To decipher the specific T_h subset of T_fh cells contributing to the establishment of durable serological memory, we employed live attenuated measles virus vaccine (LAMV) as a model and investigated T_fh responses in peripheral blood and lymph nodes (LN) in a cohort of 16 healthy adult female rhesus macaques (Figure 1A). Consistent with documented evidence of long-term persistence of the memory B cell response to LAMV, rapid recall of measles virus (MeV) IgG ensued at week 2 post LAMV, irrespective of the booster interval (Figure 1B). By week 20, MeV IgG concentrations displayed a 2.6-fold increase compared to baseline, signifying the efficacy of LAMV booster in generating persistent antibodies. These augmented levels of memory antibodies were accompanied by enhanced antibody avidity (Figure 1C), underscoring robustness of the humoral immune response elicited by LAMV.

Figure 1 with 1 supplement see all

Download asset Open asset

Tracking MeV-specific T_fh cells using the activation-induced marker (AIM) assay showed induction of circulating (c)T_fh cells at day 7 post-vaccination (Figure 1—figure supplement 1A). Confirming T_fh cell induction in response to LAMV, activated cTfh cells, identified by co-expression of ICOS and Ki-67, were transiently upregulated at day 7 (Figure 1D, Figure 1—figure supplement 1B). These responding cTfh cells were PD-1 +and their frequencies at day 7, but not day 0, significantly correlated with MeV IgG at Week 20 (Figure 1E). We next proceeded to analyze heterogeneity within cTfh cells with respect to expression of T_h1/2/17 chemokine receptors. Our temporal analysis of the responding cTfh compartment demonstrated significant induction of CXCR3 +T_fh1 cells at day 7 (Figure 1F). In contrast, we observed a decrease in frequencies of CCR4-expressing cTfh2 subsets and CCR6-expressing cTfh17 subsets. Utilizing Boolean analysis, we further identified that the majority of responding cTfh cells expressed CXCR3, indicating induction of cTfh1 cells in response to LAMV (Figure 1G).

Expanding our investigation to include phenotypic and molecular features of CXCR5⁺ PD-1⁺⁺ GC T_fh cells (Figure 1H, Figure 1—figure supplement 1C), we found that of the GC T_fh cells expressing either CXCR3 or CCR6, CXCR3 expression predominated (median %CXCR3+, 28%). Additionally, we observed that on average 7% of GC T_fh cells expressed CCR6. Notably, our analysis did not reveal substantial expression of CCR4 within GC T_fh cells (Figure 1I). Molecular analysis of sorted CXCR5 +PD-1+/++LN T_fh cells (Figure 1—figure supplement 1D) further provided insights into their functional specialization, with the induction of key transcription factors BATF and IRF4, as well as the expression of chemokine receptors CCR5 and CCR6, indicating the activation of T_h1 and T_h17 programs within GC T_fh cells (Figure 1J–K). Collectively, these findings demonstrate that both blood and GC T_fh cells, during the peak effector response following LAMV immunization, exhibit a predominant T_fh1 helper bias, with GC T_fh cells also demonstrating a T_fh17 bias.

Building on our observations made with LAMV, we next investigated the potential of fine-tuning T_fh responses towards T_fh1 and T_fh17 profiles for generation of persistent anti-HIV-1 Env antibodies. We immunized two cohorts of rhesus macaques against HIV Env using a DNA prime and protein boost vaccination strategy tailored towards either a T_fh1 (n=6) or mixed T_fh1/17 (n=6) response (Figure 2—figure supplement 2). For both vaccine regimens, animals received an initial intradermal DNA prime vaccine encoding HIV-1 Env gp160 (C.1086C) at weeks 0, 4, 8, followed by an subcutaneous HIV C.ZA 1197 MB Env gp140 protein boost vaccination at weeks 12 and 20. To tailor vaccine induced T_fh responses, we included additional T_fh polarizing factors with DNA prime vaccines encoding for cytokines IP-10 (T_fh1 regimen) or IP-10 and IL-6 (mixed T_fh1/17 regimen), coupled with corresponding gp140 protein boost formulated in either the T_h1 polarizing adjuvants, MPLA +QS-21, or the T_h1/17 polarizing CAF01 adjuvant platform, respectively. Over the course of both the priming and boost stages, we performed routine collection of both blood and rectal secretions from animals to assess the extent of systemic and mucosal humoral immunity elicited by the two regimens. Additionally, we collected inguinal LN biopsies and fine needle aspirates of LN (FNA) at baseline, during the DNA prime stage, and following both protein boost to comprehensively characterize vaccine-elicited GC T_fh responses in draining LNs. GC responses typically peak between 2 and 3 weeks post first booster with more rapid recall kinetics after the second booster (Iyer et al., 2015). Our rationale for sampling LN 3 weeks following the second booster was to assess whether the reported effects of CAF01 on sustained antigen release might result in GC persistence (Henriksen-Lacey et al., 2011; Pedersen et al., 2020).

