Resistance to Infection, Early and Persistent Suppression of Simian Immunodeficiency Virus SIV_mac251 Viremia, and Significant Reduction of Tissue Viral Burden after Mucosal Vaccination in Female Rhesus Macaques

Abstract

The efficacy of oral, intestinal, nasal, and vaginal vaccinations with DNA simian immunodeficiency virus (SIV)/interleukin-2 (IL-2)/IL-15, SIV Gag/Pol/Env recombinant modified vaccinia virus Ankara (rMVA), and AT-2 SIV_mac239 inactivated particles was compared in rhesus macaques after low-dose vaginal challenge with SIV_mac251. Intestinal immunization provided better protection from infection, as a significantly greater median number of challenges was necessary in this group than in the others. Oral and nasal vaccinations provided the most significant control of disease progression. Fifty percent of the orally and nasally vaccinated animals suppressed viremia to undetectable levels, while this occurred to a significantly lower degree in intestinally and vaginally vaccinated animals and in controls. Viremia remained undetectable after CD8⁺ T-cell depletion in seven vaccinated animals that had suppressed viremia after infection, and tissue analysis for SIV DNA and RNA was negative, a result consistent with a significant reduction of viral activity. Regardless of the route of vaccination, mucosal vaccinations prevented loss of CD4⁺ central memory and CD4⁺/α4β7⁺ T-cell populations and reduced immune activation to different degrees. None of the orally vaccinated animals and only one of the nasally vaccinated animals developed AIDS after 72 to 84 weeks of infection, when the trial was closed. The levels of anti-SIV gamma interferon-positive, CD4⁺, and CD8⁺ T cells at the time of first challenge inversely correlated with viremia and directly correlated with protection from infection and longer survival.

INTRODUCTION

There are currently over 30 million people infected with human immunodeficiency virus (HIV), and no vaccine tested in humans has shown enough promise to warrant distribution. Two HIV vaccine trials ended with different outcomes (1, 2). The Merck trial, which used a recombinant adenovirus (Ad) vector as a vehicle for HIV antigens, appeared to increase the risk of infection among vaccinees. The RV144 trial, which tested an HIV recombinant poxvirus prime/gp120 protein boost strategy, resulted in 30% efficacy in protection from infection. This protection diminished over time, and no reduction of viremia was observed in the infected vaccinees, possibly because this vaccine stimulated poor antiviral CD8⁺ T-cell responses (2, 3). The most significant correlates of protection from acquisition of infection appeared to be nonneutralizing antibodies that bind V1-V2 in Env (4).

Several preclinical trials, in which different vaccine platforms were explored to prevent infection and disease progression in rhesus macaques (RM) exposed mucosally to simian immunodeficiency virus, indicated that vaccination can provide some protection against simian immunodeficiency virus (SIV) infection and can contain virus replication (5–9). Many of them used adenoviral vectors to deliver SIV antigens (6, 9–16). A recent study showed that intramuscular (i.m.) immunization with an adenoviral vector expressing SIVsmE543 Gag, Pol, and Env antigens, combined with recombinant poxvirus or adenovirus, protected against SIV_mac251 infection (6). Env-specific antibodies were found to be critical for blocking infection, whereas multiple cellular and humoral immune responses correlated with viremia control. Anti-Gag T-cell responses correlated with viremia control (6). Control to undetectable levels of viremia early after mucosal infection was observed with a persistent SIV-recombinant RM cytomegalovirus (RhCMV), which indefinitely maintained high-frequency SIV-specific effector memory CD4⁺ and CD8⁺ T cells (7, 17). SIV viremia was not detected after CD4⁺ or CD8⁺ T cell depletion, suggesting viral clearance (7, 50).

Although promising, the most successful of these studies only investigated systemically delivered vaccines (6, 7). Since the mucosal surfaces of the intestinal and genital tracts are primary sites of HIV transmission, significant stimulation of mucosal immunity could be important, and it is achieved to a much higher degree with mucosal than systemic immunization. Combined nasal, oral, and intratracheal vaccination with replicating recombinant Ad-HIV/SIV priming followed by envelope protein boost stimulated effective systemic immunity as well as mucosal immune responses at multiple mucosal sites (9, 16). Virus-specific antibodies had a transient effect on early infection events, and reduced peak viremia was detected in the vaccinated animals (9). In another study, combined i.m. and nasal immunizations of RM with gp41-derived virosome-bound antigens elicited full protection against repeated SHIV-SF162P3 vaginal challenges (8). The protected animals showed gp41-specific cervico-vaginal IgA and IgG with transcytosis-blocking activity. In this study, mucosal protection against heterologous SHIV challenge occurred in the absence of detectable serum neutralizing antibodies.

We have shown that rectal or nasal immunization of RM with SIV DNA/recombinant modified vaccinia virus Ankara (rMVA) vaccine could induce anti-SIV IgA antibodies in rectal or vaginal secretions, but these responses were sporadic and declined over time (18,–20, 27). Systemic and mucosal virus-specific immunity provided protection from progression to AIDS, and nasal vaccination was more effective than i.m. vaccination. Formulation modifications that target the vaccine to the appropriate immune cells or reduce DNA degradation should stimulate greater mucosal humoral responses that may enhance vaccine efficacy (21). In one of our studies, high levels of anti-SIV rectal IgA were achieved with liposome delivery (19).

We recently investigated the immunogenicity of oral, intestinal, nasal, and vaginal immunizations of a vaccine composed of SIV, interleukin-2 (IL-2), and IL-15 plasmid DNAs, followed by boosts with SIV Gag/Pol/Env MVA and inactivated SIV_mac239 particles (22). All four immunizations stimulated systemic and mucosal T-cell responses against Gag and Env, albeit to different extents. Oral immunization was as effective as vaginal immunization for stimulation of SIV-specific IgA in vaginal secretions, and it generated greater rectal IgA responses than the other vaccination routes. We report here the protective efficacy of these immunizations after repeated low-dose vaginal challenge with pathogenic SIV_mac251.

MATERIALS AND METHODS

Vaccine protocol and SIV_mac251 low-dose mucosal challenge.

