Recombinant MVA-prime elicits neutralizing antibody responses by inducing antigen-specific B cells i

The RV144 HIV-1 vaccine trial has been the only clinical trial to date that has shown any degree of efficacy and associated with the presence of vaccine-elicited HIV-1 envelope-specific binding antibody and CD4+ T-cell responses. This trial also showed that a vector-prime protein boost combined vaccine strategy was better than when used alone. Here we have studied three different priming vectors—plasmid DNA, recombinant MVA, and recombinant VSV, all encoding clade C transmitted/founder Env 1086 C gp140, for priming three groups of six non-human primates each, followed by a protein boost with adjuvanted 1086 C gp120 protein. Our data showed that MVA-priming favors the development of higher antibody binding titers and neutralizing activity compared with other vectors. Analyses of the draining lymph nodes revealed that MVA-prime induced increased germinal center reactivity characterized by higher frequencies of germinal center (PNAhi) B cells, higher frequencies of antigen-specific B-cell responses as well as an increased frequency of the highly differentiated (ICOShiCD150lo) Tfh-cell subset.

Vaccine development for HIV-1 has faced unprecedented challenges. To date, several HIV-1 vaccine concepts have been evaluated for efficacy in clinical trials1,2,3,4 and a new concept, that of mosaic HIV vaccines5,6,7, is currently being evaluated in the HVTN 705 and HVTN 706 clinical trials (NCT03060629 and NCT03964415)8. However, only one of these concepts, a canarypox ALVAC vector prime with an Env gp120 boost to date has shown modest efficacy in Thailand9. However, the same vaccination strategy of ALVAC vector prime using clade C immunogen conducted in South Africa, has failed to show any efficacy.10 These clinical trials have utilized different priming vectors including plasmid DNA2,4, recombinant adenovirus 5 (rAd5)3, and canarypox ALVAC9, whereas, the ongoing clinical trials are utilizing recombinant adenovirus 26 (rAd26) and modified vaccinia Ankara (MVA) as prime and boosting vectors, respectively8. Analyses of vaccine-induced immune responses in the past clinical trials have shown the role of priming vector and their roles in the induction of protective immunity11. Several preclinical and clinical studies have also assessed rAd26 and MVA as priming immunogen and have shown both of these vectors to be comparable12. However, MVA was shown to be less efficacious than rAd26 when used in the context of a boost in the preclinical study12,13. Therefore, selection of an appropriate priming vector is crucial for the development of a successful vaccine strategy. The immune responses that a priming vector must induce need to at least match the immune responses induced by ALVAC as till date that is the only modestly successful clinical trial.

The RV144 Thai trial showed that a priming immunization with ALVAC vector encoding subunit envelope immunogen that elicited helper CD4+ T-cell responses afforded modest protection against HIV-1 acquisition, whereas, the same immunogen failed to protect when administered alone1. In RV144, vaccinees received priming immunizations with rALVAC-HIV prior to boosting with AIDSVAX B/E gp12014,15. Immunizations with AIDSVAX B/E gp120 alone in the absence of priming in VAX003 failed to prevent HIV-1 infection16. Analyses of correlates of protection for RV144 showed that binding antibody responses against V1V2 region of envelope glycoprotein11 played an important role17. The ALVAC-HIV/AIDSVAX B/E vaccine regimen did not elicit broadly neutralizing antibodies (bNAbs)18. However, the vaccines elicited antibody-dependent cell-mediated cytotoxicity responses19, neutralizing antibodies to the tier 1 viruses and Env-specific CD4+ T-cell responses14,20,21. The development of pathogen and immunogen-specific B-cell responses requires the coordinated function of highly differentiated CD4+ T cells (Tfh) and B cells within germinal centers (GC)22.