To gain insights into LN T_fh responses induced by MPLA and CAF01 platforms, we utilized complementary approaches of in-situ protein expression analysis of GCs and RNA sequencing of Env-stimulated CD95 +CD4 T cells of selected animals in each vaccine group (Figure 2A). To track immunological changes within GCs, we applied the GeoMx Digital Spatial Profiler to assess protein expression of key immune targets longitudinally at baseline (BL, prior to DNA immunization), week 2 post-protein 1 (P1w2) and week 3 post-protein 2 (P2w3) immunizations. We stained tissue sections with morphological markers (CD20, CD3, Ki-67) and collected circular regions of interest (ROI) based on co-localization of CD3 with CD20 +Ki-67+GC B cells. For each tissue, two to three ROIs were selected to ensure adequate representation of GCs within each animal. This approach allowed us to accurately identify GCs and analyze protein expression of both T and B cells within these specific regions (Figure 2B).

Figure 2 with 2 supplements see all

Download asset Open asset

Thirty-two protein targets were profiled simultaneously, along with isotype controls (Ms IgG1, Ms IgG2A, Rb IgG) and housekeeping proteins (S6, Histone H3, GAPDH). After applying quality control measures, we calculated the signal-to-noise ratio for each target which provided robust assessment of protein expression levels (). CD45 and B2M were among the most abundantly expressed proteins while LAG3 and PD-L1 demonstrated low signal expression. To address variations in ROI surface area and tissue quality, normalized protein expression (NPE) was calculated for each protein. NPE analysis revealed robust expression of key lineage proteins, CD45, CD20, CD3, CD4, and CD8, as well as GC markers PD-1 and Ki-67 within ROIs (Figure 2C). We observed notable associations between protein expression patterns within the GC microenvironment; specifically, a strong correlation between the expression of CD4 and CD20 was observed (). Furthermore, a significant association between the co-stimulatory molecule 4-1BB and CD4 expression () was observed within the GC. Association of the innate immune protein stimulator of interferon genes (STING) with both 4-1BB and CD4 expression suggested a potential interplay between innate and adaptive immune pathways within the context of GC responses. Altogether, the protein expression data provide evidence for coordinated expression of key proteins involved in GC function in both vaccine groups ().

To identify specific proteins and pathways activated in response to vaccination, we performed differential expression analysis on a per-protein basis. We modeled NPE using a linear mixed-effect model (LMM), which allowed us to account for sampling multiple ROIs per tissue within each animal (FDR p value cut-off,<0.05). Volcano plots illustrate proteins induced at P1w2 (Figure 2D) and P2w3 (Figure 2E) following vaccination. Induction of 4-1BB, ICOS, PD-1, CD44 post boost indicated activation of T cells to antigen stimulation within GCs in both vaccine groups (Figure 2F). The higher relative expression of these proteins at P2w3 was consistent with the reported effects of CAF01 in GC persistence. We found that expression of the checkpoint inhibitors CTLA4 and PD-L1 was significantly reduced post CAF01 vaccination, while CD4 expression increased. In contrast, expression of the regulatory enzyme, IDO1; checkpoint inhibitor, PD-L2; and CD256 (B7-H3), and the innate sensor, STING, was significantly higher post vaccination in the MPLA group (Figure 2D–E and G).