Twenty-eight animals, divided into four groups, received the following vaccine regimen: group 1 (n = 7), oral administration; group 2 (n = 7), intestinal delivery; group 3 (n = 7), nasal administration; and group 4 (n = 7), vaginal administration. Each animal received a total of three DNA doses on day 1, week 8, and week 24 that consisted of 1 mg pVacc6, 0.5 mg IL-2/Ig, and 0.5 mg IL-15 DNA, formulated in 1 ml of 20 mM DOTAP (1,2-dioleoyl-3-trimethylammonium-propane) and cholesterol (1:1) (Encapsula Nanoscience) (22). At weeks 48 and 50, each group received a dose of 2,2-dithiodipyridine (aldrithol-2 [AT2]) inactivated SIV_mac239 particles, equal to 300 μg total proteins and corresponding to 50 μg of p28CA, with 100 μg of heat-labile toxin (LT) of Escherichia coli (23) in phosphate-buffered saline (PBS) in a final volume of 350 μl. Immune responses to the vaccination were previously reported (22). Starting 7 weeks after the last immunization, vaccinated animals and controls were inoculated with a low dose (approximately 0.2 50% animal infectious dose [AID₅₀] in an RM vaginal titration, corresponding to 100 50% tissue culture infectious doses [TCID₅₀]) of the pathogenic SIV_mac251 virus grown in RM peripheral blood mononuclear cells (PBMCs) (a gift from Nancy Miller and Ron Desrosiers). It was administered, as cell-free virus, nontraumatically in the vagina with needleless tuberculin syringes (24). This virus stock is a highly diverse swarm of many different quasispecies of CCR5-tropic viruses (25). Challenge was repeated weekly until blinded reverse transcription-PCR (RT-PCR) tests in two independent laboratories were positive for virus in plasma. After 32 nontraumatic challenges, the animals that remained RT-PCR negative were traumatically challenged once or twice. A speculum was used so the vaginal mucosa could be visualized, and virus was injected using a 26-gauge needle into the superficial mucosa, forming a bleb of the inoculum. Bleeding was very limited (a drop or so). Animal treatment and care were carried out according to a protocol approved by the Harvard Medical School Institutional Animal Care and Use Committee.

Analysis of SIV-specific IgA in rectal and vaginal secretions and IgG in plasma.

Rectal and vaginal secretions were collected with Weck-Cel sponges as described previously (26). Saliva was collected by placing 2 dry sponges between the cheek and the gum, near the back of the mouth, for a minimum of 5 min. To collect nasal secretions, 100 μl of Dulbecco's PBS was dropped into each nostril. The nostril was pressed shut and massaged. A premoistened sponge was inserted to absorb fluid for 5 min. Sponges with secretions were stored at −70°C, and antibodies were eluted from the sponges by centrifugation as described previously (26). Antiviral and total IgA and IgG in secretions and IgG in serum were measured by enzyme-linked immunosorbent assay (ELISA) as described previously (20, 27).

Phenotyping of immune cells and intracellular cytokine staining (ICS).

Characterization of CD4⁺ and CD8⁺ T cells in PBMC and isolation of mononuclear cells (MNC) from colon-rectal mucosa and PBMC from the blood was conducted according to previously published procedures (20, 28): 10⁵ MNC and 10⁶ PBMC were incubated for 14 h with medium (unstimulated, background controls) or 1 μg/ml pools of 15-mer SIV Gag or SIV Env peptides. Positive controls were incubated with 10 ng/ml PMA (phorbol myristate acetate; Sigma) and 1 μg/ml ionomycin (Sigma). Cultures contained brefeldin A (BD Biosciences) and 1 mg/ml anti-CD49d and anti-CD28. PBMC and MNC were stained and evaluated for cytokines as previously described (18). For MNC, 200 μl of a viability dye (VIVID; Invitrogen) was added to the antibody cocktail to exclude dead cell background. CD3⁺ T cells were used as a gate for CD4⁺ or CD8⁺ cells. Data for peptide-stimulated populations are reported as percentages determined after subtracting the percentage of positive cells detected in unstimulated cultures for each sample. AT-1 (CD38) and Act-1 (α4β7) antibodies were also used in this study (29). Flow-cytometric analysis was performed using MoFlo (Dako North America, Carpinteria, CA) and FlowJo v9.1 (TreeStar, Ashland, OR).

Viral load quantitation.

Plasma SIV RNA levels were measured by real-time RT-PCR assay as described previously (30). The Lifson assay has a threshold sensitivity of 30 copy equivalents per milliliter. Interassay variation is <25% (coefficient of variation). Mean viral loads were calculated by transforming the number into its logarithmic value and averaging the logarithmic values of all animals of the group at one time point.

CD8⁺ T-cell depletion.

Three vaccinated animals that resisted 34 challenges with SIV_mac251 and nine macaques that controlled viremia to undetectable levels 4 to 28 weeks after the first positive RT-PCR were treated with a chimeric anti-human CD8α monoclonal antibody, cM-T807 (Nonhuman Primate Reagent Resource, NIH). The first dose (10 mg/kg of body weight) was given subcutaneously (day 1); the other 3 doses (5 mg/ml) were intravenous (31–33) on days 3, 7, and day 10. Cell blood counts and immune phenotyping were carried out before treatment, on days 0, 3, 7, and 10, and then weekly until CD8⁺ lymphocytes returned.

SIV RNA and DNA quantification in RM tissues.

Cell-associated SIV RNA and DNA was determined with an assay targeting the conserved sequence of Gag, as described previously (7). To characterize the levels of SIV DNA and RNA from tissues of the vaccinated animals with undetectable levels of viral loads, a nested quantitative real-time PCR method was used. MNC from tissue samples were collected by collagenase digestion followed by Ficoll-Paque (GE Healthcare) isolation as previously described (20). DNA and RNA extraction was performed in 1 ml TRIzol reagent (Ambion, Life Technologies) per 10⁶ cells by following the manufacturer's recommendations and also by using the alternative back-extraction method for DNA. Samples were dissolved in 10 mM Tris, pH 8, for the RNA and 5 mM Tris, pH 9, for the DNA. The DNA and RNA samples were tested separately. The copy number of SIV DNA and that of the gene sequence from macaque IL-15, as a reference for cell equivalents, were codetermined in a duplex-format quantitative PCR. The nesting or preamplification of the DNA or cDNA was performed for 12 cycles with primers SIVnestFO1 (GATTTGGATTAGCAGAAAGCC) and SIVnestRO2 (GACAGTGCCTGAAATCCTGGCACTAC), flanking the SIV Gag target region as described in reference 7. Following the preamplification, a duplex real-time PCR was carried out with primers and probes (IL-15 VIC probe, AGATACAGATATTCATGATACAGT; SIV 6-carboxyfluorescein [FAM] probe, TGTCCACCTGCCATTAAGCCCGA; IL-15FWD, AGTTGCAAGGTAACAGCAATGAAG; IL-15REV, GAAGACAAGATGTTGTTTGCTAGG; SIVFWD, GCAGAGGAGGAAATTACCCAGTAC; SIVREV, CAATTTTACCCAGGCATTTAATGTT) and run on an Applied Biosystems 7300 real-time PCR instrument. cDNA from RNA samples were obtained after RT-PCR by utilizing iTaq universal supermix (Bio-Rad). As a control for RNA, 18S RNA was also detected in each sample and used to normalize the SIV RNA numbers. Cell-associated SIV Gag RNA copy numbers were reported per 1 μg of total RNA. SIV Gag DNA copy numbers are reported as numbers of SIV copies per 10⁵ cells. The cell number in each DNA sample was calculated by evaluating the copy number of IL-15. For the DNA and cDNA PCRs, 10 and 5 replicates, respectively, were tested per sample.