Given the limited access to lymph nodes (LNs), several investigators have focused their analysis on circulating GC cell counterparts like circulating Tfh (cTfh)23,24. The predictive value of cTfh for vaccine-induced specific B-cell responses has been recently shown25,26. Furthermore, recent data have shown that optimal help by Tfh cells is critical for the development of bNAbs in a non-human primates (NHP) SHIV model27 possibly by regulating, among other factors, the metabolic program28 as well as the cell cycling/division of germinal center B cells in the dark zone29,30. Therefore, an optimal vaccine regimen must be able to generate robust antigen-specific CD4+ T-cell responses, including Tfh cells, which, in turn, would facilitate the development of humoral responses. A live vector prime/protein boost vaccination strategy has been widely used to generate such responses. However, the optimal priming vector for such an immunization is not known.

Here we have evaluated three priming vectors, plasmid DNA, recombinant MVA, and recombinant VSV with the same protein boost for their ability to prime for HIV-1 Env-specific antibody, CD4+ T and Tfh-cell responses. We found that recombinant MVA-prime optimally induced ICOS+ Tfh CD4+ T cells and elicited higher titers of antibody responses compared with plasmid DNA and rVSV.

Priming with rMVA and rVSV elicited higher antibody-mediated neutralization activity compared with plasmid DNA

Neutralizing antibody responses from all four groups of vaccinated monkeys were assessed against a panel of tier 1 and tier 2 viruses using both TZM-bl and A3R5 cell-based assays. Two weeks following the protein boost, neutralizing antibody responses against the HIV MW965.26 clade C, MN.3 clade B, and SF162 clade B were significantly higher in monkeys that were primed with rMVA or rVSV compared with either plasmid DNA (p = 0.0087, p = 0.0649 and p = 0.0152 against HIV MW965.26 clade C, MN.3 clade B and SF162 clade B, respectively) or protein only (p = 0.002, p = 0.004, and p = 0.002 against HIV MW965.26 clade C, MN.3 clade B, and SF162 clade B, respectively) groups (Fig. 1a, b). Although not significant, rMVA elicited higher neutralization titers for several viruses (HIV-SF162, 92RW020.5, TV.21, HIV-MN, and MW965.26) compared with rVSV that had the highest neutralization titers against HIV 1196.01, a tier 1B neutralizer (Fig. 1b). Neutralizing antibodies against tier 2 virus TV1.21 was weak in magnitude in all four groups of animals. However, monkeys in both rMVA- and rVSV-primed groups had higher titers of neutralization against this virus compared with other two groups. No neutralization was detected against 92RW020.5 (tier 1) and BG1168.1 (tier 2) viruses. An A3R5 cell-based assay was used to assess vaccine-elicited tier 2 neutralizing antibodies in these cohort of monkey, as it is known to be more sensitive than TZM-bl cell-based assay for detecting neutralizing antibodies. We did not detect any neutralization titers against Ce1176 A3, Du151.2 and CE2010 F5 viruses (Supplementary Figure 1, panels a-c). Autologous neutralizing antibodies against C.1086 virus were elicited in all three groups of monkeys that received priming immunizations with either plasmid DNA, rMVA, or rVSV. The magnitude of autologous neutralization was lower in the monkeys that were immunized only with protein without any vector priming (Supplementary Figure 1d). As expected, neutralization titers against TV1.21 were higher in the A3R5 assay than in TZM-bl cells (Supplementary Figure 1e). Both assays confirmed that rMVA and rVSV priming elicited the highest levels of neutralizing activity with rMVA been more potent for some of the tested viruses.

rMVA prime elicited highest level of autologous and heterologous binding antibodies

