells, did not show a similar trend, nor did the ALVAC trials, which mobilized weak CD8⁺ T cells. It should be noted that this trend is being debated at this time, and therefore its importance to vaccines under development remains speculative.

Unanswered questions include the following. Are the response levels (on average about 400⁺ SFU) observed in responders high enough? Are the vaccine targets the most relevant? Would polyfunctional immune responses or some other T-cell phenotype change vaccine efficacy? Do the responses need to be directed against a broader number of epitopes to be effective? What animal models are most important as predictors of human vaccine efficacy? What do the mucosal responses in Ad5-vaccinated people look like, and would such responses change the impact of the vaccine? Why are the acquisition rates different between men and women?

ADENO-ASSOCIATED VIRUS

Adeno-associated virus (AAV) is a helper-dependent parvovirus that has been well characterized as a gene therapy vector (reviewed in reference 144). The generation of the recombinant vector entails removing all viral genes except the sequences that encode the GC-rich inverted terminal repeats (ITRs). The entire genome is a mere 4.7 kb, which limits the size of the antigenic insert (between the ITRs) to roughly 3.8 kb, excluding the promoter and poly(A) tail. The removal of viral genes reduces the intrinsic immunogenicity of AAV; however, rAAV remains an attractive vaccine candidate due to the ability to produce it on a commercial scale (34), its outstanding clinical safety profile for nearly a decade (reviewed in reference 72), the ease with which it can be administered intramuscularly, and the high transduction efficiency of myofibers (reviewed in reference 72). Johnson et al. (72) demonstrated the efficacy of rAAV for delivery of SIV antigens (SIVsm/E660) in an intravenous challenge model. A portion of their study included two injections of three rAAV vectors encoding SIV genes at doses measured by DNase-resistant particles, which are equivalent to encapsulated vector genomes. The three rAAV vectors contained fragments of (i) SIV Gag, protease, and reverse transcriptase; (ii) SIV Rev and Env; and (iii) reverse transcriptase and integrase at 10¹³, 5 × 10¹², and 10¹³ DNase-resistant particles, respectively. The results showed robust cellular (tetramer-specific CTLs) and humoral (serum antibody) immune responses after a single injection. In fact, after a virulent intravenous challenge, immunized monkeys exhibited significant control of SIV replication, with viral loads almost 2 logs lower than those of the control group. The protection from challenge may be attributed to CTLs, as there were no detectable neutralizing antibodies against the SIV challenge stock observed in the sera of vaccinated animals. Interestingly, there was no increase in the observed immune response after the second immunization, suggesting the generation of antivector antibodies. In fact, the preexisting or vaccine vector-induced immunity to the capsid is still a major concern for this platform. For example, recombinant AAV type 2 (AAV2) is the most-studied serotype to date, and it has been shown that 27% of the population has neutralizing antibodies directed against AAV2 (162). Similar to strategies being employed with recombinant adenoviral vectors, less-prevalent rAAV serotypes found naturally in humans, including AAV1 and AAV5, are being examined to circumvent this issue. Alternating rAAV capsid serotypes from one immunization to the next is being explored. A phase II clinical trial organized by Targeted Genetics and IAVI utilizing rAAV encoding HIV Gag, Pol, and Δ reverse transcriptase (tgAAC09) is under way in Zambia and has been reported safe and well tolerated thus far. While initial clinical studies using rAAV appear to be overwhelmingly safe, recently a patient participating in a phase I/II gene therapy trial for the treatment of rheumatoid arthritis died after receiving a rAAV vector encoding the soluble tumor necrosis factor alpha receptor tgAAC94. The patient was also taking a tumor necrosis factor alpha blocker, Humira, which is immune suppressive. While there has been no report of a definitive determination of the cause of the patient's death, it appears likely that infection with a fungus, Histoplasma capsulatum, could be the cause (75). The role of the AAV treatment in this patient's outcome is not clear. Importantly, this study highlights the potential pitfalls of performing trials in immunosuppressed individuals whose responses are clearly different from those of immunocompetent people. The immunogenicity of rAAV vectors clearly needs improvment, and the use of molecular adjuvants and prime-boost strategies will undoubtedly be tested and will be discussed in future sections. The published data on the generation of mucosal immune responses by rAAV is limited, although it is an interesting vector due to its stability at a wide range of pH values and high temperatures.

