Virus-like vaccines against HIV/SIV synergize with a subdominant antigen T cell vaccine

Vero cells were seeded and infected with 50 IFU/cell Ad-HIVB. 48 h post infection the supernatant was harvested and VLPs were purified and concentrated to 160× the original concentration as described [22].

The Gag-P2A protein was detected in VLP samples by Western Blot using an anti-2A peptide antibody (Millipore) followed by an HRP-conjugated goat anti-rabbit Ig antibody. The membrane was developed using LumiGLO Peroxidase Chemiluminescent Substrate (KPL, 54-61-00) and the signal detected in an ImageQuant LAS 4000 (GE Healthcare Life Sciences).

Env surface expression on Vero cells was analyzed 2 days post infection with 50 IFU/cell Ad-HIVB. The cells were stained with the monoclonal antibody clones VRC01 [NIH AIDS Reagent Program (NARP)], PGT145 [International AIDS Vaccine Initiative (IAVI)] and PGT151 (IAVI). Binding of the antibodies was detected using anti-human IgG Fc-APC antibody (BioLegend, 409305) and the fluorescence of the cells acquired in an LSRII instrument (BD Biosciences). The data was analyzed with FlowJo 10 software (Tree Star, Ashland, OR).

We aimed to overcome the Env dominance in the heterologous Gag/homologous Env prime boost regimen in Andersson and Holst [11] by using both heterologous Gag and Env in prime and boost. Therefore, we designed a new Ad VLV encoding HIVconB Gag and HIV-1 JR-FL (clade B) Env (Ad-HIVB; Fig. 1a) to combine it with the MVA boost encoding SIVmac239 Gag, Pol and Env (MVA-SIV) in mice [11]. For the homologous Gag/Env prime-boost regimen we used the Ad-SIV (encoding SIVmac239 Gag and Env) and MVA-SIV VLVs from Andersson and Holst [11] (Fig. 1a). As the functionality of Ad-SIV had already been shown in the previous study, here we characterized only the Ad-HIVB VLV in vitro and in vivo.

Ad-HIVB-infected Vero cells were stained for HIV-1 Env with three different broadly neutralizing monoclonal antibodies (VRC01, PGT145, PGT151). Analysis by flow cytometry confirmed cell surface expression of Env (Fig. 1b). We further assessed VLP formation capability of the VLV by Western Blot analysis of VLPs purified from the media of Ad-HIVB-infected Vero cells. In two independent experiments Gag-P2A was detectable with the expected band size of approx. 58 kDa (Fig. 1c).

To analyze the immunogenicity of the construct we vaccinated 5 CD1 mice with Ad-HIVB and measured the vaccine-induced T-cell responses after 18 days by ICS followed by flow cytometry (Fig. 1d). Ad-HIVB induced high numbers of HIVconB Gag-specific CD8+ T-cells, which were significantly higher than the HIVconB Env-specific CD8+ T-cell responses. CD4+ T-cell responses to HIVconB Gag were comparably high, taking into account that less CD4+ than CD8+ T-cells are found in the spleen. The number of Env-specific CD4+ T-cells was slightly lower than for Gag. These results confirm antigen-expression and VLP-formation by Ad-HIVB in cells and its immunogenicity in mice.

We aimed to further improve on the heterologous and homologous prime-boost regimen by inducing CD8+ T-cell responses to accessory antigens for acute control of infection and cross-reactive CE-specific CD8+ T-cell responses. To this end, we constructed a separate adenoviral vector encoding a fragment of the mouse Ii T-cell adjuvant, SIV CEs and SIVmac239 subdominant accessory proteins Rev, Vif and Vpr (Ad-Ii-SIVCErvv) as one fusion protein (Fig. 1a). 10 CD1 mice were vaccinated with Ad-Ii-SIVCErvv and the T-cell responses analyzed by ICS after 14 days (Fig. 1e). Few, low CD8+ T-cell responses could be measured to SIV CE and SIVmac239 Rev, but all animals showed high numbers of Vif- and Vpr-specific CD8+ T-cells. CD4+ T-cell responses to Rev were only found in a few mice in high numbers while Vif-specific CD4+ T-cell responses were induced in most animals. This shows that Ad-Ii-SIVCErvv raises T-cell responses to the encoded antigens in outbred mice.

