|Home | About | Journals | Submit | Contact Us | Français|
Prime-boost immunization with heterologous vaccines elicits potent cellular immunity. Here, we assessed the influence of various TLR ligands on SIV Gag-specific T cell immunity and protection following prime-boost immunization. Rhesus macaques (RM) were primed with SIV Gag protein emulsified in montanide ISA51 with or without TLR3 (polyinosinic-polycytidylic acid (Poly IC)), TLR4 (monophosphoryl lipid A (MPL)), TLR7/8, TLR9 (CpG) or TLR3 (Poly IC) combined with TLR7/8 ligands, then boosted with replication defective adenovirus 5 expressing SIV Gag (rAd5-Gag). After priming, RM that received SIV Gag protein plus Poly IC developed significantly higher frequencies of SIV Gag-specific CD4+ Th1 responses in blood and bronchoalveolar lavage fluid lymphocytes (BAL) compared to all other adjuvants, and low-level SIV Gag-specific CD8+ T cell responses. After the rAd5-Gag boost, the magnitude and breadth of SIV Gag-specific CD8+ T cell responses were significantly increased in RM primed with SIV Gag protein plus Poly IC, with or without the TLR7/8 ligand, or CpG. However, the anamnestic, SIV Gag-specific CD8+ T cell response to SIVmac251 challenge was not significantly enhanced by SIV Gag protein priming with any of the adjuvants. In contrast, the anamnestic SIV Gag-specific CD4+ T cell response in BAL was enhanced by SIV Gag protein priming with Poly IC or CpG, which correlated with partial control of early viral replication after SIVmac251 challenge. These results demonstrate that prime-boost vaccination with SIV Gag protein/Poly IC improves magnitude, breadth, and durability of CD4+ T cell immune responses, which may have a role in control of SIV viral replication.
Induction of durable humoral and/or cellular immunity will be critical for an effective vaccine against HIV, malaria and tuberculosis (TB). Accordingly, heterologous prime-boost immunization with DNA, protein and viral vaccines in various combinations elicit potent adaptive immunity sufficient to confer varying levels of protection in pre-clinical and human efficacy trials for SIV and HIV respectively [1–4]. Amongst these vaccine platforms, DNA and especially viral-based vectors are typically the most potent and efficient for inducing CD8+ T cells, whereas protein vaccines elicit predominantly CD4+ T cells and antibody responses. However, when used in heterologous, prime-boost combination each different component of a vaccine can make a unique contribution to both cellular and humoral immunogenicity, and the overall immunogenicity of a prime-boost vaccine will depend on both the specific nature of each component and the interactions between these components . Thus, protein vaccines, which have the key advantages in terms of safety and ease of manufacture , might be designed to contribute to both humoral and cellular immunogenicity in an optimized prime-boost vaccine.
The formulation of protein-based vaccines influences magnitude and quality of antibody and T cell responses elicited by protein vaccines . First, proteins can be administered as short or long peptides, full-length proteins or particles, which can lead to differences in the potency and breadth of humoral and cellular immunity . Second, proteins can be formulated with alum, oil/water and water/oil emulsions, liposomes or nanoparticles that can act through various innate signaling pathways as well as provide improved delivery to antigen presenting cells or more prolonged antigen presentation . Finally, proteins administered with immune adjuvants that target distinct innate pathways such as TLR, nod-like receptors or retinoic acid inducible gene I can alter potency and lead to the differentiation of distinct functional (Th1, Th2, Th17) CD4+ T cell responses and improve cross-presentation [7, 10, 11]. Indeed, adjuvants that induce IL-12 and/or Type I IFN from dendritic cells (DCs) would be critical for generating Th1 immunity and CD8+ T cells [12–14]. Moreover, combining adjuvants that target distinct innate pathways such as MYD 88 and TRIF have been shown to induce strong innate cytokine production such as IL-12 in vitro from human DC  as well as improve humoral immunity in vivo 16].
As the innate responses based on TLR receptor distribution in non-human primates (NHP) are more similar to humans than rodents, we and others have used NHP to explore how different adjuvants that activate distinct innate pathways influence the induction of T cell responses and their ability to prime responses for a heterologous viral boost [17, 18]. For example, NHP immunized with HIV Gag plus montanide ISA51 and CpG or a TLR7/8 ligand elicited higher CD4+/IFN-γ-producing cells than HIV Gag montanide ISA51 with or without a TLR8 ligand. As plasmacytoid DCs express TLR7 and TLR9 but not TLR8, these data suggest that Type I IFN induced by TLR7/8 ligand or CpG, respectively was critical for enhancing the induction of IFN-γ from CD4+ T cells . Indeed, robust T cell immunity was observed in a study in which HIV Env protein was administered with CpG and ISCOMs . More recently, NHP immunized with a malarial antigen and a TLR4 ligand admixed with oil/water emulsion elicited CD4+/IL-2 but low-level IFN-γ-producing T cells . Although TLR4 and TLR8 ligands can activate monocytes and mDCs to secrete IL-12, they are relatively limited in their induction of Type I IFN compared to production by pDCs by TLR7 or TLR9 ligand stimulation [21–23]. Taken together, these data highlight the potential importance of Type I IFN for optimizing Th1 immunity and eliciting a low frequency of CD8+ T cells in NHP through cross-presentation. In this regard, we recently showed that robust Th1 responses and a low frequency of CD8+ T cell responses can be elicited in NHP using Poly ICLC as an adjuvant with protein or dendritic cell targeted protein vaccines .