Euthanasia.

Animals were euthanized because of closure of the study or earlier if they developed signs and symptoms consistent with the definition of AIDS. AIDS was defined as being SIV⁺ (detectable viremia) and experiencing one of the following criteria: 1, weight loss of >15% in 2 weeks or >30% in 2 months; 2, documented opportunistic infection; 3, persistent anorexia for >3 days without explicable cause; 4, severe, intractable diarrhea; 5, progressive neurologic signs; 6, significant cardiac and/or pulmonary signs; 7, loss of CD4⁺ T cells below 200 or 10%.

Statistical analysis.

Calculations and statistical analyses were performed using GraphPad Prism version 3 software. The two-tailed Fisher's exact test was used to compare the frequency of IgA responses between groups. Between-group comparisons were carried out by two-tailed t test or Mann-Whitney test. Within-group comparisons were done by one-way (one parameter, one time point) or two-way (one parameter at different time points in multiple groups) analysis of variance (ANOVA) followed by a Bonferroni post hoc test. Results of statistical analyses were considered significant if they produced P values of ≤0.05. The display of multicomponent distributions was performed with SPICE v5.2 (available at http://exon.niaid.nih.gov/spice/) (34).

RESULTS

Resistance to infection after repetitive vaginal exposure to low-dose SIV_mac251.

We have investigated the anti-SIV immunity stimulated by SIV/IL-2/IL-15 DNA followed by boosts with SIV Gag/Pol/Env rMVA and inactivated SIV_mac239 particles administered in four groups of RM vaccinated in the oral cavity, small intestine, nasal cavity, and vaginal mucosa (22). We found that all vaccinations stimulated systemic T-cell responses against Gag and Env to different extents, with the oral immunization providing a higher-magnitude response and the nasal immunization providing wider functional heterogeneity. SIV-specific T cells producing IFN-γ dominated these responses. The vaccinations also induced CD4⁺ and CD8⁺ T-cell responses in the rectal and vaginal mucosa with greater functional heterogeneity than in blood. Rectal T-cell responses were significantly greater in the orally vaccinated animals than in the others. The most balanced and diverse, as well as high-magnitude, vaginal T-cell responses were observed after intestinal vaccination. Significantly higher CD8⁺/granzyme B⁺ T-cell responses were observed systemically after intestinal vaccination and in rectal cells after oral immunization. Here, we report the impact of the immunity achieved with these vaccinations on SIV exposure and disease protection.

Vaccinated and naive female RM were challenged vaginally with a repeated low dose of pathogenic SIV_mac251 starting 7 weeks after the last immunization. The animals were challenged weekly until a positive SIV RT-PCR was obtained using plasma. Six additional historic female controls, challenged vaginally with the same dose of the same virus stock, were also included in the follow-up to increase the size of the naive control group to 13 members and to increase the statistical power of the study.

A different median number of challenges was required to infect the seven animals in each vaccine group and the 13 control animals. A median of 8 exposures was required in the control group, 10 in the nasal group, 13 in the vaginal group, and 16 in the oral group (Fig. 1). After 32 nontraumatic challenges, more than 50% (four of seven) of the intestinally vaccinated animals and four additional animals, one in each of the other four groups, were still RT-PCR negative. The Kaplan-Meier curve of acquisition of infection was statistically significant for the comparison of the intestinal group curve to that of the controls (P = 0.0004 by log-rank test [Mantel-Cox]). The median number of challenges required to achieve infection in the intestinal group (the number used in the analysis was 32) compared to the control or any of the other groups was also statistically significant (P < 0.03 by Wilcoxon signed-rank test). A traumatic vaginal challenge was carried out in the animals not yet infected after 32 challenges to achieve infection and evaluate disease protection in as many animals as possible. Five additional animals became infected, three in the intestinal group, one in the vaginal group, and one in the control group. Three, one in the oral, one in the intestinal, and one in the nasal group (Fig. 1b, open symbols), who resisted two traumatic challenges were excluded from the postchallenge follow-up, reducing the size of these groups to 6 animals, while 7 and 13 animals remained in the vaginal and control groups.

FIG 1.

Protection from mucosal infection. (a) Kaplan-Meier graph of the number of nontraumatic challenges received for each vaccinated group until a positive infection was detected in plasma. The nontraumatic challenges were stopped after 32 challenges, and the animals that were still uninfected received traumatic challenges. The dotted line intersects the point in each curve corresponding to 50% of the animals being infected. (b) Median number of challenges for each group and total number of challenges received by each animal, whether nontraumatic or traumatic. Three animals (249, 246, and 148) remained noninfected after 32 nontraumatic and 2 traumatic challenges, and they are shown as open symbols. Wilcoxon signed-rank test (P < 0.05) was used for the median comparison. (c) IFN-γ⁺ CD8⁺ T-cell responses in quartiles, detected 3 weeks before challenge in vaginal MNC. The quartile partition is based on the number of resisted challenges. Animals in the 1st quartile resisted up to 7 challenges, in the 2nd quartile 8 to 13 challenges, in the 3rd quartile 14 to 23 challenges, and in the 4th quartile more than 23 challenges. Prechallenge vaginal MNC were not available for the control animals that were not included in the calculation. Mann-Whitney t test was used.

