To evaluate vaccine protection in rhesus monkeys against repeated, mucosal challenge with a pathogenic SIV isolate, we immunized Indian-origin rhesus monkeys with plasmid DNA (4 mg per construct, intra-muscularly, at weeks 0, 4, and 8) and rAd5 vectors (1 × 1011 particles per construct, intramuscularly, at week 26) using vectors expressing SIVmac239 env and gag-pol. To assess vaccine protection against different SIV isolates, we challenged monkeys about 4 months after the rAd5 boost with a quasi-species of either the closely related SIV-mac251 or the genetically distant SIVsmE660.
The mucosal virus challenges were carried out weekly for 12 consecutive weeks by intrarectal inoculations with one AID50
(median animal infectious dose) of virus challenge stock. Plasma SIV RNA levels were assessed each week, and monkeys that had detectable plasma virus were excluded from subsequent virus challenges. Three parallel studies were performed: (i) a two-arm experiment using Mamu-A*01
–, and Mamu-B*17
–negative Indian-origin rhesus monkeys with an SIVmac251 challenge (n
= 40) [these monkeys did not express major histocompatibility complex (MHC) class I alleles that are associated with the development of potent cytotoxic T lymphocyte responses that confer protection against rapid SIV replication]; (ii) a two-arm experiment using Mamu-A*01
–, and Mamu-B*17
–negative Indian-origin rhesus monkeys with an SIVsmE660 challenge (n
= 50); and (iii) a two-arm experiment using Mamu-A*01
–positive Indian-origin rhesus monkeys, with an SIVsmE660 challenge (n
= 39). The control arm of each of these experiments was composed of sham-vaccinated animals. The conditions used in these virus challenge studies provided an 85% power to detect a 50% reduction in the per-challenge risk of virus acquisition for the 50-monkey experiment and about a 75% power to detect this effect size for the 40-and 39-monkey experiments (6
The endpoints in each of these experiments were acquisition of infection and peak plasma virus RNA concentrations. In the first experiment, vaccination had no impact on acquisition of infection with SIVmac251, a virus that is very close in genetic sequence to the virus used in the construction of the vaccine immunogens (). Nevertheless, the experimentally vaccinated group of monkeys had about a one-log10 lower peak plasma virus RNA level than the group of sham-vaccinated monkeys ().
Fig. 1 Protection against SIV acquisition and peak plasma virus RNA concentrations after infection. Low-dose virus challenges involved weekly intrarectal inoculations with one AID50 of virus challenge stock for 12 consecutive weeks. Plasma SIV RNA concentrations (more ...)
In the second experiment, after 12 successive challenges, experimentally vaccinated monkeys had about a 50% lower rate of infection with SIVsmE660 than the sham-vaccinated monkeys (48% versus 88%, P
= 0.001; ). This challenge virus is as distant genetically from the sequence of the virus used in the construction of the vaccine immunogens as two of the most genetically disparate clade-identical HIV-1 isolates (7
). Vaccination of these Mamu-A*01
–negative rhesus monkeys had no effect on peak plasma virus RNA concentrations after infection (). In the third experiment in Mamu-A*01
–positive rhesus monkeys, experimentally vaccinated monkeys also had about a 50% lower rate of infection with SIVsmE660 than the sham-vaccinated monkeys (). In addition, these experimentally vaccinated monkeys had about a one-log10
lower peak plasma virus RNA concentration than the group of sham-vaccinated monkeys (37% versus 75%, P
= 0.009; ). When the results of the second and third vaccine experiment were combined for analysis, the statistical significance of the difference in SIVsmE660 acquisition between experimentally vaccinated and sham-vaccinated monkeys was significant ( = 4 × 10−5
with a log-rank test). Therefore, these experiments demonstrated protection against acquisition of the SIVsmE660 isolate but not the SIVmac251 isolate. Moreover, they showed that the effects of vaccination on virus acquisition and virus load after infection are distinct, but that both effects can be manifest in the same group of vaccinated challenged monkeys. No effect of the vaccine was apparent on set-point plasma virus RNA concentration or loss of CD4+
T lymphocytes after SIV challenge in these cohorts of monkeys (figs. S1
The availability of these experimentally vaccinated monkeys in which about 50% were protected from SIVsmE660 infection provided an opportunity to explore the immune correlates associated with protection against acquisition of the virus. To examine possible correlates without imposing a bias on the findings, we evaluated vaccine-elicited cellular immune responses, innate responses, antibody responses, and host genetic factors. These studies were done on both the Mamu-A*01–negative and the Mamu-A*01–positive animals challenged with SIVsmE660. Given that these studies were done to define immune and genetic determinants of vaccine protection, only the experimentally vaccinated monkeys were evaluated. These vaccinated monkeys were divided into two groups: those that were protected and those that became infected.