In this report, we set out to establish a hierarchy of the potency of various TLR ligands on T cell immunity when admixed in a water/oil emulsion. We compared TLR ligands for intracellular receptors TLR3 (Poly IC), TLR7/8 (3M-012), or TLR9 (CpG) with the TLR4 ligand, MPL which is expressed on the surface of antigen presenting cells for their ability to prime T cell responses in NHP using SIV Gag protein as an immunogen. In addition, we assessed whether combining Poly IC with a TLR7/8 ligand had any additive or synergistic effects on T cell immunity [15, 16]. Lastly, to determine whether T cell responses after prime-boost immunization were protective, all animals were infected with SIVmac251. Taken together, the data show that SIV Gag/montanide ISA51 plus Poly IC is the most potent adjuvant for priming Th1 immunity and eliciting low-level CD8+ T cells and conferring some protection against viral challenge.
Rhesus macaques (RM) (Macaca mulatta) used in this study were housed at the Oregon National Primate Research Center in accordance with the guidelines of the Institutional Animal Care and Use Committee for Oregon Health & Science University/Oregon National Primate Research Center and the Guide for the Care and Use of Laboratory Animals (Institute for Laboratory Animal Research, 1996). Animals that developed disease states that were not manageable were euthanized according to the recommendation of the American Veterinary Medical Association (AVMA) Guidelines on Euthanasia.
SIV Gag protein, TLR ligands, 1X PBS (Invitrogen) and glycine buffers were kept on ice until immediately before emulsifying. ISA-51 VG was sterile filtered prior to use. A 2X stock of the aqueous portion of each formulation was prepared on ice by combining appropriate amounts of antigen, ligand(s) and buffer stocks in a 15 mL falcon tube. The ISA 51 VG was added to the tube in a 1:1 (v/v) ratio. This mixture then emulsified by passage between two sterile 5 mL or 10 mL Gastight glass syringes (Hamilton Co.) using a 24 gauge 1.5" sterile connector (Hamilton Co.) for 5–15 minutes.
Samples of each emulsion were analyzed by a drop test in distilled water, and for particle size distribution by laser diffraction using a Malvern Mastersizer 2000 (Malvern Instruments Inc., Southborough, MA). Drop tests were performed immediately post emulsification by adding ~20–50 µL of sample to distilled water. Formulations were considered stable if no dispersion was seen in the water after 3 minutes. Emulsions were then analyzed for particle size distribution within 1 hour of emulsifying by dispersing 150–200uL of sample in Marcol 5 paraffin oil (Exxon Co., USA, Houston, TX) and 2% (v/v) montanide 80 surfactant (SEPPIC, Inc.). A second sample of each formulation was analyzed on the day of injection (either 1–2 days post-emulsifying) to ensure emulsion stability. The average particle size for these formulations was 1.05 um +/− 0.5 um, with 90% of the sizes falling below 1.91 um. All formulations were kept refrigerated until use.
Animals (n=6/group) were immunized twice with SIV Gag protein (first dose: 400 µg, second dose: 300 µg) with 1 mg of poly IC (Invitrogen), MPL (Invitrogen), 3M-012 (3M Pharmaceuticals) and CpG (Coley Pharmaceutical Group), alone or in combination, emulsified in montanide (water-in-oil form) in a total volume of 1.2 ml. As controls, 6 RM were immunized with SIV Gag alone in montanide. Vaccines were formulated as described above . The 1st vaccination was given in three separate sites (two arms and the right leg, 0.4 ml each) subcutaneously (s.c.), and the 2nd vaccination was given in two different sites (0.6ml each) of the back (s.c.). Ten weeks after the 2nd vaccination, animals were boosted with 1010 particles of rAd5-Gag intramuscularly (i.m.) in a volume of 1 ml of PBS into the right thigh. In addition, there were 8 animals that received rAd5-Gag only with no primary immunization with SIV Gag protein. The rAd5-Gag (E1/E3-deleted Ad5/SIVmac239 Gag) vector was generated in vitro by Cre/LoxP site-specific recombination system described by Aoki et al. . Briefly, a plasmid pVRC5404 containing the SIVmac239 Gag-Pol fusion was obtained from the Vaccine Research Center at the NIH. The full length SIV Gag sequence was cloned from the plasmid by deleting the Pol sequence by an enzyme digestion with XbaI/BamHI followed by re-ligation of the vector. Following Cre/LoxP recombination the viral vector was generated and then, propagated on HEK293 cells. Viral stocks were prepared after purification on Cesium Chloride gradients followed by dialysis against GTS buffer (2.5% glycerol, 20mM Tris pH 8, 25mM NaCl) and stored at −80°C until use.
To evaluate for protective efficacy, all vaccinated monkeys, as well as 8 naïve monkeys were challenged intravenously at 30 weeks after the rAd5-Gag immunization with SIVmac251 utilizing 50 times the 50% monkey infectious dose. The SIVmac251 isolate used in this study was kindly provided by Dr. Norm Letvin (Harvard Medical School), and was a cell-free uncloned pathogenic stock that was expanded on ConA-activated human peripheral blood mononuclear cells 
PBMCs or BAL cells were stimulated in R-10 medium (RPMI 1640 supplemented with 10% FCS, 2mM L-Glutamine, 25 mM HETES, and penicillin/streptomycin) with anti-CD28 (5 µg/ml), anti-CD49d (5 µg/ml), and Brefeldin A (10 µg/ml) with or without a single pool of overlapping SIV Gag peptides (2 µg/ml). After stimulation at 37°C, 5% CO2 for 9 hours, cells were stained as previously described . Briefly, cells were washed, fixed and permeabilized and stained for CD4, CD8, CD3, CD69, IFNγ, TNFα, and IL-2 for 30 minutes at room temperature in the dark. Cells were acquired by LSRII (BD Biosciences) and DiVa software. Flow cytometric acquisition and analysis of samples were performed on at least 400,000 events. Post-acquisition analyses were performed with FlowJo. Data shown on graphs represent values of SIV Gag peptide stimulated wells from which background values (no peptides) have been subtracted. In vitro stimulation of PBMC was done in 96-well plates in triplicate.