Next, the binding antibody titers in all groups of the vaccinated monkeys against both clades B and clade C Env antigens was assessed. The panel used to assess vaccine-elicited binding antibody responses consisted of the following immunogens: C.1086 gp120, C.1086 gp140, V1V2 of 1086 C (C.1086 V1V2 tags, C.1086 V1V2 N156Q), B.63521 gp120, B.63521 V1V2 tags, and B.63521 V1V2 N156Q N160Q. Two weeks after the first priming immunizations among all three groups of monkeys that received vector priming, autologous and heterologous binding antibody titers were significantly higher in rMVA-primed group compared with the group primed with plasmid DNA (p = 0.001 and p = 0.04, respectively). Recombinant MVA-primed monkeys also had higher titer compared with rVSV-primed group. Titers for both autologous and heterologous binding antibodies were assessed 2 weeks after final priming immunizations at week 10. At this point, monkeys in plasmid DNA group received a total of three (weeks 0, 4, and 8) and rMVA group received two priming immunizations (week 0 and 8), whereas, rVSV monkeys were primed once on week 0. The priming regimen for the three vectors were determined based on previous works using these vectors.31,32,33 Post-prime titers of binding autologous, as well as heterologous antigens, were significantly the highest in the rMVA-primed group among all vector-primed groups (Fig. 1c). Antibody titers against the V1/V2 region of either C.1086 or clade B immunogen B.63521 were low. Plasma samples from all groups of monkeys were tested again at 2 weeks post-protein boost. As shown in Fig. 1c(week 32), irrespective of the type of priming immunizations, the titers of both autologous and heterologous binding antibodies were very similar across all groups.

Plasmid DNA prime induced higher Th1 type cytokines than priming with either rMVA or rVSV

Cytokine production by T cells upon antigen stimulation is a quantitative way to measure T-cell functionality. Post-prime (upper panel) and post-boost (lower panel) cytokine (IFN-γ IL-2, TNF-α)-secreting central/transitional memory T cells (TCM/TM) in the peripheral blood were quantified in all groups of monkeys (Fig. 2). For the post priming responses (Fig. 2, upper panel), plasmid DNA induced the highest levels of IFN-γ+ or IL-2+ CD4+ T-cell responses, followed by the rMVA and rVSV-induced responses (Fig. 2). Regarding the protein boost induced responses, a similar pattern was observed (Fig. 2, lower panel). Interestingly, plasmid DNA prime was able to induce both IFN-γ+ and IL-2+ significantly higher (p < 0.05; Kruskal–Wallis test) compared with post priming levels while rMVA induced selectively the IFN-γ+ CD4+ T-cell responses (Fig. 2, lower panel). This observation was also confirmed by IFN-γ ELISpot assay where the number of IFN-γ-producing spot-forming cells were significantly higher in monkeys in those two groups. A low frequency of IFN-γ+ IL-2+ CD4+ T cells across the groups was found pre and post boosting (Fig. 2). Therefore, different vectors induce different patterns of circulating cytokine-secreting CD4+ T-cell responses.

Antigen-specific memory CD4+ T cells in the periphery of rMVA-primed monkeys show a unique transcriptome profile

For an in-depth analysis of transcriptional signatures of vaccine-elicited memory CD4+ T-cell populations, we measured the expression signals of 96 genes in circulating sorted TCM and TEMcells using the Fluidigm technology. Peripheral blood mononuclear cells (PBMC) from all 24 rhesus macaques were stimulated with pooled peptides spanning 1086.C gp120 and antigen-specific CD4+ CD154+ TCM and TEM cells were sorted as described in the Methods section. TCMcells were defined as live CD3+ CD4+ CD95+ CD28+ CCR7+ T cells and TEM cells were gated as CD3+ CD4+ CD95+ CD28− CCR7− T cells. Distinct gene expression profiles were observed with or without antigen-stimulated CD4+ populations showing vaccination overall resulted in the activation of both types of memory T cells (Fig. 3a). SVM-RFE (Support Vector Machine with Recursive Feature Elimination) was used to analyze the gene expression of CD4+ CD154+ T cells among all groups. The gene expression profile between peptide stimulated cells and unstimulated cells showed differential expression of several genes important for the development of multiple T-cell lineages (Th1, Th2, Tfh, Th17), regardless of vaccine regimen (Fig. 3b), showing vaccination-induced broad T-cell responses. Furthermore, a distinct profile was found between antigen-specific (CD154+) and bulk CD4+ T cells both in TCM and TEMcompartments at post-prime (Fig. 3c) and post-boost (Fig. 3d) time points. These differences show priming and boosting have differential effects on the TCM and TEM compartments which may be important for the development of the broad T-cell responses observed. Because rMVA immunization provided the best neutralization titers, the gene expression of cells from rMVA group (post-boost time point) was compared with the other groups (Supplementary Table 2). Our analyses revealed an upregulation of genes such as IL-6ST, GP130, Bcl-6 and IL-4 (Tfh factors), IL-10, IL-13 (Th2), and IL-17 (inflammatory-Th17) selectively in CD4+ CD154+ T cells from the rMVA group (Supplementary Table 2). Therefore, our gene analyses data support the neutralizing and antibody titers data further suggesting that rMVA immunization favors the development of neutralizing antibodies compared with other vaccine platforms tested in this study.

