Background
The ideal human immunodeficiency virus (HIV) vaccine should likely activate both the humoral and the cellular arm of the immune system; enabling antibodies to neutralize incoming virus particles while CD8+ T cells would kill already infected cells and limit the spread of the virus.
The substantial role of CD8+ T cells in the control of HIV type 1 (HIV-1) infection is supported by the correlate of protection with specific HLA class I allotypes [
1‐
4], a temporary decline in peak viremia coinciding with appearance of CD8+ T cell responses during acute infection [
5,
6], and occurrence of escape mutants in response to specific CD8+ T cells [
7,
8]. In humans, protective CD8+ T cell responses have been found to correlate with the targeting of Gag [
9,
10] and possibly of Nef [
11], while inversely correlating with responses against envelope (Env) [
10]. These correlations have been substantiated by additional population studies that have generated in-depth data by carefully screening responses to the HIV-1 proteome and investigating protective responses towards individual epitopes [
12‐
17]. Together, these studies indicate that it is theoretically possible to successfully target HIV-1 by inducing T cells targeting specifically effective epitopes. However, such protective T cell specificities have only been raised infrequently after immunization with full length HIV-1 antigens [
18] and previous vaccination attempts that aimed to induce CD8+ T cells have failed to control HIV-1 replication overall [
19]. Despite their overall clinical failure, these studies did provide evidence of improved virus control after polyspecific targeting of defined regions of Gag, Vif, Nef, and Pol [
18,
20]. A shift to more focused constructs aimed to direct CD8+ T cells specifically to carefully selected regions has therefore been a logical step forward. Conserved regions vaccines [
14,
21] and conserved elements (CE) vaccines [
22] each represents approaches aiming to overcome the challenge of HIV-1 diversity and raise responses to rarely targeted, yet protective specificities [
23].
While T cell based vaccines have yet to show efficacy in clinical trials, the RV144 trial demonstrated the possibility of reducing infection risk by raising Env specific antibodies with a virus-like particle (VLP) encoded virus vectored prime and a protein boost vaccine [
24]. A recent interest in the coordinated responses against Gag and Env has emerged. Studies by Schell et al. suggested that Gag benefits a response based on protective antibodies in a vaccine secreting VLPs [
25,
26]. Additionally, the immune response against Gag has been shown to support antibody responses against Env displayed on VLPs by providing an additional source of intrastructural T cell help [
27,
28].
Due to the additive and potentially synergistic value of using Gag and Env for effective antibody and T cell mediated protection, we sought to study the induction of potentially protective Gag and Env responses in VLP based virus vectored immunization regimens. To investigate the possibility of inducing T cell responses against conserved or cross-reactive epitopes in the Gag polyprotein, while preserving antibody responses towards Env trimers in such a virus vectored VLP vaccine construct, we used a prime-boost regimen with heterologous Gag together with homologous Env and compared it to a homologous Gag/Env regimen. We did this by utilizing a combination of adenoviral vectors co-encoding SIVmac239
env and either HIV-1 or SIVmac239
gag sequences as primers for modified vaccinia Ankara (MVA) vectors encoding SIVmac239
env and SIVmac239
gag. We included a non HIV sequence as previous studies have found no benefit of prime-boost regimens using Gag from different HIV clades while improved breadth was found using different HIV Env clade sequences [
29,
30]. Using SIV and HIV Env provides a diversity resembling the successful heterologous Env sequences and results in a vaccine design immediately testable in a non-human primate model. Here we show that outbred CD1 mice immunized with homologous Gag/Env had an increased breadth and functionality of T cell responses against heterologous
gag sequences as well as a surprising increased magnitude of Env specific antibody responses. Mice immunized with homologous Gag selectively expanded Gag specific T cells following the booster immunization whereas mice immunized with heterologous Gag selectively expanded Env specific T cells following the boost. These data highlights the importance in the selection of
gag sequences in VLP encoding virus vectored immunization regimens.
Methods
Mice
Female CD1 mice at the age of 6–8 weeks were obtained from Scanbur (Denmark). The mice were allowed to acclimatize for one week prior to the initiation of an experiment. All experiments were performed according to national guidelines and experimental protocols approved by the national animal experiments inspectorate (Dyreforsøgstilsynet).
Adenoviral vaccine production
HIV-1 clade B consensus (HIV-1 CON B) or SIVmac239
gag was encoded after a CMV promoter, followed by a self-cleavable P2A peptide and then by SIVmac239
env and SV40 polyA. The expression cassette was cloned into a human adenovirus type 5 backbone and produced, purified and titered as described [
31]. A modified vaccinia Ankara (MVA) vaccine encoding SIVmac239
gag,
pol,
env (truncated at aa 733) (MVAgpe) [
32] was kindly provided by Dr. Patricia Earl (Laboratory of Viral Diseases, NIH). The vaccine was amplified and titered in primary chicken embryoblasts according to the protocols in Kramer et al. [
33].