The following antibodies were used in this study; CD3 Pacific Blue/AX700 (Clone SP34-2), CD4 AmCyan/PerCP Cy5.5 (Clone L200), CD8a AmCyan/PerCP Cy5.5/ PECy7 (Clone SK1), CD14 FITC/PECy7 (Clone M5E2), CD95 PE/APC (Clone DX2), CCR5 APC (clone 3A9), TNFα FITC (Clone MAB11), IL2 PE (Clone MQ1-17H12), and IFNγAPC (Clone B27). Fluorophore-conjugated mAbs were obtained from BD Biosciences. In addition, CD28 ECD (Clone CD28.2) and CD69 ECD (Clone TP1.55.3) were obtained from Beckman Coulter. Unconjugated purified CCR7 and CCR7 FITC (Clone 15053) were obtained from R&D Systems. Purified CCR7 was conjugated with biotin using Pierce Biotinylation kit obtained from Thermo Fisher Scientific. CCR7-biotin was detected using Pacific blue-conjugated streptavidin (Invitrogen).
Mamu-A*01 MHC class I tetramer specific for the SIVmac239 immunodominant epitope Gag-CM9 (CTPYDINQM) conjugated with PE was purchased from the Wisconsin National Primate Research Center (University of Wisconsin, Madison, WI). One hundred µl of whole blood was incubated with CM9-PE for 20 minutes at room temperature and washed twice. Cells were stained for CD3, CD4, CD8, CD28 and CCR7 for 30 minutes at room temperature. After washing and fixing, stained cells were analyzed as described for ICS.
PBMC (2×106 cell/ml) were placed in 24-well plate containing R-10 medium, and CM9 peptide was added at a final concentration of 2 ug/ml. Following culture at 37°C, 5% CO2 for 2 days, the medium was replaced with fresh R-10 containing IL-2 (100 units/ml) and cultured for an additional 3 days. The medium was changed every 2 days. The cells were harvested at day 7, washed, and stained with tetramers as described above.
SIV viral RNA levels were determined by real time RT-PCR assay described previously . Plasma samples were run in parallel with SIV Gag RNA standards.
All group comparisons on ICS data were done using Mann-Whitney test. Paired t-test was done for testing an increase in total TNF-α+ T cell frequency in each animal before and after the boost within a group. P values of <0.05 were considered significant.
The goals of this study were to assess how various TLR ligands given alone or in combination influence the induction of SIV Gag-specific T cell responses when given with SIV Gag protein emulsified in an oil-water emulsion montanide ISA51. We also determined how these primed responses affected SIV Gag-specific T cell responses after boosting with rAd5-Gag and after SIV challenge, and whether the overall vaccine-elicited responses exerted a role in mediating viral control following this challenge. The immunization schedule of each experimental group of NHP is summarized in Table 1. For the protein vaccines, purified SIV Gag protein was emulsified in montanide ISA51 with or without TLR3, TLR4, TLR7/8 or TLR9 ligands and administered subcutaneously at weeks 0 and 10. One additional group of RM received a mixture of both TLR3 and TLR7/8 ligands to determine whether activating two distinct innate signaling pathways improved T cell immunity. All groups of primed RM were then boosted with rAd5-Gag intramuscularly at week 20. To control for the influence of priming on immunity after the rAd5-Gag boost, an additional group of eight naive RM received rAd5-Gag at the same time. At 30 weeks after the rAd5-Gag immunization, all vaccinated RM, as well as 8 naïve RM, were challenged intravenously with fifty times the 50% monkey infectious doses of pathogenic SIVmac251.
To assess differences in the potency of adjuvants we measured the frequency of SIV Gag-specific CD4+ T cell responses for all individual animals immunized with SIV Gag protein with or without the TLR adjuvants and after the rAd5-Gag boost from PBMCs (Fig. 1A) and BAL (Fig. 1B). Statistical analysis comparing the different vaccine groups at each stage of prime-boost immunization is shown in Figure 2A. RM primed with the TLR3 ligand, Poly IC with or without the TLR7/8 ligands exhibited the most robust PBMC responses in the majority of the animals. Several RM that received the TLR7/8, or TLR9 ligands also elicited CD4+ T cell responses whereas only one RM primed with the TLR4 ligand or Gag protein alone showed detectable PBMC responses. Peak responses were typically noted 2 weeks after the second prime for the RM that received a TLR ligand with a range from the undetectable to 4.57% of the circulating memory T cells. TLR3 and TLR9 ligands elicited significantly higher SIV Gag-specific CD4+ T cell responses when compared with RM that received SIV Gag/montanide ISA51 only (Fig. 2A), and this significant difference was maintained through week 19 (immediately prior to the rAd5-Gag boost).