We investigated whether correlates of protection from infection could be established. On the day the first challenge was administered, no SIV-specific IgG was detected in serum, and only two animals had anti-SIV antibodies (IgA) in vaginal secretions (not shown). Using a quartile analysis based on the number of resisted challenges, statistical significance was observed when the levels of CD8⁺/IFN-γ⁺ T-cell responses in vaginal biopsy specimens before challenge in animals of the 4th quartile were compared to those of animals of the 1st (up to 7 challenges; P = 0.05 by unpaired t test) and 2nd quartiles (up to 13 challenges; P = 0.008). They were not statistically significant against the third quartile (up to 23 challenges) (Fig. 1c).

These results indicate that intestinal immunization with the vaccine provided better protection from infection than the other immunization routes. It is possible that with larger groups, statistical differences could be detected between the medians of other groups, in particular when the oral group is compared to the controls.

Kinetics of viral replication and of neutralizing antibodies after infection.

Plasma viral loads in the infected macaques were monitored by RT-PCR at several time points after infection (Fig. 2a). Overall, mucosal vaccination significantly reduced viral loads compared to controls (Fig. 2a). However, when individual groups were compared to each other, virus replication in the control group was found to be comparable to that in animals vaccinated intestinally or vaginally. In contrast, 50% of the infected animals in the oral and nasal groups fully controlled viral replication by week 16 postinfection. The same control was achieved in 16.5% (one of six) of the animals vaccinated intestinally, in none of those immunized vaginally, and in 23% of the controls at weeks 28 and 60. None of the vaccinated animals that suppressed viremia carried protective alleles, although control animal 301 was MamuA*01 positive.

FIG 2.

Postinfection viremia and neutralizing antibodies. (a) Dynamics of viral loads during infection: comparison between the mean viremia of the vaccinated animals and of the control macaques during the course of infection (top left graph). The values of viral loads for each individual animal and the time points for individual group averages (black line in each graph) are represented for all groups as logarithmic values. Mann-Whitney t test was used to compare mean viremia of controls and vaccinated animals (P = 0.01). Error bars represent standard errors of the means (SEM). (b) Plasma viral loads at the peak of viremia (left graph) and area under the curve (AUC) from week 16 to 60 (right graph) for each animal, calculated using the logarithmic values of the viral loads. One-way ANOVA and a Bonferroni post hoc test were used to compare groups (P = 0.05 and 0.03, respectively). (c) Logarithmic geometric means and SEM of serum neutralizing antibody levels, measured against SIV_mac251, are reported for each group (top left graph) and for each animal of all groups. Open symbols and dotted lines are used for animals that control viral replication to undetectable levels. Two-way ANOVA was used for multiple comparisons. No significant differences were detected.

A more detailed analysis of viremia revealed that when the vaccine was given by the nasal route, a significant reduction of peak of viremia was observed compared to the control group (Fig. 2b). Furthermore, the chronic viremia (area under the curve from week 16 to 60) in animals that were nasally vaccinated was significantly lower than that of control macaques (Fig. 2b). Animals vaccinated vaginally showed only limited control of viral replication. Given the control observed in preventing infection, it was surprising that intestinal vaccination produced poorer control of viremia than did oral or nasal vaccination; viremia in the intestinal group was similar to that in controls. It is possible that immune responses that are critical for control of viral replication differ in various bodily compartments compared to those required for generating sterile protection from infection.

Levels of postinfection neutralizing antibodies in all of these animals paralleled the level of viremia, excluding a role of these antibodies in viremia control (Fig. 2c). Although differences were not statistically significant, possibly because of substantial standard errors at each time point, higher titers of neutralizing antibodies (Fig. 2c) as well as serum and mucosal SIV binding antibodies (not shown) were observed in animals with high viral loads, indicating that postinfection humoral responses are driven and sustained by viral antigen. The animals that were infected nontraumatically had significantly lower viral loads (P = 0.0095 by Mann-Whitney test) than those infected traumatically; however, no significant difference in the neutralizing antibodies of these two groups was observed.

These results suggest that oral or nasal vaccination with this vaccine modality induced immune responses that were more effective at controlling virus replication than those generated by vaginal or intestinal vaccination.

Vaccine-induced responses associated with control of viral replication and longer survival after infection.

Time from infection to disease progression was evaluated as a measure of vaccine efficacy (Fig. 3a). Lower viral loads have been associated with longer survival (6–9, 16, 17, 20, 27, 35–38). The animals that received oral or nasal vaccination and managed a better viremia control were also better protected from disease progression than the other animals. When the study was closed, 100% of the orally vaccinated animals and 83% (5 of 6) of the nasally vaccinated animals were AIDS free (Fig. 3a). As the number of resisted challenges and date of infection varied among animals, the range of AIDS-free survival at the time of the trial closure also varied and reached up to 84 weeks in the oral group and up to 80 weeks in the nasal group. However, a more rapid disease development, with median time to AIDS of 24 weeks, was observed in animals vaccinated intestinally, for which a better resistance to infection but poor viremia control had been observed, indicating that intestinal immunization provided limited protection from disease progression (Fig. 3a). Nine of 13 controls (69%) had died by week 75, while the remaining 4 were euthanized at the closure of the trial. The log-rank (Mantel-Cox) test was used for comparison among curves (P = 0.05). The comparison of individual survival curves of vaccinated animals with controls was significant when orally vaccinated animals were compared to controls (P = 0.01). The median survival was not available for nasally and orally vaccinated animals, as the trial was closed before the group median was reached. However, the survival of more than 75 weeks for most of the animals, combined with the CD4⁺ T cell values of the same animals, indicates that it is highly likely that the median of these groups would have been 80 weeks or higher, the number needed to reach statistical significance (median survival of non-AIDS-related euthanasia at the time of trial closure was 75 weeks). As expected, there was a significant inverse correlation between the level of viremia and time to AIDS (Fig. 3b).

FIG 3.

Protection from disease progression. (a) Kaplan-Meier curves of survival during infection, plotted for each group. The dotted line intersects the point in each curve corresponding to 50% of the animals still surviving. A log-rank test (Mantel-Cox) was used for comparison (P = 0.05). (b) Correlation between the levels of viral loads measured early in the chronic phase (16 weeks postinfection) and during survival (P = 0.001 and r = −0.66 by Spearman test). (c) Correlation between anti-Gag⁺ Env CD4⁺/IFN-γ and CD8⁺/IFN-γ T-cell responses detected in PBMCs on the day of first challenge and control of early chronic viremia (week 16 postinfection viral loads). The Spearman test was used to determine statistical significance (P = 0.0034 and r = −0.56 for anti-Gag⁺ Env CD4⁺/IFN-γ and P = 0.0045 and r = −0.54 for CD8⁺/IFN-γ).