Vector priming induces LN immune dynamics favoring the development of vaccine-induced B-cell responses

Follicular helper CD4+ T cells (Tfh) represent a population with unique phenotype, characterized by high expression of CXCR4, CXCR5, PD-1, ICOS, and low expression of CCR7, CCR6, and CD127, and localization in follicular areas and particularly GCs34,35,36. Circulating counterparts of Tfh (cTfh) have been described based on the expression of receptors like CXCR5, PD-1, ICOS24. Thus in this study, the expression of several molecules that are important for Tfh function were analyzed on CD4+ TCM cells (CD3+ CD4+ CD28hiCD95hi) that had low expression of CXCR3, a marker of Th1-like LN Tfh, functionally distinct from “classic” germinal center Tfh cells.37 Regardless of vaccine modality, the majority of CD4+ TCM in LN showed high expression of CCR7 and low expression of PD-1 (Fig. 4a, upper panel). The relative frequency of bulk GC-Tfh (CCR7lowPD-1high) was significantly lower in comparison with CCR7highPD-1lowpopulation and it did not change after prime or boost immunization within each group (Fig. 4a, lower panel). We have previously showed that NHP Tfh cells express a high ICOS phenotype while the expression of SLAM (CD150) can further define Tfh subpopulations with different capacity for in vivo proliferation36. Furthermore, ICOShighSLAMlow was the Tfh population with higher capacity for IL-4 production36 a cytokine critical for the GC B-cell development38. Therefore, we sought to analyze the expression of these receptors on the Tfh cells from different vaccinations. The rMVA group had a significantly higher frequency of ICOShighSLAMlow Tfh cells (p = 0.0087) even compared with rVSV group (Fig. 4b, upper panel). In line with this, the expression of ICOS per cell, judged by the mean fluorescence intensity-MFI, was also significantly higher in the rMVA group at 2-week post-prime compared with the Tfh cells in other groups (Fig. 4b, lower panel). Furthermore, a trend for higher expression of CXCR4 per cell was found on ICOShighSLAMlow Tfh in rMVA compared with other groups (Supplementary Figure 2). Although LN cells from pre-vaccinated animals were not available for the rMVA group owing to technical difficulties in LN collection, our analysis showed no difference between the groups when the bulk non-Tfh and Tfh-cell populations were investigated (Fig. 4a), indicating that the higher frequency of ICOShighSLAMlow Tfh cells, as well as the expression of ICOS per cell, are possibly related to rMVA vaccination.

The frequency of immunogen-specific Tfh CD4+ T cells was analyzed following in vitro stimulation with peptide pool spanning the entire 1086.C gp120 and detection of CD154 mobilization. The group primed with rVSV showed the highest frequency of antigen-specific Tfh CD4+ T cells followed by the rMVA group, particularly after boosting (Supplementary Figure 3a). A similar profile was found when antigen-specific CXCR5hi cTfh cells were analyzed (Supplementary Figure 3b). We further analyzed the frequency of IFN-γ responses in the antigen-specific Tfhs. Again, rVSV was the group characterized by the highest frequency of Tfh cells able to produce IFN-γ (Supplementary Figure 3c). No differences were found when the production of IL-4 was analyzed in all groups. Overall, our data suggest that rMVA prime induces germinal center immune dynamics favoring the development of vaccine-induced B-cell responses even after one priming immunization.