SIV-specific cellular immune responses in the experimentally vaccinated monkeys were evaluated by interferon-γ (IFN-γ) enzyme-linked immunospot (ELISpot) and intracellular cytokine staining; total CD4+
T cell populations were characterized by phenotypic profiling. Unfractionated peripheral blood mononuclear cells (PBMCs) evaluated in a pooled peptide ELISpot assay for SIVmac239 Env-, Gag-, and Pol-specific cellular immune responses on the day of challenge demonstrated no differences between the infected and the uninfected monkeys (). PBMCs sampled at week 46, 16 weeks after the rAd boost, were also evaluated by intracellular cytokine staining after pooled peptide stimulation for SIV Env- and Gag-specific T cell responses. These studies were performed with both SIVmac239 and SIVsmE543 peptide pools. No statistically significant differences were noted between the infected and the uninfected monkeys in the magnitude of either the CD4+
or the CD8+
T cell responses (). An evaluation of the polyfunctionality of these cellular responses also did not differentiate between the monkeys that became infected and the monkeys that did not become infected (). Finally, a phenotypic characterization of the total CD4+
for expression of cell surface molecules associated with memory did not differentiate the monkeys that became infected from the monkeys that did not become infected (fig. S3
Fig. 2 Vaccine-induced pooled peptide ELISpot responses to SIV proteins. Monkey PBMCs were assessed in a pooled peptide ELISpot assay for SIVmac239 Env-, Gag-, and Pol-specific cellular immune responses on the day of challenge. Each plotted data point represents (more ...)
Fig. 3 Vaccine-induced virus-specific cellular immune responses. (A and B) PBMCs isolated from Mamu-A*01–negative (A) and Mamu-A*01–positive (B) monkeys after the boost immunization were exposed to pools of overlapping peptides spanning the Gag (more ...)
Mediators of innate immune responses were also evaluated in these monkeys 4 weeks after the rAd boost. Natural killer (NK) cell activity, assessed by determining CD3−
PBMC expression of molecules associated with NK cell activation and function, was comparable in the monkeys that did and did not become infected (fig. S4
). Further, Luminex technology was used to assess plasma concentrations of a diversity of cytokines 4 weeks after the rAd boost. No differences were documented in the patterns of cytokine expression in the vaccinated monkeys that became infected and in the vaccinated monkeys that did not become infected (fig. S5
). The anti-Env antibodies generated by these vaccinated monkeys were assessed with binding, antibody-dependent cellular cytotoxicity (ADCC), and neutralization assays. To evaluate serum binding to the challenge virus, we determined ED50
(median effective dose) values for serum binding to SIVsmE660 gp140 using serum sampled from both Mamu-A*01
–positive and Mamu-A*01
–negative monkeys that were challenged with SIVsmE660 at both week 34 and day of challenge time points. No statistically significant differences in these values were observed between the vaccinated monkeys that became infected and those that did not become infected (). The avidity of the anti-Env antibodies in these two groups of monkeys, measured by sodium thiocyanate (NaSCN) sensitivity of antibody binding to Env protein, did not differ (). Further, ADCC activity, as measured with recombinant Env protein–pulsed target cells, was not different in these two groups of animals ().
Fig. 4 Vaccine-induced antibody binding to SIVsmE660. (A) Sera sampled 2 weeks after the rAd5 boost (week 34) and from the day of first SIVsmE660 viral challenge (week 53) were assayed by ELISA for the level of antibody binding to SIVsmE660 gp140. Values are (more ...)