Since induction of T cell responses in extra-lymphoid effector sites may play a critical role in controlling infections such as TB or HIV, we assessed T cell responses in BAL, a mucosal effector compartment that can be repeatedly sampled for T cell response quantification in RM. In BAL, CD4+ T cell responses were much higher in magnitude than in PBMC for all RM, peaking at 4 weeks after the second immunization and declining slowly (Fig. 1B and and2A).2A). Similar to the PBMC responses, the TLR3 ligand, Poly IC significantly increased SIV Gag-specific CD4+ T cell responses in BAL compared to SIV Gag/montanide ISA51 alone (Fig. 2A). However, the SIV Gag-specific CD4+ T cell responses in BAL were not significantly increased by the other TLR ligands, including the TLR9 ligand. In addition, it was noted that co-administration of both TLR3 and TLR7/8 ligands with SIV Gag protein did not increase the magnitude of SIV Gag-specific CD4+ T cell responses in PBMC or BAL over the increase observed with administration of SIV Gag protein with either of these TLR ligands alone. Together, these data demonstrate that the TLR3 ligand, Poly IC, is the most potent of the adjuvants tested for eliciting robust SIV Gag-specific CD4+ T cell responses as well as efficient dissemination of memory/effector cells to a representative mucosal effector site.
In contrast to the potent priming of CD4+ T cell responses, only a small number of RM immunized with any of the TLR adjuvants elicited detectable SIV Gag-specific CD8+ T cell responses in PBMC. Such responses were most notable in the group that received the TLR3 ligand, Poly IC (Fig. 3A and and4A).4A). Although smaller in magnitude compared to the CD4+ T cell responses, significantly higher CD8+ T cell responses were detected in BAL compared to peripheral blood from RM immunized with the TLR3 or TLR9 ligands compared to RM immunized with SIV Gag protein/montanide ISA51 alone (Fig. 3B and and4B4B).
All RM primed with SIV Gag protein/montanide ISA51 with or without the TLR adjuvants were boosted with rAd5-Gag at week 20. It should be noted that rAd5 is among the most potent human viral vector for inducing CD8+ T cells in NHP . As a control for the influence of the SIV Gag protein priming, we immunized an additional group of 8 naïve (unprimed) RM with the same rAd5-Gag vector. RM primed with SIV Gag/montanide ISA51 plus the TLR3 with or without the TLR7/8 ligand showed a striking, albeit transient, increase in Gag-specific CD4+ T cell cytokine responses at week 22 in peripheral blood compared to the control RM that were primed with SIV Gag/montanide ISA51 without an adjuvant or no priming at all (Fig. 1A and and2B).2B). In contrast, while there was more limited enhancement of CD4+ T cells in BAL following the rAd5-Gag boost (Fig. 1B and and2B)2B) compared to the high level of responses present prior to the boost, RM that received TLR3, TLR9, or TLR3 plus TLR7/8 ligands maintained statistically higher responses in BAL, compared to those that received SIV Gag/montanide ISA51with no adjuvant at peak and memory time points (Fig. 2B).
The data presented above characterized the magnitude of SIV Gag-specific T cells using production of TNF-α since it is expressed in the majority of antigen specific CD4+ and CD8+ T cell responses. To further extend the analysis we determined the quality of the SIV Gag-specific CD4+ T cell responses based on the production of IFN-γ, IL-2 and TNF-α in any combination at the single cell level (Supplemental Fig. 1 and 2). As shown in Figure S2, CD4+ T cells producing all three cytokines comprised the largest percentage of the total cytokine responses in RM primed with SIV/Gag montanide ISA51 plus the TLR3, alone or with TLR7/8, and this polyfunctionality was not changed by rAd-Gag boost and remained durable. By contrast, cells co-producing only TNF-α and IL-2 comprised ~ 60% of the total cytokine responses in blood of RM immunized with SIV Gag/montanide ISA51 and the TLR4 ligand, and cells producing IL-2 or TNF-α alone dominated when only SIV Gag/montanide ISA51 was used. Thus, TLR3 ligand increases both the magnitude and generation of multi-functional CD4+ T cells during prime, which can be maintained after the boost by rAd5-Gag.
Although there were low to undetectable CD8+ T cell responses following the primary immunization, there was a striking increase in the magnitude of SIV Gag-specific CD8+ T cell responses after the rAd5-Gag boost in both PBMC (Fig. 3A and and4B)4B) and BAL (Fig. 3B and and4B)4B) in RM primed with TLR3, 7/8 and 9 ligands or TLR3+TLR7/8 compared to having no adjuvant in the prime or rAd5-Gag only. Like CD4+ T cells, the higher CD8+ T cell responses were maintained in both blood and BAL in these RM up to 30 weeks after the boost. These data show that, despite the low frequency of cross-primed CD8+ T cells after primary immunization, the TLR adjuvants enhanced CD8+ T cell responses after the rAd5-Gag boost.
Overall analysis of Gag-specific T cell responses in PBMC and BAL in RM primed with TLR3 or TLR3 + TLR7/8 vs the rest of RM primed with TLR4, TLR7/8, or TLR9 showed that TLR3 is generally better at priming T cells (CD4>CD8) than the other adjuvant at the dose used, and is maintained after the boost by rAd5-Gag (Supplemental Fig. 3).