In the two groups receiving only the priming vector Ad-HIVB with or without Ad-Ii-SIVCErvv, no significant differences were detectable in CD8+ T-cell responses towards Gag or Env epitopes (Fig. 2a). The CD8+ T-cell response to HIVconB Gag was significantly stronger than to HIVconB Env in both groups and the Ad-Ii-SIVCErvv group showed slightly broader responses to the different Gags and SIV CE. Including the MVA-SIV boost, HIVconB Gag-specific CD8+ T-cell responses were significantly higher in the group without Ad-Ii-SIVCErvv, while the SIV CE- and SIVagm Gag-specific responses were significantly higher in the Ad-Ii-SIVCErvv group (Fig. 2b). In the boosted Ad-Ii-SIVCErvv group we observed significantly more HIVconB Env-specific CD8+ T-cells than without the boost (p = 0.0468), which was not observed for the group without Ad-Ii-SIVCErvv. As a result, in the group without Ad-Ii-SIVCErvv, CD8+ T-cell responses to HIVconB Gag were significantly higher than to HIVconB Env. Accordingly, the completely heterologous prime-boost regimen could overcome the immunodominance of Env that has been observed when only varying Gag [11], but it could not select for cross-reactive or CE-specific CD8+ T-cell responses. In combination with Ad-Ii-SIVCErvv the CD8+ T-cell responses to HIVconB Gag were even slightly lower than to HIVconB Env, indicating a reappearance of Env dominance. However, addition of Ad-Ii-SIVCErvv to the vaccination regimen increased the breadth of CD8+ T-cell responses to the different Gags tested, which could potentially be explained by the included CEs priming T-cell specificities with broadened recognition of Gag sequences.

Without the boost, the highest CD4+ T-cell responses were induced to HIVconB Gag. In the group without Ad-Ii-SIVCErvv these responses were significantly higher than HIVconB Env-specific CD4+ T-cell responses (Fig. 2c).

Boosting vaccination with MVA-SIV had only a moderate effect on most specificities but significantly increased the CD4+ T-cell responses to SIV CE in the Ad-Ii-SIVCErvv group (p = 0.0251) and SIVmac239 Env in both groups (p < 0.001) (Fig. 2d). This resulted in CD4+ T-cell responses dominated by SIVmac239 Env in both groups and slightly broader responses to the Gags/CE in the Ad-Ii-SIVCErvv group.

Vaccine-induced antibody responses to the 2 Env protein trimers used for immunizations (SIVmac239 and HIV-1 clade B Env) and additionally to HIV-1 clade C, to test cross-clade-reactivity, were tested by ELISA (Fig. 3). In all 4 vaccination groups cross-reactive antibody responses to Env were elicited, indicating broad lentivirus-specific responses induced by the VLVs. Boosting with MVA-SIV had little effect, but in combination with Ad-Ii-SIVCErvv in the prime yielded responses against all 3 tested Envs that were significantly higher than in the boosted group without Ad-Ii-SIVCErvv.

The T-cell and antibody responses measured in the boosted groups were analyzed for correlations using Spearman correlation and the obtained p-values were adjusted for multiple comparisons using the Holm–Sidak method.