Purification of vaccine encoded VLPs for further characterization
Vero cells were infected with 50 plaque forming units (PFU)/cell of either Ad5 vaccine, supernatants harvested 48 h post infection, and VLPs concentrated as previously described [
31]. Pellets were resuspended in PBS at 280 X of the original concentration.
Western blot analysis
VLPs purified from Ad5 infected vero cells were prepared as previously described [
31] and characterized by western blot. Env was detected with the SIVmac251 gp120 specific monoclonal antibody (mAb) KK46 [
34] [NIH AIDS Research and Reference Reagent Program (NARRRP)] followed by HRP coupled goat anti-mouse immunoglobulin antibody (Dako). The same blot was analysed for the presence of Gag using HIV-1 anti-p24 mAb 183-H12-5C [
35] (NARRRP), and goat anti-mouse immunoglobulin antibody (Dako). SIVmac239 gp130 [
36] and SIVmac251 BK28 pr55 Gag were loaded as positive controls (NARRRP). The blots were developed using ChemiLucent Detection System kit (Pierce). Analysis was performed using Image Studio Lite software (LI-COR Biosciences).
Cell surface expression analysis
Env expression was analysed on the surface of vero cells 2 days after infection with 50 PFU/cell of either Ad5 vaccine. Cells were stained with the ITS52, ITS03, and ITS40 mAbs [
37] (kindly provided by Dr. Mario Roederer, VRC, NIAID, NIH). Binding of the mAbs was detected using anti-human IgG Fc-APC antibody (BioLegend), and the cells were acquired using an LSRII instrument (BD Biosciences) and analysed with FlowJo software (Tree Star, Ashland, OR).
Immunizations
Groups of mice (5 or 10 per group) were immunized intramuscularly (i.m.) with 2 × 108 IFU with either Ad5 vaccine. Where indicated, mice were boosted i.m. 59 days after priming with 1 × 107 IFU of MVAgpe. Vaccines were applied in a total volume of 50 μl PBS in the quadriceps muscle, changing legs at each immunization time-point.
Antibody response measurements
Antibody responses against SIVmac239 Env were determined in a concanavalin A (ConA) (Sigma-Aldrich) enzyme linked immunosorbent assay (ELISA) as adapted from [
38]. SIVmac239 Env was produced in 293FT cells by co-transfection of a shuttle plasmid encoding SIVmac239 Env and the plasmid pSG3
Δenv [
39] (NARRRP). ELISA was performed as described elsewhere [
31]. Avidity measurements were performed simultaneously with antibody response measurements as adapted from [
38], using 0.1 M sodium citrate buffer (pH 3.0). The avidity index was calculated by dividing the titer obtained with sodium citrate treatment by the titer obtained without sodium citrate treatment, and multiplied by 100.
SIV Env pseudovirus production and titration
The Env constructs for SIVmac239 and SIVsmE660.11 pseudovirus production were obtained from Dennis Burton (The Scripps Research Institute) and David C. Montefiori (Duke University), respectively. Production and determining TCID has been described elsewhere [
40].
Neutralization assay
Neutralization of pseudoviruses was measured in TZM-bl cells following the protocols in Montefiori [
40]. To account for unspecific binding to either pseudovirus, pooled pre-immunization serum samples were analysed at a dilution of 1:20. SIVmac251 antiserum [
41] (NARRRP) was included as a control. Neutralization against SIVmac239 was assayed in pools of 4–5 mice per group. The SIVmac239 pseudovirus did not achieve the RLU of 10 times the background (5.2 times the RLU of the background), but was included in the analysis since it was clear that there was no neutralization in neither sample that was tested.
Intracellular cytokine staining
Intracellular staining was performed on splenocytes using a standard protocol [
42] using 0.67 μg/ml HIV-1 CON B Gag pool, SIVmac239 Gag pool, SIVagm vervet Gag pool, SIVmac239 Env pool, CE pools for the three different Gags respectively) (all from NARRRP) at 37 °C and 5% CO
2. Based on definitions by [
22], CE pools were prepared from peptides spanning the CE from whole Gag peptide sets. The following antibodies were used for detection (Biolegend): CD8_PerCP.Cy5.5, CD4_FITC, B220_Pacific Blue, CD44_APC.Cy7, IFN-γ_APC, TNF-α_Pe.Cy7. The cells were acquired using an LSRII instrument (BD Biosciences) and analysed with FlowJo software (Tree Star, Ashland, OR). One mouse was excluded from the T cell analysis due to the collection of too few events.
Statistical analysis
Nonparametric Mann–Whitney tests were performed for analysis of differences between the groups. Spearman rank testing followed by Holms correction for multiple comparisons was used to assess correlations. The ID50 values in the neutralization assay were calculated in Graph Pad Prism using an asymmetric curve fitting (5 parameters). Statistical analyses were performed using either the R statistical software or Graph Pad Prism. p values <0.05 were considered significant.