ALPHAVIRUSES AND RHABDOVIRUSES

Alphaviruses are positive-stranded RNA viruses with a broad host range that cause a variety of mosquito-transmitted diseases, including Eastern equine encephalitis, Western equine encephalitis, and Venezuelan equine encephalitis (VEE), in the United States. While the majority of alphaviral infections are asymptomatic in humans, flu-like symptoms, including fever, headaches, and body aches, can occur. The virus is also capable of invading the central nervous system, leading to encephalitis and perhaps even death. Modifications to the virus for the generation of vaccine vectors has been well studied (reviewed in references 118 and 130). To generate vaccine replicon particles, the nonstructural protein genes and natural subgenomic promoter (26S) are retained, while the structural protein genes are replaced with the antigenic gene of interest. Current alphavirus replicon particle vaccines have been derived from the Sindbis virus, Semliki Forest virus (91), and VEE virus. This process creates a single-cycle replication vector that is translated upon entry into the cytoplasm of the target cell, allowing for intracellular processing and subsequent efficient presentation by MHC-I. The cytoplasmic amplification of these vectors includes a double-stranded RNA intermediate that activates both the innate and adaptive immune systems, strengthening this platform as a vaccine candidate delivery system (85). Indeed, the replication of alphavirus replicons results in extremely high levels of antigen expression (157, 166, 171); however, this replication is generally toxic to the infected cell, thus preventing long-term antigen expression (48). The antivector immune response against alphavirus replicons has so far been minimal, allowing for readministration, a major benefit for a vaccine (118). Similarly, seroprevalence is low in humans, and the replicons can be engineered to be lymph node trophic, enhancing antigen presentation through efficient targeting to, and expression in, the immune inductive site. A possible drawback to the alphavirus platform includes the requirement of the Rep protein for high-level antigen expression. This antigen may be prone to immune surveillance by CTLs. However, this platform has many positive features. For example, previous studies report that the primary cellular targets of replicon particles have been shown to be those of the CD14⁺ monocyte lineage in rhesus macaques (55) and dendritic cells in human peripheral blood mononuclear cells (51). In nonhuman primates, vaccination with 2 × 10⁸ PFU of chimeric SINenv and VEErep alphavirus replicon particles expressing SIVp55Gag and/or HIVΔV2gp140 alone resulted in high titers of anti-HIV neutralizing antibodies, while moderate anti-Env IFN-γ ELISPOT responses were also observed (163). In a separate study, 1 × 10⁷ infectious units of each of three VEE virus replicon particles expressing the SIVsmH-4 matrix capsid region of Gag, gp160, or unanchored gp140 were administered three times to rhesus macaques, followed by a mucosal challenge with SIVsmE660 (73). Macaques immunized with SIV-VEE virus replicon particles produced neutralizing antibodies and had reduced peak viral loads (1 log) postchallenge. In correlation with the decreased viral loads was an increase in the peripheral CD4⁺ T-cell counts in immunized macaques, observed at the viral set point. While not measured in this study, immune responses at mucosal sites, namely the preservation of CD4⁺ T cells, will be interesting to analyze. Overall, however, this platform has progressed very slowly. To date, there is a limited amount of clinical data, and studies to support the efficiency of this platform are remain important.