Without Ad-Ii-SIVCErvv the SIVmac239 Gag- and HIVconB Env-specific CD8+ T-cell responses as well as antibody responses to SIVmac Env and HIV-1 clade C Env correlated significantly (Table 2a). In the Ad-Ii-SIVCErvv-vaccinated group CD4+ T-cell responses to each of the following peptide pools were significantly correlated with the CD8+ T-cell response to the same pool: HIVconB Gag, SIVmac239 Gag, SIV CE, SIVagm Gag, HIVconB Env and SIVmac239 Env (Table 2b). Besides, CD4+ and CD8+ T-cell responses to SIVmac239 Gag correlated significantly with CD4+ and CD8+ T-cell responses to SIVagm Gag. Antibody responses to HIV-1 clade B Env and HIV-1 clade C Env were correlated.

High Gag- and Env-specific CD8+ T-cell responses were present in both groups, but responses to Gag were higher when Ad-Ii-SIVCErvv was administered additionally (Fig. 4a). Env-specific CD8+ T-cell responses in the group without Ad-Ii-SIVCErvv were significantly stronger than Gag-specific responses in the same group and Env-specific responses in the Ad-Ii-SIVCErvv-vaccinated group. As a result, the ratio of Gag- to Env-specific CD8+ T-cells was significantly higher in the Ad-Ii-SIVCErvv group (p = 0.0021). In the Ad-Ii-SIVCErvv group some low CD8+ T-cell responses were detectable to Rev and moderate and comparably high responses to Vif and Vpr, with the latter 2 being significantly higher than the responses to these antigens in the group without Ad-Ii-SIVCErvv.

To evaluate the functionality of the vaccine-induced memory CD8+ T-cells we analyzed which percentage of IFNγ+ CD8+ T-cells simultaneously produced TNFα (Fig. 4b). Only samples with more than 10⁴ IFNγ+ CD8+ T-cells per spleen were considered, which was the case for all Gag- and Env-specific CD8+ T-cell responses and for the majority of Vif- and Vpr-specific CD8+ T-cell responses. Nearly all mice showed more than 50% double-positive cells for the tested vaccine antigens with geometric means between 65% and 85%, indicating that the quality of induced CD8+ T-cells was good and did not differ distinctly between groups.

When mice were vaccinated with the homologous prime-boost without Ad-Ii-SIVCErvv, higher numbers of CD4+ T-cells reacted to Gag and significantly more CD4+ T-cells reacted to Env than in mice vaccinated with the additional prime (Fig. 4c). In the group without Ad-Ii-SIVCErvv, background CD4+ T-cell responses to the accessory antigens were observed and addition of Ad-Ii-SIVCErvv had only a remote effect on CD4+ T-cell responses towards Rev and Vif. Furthermore, responses to Vpr were not affected by including Ad-Ii-SIVCErvv indicating variation around the background response levels in the assay.

The percentage of IFNγ TNFα double-positive cells for Env-specific CD4+ T-cells was consistently positive, with a geometric mean above 60% for both groups and no detectable difference (Fig. 4d).

In order to assess antibody responses, we calculated the titer as well as the area under the curve (AUC), which both did not differ significantly between the groups (Fig. 5a, b). In addition, the avidity of antibodies did not differ between the groups (Fig. 5c).

The induction of different IgG isotypes was further analyzed and is depicted in Fig. 5d as the AUC ratio to whole IgG serum antibodies detected using an isotype-unspecific pan-IgG secondary antibody. The ratio of IgG2a antibodies to total IgG or IgG1 has been proposed as a functional association with protective antibodies in some models in inbred mice, however, it was quite evident that CD1 mice were heterogenous for IgG2a and IgG2c, which made it impossible to produce meaningful ratios including either isotype. Nevertheless, more IgG2a and IgG2c than IgG2b were induced with no significant differences between the groups while no IgG3 was detected.

In the group without Ad-Ii-SIVCErvv the AUC and titer of antibody responses correlated significantly (Table 3a). In the group with Ad-Ii-SIVCErvv, Env- and Vpr-specific CD4+ T-cell responses showed significant correlation (Table 3b). In addition, both the AUC and titer of the antibody responses correlated significantly with the avidity index.