Three to 4 RM in each vaccine group expressed the Mamu-A*01 MHC class I allele, allowing assessment of the frequency of the CD8+ T cell response to the immunodominant Mamu-A*01-restricted SIV Gag CM9 epitope  during the course of prime-boost immunization. As shown in Figure 5, tetramer positive CM9 responses were undetectable in blood or BAL in Mamu-A*01 RM that were primed twice with SIV Gag/montanide ISA51 and any of the TLR ligands. However, after the rAd5-Gag boost, there was a dramatic increase in the frequencies of CM9-specific CD8+ T cells in most of the RM that received a TLR ligand with the SIV Gag/montanide ISA51 (Fig. 5). To examine whether CM9-specific CD8+ T cells were cross-primed at frequencies below the limit of detection by direct tetramer staining or ICS, we stimulated PBMCs from prior to vaccination (Day 0) and after priming immunizations, but prior to rAd5-Gag boost (Day 98), in vitro for 7 days with CM9 peptide and then stained with CM9-tetramer. Figure 5C shows CM9 tetramer staining from PBMCs obtained from a representative RM prime with Gag protein + poly IC at pre-vaccination (Day 0) and post-prime (Day 98), compared to a positive control (post-rAd5-Gag vaccination). CM9 tetramer staining was observed in fresh PBMC only after the rAd5-Gag boost (week 26). However, after in vitro expansion, a discrete CM9 tetramer+ population was apparent in PBMC obtained after priming, but not before priming. Indeed, all the Mamu-A*01 animals that received Gag protein + TLR3 (n=3) or TLR3 plus TLR7/8 (n=4) ligands developed positive responses to the CM9 epitope in PBMC after the in vitro expansion, along with 2/3 RM in the groups that received the TLR7/8 or TLR9 ligand adjuvants. Of note, one RM primed with the TLR4 ligand also had low but detectable CM9 responses following in vitro expansion. Thus, while the numbers of RM in each group are too few to demonstrate statistical differences between the groups, the data suggest that the CM9 responses after the rAd5-Gag boost are influenced by a very low frequency of cross primed responses after the protein immunization with the various TLR ligands, but not with SIV Gag protein alone.
We next focused on how the TLR ligands influenced the breadth of SIV Gag T cell immunity. In this regard, a recent study showed that the type of adjuvant given with a protein antigen can influence the specificity and breadth of the cellular responses and, therefore the vaccine efficacy . At 15 weeks after the rAd5-Gag boost, PBMCs were stimulated with 13 pools of peptides (ten consecutive 15-mer peptides overlapping by 11 amino acids) covering the entire sequence of the SIV Gag protein. Cytokine–secreting CD4+ and CD8+ T cells were evaluated by ICS. PBMC from the majority of RM that received TLR ligands during priming responded to multiple SIV Gag peptide pools, showing higher diversity in T cell responses for both CD4+ and CD8+ subsets (Fig. 6). In particular, the TLR3 ligand, Poly IC elicited a mean of 6.8 (CD4+) and 5.3 (CD8+) detectable SIV Gag pool responses per RM, whereas RM that received SIV Gag/montanide ISA51 alone elicited a mean of only 1.5 (CD4+) and 1.3 (CD8+) pools per animal. These data show that the Poly IC induces a significant 4.0-fold in breadth of T cell responses. In addition, the TLR9 ligand CpG and the combination of the TLR3 and TLR7/8 ligands also elicited a significantly higher number of responses to SIV Gag pools for CD4+ T cells, and the combination of TLR3 and TLR7/8 ligands also induced a higher number of positive CD8+ responses. However, it is again important to note that the administration of the TLR3 and TLR7/8 ligands together with the Gag protein provided no additional enhancement of response breadth over the use of the TLR3 ligand alone.
In assessing the specificity of T cell responses to the 13 SIV Gag pools, we noted that the vaccine-elicited CD4+ T memory cells recognizing peptide pool G (amino acids 241–291) were observed in 75% (18 out of 24) of RM that were immunized with Gag plus a single TLR ligand, but not in any of the RM that were primed with SIV Gag/montanide ISA51 or given rAd5-Gag alone (Supplemental Fig. S4). The pool G consecutive peptides are located in Gag-p27, and recently has been identified as a common target (“hotspot’”) for SIV Gag-specific CD4+ T cells . For CD8+ T cells, responses to the pool G were not as common as for CD4+ T cell responses in all vaccinated animals except for those that received TLR3 ligand, of which all 6 animals showed responses to the pool G (Supplemental Fig. S4). In particular, CD8+ T cell responses to the pool G in the TLR3 ligand group were higher magnitude compared to the responses to the other pools although this region is known to contain only subdominant CD8+ T cell epitopes [32, 33]. These results suggest that a TLR3 ligand enhances cellular responses to subdominant epitopes and changes the immunodominance in SIV Gag protein (Supplemental Fig. S4). As expected, for all Mamu-A*01 animals, we detected CD8+ T cell responses to pool E (Supplemental Fig. S4) which contains the dominant SIV Gag181–189 CM9 epitope. Moreover, further screening of PBMC with CM9-mer peptide by ICS analyses yielded similar results, showing that the responses to pool E in Mamu-A*01 monkeys were mainly due to a recognition of CM9 peptide (data not shown) by CD8+ T cells.
We next assessed whether prime-boost immunization altered the anamnestic T cell response following a highly pathogenic infectious challenge with SIVmac251. As shown in Figure 7A, priming with SIV Gag protein/montanide ISA51 with or without any of the TLR ligands had little effect on the frequency of SIV Gag-specific CD4+ T cells in peripheral blood 2 weeks following SIVmac251 challenge. However, SIV Gag protein priming in the presence of the TLR3 ligand Poly IC (alone or with TLR 7/8 ligand) or the TLR 9 ligand resulted in significantly higher frequencies of SIV Gag-specific CD4+ T cells in BAL 2 weeks following SIVmac251 challenge compared to RM that only received rAd5-Gag or were primed with SIV Gag protein/montanide ISA51 alone.