Discussion
The effect of the VLP scaffold Gag on the strength of virus encoded VLP vaccine induced T cell and antibody response has not been investigated previously, nor has the effect of using different gag sequences in heterologous prime-boost VLP encoding vaccines. In our analysis of antibody responses we found significantly higher Env immunoglobulin titers induced by the homologous Gag prime-boost regimen. In contrast, the two different immunization regimens induced similar levels of neutralizing antibodies against the neutralization sensitive tier 1 SIVsmE660.11, and also similar avidity of anti-Env antibodies. More interesting, in addition to higher Env binding titers, the analysis of T cell responses revealed a superiority of the homologous Gag regimen; also at inducing broader CD8+ T cell responses against Gag. Interestingly, and perhaps retrospectively explainable, it appeared that while Gag showed expected CD8+ T cell immunodominance over Env, homologous Env showed immunodominance over heterologous Gag in a prime-boost regimen.
Our investigation of conserved T cell responses and the selection of CE peptide pools for analysis has its foundation in studies by Kulkarni et al. [
22,
23,
48], who have designed an antigen combining only conserved parts of p24 of Gag that normally correlates with increased virus control, and are found to include epitopes that cannot be mutated without virus fitness loss. Our homologous full-length Gag prime-boost regimen increased both the magnitude and breadth to full length Gag variants and CE, with the induced CE responses generally correlating with the overall responses to different virus variants. Bimodal responses were observed in particular for CE for both CD4+ and CD8+ T cells, reflecting the harbouring of different MHC alleles in CD1 mice. Heterologous Gag immunization did induce some CE responses, albeit not the cross-reactive responses that we had expected, reflecting the preferential Env specific T cell expansion in this group that appeared to come at the expense of negligible boosting of Gag specific responses. Compared to our results, Kaufman et al. reported an intermediate effect where heterologous full length insert in a prime-boost regimen neither increased, nor decreased the responses against conserved epitopes of Gag, but this included the weakly immunogenic Ad35 vector in the prime-boost regimen and no
env was encoded [
29]. The inclusion of Env in our prime-boost vaccinations likely makes a profound difference as the MVA immunization only boosted homologous CD8+ T cell responses in our study, and it would appear that homologous responses must have blocked expansion of potentially cross-reactive T cells recognizing the heterologous boosted antigens. A mechanism for how such preferential boosting of homologous antigens can increase breadth, as compared to a heterologous boost where conserved sequence specific T cells are preferentially re-stimulated, may have been provided by Kelly et al. [
49] who found dominant and highly immunogenic epitopes to be effective at selecting for cross-reactive T cell responders, and Varela-Rohena et al. [
50] who observed increased cross-reactivity of T cell receptors mutated for increasing affinity.
With regards to antibodies, both of the prime-boost immunization regimens analysed here had increasing antibody titers after being boosted. The priming vaccine encoding HIV-1 CON B gag was inferior in priming Env responses, having significantly lower antibody titers 7 weeks after priming and also 10 weeks post priming. These results implicate the benefit of SIVmac239 Gag as a VLP scaffold for induction of a binding antibody, although we could not observe any clear differences in the VLPs or transduced cells likely to explain this benefit. We also observed a fourfold difference in the geometric mean titers between the boosted groups; however, both the avidity of the anti-Env responses and neutralizing responses against SIVsmE660.11 were roughly similar.
An intriguing aspect of our findings is if the immunodominance interplay between Gag and Env and potentially other antigens in prime-boost regimens is also playing a role after infection in vaccinated hosts. We have observed vaccine induced changes in post-exposure epitope targeting in inbred animals challenged with lymphocytic choriomeningitis virus [
51], but this study is the first to report Env immunodominance over Gag at the whole antigen level in outbred animals. In SIV models, immunodominant responses towards easily mutated epitopes have been found to delay responses towards other epitopes that are harder for the virus to mutate, thus reducing the effect of vaccination [
52]. In summary, our study demonstrates that interantigenic immunodominance is a critical parameter in prime-boost regimens. Future optimization of this and related regimens should include both heterologous Env and homologous SIV Gag antigens before nonhuman primate efficacy trials.
Conclusions
Our study demonstrates the profound impact the choice of Gag can have, when used as a VLP scaffold in prime-boost regimens, on both antibody responses and T cell responses. The broad and increased T cell responses induced by homologous Gag immunization also implicate that these widely sought after responses can be induced by full length protein in most animals, while obtaining humoral responses against Env. Future studies should include analysis of Gag and CE responses to more distant Gag sequences, to provide an even broader knowledge of the perspective of homologous virus encoded VLP vaccines.
Authors' contributions
ACA contributed to study design, performed experiments, data analysis, contributed to interpretations of results and wrote the manuscript. PJH contributed to study design, interpretations of the results and commented on the manuscript and data presentation. Both authors read and approved the final manuscript.