VESICULAR STOMATITIS VIRUS

A member of the Rhabdoviridae family, vesicular stomatitis virus (VSV) is a negative-stranded RNA virus that primarily infects livestock. In most of the world, infection of humans is rare and asymptomatic but can cause flu-like symptoms. VSV is an attractive platform for clinical use, as it can be delivered at mucosal sites, carry multiple transgenes, is easily manufactured, and has low seroprevalence in humans (reviewed in references 88 and 121). One potential drawback to VSV vaccine vectors is the development of antibodies against VSV surface glycoproteins in immunized hosts after one immunization. Therefore, effective boosting may require substituting a glycoprotein from a different VSV serotype. In nonhuman primates, immunization with rVSV vectors encoding HIV Env and SIV Gag protected against disease after challenges with SHIV 89.6p (41, 43). Furthermore, the direct comparison of intramuscular versus intranasal administrations of equal doses of HIV Env and SIV Gag (5 × 10⁶ PFU each) to rhesus macaques demonstrated that intranasal vaccination elicited higher cellular immune responses, as measured by IFN-γ ELISPOT assays, Cr release assays, and tetramer analysis (43). Interestingly, there was no difference in the humoral immune responses elicited by either vaccination strategy at systemic or mucosal sites (43). The intranasal delivery of VSV vectors does not seem to cause neurovirulence in nonhuman primates (71). It will be interesting to see if the mucosal administration of this promising platform can elicit cellular immune responses at mucosal sites.

POXVIRUS VECTORS

Poxviruses are large DNA viruses that evolved as vectors based on experience with the smallpox vaccine (reviewed in reference 93). As vaccine vectors, they are able to carry multiple and large genes (∼25 kb) (97) that are stably expressed. These vectors are also capable of inducing immune responses similar to those observed with the pathogen. Several studies have shown that vaccinia virus vectors can be pathogenic in immunocompromised people, thereby refocusing efforts on developing a more-attenuated, modified vaccinia virus Ankara (MVA) as a vector system (105). MVA has been shown to be safe for mice, monkeys, and humans, as it cannot complete an entire life cycle in humans or primates (41) while still being capable of inducing strong cellular and humoral immune responses in mice and primates. Numerous poxvirus-based vaccines have been tested in rhesus macaques after they were challenged with SIV, SHIV, and HIV-2 isolates (2, 7, 19, 47, 57, 59-62, 114) and have produced various degrees of cell-mediated immune responses and protection against viral challenge.

In a study conducted with rhesus macaques, SHIV DNA plasmid-producing noninfectious viral particles were delivered intranasally, with a rMVA SIV Gag-Pol and HIV Env boost. Coimmunization with an interleukin-2 (IL-2)/Ig plasmid elicited both mucosal and systemic humoral and cellular anti-SHIV immune responses (22). Intranasal immunization with other platforms had previously been reported to be the only route of mucosal vaccination that results in disseminated humoral and cellular immune responses (82, 83) and increases the level of vaginal responses compared to that for rectal immunization (20, 83). Indeed, as nasal immunization also induced greater systemic antibody responses and thus generated both systemic and mucosal immunity, the intranasal route may precede other mucosal routes (83). Mucosal SHIV-specific responses were detected in the IL-2/Ig group, as measured by the levels of IgA in rectal secretions and tetramer-positive T cells in rectal biopsies. Upon rectal challenge with SHIV 89.6p, all of the animals became infected, but the group coimmunized with IL-2/Ig was protected from CD4⁺ T-cell loss and disease progression to AIDS. Interestingly, a different study showed that tetramer-positive CD8⁺ T-cell responses were induced at mucosal sites by live attenuated poxvirus NYVAC encoding SIV Gag, Pol, and Env, whether it was administered by the intranasal, intramuscular, or intrarectal route (140).