Since the orally and nasally vaccinated animals were better protected from disease progression, we investigated which immune responses stimulated by the vaccination could explain the outcome. Mucosal vaccination, independently of the route, stimulated responses that were mainly dominated by the production of IFN-γ (22). The levels of systemic SIV-specific CD4⁺ and CD8⁺ T cells producing IFN-γ, measured at the time of first challenge, inversely correlated with the levels of viral load observed 16 weeks postinfection (Fig. 3c). The same was true for week 28 and the last time point available (week 60 or earlier if the animal died before week 60; data not shown). These results confirm what we observed before with nasal vaccination alone and extend to the same responses stimulated with other mucosal vaccinations: preexisting CD4⁺ and CD8⁺ IFN-γ⁺ T-cell responses are important to limit virus replication and prolong the asymptomatic phase of infection (20).

SIV-specific immune responses during early chronic infection.

The analysis of the systemic SIV-specific immune responses measured during infection in PBMC revealed no significant differences among groups regardless of the route of immunization, and their size was usually directly proportional to viremia (not shown). However, a change of the quality of the response was observed when PBMC CD8⁺ T-cell responses stimulated by the vaccine where compared to those observed during infection, with a statistically significant shift from IFN-γ responses to responses dominated by IL-2 and tumor necrosis factor alpha (TNF-α) production. In all cases, the comparison is between values observed on the day of first challenge to those observed 12 weeks postinfection (CD4⁺/IFN-γ⁺ responses, oral group, P = 0.0012; CD8⁺/IFN-γ responses, oral group, P = 0.0012; CD8⁺/IFN-γ responses, intestinal group, P = 0.03; CD8⁺/IFN-γ responses, nasal group, P = 0.038; CD4⁺/TNF-α⁺ responses, nasal group, P = 0.037; CD8⁺/TNF-α⁺ responses, oral group, P = 0.02; CD8⁺/TNF-α⁺ responses, vaginal group, P = 0.01; CD8⁺/IL-2⁺ responses, oral group P = 0.00047) (Fig. 4a). No significant differences were observed in the magnitude of PBMC CD4⁺ and CD8⁺ T-cell responses of different groups despite higher viral loads in intestinally vaccinated animals and controls, suggesting that their responses were less effective in controlling viral replication. A significant increase of SIV-specific, IL-2-producing CD4⁺ T cells observed on week 12 compared to first day of challenge was also observed in the rectal mucosa of all vaccinated groups except for the oral group (intestinal, P = 0.005; nasal, P = 0.001; vaginal, P = 0.04; Mann-Whitney t test). The immune responses stimulated by the vaccination appear to be of a different quality than those stimulated by the infection. We do not know if this means that new responses are primed or if the secretion pattern of antigen-specific cells changes and is influenced by the nature of the SIV infection and its effects on the immune system.

FIG 4.

Immune responses after infection. (a) PBMC, (b) rectal, and (c) vaginal SIV Gag⁺ Env-specific T-cell immune responses early in chronic infection (12 weeks postinfection, the only postinfection time point available for rectal and vaginal MNC). The pie plots illustrate the diversity and the amount (group average percentages are given under the pie) of the CD4⁺ or CD8⁺ T-cell responses, detected as the ability to produce IL-2, IFN-γ, or TNF-α or multiple cytokines. Sampling in the vaginal and rectal mucosa was carried out 3 weeks before vaginal challenge to allow for the healing of the vaginal mucosa. Comparisons of the responses within each group were done using a Mann-Whitney t test, and for each cytokine, single-, double-, and triple-positive cells that express that cytokine are included in the numbers used for statistical analysis. The arched black line marks populations that are significantly larger than the same population at another time point. The red numbers indicate values for total populations that are significantly higher before challenge than the corresponding values at week 12 (oral and nasal groups) and are higher than those for the same population at the same time point in the other groups (intestinal group).

Significantly higher percentages of total CD4⁺ T-cell responses were detected on week 12 in the rectal and vaginal mucosa of the intestinally vaccinated animals than in the other groups (P = 0.01 and 0.0003, respectively, by ANOVA), suggesting that the viral replication that occurs in the intestine could expand the preexisting responses stimulated at this site by the vaccination to a larger degree (Fig. 4b and c, red numbers). A significant reduction of total CD4⁺ responses was observed in the rectal mucosa of orally vaccinated animals (P = 0.01) and in the vaginal mucosa of nasally vaccinated animals (P = 0.05), a possible reflection of the lower viral loads observed in these animals. Measurements of intestinal viral loads and small intestine responses, which were not planned for this trial, could be important to determine if the accelerated disease progression observed in intestinally vaccinated animals correlates with the higher intestinal levels of SIV-specific CD4⁺ T cells, which are known to be preferentially infected by the virus (39).

Evaluation of immunological parameters associated with vaccine efficacy.

Several immunological parameters measured early after infection can be predictive of protection from disease progression (18, 20, 27). These markers could be very helpful in the assessment of vaccine efficacy in human trials, where time to disease progression is longer than that in RM and therapeutic intervention by itself may delay disease progression.

The CD4⁺ central memory (C_M) T-cell population is the main viral target, and loss of this population becomes evident early in infection, while the total CD4⁺ T-cell population declines later and in human trials could be affected by therapy. Therefore, preservation of the CD4⁺ C_M T-cell population is a good prediction marker of protection from disease (37). Loss of the CD4⁺ C_M T-cell population during early chronic infection (16 weeks postinfection) was significantly reduced when all vaccinated animals were compared to controls, regardless of the vaccination route (Fig. 5a). Only the values observed in animals vaccinated nasally reached statistical significance compared to those of controls (Fig. 5a). The averages for orally and nasally vaccinated animals were similar but with a larger standard error. When we looked at the CD4⁺ T-cell population, the analysis of the CD4⁺ C_M T-cell population at week 16 provided an appropriate predictor of the long-term changes in this population and of survival (Fig. 5b). Vaginal or intestinal vaccination did not prevent the loss of CD4⁺ T cells during chronic infection, since similar levels of CD4⁺ T cells were observed in these groups and in controls (Fig. 5b).