Across vaccine regimens, significant differences were observed at P1w2 (Figure 2H). Significantly higher expression of Ki-67 with MPLA compared to CAF01 was suggestive of a more rapid GC response (Figure 2H). Furthermore, the checkpoint inhibitory receptors LAG3, CTLA4, as well as IDO1, exhibited higher expression levels with MPLA at P1w2. Importantly, no significant differences were observed at BL, validating the specificity of immune activation to vaccination. Altogether, the in situ protein analysis indicated robust induction of GC responses across vaccine platforms with distinct qualitative and quantitative effects initiated by MPLA versus CAF01 suggesting potential differences in regulatory mechanisms and immune activation pathways between the vaccine regimens.

To assess the extent of T_h1/T_h17 programming in CD4 T cells induced by vaccination, we profiled the transcriptome of CD95 +CD4 T cells isolated from the LN. We sort purified and RNA sequenced naive (CD95-) and CD95 + CD4 T cells at BL, P1w2, and P2w3 following overnight stimulation with HIV-1 Clade C Env peptide pools. A total of 18 samples per group, which included three biological replicates per subset, were sequenced in parallel resulting in over 25 million high-quality reads per sample (Figure 3A, Figure 3—figure supplement 1A-B). The transcriptomes of naive versus CD95 + CD4 T cells post immunization, in alignment with their distinctive biological states, exhibited clear demarcations along distinct dimensions within the principal component analysis (PCA) plot (Figure 3B). Transcription factors (TF) involved in regulating distinct cellular differentiation programs were enriched in CD95 + CD4 T cells at P1w2 and P2w3. These included HOPX, IKZF2, PRDM1, RUNX2, associated with effector differentiation. Additionally, TF such as BATF, BCL6, BHLHE40, EGR2, and TOX2 involved in T_fh differentiation were also identified. Furthermore, we identified genes regulating T_h1 (STAT1, TBX21, EOMES), T_h17 (AHR), and Treg (FOXP3, NR4A2) programs. Moreover, the expression of interferon regulatory factors (IRFs) controlling T cell differentiation (IRF1, IRF5, IRF8) was observed, demonstrating that CD95 + cells post vaccination encompassed effector and fully differentiated CD4 T cell subsets (Figure 3C).

Figure 3 with 1 supplement see all

Download asset Open asset

We first assessed temporal changes in gene expression by focusing on differentially expressed genes (DEGs) within CD95 + cells at P1w2 and P2w3, compared to BL (Venn diagram in 3D). Notably, a significantly higher level of transcriptional perturbation was observed in response to CAF01, particularly at P2w3, within both naive and CD95 + CD4 T cells (top bubble plot in 3D, Figure 3—figure supplement 1C), suggesting that CAF01 induces extensive gene expression changes in a cell-extrinsic manner. By specifically examining genes induced exclusively in CD95 + cells, but not naive cells, at P1w2, we made several observations. The T_h1 transcriptional regulator STAT1, as well as IFNγ-induced genes such as IRF7, IRF8, IFI30, XAF-1, and CXCL11, were among the most highly expressed genes in MPLA (bottom bubble plot, 3D), highlighting the activation of T_h1 programs following vaccination. Indeed, interrogation of top 3 enriched pathways utilizing the Reactome database (Fabregat et al., 2017) revealed that genes encompassing IFNγ, IL-7, and IL-10 signaling networks were induced with MPLA at P1w2. In the CAF01 group, we observed enrichment of co-stimulatory molecules TNFRSF8 (CD30) and TNFRS13B (TACI), expressed by activated T cells, along with expression of the anti-apoptotic molecule BCL-2 and the transcription factor Foxp3.

Analysis of genes downregulated showed pathways regulating cell cycle inhibitors p21 and p53 were enriched with MPLA while pathways regulating cellular responses to stimuli and stress were downregulated with CAF01 (Figure 3E). Collectively, transcriptional profiles demonstrated activation of effector programs within CD4 T cells with common and unique pathways induced across vaccine platforms.