A recombinant canarypox vector (ALVAC 1452) has been tested in multiple clinical trials with different doses, formulations, and prime-boost regimens by the HVTN and others. The results of a phase II study of ALVAC and VaxGen's gp120 protein (HVTN 203), demonstrated by IFN-γ ELISPOT assays, were observed in less than 36% of the vaccinees; therefore, the planned phase III trial (HVTN 501) was cancelled. Results from another phase II trial, outside of the United States, which tested ALVAC with or without a rgp120 booster (HVTN 026), did not show a difference in cellular immunity between the vaccinated and placebo groups, as measured by IFN-γ ELISPOT assays, lymphocyte proliferation assays, or Cr release assays (35). Increasing the dose of the ALVAC vector (HVTN 039) or the addition of lipopeptides (HVTN 042) did not enhance the number of responders, 5 of 52 and 3 of 77, respectively (37). A phase III clinical trial in Thailand involving the same ALVAC vector expressing subtypes E and B Env (gp41), Gag, protease plus gp160, or Chiron gp120B/E boosting is currently under way (76). In this study, CTL responses have been observed in less than 30% of the recipients, and antibodies, but not broadly neutralizing antibodies, have been observed. This finding may have been expected, as the preceding phase I/II study results showed no statistical difference between cellular responses in the ALVAC prime plus gp120 or gp160 protein boost group and the placebo group (149). Improving the immunogenicity of these experienced vectors is a major focus of the poxvirus field.

HERPESVIRUS

Unlike the viral vectors described above, herpes simplex virus type 1 (HSV-1) is able to persist for essentially the lifetime of the infected host. Active HSV-1 infection occurs at mucosal surfaces, including the mouth and genitals, and the virus remains latent in the nervous system, where it can be periodically reactivated. In addition, both cellular and humoral immune responses are generated in response to HSV-1 infection. These features of HSV-1 highlight its potential utility as a vaccine vector in regard to long-term antigenic expression, immunogenicity, and tropism for mucosal surfaces. Both replication-competent and replication-defective recombinant HSV-1 vectors have been studied as vaccines for HIV-1 in nonhuman primate models. In a study by Murphy et al. (107), the replication capability of HSV-1 did not elicit statistically different results, as measured by antibody levels, CTL activity, or protection against challenge. In fact, two rhesus macaques were protected against a SIVmac239 rectal challenge, one immunized with a replication-competent, and the other a replication-deficient, HSV-1 encoding SIV Env and Nef. However, the replication-competent group had peak viral loads 1.2 logs lower than those of the infected control animals. Clearly, the protection against mucosal challenge and the tropism of HSV-1 for mucosal surfaces warrant further examination of immune responses elicited by the mucosal delivery using this platform.

BACTERIAL VECTORS

The use of recombinant bacteria as vectors for gene delivery has been extensively reviewed elsewhere (94). Here we will touch on two particular platforms that have been used for HIV-1 vaccines in nonhuman primate models, Listeria monocytogenes and Salmonella.

NUCLEIC ACIDS

While RNA vaccines have been important in the area of immune therapy for cancer, work in the particular area of an HIV-1 vaccine has been limited. Accordingly, we focus here on the DNA vaccine approach.

DNA has been studied as a vaccine platform for over 15 years. Studies with mice immunized with plasmid antigens yielded promising levels of immunity that, unfortunately, have not been well translated in initial clinical studies. Currently, there is a significant amount of work in the field devoted to improving this platform, as it is appealing on many levels, including large-scale vaccine production, safety, repeat administration (no preexisting or vaccine-induced vector serology), and storage, as a cold chain is not required. DNA vaccines for HIV-1 have been reviewed elsewhere (66). Here we will focus primarily on the optimization of future DNA vaccines.

DNA/VECTOR PRIME-PROTEIN BOOST

The administration of a DNA prime and a subsequent protein boost effectively induces both cellular and humoral immunity. In rhesus macaques, a polyvalent DNA encoding Env from multiple subtypes and a Gag gene prime plus a homologous gp120 protein boost elicited modest cellular (around 500 SFC) and robust humoral (∼10⁶ SFC) responses (116). Upon rectal challenge with SHIV-Ba-L, the majority of immunized macaques were protected from infection, while the rest had lower viral loads than the unvaccinated group. A similar human ALVAC-HIV-1 recombinant vaccine expressing Gag, Pol, and gp120 or a Gag, Pol, and gp160 prime plus a homologous Env protein boost resulted in decreased viral load levels in the blood and in mucosal sites while simultaneously protecting macaques from peripheral CD4⁺ T-cell loss after challenge with SHIV_KU2 (115). This approach is being studied in a phase I clinical trial by S. Lu (95) and colleagues, using five different gp120 isolates and one Gag DNA prime plus a boost with heterologous gp120 protein. Interestingly, anti-gp120 antibody titers comparable to those of patients with chronic HIV-1 infection have been reported. In addition, anti-Env cellular responses were induced by this combination.