FIG 5.

Immunological parameters associated with vaccine efficacy. (a) Circulating C_M (CD95⁺/CD28⁺) CD4⁺ T cells are given as percentages of total CD4⁺ T cells in vaccinated (black) and control (green) animals (left; P = 0.005) and for each animal in the groups (right; P = 0.02). Values observed on week 16 after infection in PBMC are reported for C_M CD4⁺ T cells. (b) PBMC CD4⁺ T-cell levels reported for each group as average percentages (left) or average absolute counts/ml (right) during the course of the infection. (c) Percentages of circulating α4β7^high+/CD4⁺ T cells in vaccinated (black) and control (green) groups (left; P = 0.0001 by Mann-Whitney t test) and in each group (right; P = 0.0001 by one-way ANOVA followed by a Bonferroni posttest, which reveals which between-group comparisons contribute to the significant ANOVA result) early during chronic infection (16 weeks postinfection). (d) Levels of immune activation in CD4⁺ and CD8⁺ T-cell populations during infection. Geometric means of the percentages of CD38^high+/HLA-DR⁺ CD4⁺ (left; P = 0.0001) or CD8⁺ (middle; P = 0.005) C_M T cells in PBMC in each group are reported. Shown are comparisons of the geometric mean of CD38^high+/HLA-DR⁺ CD4⁺ or CD8⁺ C_M T-cell percentages in each group at the peak of immune activation 4 weeks after infection (right; P = 0.03). Mann-Whitney t test was used for the pair average comparisons. Two-way ANOVA and a Bonferroni post hoc test were used for multiple comparisons. Error bars represent SEM. (e) Correlation between levels of immune activation detected as percentages of CD38^high+/HLA-DR⁺ CD8⁺ C_M in PBMC 20 weeks postinfection and early control of the viral replication 16 weeks postinfection (left; P = 0.0001 and r = 0.79), preservation of the CD4⁺ C_M T-cell population 16 weeks after infection (middle; P = 0.03 and r = −0.38), and long-term survival (right; P = 0.0006 and r = −0.55). The Spearman test was used to test the statistical significance of the correlation.

CD4⁺ T cells expressing α4β7 integrin measured in blood have been shown to correctly represent the size of this population in the intestine, which is significantly depleted during the acute stage of the infection (23). Thus, loss of this particular subset in blood serves as a surrogate marker for monitoring the intestinal CD4⁺ T-cell depletion (23, 40–43). There was a significant preservation of the CD4⁺ C_M α4β7⁺ T-cell population in the early stages of chronic infection (16 weeks postinfection) in vaccinated animals compared to controls (Fig. 5c), and this was true also at later time points (not shown). Furthermore, significantly higher levels of CD4⁺ C_M α4β7⁺ T cells were measured in each of the four vaccinated groups compared to the control animals (Fig. 5c, right). The animals vaccinated intestinally, which had the worst postinfection outcome of all of the vaccinated animals, had a greater depletion of this population than the other groups, even though the comparison was statistically significant only against the values of the oral group, probably because of the small size of the groups. This finding supports a higher viral load in the intestine of these animals than in the other vaccinated groups.

Immune activation occurs more significantly during viral infection in species that progress to AIDS and is associated with HIV-1 pathogenesis (44–46). If a vaccine provides protection, it should also reduce virus-driven immune activation during the asymptomatic phase of the infection. We evaluated the magnitude of immune activation in CD4⁺ or CD8⁺ C_M T-cell populations by measuring the CD38 marker, whose increased expression has been link to T-cell activation, and the HLA-DR marker (47). Significantly lower levels of immune activation were detected during the course of the infection in the orally and nasally immunized animals than in the controls (Fig. 5d). As expected, based on disease outcome and evaluation of CD4⁺ C_M and of α4β7⁺/CD4⁺ C_M T-cell populations, the highest levels of immune activation were observed in controls, followed by those observed in the intestinally vaccinated macaques. Reduction of the immune activation could be detected early on, as levels of immune activation measured 4 weeks postinfection, right after the peak of viremia, were significantly lower in CD4⁺ and CD8⁺ T cells of the orally vaccinated animals than in the other groups (Fig. 5d). Immune activation remained low in this group during the entire postinfection follow-up. Animals that were nasally immunized managed to reduce the immune activation detected at the peak of viremia to levels similar to those observed in the oral group during the chronic phase of the infection (Fig. 5d).

Virus replication is known to drive the level of immune activation, and both play an important role in CD4⁺ T-cell depletion and viral pathogenesis. Since in previous trials we and others observed a correlation between level of viremia, immune activation, and disease progression (20, 27), we evaluated statistically the extent of these correlations in the animals in this study and found that immune activation of CD8⁺ C_M T cells positively correlated with viral loads 16 weeks postinfection and inversely with CD4⁺ C_M levels measured on week 16 postinfection and length of survival (Fig. 5e). Therefore, affecting viremia and immune activation by vaccination will result in prolonged survival. These data also support the fact that measurements of viral loads, CD4⁺ C_M levels, and immune activation each are independently sufficient to predict outcome, and that the simplest of all, evaluation of viremia, may be by itself a sufficient and significant marker to evaluate vaccine efficacy in clinical trials.

Role of the anti-SIV CD8⁺ T-cell responses in nonviremic animals.

CD8⁺ T-cell responses have been directly implicated in the control of viral replication in humans and in the RM model (31–33). Nine of the infected animals, seven vaccinated and two control macaques, fully suppressed viremia after SIV was detected in the blood at one or more time points. We hypothesized that the immune responses present in these macaques played a role in controlling viral replication to undetectable levels or achieved clearance of the infection. The latter possibility is less likely, since viral clearance after detection of systemic viremia has been infrequently reported (50). Instead, CD8⁺ T-cell responses have been directly implicated in the control of viral replication in humans and in the RM model. In this model, a rebound of suppressed viremia was observed in blood of SIV-infected animals after treatment with an anti-CD8 monoclonal antibody that depletes CD8⁺ T cells (48). To evaluate whether the undetectable viremia observed in our animals depended on SIV-specific CD8⁺ T-cell control or was an indication of viral clearance, macaques were depleted of the CD8⁺ T-cell population and viral loads were monitored. We found that while no rebound was observed in the vaccinated animals, viremia rebounded in the two naive controls (Fig. 6).

FIG 6.

