We monitored immune responses using peptide sets that fully span diverse natural Env proteins, so we could directly determine the number responses we generate to actual HIV-1 Env proteins (Santra et al., 2008
). In the past we have used 10 proteins for these purposes that were selected to be representative of the global diversity of HIV-1 (Santra et al., 2008
). To determine if a subset of 5 of these 10 proteins would be adequate for evaluating immune breadth in this and future studies, we compared the potential T cell epitope (PTE) or 9-mer coverage of a series of Env immunogen sequences to different sets of envelopes. The series of Env immunogen sequences in this comparison included the single strains HXB2 and MN, the M group Consensus (CON-S) sequence, and finally the 2-valent and 3-valent mosaic sequences (). For each immunogen sequence, we calculated the coverage of all 9-mers in the collection of global envelope sequences in the current Los Alamos database (Database), in a set of 10 divergent natural envelope sequences used to estimate the breadth of vaccine coverage in a previous nonhuman primate study (Santra et al., 2008
), and a subset of 5 of those 10 natural divergent envelope sequences. The sets of 5 and 10 Env sequences were comparably representative of 9-mer coverage based on the global HIV database, suggesting that peptides derived from this set of 5 Env sequences would provide a reasonable framework for estimating the breadth and depth of vaccine coverage of circulating HIV-1 isolates worldwide (). We therefore decided to use these 5 Env peptide series to monitor T cell responses in the present study. These 5 diverse HIV Env protein sequences included one each from clade A, B and G and two from clade C (GenBank accession numbers: A1, AY371163; B1, AY561237; G1, AY371121; C1, AY563169, and C3, AF457054, respectively (see Santra 2008
) for details).
Fig. 1 The coverage of potential T cell epitopes (PTEs) by different vaccine candidates. Shown is the coverage of PTEs by different vaccine candidates as would be detected by the 5 Env proteins used to monitor this study (red), the 10 proteins used in a previous (more ...)
reveals important differences in the potential coverage that might be induced by these various immunogens. Natural proteins had very low levels of coverage of potential T cell epitopes (PTEs), with MN matching only approximately 17% of the PTE sequences found in the global database of circulating strain M group strains (, blue bar). The CON-S immunogen provided 26% coverage of PTEs, or a 1.6-fold increase relative to MN (, blue bar). In contrast, a 3-valent mosaic Env immunogen increased the coverage of PTE sequences to 44% (, blue bar).
A vaccination study in nonhuman primates was initiated to evaluate the breadth and depth of cellular and humoral immune responses elicited by 2-valent and 3-valent HIV-1 Env mosaic immunogens. Thirty Mamu-A
negative Indian-origin rhesus monkeys were divided into three groups: two experimental groups, each consisting of 12 monkeys, and a control group consisting of 6 monkeys. One experimental group received 2-valent mosaic Env immunogens, the other experimental group received 3-valent mosaic Env immunogens, and the control group received empty vectors. Monkeys received priming immunizations by the intramuscular route at weeks 0 and 4 with a total of 4 mg plasmid DNA expressing 2-valent mosaic Env, 3-valent mosaic Env, or a sham construct. The monkeys that received mosaic Env plasmid DNA immunogens also received a single clade B Gag plasmid DNA immunogen. At week 8, monkeys were boosted by intramuscular immunizations with 1011
particles of recombinant adenovirus serotype 5 (rAd5) expressing the same mosaic Env and single clade B Gag genes. The vaccination strategy was adapted from our earlier study of the immunogenicity of a natural strain and an M group consensus Env immunogen (Santra et al., 2008
); we did this to facilitate a direct comparison of the data generated in the two studies. These studies did, however, differ in that Gag immunogens were included in the current study, but not in the prior experiment comparing HXB2 and the M group consensus Env immunization. Gag immunogens were added to the current study because they would also be administered if mosaic immunogens were to be evaluated in human populations. It should be noted, however, that if strong T cell responses were made to Gag, these responses might divert some of the T cell responses away from Env (Ferre et al., 2010
The breadth of the vaccine-elicited cellular immune responses was evaluated by assessing peripheral blood T lymphocyte recognition of the set of 5 different natural Env sequences discussed above, using a peptide/IFN-γ ELISpot assay and matrix epitope mapping. Peripheral blood lymphocytes (PBL) from each vaccinated monkey were evaluated for IFN-γ ELISpot responses to pools of 15-mers overlapping by 11 amino acids spanning each of the 5 envelope sequences. Epitopes of each of the 5 indicator envelope proteins recognized by PBL sampled from each of the experimentally vaccinated monkeys were enumerated as previously described (Santra et al., 2008