DNA PRIME-rAd BOOST

rAd5 is also being used as a boost for a unique DNA prime in large-scale clinical trials being conducted by the NIH and VRC. In a phase I trial, DNA and Ad5 were administered separately to 86 recipients at increasing doses. The vast majority of recipients had an anti-HIV-1 T-cell response, the first high responses in published reports (30, 52). Four DNA constructs, clade B Gag-Pol-Nef fusion protein, and clades A, B, C Env glycoproteins (31, 79) elicited an antibody and long-lasting Env-specific CD4⁺ and CD8⁺ T cells. Comparatively, a higher percentage of responders with a greater magnitude of immune responses was observed in recipients of rAd5 (clade B Gag-Pol fusion protein and clades A, B, C Env glycoproteins). Although impressive levels of cellular and humoral immunity were observed in the majority of subjects, preexisting immunity to Ad5 lowered T-cell responses approximately threefold compared to those of seronegative subjects. Based on the safety of, and encouraging results from, this study, a phase II trial has been initiated (124). Recent studies have shown a dramatic increase in HIV-1-specific immune responses with the combination of these two platforms. The DNA prime with a rAd5 boost resulted in a greater-than-1,000-fold increase in antibody and a greater-than-5-fold increase in T-cell responses against Env (124). Data from macaques immunized with this strategy and subsequently challenged with SIV show long-term preservation of the total number and function of central memory CD4⁺ T cells in the peripheral blood as a result of vaccine-generated protection against high levels of virus replication (86, 98). Efficacy trials of this approach are being planned through a HVTN/VRC collaboration.

DNA PRIME-rAAV BOOST

In the previously mentioned rAAV nonhuman primate study by Johnson et al. (72), a DNA prime-rAAV boost protocol elicited potent responses compared to those of other regimens using rAAV vectors alone. The study utilized either a DNA or rAAV prime with a mixture of three plasmids encoding the same SIV subgenomic fragments of (i) Gag, protease, and reverse transcriptase, (ii) reverse transciptase and integrase, and (iii) Rev and Env. Both groups received the same rAAV Rev-Env boost. Despite the weakly immunogenic DNA prime, the monkeys with the lowest viral loads and therefore the best control of SIV replication postchallenge were in the DNA prime and rAAV boost group.

DNA PRIME-VSV BOOST

While preexisting immunity to VSV is not an obstacle for humans, the ability to boost with the same rVSV strain is limited due to antivector immune responses. Therefore, the use of VSV in a heterologous prime-boost regimen with plasmid DNA is an attractive strategy for eliciting more-potent immune responses. In A01-negative rhesus macaques, intramuscular immunization with 5 mg of each DNA plasmid encoding SIV Gag and the cytokine adjuvant IL-12, followed by intranasal delivery of 5 × 10⁶ PFU of each rVSV encoding HIV Env (gp120) and a SIV Gag booster elicited both cellular and humoral immune responses greater than either platform alone (42). In particular, following a high-dose (300 monkey infectious doses) intravenous SHIV 89.6p challenge, this prime-boost regimen resulted in a 2.4-log reduction in viral loads compared to that for the DNA prime (1.5 log) or rVSV boost (1.1 log) given alone (42). In addition to there being no clinical disease observed postchallenge (42), there was enhanced preservation of CD4⁺ T cells in the peripheral blood of these animals.