Fig. 5 Vaccine-induced antibody-dependent cell-mediated cytotoxicity responses to SIVmac251gp120. CD3−CD20−CD56+ human PBMCs (NK cells) were incubated with SIVmac251gp120-coated CEM-NKr-CCR5 target cells and plasma sampled from the monkeys on (more ...)
Before initiating studies of vaccine-induced neutralizing antibody responses in these vaccinated monkeys, we assessed the neutralization sensitivity of the challenge viruses. Although the neutralization resistance of the SIVmac251 challenge stock has been well documented, the neutralization phenotype of the SIVsmE660 stock had not been well characterized. To evaluate the neutralization sensitivity of this SIVsmE660 virus stock, we assayed pseudovirions generated from Env proteins of selected members of the virus quasi-species in TZM-bl cells, and we also assayed the challenge stock itself in TZM-bl and rhesus monkey PBMCs. Although some of the SIVsmE660 Env-based pseudovirions were easy to neutralize (fig. S6A
), the CR54-PK-2A5 pseudovirion displayed a more difficult to neutralize (tier 2–like) phenotype (fig. S6B
). The challenge stock virus quasi-species was easy to neutralize in TZM-bl cells (fig. S7A
), but was tier 2–like in rhesus monkey PBMCs (fig. S7B
). Therefore, this SIVsmE660 challenge virus exhibited intermediate sensitivity to neutralization compared to the resistance to neutralization of most SIVmac239/251 stocks.
To evaluate serum neutralization in pseudovirion-based assays, we generated pseudovirions using two different transmitted/founder env genes selected from a separate cohort of monkeys that were infected intrarectally; one pseudovirion had a tier 1– and the other a tier 2–like phenotype (). Assays were performed to assess neutralization by serum sampled from these monkeys at peak immunity and day of challenge time points. Comparable neutralization of the tier 1–like pseudovirion was demonstrated by serum from the vaccinated monkeys that became infected and those that did not become infected (). The sera sampled on the day of challenge from these two groups of monkeys also mediated comparable neutralization of the tier 2–like pseudovirion when assessed by determining ID50 (50% inhibitory dilution) values. However, a significantly lower titer-neutralizing antibody response was measured in the day-of-challenge serum of the Mamu-A*01–negative monkeys that became infected when assessed with the tier 2–like pseudovirion and expressed as percent neutralization with a 1:10 serum dilution (P = 0.03) ().
Fig. 6 Neutralization of tiers 1 and 2 SIVsmE660 Env pseudoviruses. (A to D) Serum samples were obtained from Mamu-A*01–negative (A and B) or Mamu-A*01–positive (C and D) monkeys 2 weeks after rAd boost immunization or on the day of challenge (more ...)
A 1:50 dilution of plasma sampled from these monkeys at the time of the peak vaccine-elicited immuneresponse was assessed in a neutralization assay with human PBMCs as target cells and the challenge stock of SIVsmE660 as the replicating virus. A statistically significant decrease in virus replication was observed with plasma from Mamu-A*01–negative vaccinated monkeys that became infected compared with plasma from Mamu-A*01–negative vaccinated monkeys that did not become infected (P = 0.008)().Although a similar analysis of the plasmas sampled on the day of challenge did not show a statistically significant difference in neutralization between these groups of monkeys in this PBMC-based assay, the relative levels of neutralizing antibodies in the plasma of each monkey in this assay were moderately correlated with those from week 38 (estimated Spearman correlation = 0.40; P = 0.05). No correlate of protection against acquisition of SIVsmE660 was observed when the vaccine-elicited immune responses in the Mamu-A*01–positive rhesus monkeys were assessed in either the pseudovirion- or the PBMC target cell–based neutralization assay.
Fig. 7 Neutralization of SIVsmE660 in human PBMCs. CD8+ T cell–depleted, concanavalin A–stimulated human PBMCs were infected at a low MOI with SIVsmE660. Infected PBMCs were subsequently cultured in the presence of a 1:50 dilution of plasma collected (more ...)