). To determine if there were differences in the epitope recognition by PBL of monkeys vaccinated with HXB2, the M group consensus, or the 2-valent and 3-valent mosaic immunogens, we evaluated the number of epitope peptides recognized per natural Env sequence for all immunized animals (). To compare the immunogenicity of each of the four vaccines, we used a quasi-binomial regression model to study the number of vaccine-elicited positive responses per animal per peptide series, using a strategy similar to analyses described in previous studies (Santra et al., 2008
). No immunogen elicited a response that was biased toward a particular clade or strain of virus. The HXB2 vaccine elicited significantly fewer epitope peptide responses per peptide series than the other three immunogens. CON-S had an odds ratio (OR) of 3.09, p
=8.3 × 10−7
relative to HXB2; the 2-valent mosaic, an OR of 3.95, p
=6.4 × 10−10
; and the 3-valent mosaic, an OR of 4.12, p
=2.0 × 10−10
. The immunogens can be rank ordered on the basis of odds ratio of the number of epitope responses they elicited from lowest to highest: HXB2
CON-S<2-valent mosaic<3-valent mosaic, and this ordering is consistent with the median number of epitope-specific responses per strain elicited by each vaccine; 1 for HXB2, 2 for CON-S, 3 for 2-valent mosaic, and 4 for the 3-valent mosaic (). Further statistical comparisons were consistent with our previous findings (Santra et al., 2010
) that the CON-S immunogens elicited fewer epitope peptide-specific responses than the mosaic vaccines, but in this study the p
-values showed only a trend: 2-valent mosaic compared to CON-S had an OR 1.28 (p
=0.08), and 3-valent mosaic compared to CON-S had an OR of 1.33 (p
=0.04). Of note, in this analysis we are comparing results from two different studies (the HXB2 and CON-S data was from Santra et al., 2008
, the 2 and 3 valent mosaics from the present study). The numbers of epitope peptide-specific responses elicited by the 2- and 3-valent mosaics were not statistically distinguishable.
Fig. 2 Breadth of Env-specific T lymphocyte responses in the 12 individual monkeys from each experimental group, showing number of peptide epitopes recognized per protein. (A) The breadth of the vaccine-elicited cellular immune responses was determined by assessing (more ...)
We also evaluated the depth of T lymphocyte responses, the diversity of sequences of a given epitope that might be recognized by T lymphocytes of the vaccinated monkeys (). This was estimated by counting the responses made to all variant peptides from every epitopic region of the Env protein for each vaccinated monkey. It was very common for the vaccine-elicited responses to be strain specific, i.e., only 1 of 5 variant peptides was recognized by the vaccine-elicited T cell population (). However, the number of T cell responses that recognized an epitope in two or more divergent strains was increased in the monkeys that received polyvalent mosaic immunogens. We first did a Fisher’s exact test comparison of a 2 × 4 contingency table to compare the number of epitope-specific responses that recognized only 1 of the 5 variant strains to the number that recognized 2 or more strains; the p-value was 0.0006, indicating there were vaccine-associated differences in the depth of the elicited responses. As shown in , more than 60% of the responses to the single valent CON-S and HXB2 vaccines recognized only 1 of the 5 strains. In contrast, the polyvalent mosaic vaccines elicited more cross-reactive responses (Fisher’s exact p=4 × 10−5). At the other end of the spectrum, there was a 4-fold greater representation of responses that recognized epitopes derived from all 5 evaluated virus strains in mosaic-vaccinated animals than in animals vaccinated with the single antigens CON-S or MN. (12% compared to 3%). There was no clear difference in the depth of T cell responses elicited by the 2-and 3-valent mosaic immunogens.
Fig. 3 Depth of cross-reactive T cell responses elicited by various vaccines. The peptides used for the ELISpot assays were designed to maintain their alignment, so they could be directly compared between strains. For each animal, if a vaccine response was found (more ...)
Finally, we assessed the effect of increasing the valency of the mosaic immunogens on humoral responses. Sera sampled from immunized monkeys at week 13 were assessed for antibody binding to purified BaL.01 and Du156.12 gp120 and 93TH057 gp120 core proteins by ELISA as previously described (Wu et al., 2010
) Both the 2- and 3-valent mosaic immunogens elicited comparable titers of binding antibody responses to the Env proteins used in the assays (). We also assessed the neutralizing antibody responses in the sera of immunized monkeys to a limited number of Tier 1 (viruses that are highly sensitive to antibody-mediated neutralization) and Tier 2 (viruses that are moderately sensitive to antibody-mediated neutralization) viruses (Seaman et al., 2010
). Interestingly, the 3-valent mosaic immunogens elicited significantly higher titer neutralizing antibody responses to tier 1 viruses than the 2-valent mosaic immunogens (). However, the 2-valent and 3-valent mosaic immunogens elicited comparable, very weak or undetectable neutralizing antibody responses to tier 2 viruses (). These findings are comparable to the antibody responses elicited by natural strain and consensus Env immunogens (data not shown).
Fig. 4 Vaccine-elicited binding antibody and neutralizing antibody responses. (A) Five weeks post-rAd5 boost serum samples were collected from monkeys vaccinated with either 2-valent or 3-valent mosaic immunogens. Antibodies from these serum samples were assessed (more ...)