The peak post-challenge SIV Gag CD8+ T cell responses in peripheral blood and BAL were significantly greater in all vaccinated RM compared to unvaccinated controls; however, these responses were no different whether or not the RM were primed with SIV Gag protein/montanide ISA51 alone or any of the adjuvants (Fig. 7B). Similarly, post-challenge frequencies of CM9-specific CD8+ T cells specific CD8+ T cells in the Mamu*A01 RM were increased in vaccinated vs. unvaccinated RM, but did not appear to be consistently augmented (over responses in RM given rAd5-Gag vaccination alone) by administration of a SIV Gag protein/montanide ISA51 prime, with or without any of the adjuvants. Thus, while peak post-challenge frequencies of SIV Gag-specific CD4+ T cells were augmented by a SIV Gag protein prime in the presence of TLR3 and TLR9 ligands (primarily in BAL), such priming (with or without adjuvants) had no consistent effect on SIV Gag-specific CD8+ T cell responses post-challenge.
It has been previously reported that NHP vaccinated with rAd26 followed by rAd5 encoding SIV Gag showed reduction in peak and set point plasma viral load after SIVmac251 challenge. These data demonstrated that a SIV Gag-only T cell-based vaccine could achieve partial immune control of a highly pathogenic SIV infection, and that protection correlated with the magnitude and quality of SIV Gag-specific CD8+ T cells . These data would predict that the SIV Gag protein/adjuvant priming used in our study, which skewed the responses toward CD4+ rather than CD8+ T cells prior to the rAd5-Gag boost, would have little to no protective effect after pathogenic SIV challenge. Somewhat surprisingly, we observed a significant ~ 1-log reduction of peak viral load at day 14 post- SIV challenge in the RM primed with SIV Gag/montanide ISA51 with the TLR3 ligand Poly IC, compared to unvaccinated control RM. Smaller, but significant, reductions in peak viremia were observed in RM primed with SIV Gag/montanide ISA51 and Poly IC in combination with TLR7/8 ligand and with TLR9 ligand alone (Fig. 8A and B). These differences in viral replication levels were transient, as plateau phase plasma viral loads (day 70 post-challenge) did not differentiate vaccinated and unvaccinated groups.
We next assessed the relationship between peak acute phase plasma viral load and the magnitude of the SIV Gag-specific CD4+ and CD8+ T cell responses in blood at BAL, both immediately prior to, and 2 weeks following SIVmac251 challenge (Fig. 8C). Strikingly, there was no correlation between peak acute phase viremia and the SIV Gag-specific CD8+ T cell responses in either site and at either time point. There was also no correlation between peak acute phase viremia and the magnitude of the pre-challenge SIV Gag-specific CD4+ T cell response in blood. However, we observed a weak (rho = −0.35), but significant (P = .021), negative correlation between the magnitude of pre-challenge SIV Gag-specific CD4+ T cell response in BAL and peak acute phase viremia, and a considerably stronger negative correlation between the magnitude of the 2 week post-challenge SIV Gag-specific CD4+ T cell response in both blood and BAL and peak acute phase viremia (rho/P = −0.49/0.0008 and −0.50/0.0006, respectively). Taken together, these data suggest that strong SIV Gag-specific CD4+ T cell recall responses may have contributed to modest control of early viral replication.
Adjuvants are used to improve the magnitude, breath, quality and durability of adaptive immunity to specific antigens  . Accordingly, adjuvants such as alum or oil/water formulations have been used in humans in combination with viral or bacterial antigens to improve humoral immunity which is the primary correlate of protection for certain infections. Based on the importance of T cell immunity for mediating protection against TB, certain stages of malaria infection and controlling HIV viral load, there is a critical need to develop non-live vaccine formulations that can be used as part of prime-boost immunization regimen vaccines and adjuvants to optimize such responses. In this regard, based on greater similarity in the innate response and TLR expression between RM and humans compared to mice, we compared a variety of adjuvants that signal through distinct TLR signaling pathways.
We previously showed that potent Th1 responses are induced following immunization with HIV Gag protein/montanide ISA51 mixed with TLR7/8 ligand or CpG . These data suggested that induction of IL-12 or Type I IFN  by these adjuvants were mediating Th1 differentiation . The data presented here extends these findings and provides direct comparative evidence for enhanced potency of Poly IC. The results are consistent with prior studies in RM showing that Poly IC or Poly ICLC given with HPV  or a malaria protein  without montanide induces Th1 responses after multiple immunizations. Thus, in this study, formulating Poly IC with montanide ISA51 provides a depot of the vaccine thereby enhancing potency and efficiency since robust Th1 immunity was induced by a single immunization. In terms of the quality of the responses, we show that durable multi-functional Th1 responses are induced by Poly IC, TLR 7/8 and CpG . Such responses may be optimal for protection against TB  or Leishmania . The enhanced potency of Poly IC observed in this study may be due to secretion of IL-12 from DC in response to TLR3 and Type I IFN via melanoma differentiation-associated gene (MDA)-5 from DC or stromal cells . By contrast, as TLR9 expression is limited to pDC and B cells in NHP, only Type I IFN-γ from pDC would be induced. TLR7 is also expressed in pDC and TLR8 in mDC, thus inducing Type I IFN and IL-12, respectively. Thus, if there was a dual requirement for Type I IFN and IL-12, it is not clear why the TLR7/8 was not as effective as Poly IC in this study. As Poly IC induces substantially more IL-12 from human DCs than TLR7/8 [21, 23] it is possible that the amounts of these cytokines induced by the adjuvants or the duration at the site of priming in vivo are critical .