CD8⁺ T-cell depletion and SIV detection in serum and tissues. (a) Percentage of CD8⁺ T cells during and after CD8⁺ T cell depletion in PBMC of three animals (numbers 148, 246, and 249) that resisted 32 nontraumatic and two traumatic challenges and nine animals that had controlled viremia to undetectable levels between 4 and 28 weeks postinfection (numbers 255, 253, 153, 155, 150, 250, 147, 175, and 301). The x axis represents days since first anti-CD8 antibody injection (day 1). Positive RT-PCR time points are indicated for the two animals (numbers 301 and 175) that became viremic after CD8⁺ T-cell depletion. SIV copies/ml were 48,000, 56,000, and 45,000 for animal 175 and 12,000, 7,700, and 280,000 for animal 301. All of the other animals were consistently negative during the time course. (b) Detection of SIV gag DNA in mesenteric lymph nodes, tonsils, and spleen and in jejunum, ileum, and colon of 4 nonviremic animals, two vaccinated nasally, one orally, and one intestinally, and two viremic animals, one orally and one intestinally vaccinated. The color for each animal is according to its vaccine group, as coded before. DNA copy numbers were normalized according to the detection of IL-15 copies in the same amount of DNA. Each tissue symbol represents averages from multiple fragments of tissue or lymph nodes, varying from 3 to 5 for each. (c) Detection of SIV Gag RNA in mesenteric lymph nodes (LN) and PBMC of nonviremic and viremic animals. RNA copy numbers were normalized according to the detection of 18S RNA in the same amount of RNA for each sample. Only samples from one viremic animal were tested.

To further evaluate whether residual SIV-specific, CD8⁺ T-cell activity present in tissues after antibody-mediated CD8⁺ T-cell depletion was responsible for the absence of viremia rebound in the vaccinated animals, we evaluated DNA and RNA copy numbers in tissues obtained at necropsy from four of the animals that remained nonviremic (numbers 246, 153, 250, and 150). Positive controls were provided by tissues of two animals that were viremic (numbers 247 and 260), with 400,000 and 260,000 copies of SIV/ml on last measurement, 13 and 6 weeks before euthanasia due to AIDS, respectively. Ten independent reactions were carried out in each DNA sample and 5 in each RNA sample (7). A significantly lower number of copies was detected in the gastrointestinal tissues of nonviremic animals than in the viremic ones (Fig. 6b). Two animals, one vaccinated orally and one intestinally, had detectable copies in their lymph nodes, although at a significantly lower rate (more than 2 logs lower) than the viremic animals. These copies could be due to survival of cells infected at the time of viremia that subsequently became quiescent, avoiding immunological clearance. RNA analysis revealed no significant detection of SIV expression in the lymph nodes and PBMC of the same animals (Fig. 6c). These results support the fact that the absence of viremia in the vaccinated animals was unlikely to be due to CD8⁺ T-cell-mediated viremia control and that, more likely, clearance of infection in the vaccinated, infected animals had occurred. Even if one considered the issue of occurrence of complete clearance questionable and postulated the persistence of some residual infectivity that became detectable over a longer period of time if the animals were not euthanized, it appears that the control of virus replication achieved in these animals is clearly significantly higher and qualitatively different from that observed in the nonvaccinated animals that became viremic after CD8⁺ T cell depletion or that observed with another vaccine candidate.

Exposure to SIV during multiple low-dose challenges and stimulation of cell-mediated immunity.

As seven vaccinated animals remained RT-PCR negative after 32 low-dose vaginal challenges with SIV_mac251, we questioned whether virus exposure during challenges could have expanded SIV-specific preexisting immune responses. To address this question, immune responses evaluated before the first and after 32 challenges were compared in vaginal MNC. Higher SIV-specific CD4⁺ and CD8⁺ T-cell responses were detected in the majority of the seven vaccinated animals, indicating that a challenge-mediated stimulation of virus-specific immunity did occur (Fig. 7). A small increase in the anti-Gag CD4⁺ T-cell responses was also observed in the nonvaccinated animal (number 180), which also resisted 32 challenges. Anti-Gag cell-mediated responses expanded more efficiently than anti-Env responses, with an increase of anti-Gag responses in the CD4⁺ and CD8⁺ T-cell populations of six of the seven vaccinated animals. Higher levels of Env-specific CD4⁺ or CD8⁺ T cells were observed in three of the seven animals (Fig. 7a). Anti-Env IgA antibodies in vaginal and rectal secretions were also expanded in some animals, and this was more significant for the rectal secretions (Fig. 7b). Increased IgA levels in rectal secretions may result from recirculation of IgA-producing cells between the two compartments or very short-lived virus dissemination to the intestine. No anti-SIV IgG was detected in serum or secretions of these animals. We conclude that repeated low-dose vaginal challenge can expand antiviral responses. This finding should not be viewed as a confounding variable in the evaluation of vaccine efficacy, as this event can also occur in humans in clinical trials, cannot be easily controlled for, and can affect the results of the trials.

FIG 7.

Virus-specific immune responses stimulated during multiple low-dose challenges. (a) Total anti-SIV CD4⁺ and CD8⁺ Gag or Env responses detected before challenge (open bars) or after 32 nontraumatic vaginal exposures to SIV_mac251 (closed bars) in vaginal MNC of animals that remained RT-PCR negative after 32 nontraumatic challenges. (b) Anti-SIV gp140 IgA-specific activity (ng gp140-specific IgA per μg total IgA) detected before challenge (open symbols) or after 32 nontraumatic vaginal exposures to SIV_mac251 (closed symbols) in rectal and vaginal secretions of animals that remained RT-PCR negative after 32 nontraumatic challenges. The SIV antibody concentration in secretions is expressed as specific activity (ng SIV-specific IgA per microgram of total IgA). Asterisks indicate animals that resisted two traumatic challenges that were administered after the sampling reported here.

DISCUSSION

Vaccine-induced protection against many pathogens that infect mucosal surfaces often relies on the generation of mucosal immunity. The superior effectiveness of the oral polio vaccine compared to the i.m. vaccine is an example, although the oral polio virus vaccine is also made of a live-attenuated virus, while killed poliovirus is used in the i.m. vaccine. Mucosal delivery of a vaccine stimulates more significant immune responses at different mucosal sites than systemic delivery (9, 18–20, 22, 27, 49). Exploring mucosal delivery of vaccine candidates might be particularly important in the case of protective immunity against HIV infection, where infection clearance is difficult once the infection has become systemic.