We also evaluated genetic determinants that might have contributed to the protection from infection observed in the monkeys. We have recently shown that the expression of particular TRIM5
alleles (encoding the host restriction factor TRIM5) by rhesus monkeys is associated with control of SIVmac251 replication both in vitro and in vivo (8
). Two alleles are codominantly expressed at the TRIM5
locus of the rhesus monkey. If one or both of those alleles are from the group of alleles 6 to 11 (permissive), SIV replication occurs at a higher level than if both are from the group of alleles 1 to 5 (restrictive). To evaluate the contribution of the expression of particular TRIM5
alleles to protection against infection, we retrospectively determined the TRIM5
genotypes for all of the monkeys in these studies (tables S2
). We separated the common variants of TRIM5
into two groups: (i) restrictive (1-5/1-5
) and (ii) permissive (1-5/6-11
, or 6-11/TRIMCypA
). We excluded monkeys that had two TRIMcypA
) from this analysis because TRIM5
alleles were deleted in these monkeys (3 of 89 monkeys).
When the sham-immunized monkeys were evaluated together as a single group regardless of their Mamu-A*01 status, monkeys with at least one permissive TRIM5 allele were more likely to acquire SIVsmE660 infection than those monkeys expressing restrictive alleles () (P = 0.0023). Similarly, when vaccinated monkeys (Mamu-A*01− and Mamu-A*01+) that were challenged with SIVsmE660 were assessed as one cohort, we observed a statistically significant association between the expression of two restrictive alleles and protection against virus acquisition () (P = 0.0013). Therefore, the expression of TRIM5 alleles 1 to 5 was associated with protection from infection after low-dose SIVsmE660 intrarectal exposures in rhesus monkeys regardless of vaccination status. Because this was a retrospective analysis, there were not enough monkeys that were homozygous for TRIM5 alleles 6 to 11 to differentiate between the effect of having one versus two permissive alleles on mucosal acquisition of infection. A protective effect of the vaccine was observed in vaccinated monkeys regardless of their TRIM5 genotypes ().
Fig. 8 Effect of TRIM5 genotype on SIV mucosal infection in naïve and vaccinated monkeys. The effect of restrictive or permissive TRIM5 alleles on the percentage of monkeys that remained un-infected after each weekly intrarectal SIVsmE660 exposure is (more ...)
Fig. 9 Effect of vaccination on SIV mucosal infection. The effect of vaccination on the percentage of monkeys with either restrictive or permissive TRIM5 genotypes that remain uninfected after each weekly intrarectal SIVsmE660 exposure is shown in Kaplan-Meier (more ...)
Finally, using a logistic regression analysis, we assessed the relative contributions of vaccine-elicited immune responses and genetic determinants of protection from infection in vaccinated monkeys (details of analysis in the Supplemental Material
). SIVsmE660 infection in vaccinated monkeys was the outcome variable used in the logistic regression models. In the Mamu-A*01
–positive animal study, the permissive TRIM5
phenotype was the most significant covariate (; P
= 0.008, estimated odds ratio (OR) of infection = 18). Once the permissive TRIM5
phenotype was in the model, no other covariate added significantly to the model. In the Mamu-A*01
–negative animal study, the first covariate added was lack of SIVsmE660 neutralization as measured in PBMCs at week 38, the time of peak vaccine-induced immune responses (P
= 0.028, OR = 18.6). The second covariate added was low SIVsmE660 Env-specific CD4+
T cell responses (P
= 0.032, OR = 9.8). No other covariates were subsequently significant. Therefore, low levels of virus neutralization and low Env-specific CD4+
T cell responses were strong predictors of infection in the Mamu-A*01
–negative animals. When both the Mamu-A*01
–positive and the Mamu-A*01
–negative monkeys were grouped together for analysis, the first covariate added to the model (P
= 0.005, OR = 31.7) was lack of restrictive TRIM5
alleles. Other statistically significant covariates included Env-specific CD4+
T cell responses and neutralizing antibody responses as measured with the pseudovirion/TZM-bl cell assay ().
Best models for protection and infection after vaccination. OR, odds ratio.