Consistent with this, CD95⁺ CD4 T cells at P1w2 and P2w3 exhibited notable clustering patterns across vaccine regimens. Particularly, tighter, and more distinct clusters were observed at P1w2 compared to P2w3 notably with CAF01 (Figure 3F, Figure 3—figure supplement 1D-E). Exploration of genes regulating T_h1 and T_h17 programs within CD95 + CD4 T cells at P1w2 showed that CD95 + CD4 T cells in both MPLA and CAF01 groups exhibited overlapping expression profiles, with genes regulating major classes of cytokine-cytokine receptor interactions represented. The MPLA group exhibited higher expression levels of IFN-induced chemokines CXCL10 and CXCL11, IL1 family members IL18 and IL1RAP, IL15 family members IL15 and IL15RA, as well as TNF cytokine family members FAS, TNFα, TNFRSF8 (CD30), and TNFRSF9 (4-1BB) (Figure 3G–H, Figure 3—figure supplement 1F).

In contrast, the CAF01 group demonstrated expression of the T_h17 receptor IL23R, along with TNF family members FASLG, TNFRSF13B (TACI), TNFRSF11 (RANKL), and TNFRSF12 (DR3) within CD95 + cells with higher relative expression of IL17A observed with CAF01, indicative of T_h17 responses (Figure 3I, Figure 3—figure supplement 1G). Collectively, this gene-level analysis indicated strong engagement of T_h1 molecular programs within LN CD4 T cells with MPLA.

After confirming the robust induction of GC responses by both vaccines and observing distinct molecular programs in CD4 T cells, we proceeded to investigate the phenotypic and functional characteristics of T_fh cells. We examined the GC T_fh cell subset characterized by the expression of CXCR5 and high levels of PD-1, using the gating strategy outlined in Figure 4—figure supplement 1A. Our analysis of the lymph nodes revealed an increase in the frequencies of GC T_fh cells following each protein boost in both groups compared to baseline. Importantly, there were no statistically significant differences observed between the two adjuvant platforms, as shown in Figure 4A. We further examined the frequencies of T_fh cells characterized by the expression of CXCR5 and PD-1 (CXCR5 +PD-1+), which also exhibited a significant increase following the first protein boost in both groups (Figure 4—figure supplement 1B).

Figure 4 with 1 supplement see all

Download asset Open asset

To gain insights into the T_fh helper profiles within the GC, we assessed the expression of CXCR3 (T_fh1) and CCR6 (T_fh17) on the GC T_fh cells (flow plot in B). Both adjuvants induced GC T_fh1 cells, as evidenced by the expression of CXCR3. However, a selective increase in GC T_fh1/17 cells was observed with CAF01, while the frequencies of GC T_fh17 cells did not show a significant increase following immunization (Figure 4C).

At P1w2, our investigation into the relative frequencies of polarized GC T_fh subsets revealed significant insights. We observed that most GC T_fh cells retained a non-polarized state (with respect to expression of CXCR3 and CCR6), with no substantial distinctions noted between the CAF01 and MPLA groups. Notably, the T_h1 GC subset demonstrated a significant increase in the MPLA group, while the T_h17 subset exhibited a higher prevalence in the CAF01 group (Figure 4D). This observation was further supported by Boolean analysis, which revealed a greater proportion of the T_h17 subset associated with CAF01, and a higher proportion of the GC T_fh1 subset with MPLA at P1w2, as illustrated in Figure 4E.

Following the observed induction of LN T_fh1 responses, we quantified the Env-specific CD4 T cell response using intracellular cytokine staining (ICS). The analysis revealed higher relative frequencies of Env-specific T_fh cells and Env-specific CXCR5- CD4 T cells producing IFNγ with MPLA compared to CAF01 (Figure 4—figure supplement 1C). To obtain a more comprehensive understanding of the cytokine profile within the LN T_fh1 and T_fh17 subsets, we assessed cytokine production after PMA/Ionomycin stimulation (Figure 4F). By utilizing the discrete expression patterns of CXCR3 and CCR6 on CXCR5 +CD4 T cells in the LN, in conjunction with the canonical cytokines IFNγ (T_h1), IL-17 (T_h17), and IL-21 (T_fh), we made several important observations. Firstly, we found that within the T_fh1 subset, IL-21 +IFNγ+co-producing cells were more abundant, and their levels were significantly increased with MPLA stimulation (Figure 4G). Conversely, within the T_fh17 subset, we observed that IL-21 +IL-17+co-producing cells were more abundant, and their levels were significantly higher in the CAF01 group. Notably, the T_fh1/17 subset demonstrated the highest degree of polyfunctionality, as evidenced by the production of multiple cytokines.