B-cell development inside the germinal center is facilitated by rMVA prime

Next, we investigated the dynamics of B cells in LNs obtained at pre-immunization, post-prime and post-boost time points from all groups of animals. Similar relative frequencies of bulk memory (IgDloIgMlo) B cells were found across the animal groups tested. A similar profile was found when the expression of surface IgG on memory B cells was analyzed (Supplementary Figure 4). Using PNA, a marker of GC B cells36, we found a significant induction of the GC B cells post priming (Week 2) selectively in the rMVA group (Fig. 4c). Furthermore, this induction was associated with increased frequency of immunogen-specific B-cell responses as measured by the binding of a gp120-specific probe post-prime in the rMVA-primed group (Fig. 4d). To confirm the GC activity in monkeys primed with rMVA, we measured plasma CXCL13 levels, a surrogate of GC activity39, in all four groups of monkeys. Plasma CXCL13 levels were measured in all four groups of monkeys at pre-immunization, post-1st prime (week 2) and post-boost time points (week 32). As shown in Fig. 5, two weeks post-prime at week 2 both rVSV- (twofold increase) and rMVA-primed (15-fold increase) monkeys had higher plasma CXCL13 levels than plasmid DNA and protein control group. Monkeys primed with rMVA had a significant increase in plasma CXCL13 level two weeks post-first prime at week 2 (p = 0.008). However, the post-boost plasma CXCL13 level in rVSV-prime group contracted to the pre-immunization level whereas, in rMVA-primed monkeys the post-boost level remained at the post-prime level indicating GC activity as a result of priming with rMVA.

We have investigated biological factors associated with the development of B-cell responses in NHPs vaccinated with different vectors/carriers encoding a clade C HIV-1 Env. The development of antigen-specific antibodies requires the coordinated interaction of highly differentiated CD4+ T and B cells within the germinal center. The quality of provided help by Tfh cells directs the fate of dividing B cells within the dark zone of the germinal center and the development of immunoglobulin mutations directly related to the neutralization potential of antigen-specific antibodies.22 Analysis of plasma Env-specific antibodies showed that rMVA was the regime inducing the highest titers compared with other vectors for both autologous and heterologous Env strains. Furthermore, rMVA boost induced similar or higher neutralizing activity compared with other vectors. Post-boost induced humoral responses can result from the reactivation of memory responses formed during the priming and possibly the generation of new responses during the boosting phase. The titers of the specific antibodies after priming indicate that rMVA generates higher antigen-specific memory B-cell responses. However, it should be pointed out that the vaccinated groups did not receive the same number of priming doses, which can affect the outcome of the immune dynamics under investigation. Despite the lower titers post priming, rVSV boosting led to the highest fold change of antibody titers compared with other vectors. Therefore, we can hypothesize that the quality and/or quantity of rVSV-induced memory Tfh/B cells during priming is different compared with the other vectors. The vaccination scheme that was used in the current study could be, at least in part, a critical determinant for these differences. Further investigation is needed to determine whether the different neutralization activity observed among the vectors, reflects different affinity/avidity of relevant antibodies between the groups and/or different elicited epitope antibody specificities.