DNA PRIME-MVA BOOST

One of the most-studied prime-boost regimens is a DNA prime and a poxvirus boost (123, 132). Interestingly, the T-cell responses generated with this heterologous strategy produce immune responses 10 times higher than either platform given separately (106). In fact, an important clinical trial by Robinson and colleagues is under way through the HVTN, utilizing a multivalent DNA (Gag, Pol, Env, Tat, Rev, Vpu) prime followed by a MVA boost. Initial results for humans show exceptional safety and tolerability but undetectable HIV-1-specific cellular (responders = 0/28, as determined by IFN-γ ELISPOT assay, and 0/24 by Cr release assay [37]) or humoral responses after a DNA prime (106). However, based upon data for rhesus macaques, the animals with minimal responses detected after a DNA prime were able to generate potent immune responses post-MVA boost (138), suggesting that the same may be true for humans. Preliminary reports suggest that the poxviral component increases the levels of immunity observed in these studies. In addition, data from another rhesus macaque study using Gag, Pol, and Env DNA prime plus MVA boost resulted in the control of viremia after a mucosal challenge with SHIV 89.6p (6).

DNA PRIME-HSV-1 BOOST

A recent study with rhesus macaques compared the immune responses elicited by immunization with recombinant DNA and HSV-1-encoding SIV genes in a heterologous DNA prime-HSV-1 boost strategy to those from homologous HSV-1 immunizations (77). Despite the robust cellular and modest humoral immune responses observed in the immunized macaques, all became infected upon an intravenous challenge with SIV239. Interestingly, the macaques immunized with the DNA prime-HSV-1 boost regimen had greater levels of T-cell responses than those of macaques immunized with HSV-1 alone. The observed levels of IFN-γ secretion for the prime-boost group were greater than or equal to those observed for unvaccinated macaques infected with wild-type SIV. In fact, while the DNA prime-HSV-1 boost regimen also produced a higher percentage of tetramer-postive IFN-γ-secreting cells, the group that received the homologous HSV-1 immunizations had more α4β7-positive cells, a marker of T cells homing to the mucosa. We look forward to future studies addressing the ability of HSV-1 to induce mucosal immune responses.

CONCLUSIONS

It is a tumultuous time for HIV vaccine development. There are few promising leads in the vaccine toolbox for antibody-based approaches. The STEP study has thrown cold water on the cadre of CTL-based platforms. These approaches (summarized in Table 2) provide the broadest array of T-cell-driving vectors in the history of vaccinology. In the near future, additional prime-boost studies and the use of adjuvanted DNA, designer Ad vectors, and DNA electroporation strategies will be expanded in the clinic. The world's first glimpse of the lack of efficacy of a pure T-cell-based approach supports new research goals for the CTL-based HIV vaccine field. (i) More-stringent primate models for determining vaccine efficacy in decision studies should be encouraged. (ii) We should demand greater CTL responses in humans, greater than an average of 300 to 400 10⁶ peripheral blood mononuclear cells, from next-generation is approaches. (iii) The induction of T-helper responses T-cell desirable and likely important, but we need to learn a great deal more about their contributions to immune control. (iv) The importance of polyfunctional immune responses, in addition to the proliferative capacity of vaccine-elicited T-cell responses, should be studied, and benchmarked as they relate to vaccine outcome in macaques and immune control in humans. (v) Novel and clean hypotheses regarding vectors entering clinical trials should be encouraged, and immune evaluations should be designed on the basis of these hypotheses and serve as benchmarks for progress in the clinic. (vi) Most importantly, rather than redividing the limited DAIDS funding pie, thus depriving important programs of needed resources, Congress should support the NIH by increasing funding for basic HIV vaccine research and pilot programs for early clinical testing. New ideas and concepts cannot be advanced effectively based on the current levels of funding for R01's, P01's, and developmental resources.

The Merck STEP study should be a catalyst for propelling the next generation of vaccines forward by raising the bar for what we expect from them. Such an occurrence will firmly add the CTL platform with all its potential promise to the armitarium of vaccinologists and, with support, hard work, and luck, welcome in the heyday of the age of designer vaccine platforms.