We further analyzed the cytokine profiles by gating IFNγ and IL-21 co-producers versus IL-21 single producers. This analysis revealed that both subsets, T_fh1 and T_fh17, produced IL-2 and TNFα. However, IL-17 production was predominantly observed in IFNγ-negative, IL-21 + cells (Figure 4—figure supplement 1D). To assess systemic levels of IL-21, we measured serum IL-21. We found a significant induction of IL-21 in both vaccine groups at day 7 post-immunization, with no observable differences observed across the adjuvant platforms (Figure 4—figure supplement 1E). This suggests that the induction of IL-21 is a common feature of both vaccine formulations. Overall, these findings suggest that MPLA preferentially induces the production of IL-21 + IFNγ+ cells within the T_fh1 subset, while CAF01 promotes the generation of IL-21 + IL-17+ cells within the T_fh17 subset. Furthermore, the T_fh1/17 subset exhibits the most diverse cytokine production profile among the subsets examined.

Next, we investigated whether the effects of MPLA versus CAF01 adjuvants were also evident in cTfh responses. We observed a significant increase in proliferating cTfh cells, with a predominant induction of cTfh1 cells, similar to the phenotype observed in the LN, when MPLA was used as the adjuvant (Figure 4H). In contrast, no evidence of a T_fh17 or T_fh1/17 bias in cTfh cells was observed with CAF01 (data not shown). To evaluate Env-specific T_fh cell abundance, we utilized the AIM assay (Figure 4—figure supplement 1F). The analysis showed a higher relative induction of Env-specific T_fh responses with MPLA at P1w1 but not P2w1 (Figure 4I), which was further corroborated by ICS assays (Figure 4J). The frequencies of CXCR5 + cells co-producing IL-2 and IFNγ, as well as cells co-producing IL-21 and IFNγ, indicated that higher numbers of IFNγ+Env-specific cTfh cells were induced with MPLA at P1w1 and P2w1 (Figure 4K). Therefore, formulation of gp140 with MPLA induced strong T_h1-polarized GC T_fh responses, characterized by a higher magnitude of IFNγ+anti Env T_fh cells.

We next sought to determine whether induction of higher frequencies of Env-specific T_fh1 cells with MPLA could effectively enhance B cell proliferation and differentiation, in turn, promoting the development of Env antibodies with heightened affinity and increased durability. To this end, we performed a wide repertoire of quantitative and functional assays to assess the magnitude, durability, affinity, avidity, reactivity to Group M consensus protein, and effector functionality of vaccine-induced humoral responses to Env. Longitudinal measures of binding antibody concentrations in sera against C.1086 Env showed rapid and equivalent recall in all animal’s post P1, with antibodies reaching peak levels after the second protein boost. While antibody kinetics were comparable across vaccine groups, significantly higher antibody magnitude was elicited by MPLA at week 8 post second boost and until 30 weeks after final immunization (Figure 5A). At this protracted memory time point, serum antibody concentrations were, on average, 2.5-fold higher with MPLA (median [ng/ml]; MPLA, 5566 versus CAF01, 2155) (Figure 5B). To assess CD4 correlates of antibody durability, we asked whether the magnitude of vaccine-elicited GC T_fh subsets after each boost predicted antibody concentrations at week 30. Frequencies of GC T_fh cells (at P1w2) were positively associated with antibodies at week 30 (r=0.6, p<0.05, data not shown). Two additional key cellular determinants of antibody durability emerged from our analysis: GC T_fh1 cell magnitude (p<0.05, r=0.65) (Figure 5C) and frequencies of IFNγ+IL-2+Env-specific LN T_fh cells (p<0.05, r=0.62; Figure 5D). These data in concert with IFNγ-regulated molecular programs induced in LN CD4 T cells support a role for GC T_fh1 induction in antibody persistence.

Figure 5

Download asset Open asset

Using a binding antibody multiplex assay (BAMA), we further confirmed that stronger memory responses were elicited by MPLA, evidenced by higher median serum IgG antibody concentrations to C.1086 gp140 and gp120 antigens measured as area-under-the curve (AUC) over time (Figure 5E–F). Thus, utilizing two orthogonal approaches, the data showed significantly higher and protracted serum antibody responses against autologous Env with MPLA, indicative of efficient T_fh cell mediated B cell recruitment into the long-lived plasma cell pool.