Analysis of circulating antigen-specific (cytokine+) CD4+ T cells revealed that plasmid DNA vaccination, associated with the lowest neutralization activity, induced the higher frequency of IFN-γ+ and IL-2+ CD4+ T cells both at post-boost and post-prime time points. Our data indicate a dissociation between the elicitation of classic circulating Th1 CD4+ responses and the generation of neutralizing antibodies. That was more evident in the DNA vaccinated animals indicating that plasmid DNA elicited vaccine responses are more skewed towards a Th1 phenotype, at least in this experimental settings. Compared with rMVA, vaccination with rVSV induced higher levels of CD154+ IFN-γ+ Tfh cells in the LN both at week 2 and week 32. This profile was also associated with higher circulating IFN-γ+ CD4 T cells at week 32 in rVSV compared with rMVA vaccinated animals. Tfh represent a highly heterogeneous group of cells with presumably different functions38. Previous work has shown that CXCR3 + (Th1 like) Tfh cells could support B-cell responses, although the impact of IFN-γ in the in vitro IgG production by B cells was modest and significantly lower compared with the impact of IL-2137. Furthermore, a recent report showed that IL-21 and IFN-γ were produced by different human tonsillar Tfh-cell subsets40. Therefore, the co-expression of IFN-γ with other mediators that could impact the B-cell responses (e.g., IL-21, IL-4) is critical in order to better understand the role of the IFN-γ+ Tfh cells. In our experimental settings the prevalence of IFN-γ+ Tfh cells was not correlated with better antigen-specific LN B-cell responses in the rVSV group. Whether this is owing to a particular cytokine production profile of Tfh cells (our gene sequencing analysis suggests a higher IL-4 expression in cTfh cells induced by rMVA, a vector associated with a lower frequency of IFN-γ+ Tfh cells) or the altered activity of other mechanisms regulated by IFN-γ (e.g., production of BAFF41) needs further investigation.

Upon stimulation, distinct T-cell pathways for Th1, Th2, Tfh, and Th17 cells could be identified by up- and downregulation of selected genes that are important for different T-cell development regardless of vaccine regimen. Our data highlight a select signature of genes capable of distinguishing TCM and TEM populations. TCM and TEM populations differed in CD154 (CD40L) expression when comparing post-prime and post-boost samples. However, CD154 expression is not independently sufficient to distinguish these populations. Direct comparison of post-prime and post-boost populations demonstrated an increase in TCMCD154+ cells post-boost, suggesting the immune system may yield an improved anamnestic response to subsequent exposure to 1086.C Env. Still, rMVA induced a pattern closer to Tfh further supporting rMVA prime in line with its profile of cytokine-induced CD4 T-cell responses and antigen-specific neutralization activity.

In addition to circulating cells, analysis of LN-derived cells, when available, showed that rMVA priming was associated with the highest levels of germinal center (PNAhi) B cells and ICOShiSLAMlo Tfh cells, which are characterized by the highest capacity for IL-4 production36, a cytokine critical for the maturation/switching of B cells42. This profile is concordant with the higher autologous and heterologous antibody-binding titers found after priming with rMVA compared with other groups. Furthermore, the increased post-prime percentages of GC B cells and antigen-specific B cells found in rMVA animals are in line with the high expression of ICOS on Tfh cells post-prime in this group of animals. rMVA vaccination induced the higher levels of ICOS expression on Tfh cells, a receptor possibly mediating the mutual regulation between Tfh and B cells43. Supporting the favorable activity of rMVA vaccination for the development of humoral responses is the induction of significantly higher levels of bulk GC B cells and antigen-specific IgGhi B cells in the follicles even at week 2 post priming. Therefore, the direct comparison between the groups at week 2 (after one priming dose) indicates that rMVA vaccination initiates an early germinal center reactivity favoring the development of antigen-specific antibodies compared with other regimens. Circulating CXCL13, the ligand for CXCR5 and a critical regulator for the trafficking of CD4+ T cells into the germinal center44, has been proposed as a surrogate for GC reactivity associated with vaccine outcome39. We found that only rMVA was able to continuously induce and maintain higher levels of CXCL13 compared with other vectors, in line with the higher frequencies of ICOShiSLAMlo Tfh cells post boosting/post priming. This is in line with previous reports showing that vaccination of mice with rMVA was able to induce tertiary lymphoid structures enriched in CXCL13 and capable of supporting T-cell priming.45,46 Overall, our data show that (i) different vector-based vaccines are associated with differential development of germinal center reactivity and generation of neutralizing antibody responses and (ii) indicate that the germinal center dynamics during the primary phase is a critical factor for the development of the humoral responses after boost. The molecular and cellular mechanisms responsible for this outcome are not known and need further investigation. The use of the NHP model and relevant vaccination strategies could provide guidance for the development of “effective” germinal center reactivity and possible targets for in vivo manipulation aiming to further strengthen the production of bNAbs.