While protein-based vaccines are effective for eliciting potent antibody and CD4+ T cell responses, their ability to induce CD8+ T cell through cross presentation is less efficient and robust. Formulation and delivery of the protein and the presence of Type I IFN are critical variables of cross-priming . While there were low to undetectable SIV Gag-specific CD8+ T cells in the blood of most of the RM immunized with SIV Gag/montanide ISA51 and any of the TLR ligands, Poly IC was the most effective for inducing SIV Gag-specific CD8+ T cells, as responses using this adjuvant were detected in 6/6 RM in cells from BAL fluid, compared to RM given SIV Gag protein in montanide ISA51 alone or with any of the other adjuvants. Notably, in Mamu-A*01 RM, these responses did not include directly detectable responses to the Mamu A*01-restricted Gag CM9 epitope, an epitope that in the context of viral vectors is strongly immunodominant . While in vitro expansion revealed low-level CD8+ T cell priming to CM9 in animals receiving SIV Gag protein and either TLR3, TLR7/8 or TLR9 ligands in montanide ISA51, these data suggest that epitope specificity and immunodominance hierarchies of CD8+ T cell responses elicited by cross-presentation of adjuvanted protein differs from that of viral vectors. Given that prime-boost strategies for eliciting potent CD8+ T cell responses depend on epitope overlap between the prime and the boost, these differences may have important implications for the prime-boost combinations of protein and viral vectors, as discussed below. We would note that our previous NHP study, we did elicit low CD8+ T cell responses with the TLR7/8 ligand (3M-012) formulated in the same way (water-in-oil-emulsion with montanide ISA51) with HIV Gag. It remains possible that the HIV Gag p41 protein used in this earlier study was more particulate than the SIV Gag used in this report which we have shown to be critical for optimizing cross-priming with a TLR7/8 ligand . One way of improving cross-presentation of an exogenous antigen by APCs would be simultaneous loading and activation of APCs that can be done through covalently linking antigen to the adjuvant or by co-encapsulating antigen and adjuvant into microparticles, showing satisfying results [13, 44–46]. Alternatively, we showed that delivery of HIV Gag p24 to DCs by targeting DEC-205 also induced CD8+ T cells when given with Poly ICLC. Nevertheless, based on all these studies with various formulations and adjuvants that induce Type I IFN, it is clear that generation of CD8+ T cells with protein-based vaccines in NHP is not as efficient as, and quantitatively and qualitatively different from those elicited by viral vectors.
Heterologous prime-boost immunization with DNA, protein and viral vectors are currently being tested in humans against HIV, malaria and TB. The major advantage of such regimens is to elicit higher responses than either vaccine modality alone or homologous prime-boost immunization. It has been reported in RM that potency of primed SIV Gag T cell responses influence the magnitude and quality of T cells after a boost . Here, we show that after the rAd5-Gag boost, there was no durable increase in the frequency of the total SIV Gag CD4+ T cells in any of the vaccine groups relative to what was present before the boost. Notably, at 2 weeks after the rAd5-Gag boost, the magnitude of the CD4+ T cell response in RM primed with SIV Gag and Poly IC was similar to peak responses after the second prime and no further accumulation of CD4+ T memory cells was detected in BAL. In addition, there was little change in response quality (e.g., polyfunctionality) following the rAd5-Gag boost. Thus, Ad5-Gag is not particularly effective at boosting CD4− T cell responses and does not appear to significantly change the functional profiles of the SIV Gag-specific CD4+ T cells established during the priming phase. Importantly, the SIV Gag-specific CD4+ memory T cells established by the most effective primes were durable, particularly in an effector site (BAL). These data may have relevance for protection against infections such as TB in which effector/memory cells may be critical at the site of initial infection for early control.
In contrast to CD4+ T cell responses, the magnitude of the SIV Gag-specific CD8+ T cell response significantly increased following the rAd5-Gag boost in RM primed with SIV Gag protein with TLR3 ligand (along and in combination with TLR7/8 ligand) or TLR9 ligand compared to RM primed with SIV Gag protein alone or not primed at all. Since primary SIV Gag CD8+ T cell responses were below the limit of detection in blood and BAL in many of these RM, it is possible that the SIV Gag-specific CD4+ T cells elicited during priming were influencing SIV Gag-specific CD8+ T cell expansion after the boost. Alternatively, a low frequency of SIV Gag-specific CD8+ memory T cells generated during priming may have been responsible for the expansion after the rAd5-Gag boost. This latter possibility is supported by the demonstration that CD8+ T cells specific for CM9 peptide were detectable after in vitro culture of PBMCs after priming in Mamu-A*01 RM. In addition, RM that failed to elicit CD8+ T cell responses after the rAd5-Gag showed limited expansion of CD8+ memory T cells by CM9-tetramer staining. Overall, these data suggest that generation of robust antigen-specific CD4+ T cell responses may not be sufficient to enhance boosting of CD8+ T cells with a viral vector such as rAd5 without some primed CD8+ T cell responses. Indeed, such responses may in fact be below the limit of detection by ICS assays or staining with tetramer. This has clear implications for using protein-based vaccines as a prime prior to a viral boost to optimize CD8+ T cell immunity. Notwithstanding the potency of CD4+ T cells induced by our protein/TLR3, TLR7/8 or TLR9 ligands, based on the low frequency of CD8+ T cells it seems prudent to use viral based vectors as a prime in settings where CD8+ T cells are the primary correlate.