To elucidate which mucosal immunization is the most effective to stimulate protective virus-specific responses at sites of HIV-1 entry and replication, we compared the effectiveness of the immunity stimulated by four different mucosal immunization routes using a vaccine composed of an SIV DNA plasmid that produces noninfectious SIV particles in combination with rMVA and inactivated SIV particles (22) in preventing SIV infection or delaying the occurrence of AIDS after repeated, low-dose SIV_mac251 vaginal challenge.

Systemic infection, as indicated by a positive serum RT-PCR, did not occur in more than 50% of the intestinally vaccinated animals despite 32 nontraumatic vaginal challenges. This was true for 14 to 15% of the animals in the other groups, and the difference between median numbers of challenges required in the intestinal group and those of the others was statistically significant. Anti-SIV CD8⁺ T-cell responses in vaginal MNC before the first challenge in the animals that resisted 32 challenges appeared to be critical for protection from infection, regardless of the vaccination group. Antibodies did not appear to be involved in protection from infection in this group, as their levels were low and sporadic after immunization and had declined to undetectable levels by the time challenges were initiated (22). This result does not exclude the role of antibodies in protection from infection but simply supports the fact that cell-mediated immunity can also play a role in virus clearance at the time of exposure. It will be interesting to investigate whether achieving higher levels of antibody with a higher dose of particle or Env protein immunization will further increase the rate of protection from infection. Traumatic exposure produced infection in some of these animals, but in three, RT-PCR remained negative even after two traumatic challenges. Given that the amount of virus used for each challenge is likely to be higher than that present in human secretions during natural exposure, it is possible that the immunity generated by this vaccine can achieve even better protection if tested in clinical trials.

When we questioned whether the repeated exposure to SIV_mac251 was able to stimulate virus-specific immune responses at the site of challenge, we found that this was true in many animals. The fact that a very small amount of viral protein was present in the challenge inoculum, that there was no adjuvant administered with it, and that primary anti-SIV CD4⁺ T-cell responses could be stimulated in one of the naive animals that had no preexisting anti-SIV immunity suggests that a short-lived mucosal infection that did not become chronic, and therefore was not detected by RT-PCR, occurred after some of the challenges and produced enough viral antigen to initiate or restimulate the immune response against SIV. Data of exposed seronegative individuals that are RT-PCR negative in blood but positive for anti-HIV IgA and cell-mediated responses suggest that a contained infection happened in these individuals and could happen during human trials.

None of the orally vaccinated animals and one of those nasally vaccinated developed AIDS during the course of the trial. The oral and nasal vaccinations appeared better suited than the intestinal and vaginal vaccinations for inducing virus-specific responses that, although not sufficient for blocking virus entry, could control viral replication and delay disease. The fact that greater systemic T-cell and rectal IgA responses were detected in these groups appeared to be important in this protection. As the role of nonneutralizing antibodies with virus entry-blocking activity seems to gain importance in light of the RV144 trial results and the SIV-adenovirus studies, it would be interesting to evaluate whether the antibodies stimulated by the vaccine evaluated here have this activity and whether these responses correlated with the protection of infection achieved with this vaccine.

Discordant results between protection from infection and from disease progression were observed in the animals intestinally vaccinated, as the immune responses were efficient for preventing infection but did not control virus replication or progression to AIDS. One possible explanation of these outcomes is that the intestinal vaccination stimulates qualitatively different anti-SIV CD4⁺ and CD8⁺ T cells in the vaginal, intestinal, and systemic compartments, with the first being efficient at blocking the small viral amount at the exposure site and the others being insufficient to contain the infection when it spreads to the intestinal tract and periphery. Indeed, a significance between levels of CD8⁺/IFN-γ⁺ T cells in the vaginal mucosa and number of resisted challenges compared to those of animals that resisted fewer challenges was found (Fig. 1c). The same responses were at the lowest level compared to the other vaccinated groups when analyzed in PBMC. Considering the timing of vector vaccination (weeks 1 to 33) and the beginning and duration of the challenge (week 57 plus 32 additional weeks), it is unlikely that innate responses stimulated at the intestinal level played a role in protection at the vaginal level, as they tend to be local at the site of immunization and limited in time. Vaginal vaccination, which is most likely to produce innate responses in the vaginal mucosa, did not result in protection. Measurements of DNA viral loads in the intestinal mucosa after infection, which were not planned for this study, could shed light on what is occurring at this site early after infection.

One of the most outstanding results achieved with this vaccine regimen was the complete suppression of viral replication observed in 25% (7/28) of the vaccinated animals. Only a brief transient systemic viremia after infection, which was rapidly controlled and remained undetectable even after CD8⁺ T cell depletion, could be detected in these animals. A similar short-lived viremia in human vaccine recipients could easily be missed, as it is unlikely that the follow-up in humans is as intensive as it is in experimental animals. This result, in combination with the tissue analysis at necropsy after CD8⁺ T-cell depletion, strongly supports viral clearance or substantial reduction of viral burden to a very significant degree and suggests that if an appropriate immune response is preexisting, it is possible to gain full control of virus replication before the infection becomes chronic and latency is established. Interestingly, complete suppression of viremia after small spikes of viral loads and lack of viral rebound after CD8⁺ T-cell depletion has also been observed with an RhCMV vaccine, confirming that clearance of SIV infection can be achieved with an appropriate vaccine regimen (7, 17, 50). In this case, effector responses against epitopes different from the immunodominant epitopes recognized by the responses stimulated by SIV infection were detected after the immunization (51). The very small amounts of cells obtained from biopsy specimens prevented more-detailed mechanistic studies to identify features of protective mucosal responses, such as analysis of specific epitope recognition and type of memory (central or effector). Different approaches not planned for this study, such as surgical removal of lymph nodes and fragments of intestine or collection of large amounts of tissues at euthanasia after vaccination, are required to carry out these investigations.

Based on the efficacy of protection observed in this trial by the different immunization routes, the oral route appears the more suitable candidate for vaccine delivery. Protection by the oral immunization route was similar to that observed after nasal vaccination. However, nasal vaccination is generally less safe, as the accessibility of vaccine components to the brain from the nasal cavity can result in undesirable side effects. Further improvement and a more thorough understanding of the immune responses generated by the approach here, with achievement of better antibody responses, should additionally increase the efficacy of this vaccine.