Immunization with SIV Gag/montanide plus Poly IC followed by rAd5-Gag boost induced highly diverse CD4+ and CD8+ T cell responses relative to rAd5-Gag vaccination alone. Generating a diverse cellular immune response may indeed be desirable for an HIV vaccine. A recent study extended this finding by demonstrating that the breadth of Gag-specific cellular immune responses elicited by vaccination correlated with control of set-point viral loads after challenge . Heterologous prime-boost immunization with rAd26/rAd5-Gag generates an average of 8.6 detectable Gag epitopes per RM based on assessment by ELISPOT. In the present study, vaccination with SIV Gag plus Poly IC followed by rAd5-Gag generates an average of 6.8 CD4+ responses and 5.5 CD8+ T cell responses specific for SIV Gag (assuming 1 epitope per peptide pool). Importantly, all but one in this group have a common CD4+ T cell response specific to the SIV Gag pool G, which was recently described as a common target of CD4+ T cell responses elicited by a successful vaccine against SIV . In addition, this study also showed adjuvant-driven changes in immunodominance in SIV Gag protein, and illustrates the profound influence of vaccine formula on immune hierarchy and vaccine efficacy. The major factors contributing to immunodominance are still unknown. A recent study has shown that a protein delivered by rAd5 vector changed the epitope hierarchy different from what has been seen when this molecule was delivered as a recombinant protein . It has been suggested that the microenvironment in which the antigen is presented leads to differential presentation so that other epitopes are presented.
For any vaccine, the final arbitrator of utility is not immunogenicity, but efficacy. T cell targeted vaccines mediate protection by establishment of anti-viral effector or effector memory cells in sites of early viral replication or by elicitation of a memory T cell response that upon subsequent infection generates a robust anamnestic anti-viral effector response . Typically, CD8+ T cell responses are thought to be the primary effectors for viral infection. Although we found that a SIV Gag protein prime adjuvanted with TLR3 (±TLR7/8) and TLR9 ligands durably increased the overall frequency of SIV Gag-specific CD8+ memory T cells after the rAd5-Gag boost, the level of these responses did not predict the limited protection observed in the RM vaccinated with these regimens after SIVmac251 challenge. Moreover, the TLR3 (±TLR7/8) and TLR9 ligand-adjuvanted SIV Gag protein prime did not enhance (over rAd5-Gag alone) the anamnestic SIV Gag-specific CD8+ T cell response to challenge. These adjuvants had a stronger effect on the ability of SIV Gag protein to prime for durable, high frequency SIV Gag-specific CD4+ T cell responses and the ability of the RM vaccinated with these vaccines to respond to SIV challenge with a robust anamnestic SIV Gag-specific CD4+ effector response. In contrast to the SIV Gag-specific CD8+ T cell responses, the magnitude of the SIV Gag-specific CD4+ T cell responses, particularly the anamnestic responses after challenge, correlated with, and appeared to explain the modest protection against peak viremia observed in these vaccinated RM. These data suggest that robust SIV Gag-specific CD4+ T cell anamnestic response can contribute to efficacy, and do not necessarily enhance infection by increased provision of viral targets .
The mechanisms by which SIV Gag-specific CD4+ T cell response would contribute to the observed early reduction in peak viremia are unclear. Traditionally, CD4+ T cells are known for their ability to help B cells and CD8+ T cell responses. However, direct participation of CD4+ effector cells in cell-mediated anti-microbial immunity is increasingly appreciated. Emerging data suggest that class II-restricted CD4+ T CTL can contribute to protective responses against HIV, as well as other viral and bacterial infections and tumors [49–51]. Therefore, vaccines aimed at CD4+ T cell responses should be incorporated in development of a future AIDS vaccine in conjunction with a strategy optimizing antibody and CD8+ T cell responses.
Overall, these data demonstrate a hierarchy among TLR adjuvants administered with a protein vaccine in a water/oil emulsion. TLR ligands targeting intracellular receptors such as TLR3, TLR7/8 and TLR9 are more potent for T cell immunity than the emulsion alone or with the TLR4 ligand. These data show that the intracellular TLR ligands, likely through enhanced production of type I IFN may be critical for improving Th1 immunity. Furthermore, we failed to show any synergy using TLR3 and TLR7/8 ligands in vivo over either one alone for T cell immunity. An important question for vaccine development would be what type and quality of the immune responses are a better indicator of efficacy. To this end, we showed SIV Gag protein vaccination with Poly IC is able to induce more efficacious CD4+T cell responses in a prime-boost regimen and expansion post HIV infection. Moreover, Poly IC is also a potent adjuvant for eliciting humoral immunity [18, 37, 52, 53]. Thus, Poly IC may be useful as a vaccine adjuvant where both antibody and T cell immunity is critical such as in HIV, malaria and TB.
We thank Kelly Rausch, John Turner, Shannon Planer, and Tonya Swanson for expert husbandry and technical assistance with the vaccinations. We are also grateful to Nate Whizin for technical assistance and help with the LSR, to Nancy Wilson at Wisconsin National Primate Research Center for tetramer manufacture, and to Greg Spies at FHCRC for rAd5-Gag construct.
Grant Support: This work was supported by the Bill & Melinda Gates Foundation Collaboration for AIDS Vaccine Discovery.
All authors have no financial conflicts of interest.