We next sought to understand if induction of a T_fh1 profile was also associated with improved Env reactivity to Group M consensus proteins. Therefore, we tested serum reactivity against an antigen panel comprising of consensus gp140 proteins and found that antibodies with higher cross-clade reactivity were induced with MPLA relative to CAF01 (Figure 5G–H). We extended these analyses to assess reactivity to V1V2 loops of 1086 C and noted that antibodies with significantly higher specificity to these important regions within Env were induced in both vaccine regimens (Figure 5I).

Having established induction of robust serum Env antibodies with MPLA, we next determined whether Env-specific IgG reactivity would also be higher within rectal secretions. After normalization for total IgG levels within rectal secretions, the data showed that Env antibodies were effectively recalled in most animals at week 2 post second protein boost and persisted in all animals 8 weeks after final immunization. However, akin to the systemic compartment, gp140 IgG in rectal secretions were significantly higher with MPLA at memory time point (Figure 5J), with sixfold higher levels in MPLA relative to CAF01 at week 8 after final immunization (Figure 5K). In both vaccine regimens, the induction of Env-IgA antibodies was poor, preventing a quantitative assessment of IgA responses in secretions.

Since antibody effector functions, facilitated by interaction of Ig constant region with cognate Fc receptors on innate cells, mediate protection from acquisition and viral control (Carpenter and Ackerman, 2020), we assessed the IgG subclass profile at week 2 post second protein boost. Both CAF01 and MPLA +QS-21 induced Env-specific IgG1, and modest levels of IgG2, and IgG4 isotypes with higher IgG1 in all animals, with exception of one animal with higher IgG4 (Figure 6—figure supplement 1).

Significantly higher IgG1 was induced by MPLA relative to CAF01 (Figure 6A), while IgG2 and IgG4 antibody subclasses were comparably induced (Figure 6B–C). The improved IgG1 subtype antibody responses with MPLA prompted us to determine whether a corresponding increase in antibody effector functions might also be observed. Measurement of antibody-dependent phagocytosis (ADP) suggested enhanced effector functions with MPLA (Figure 6D–E) which strongly correlated with Env IgG1 but not IgG2 or IgG4 (Figure 6F).

Figure 6 with 1 supplement see all

Download asset Open asset

Finally, we observed higher levels of neutralizing antibodies against the tier 1A MW965.26 at peak (2 weeks post second protein boost) with MPLA relative to the CAF01 platform (Figure 6G). Antibody avidity measurements revealed relatively higher avidity with MPLA against autologous Env and to consensus Env proteins but not V1V2 loops (Figure 6H–L). Collectively, these findings provide evidence for significantly improved antibody quality associated with induction of GC T_fh1 cells.

Finally, we sought to determine the degree to which T_fh responses elicited during the priming phase influenced antibody responses post first protein boost. To this end, we analyzed blood samples at weeks 0, 1, 2, and 4, while also evaluating FNA of the draining LN at week 2 after the 3rd DNA prime (Figure 7A).

Figure 7 with 1 supplement see all

Download asset Open asset

Animals were immunized with C.1086 DNA expressing IP10 (DNA_IP10) which primes stronger GC T_fh responses relative to DNA-alone (Verma et al., 2019). To further enhance GC T_fh responses, beyond that induced by DNA_IP10, we engineered DNA_IP10 to co-express IL-6, a key cytokine promoting T_fh differentiation. Our vector achieved coordinate expression of Gag and Env antigens together with IP10 +IL6 as confirmed in transfected 293T cells (Figure 7B). Following three rounds of intradermal immunization with DNA_IP10 or DNA_IP10+IL6, we assessed proliferating cTfh cells (Ki-67⁺ PD-1⁺ of CXCR5 +CD4 T cells in blood, Figure 7C). We observed a significant increase at weeks 1 and 2 with DNA_IP10+IL6 but not DNA_IP10 indicating that IP10 +IL6 incorporation promoted T_fh differentiation. Additionally, T_fh induction was accompanied by notable T_fh1 skewing within proliferating cTfh cells (median cTfh1 at day 0 (% CD95), 0.89% vs 1.48% at day 7 and 14). Measurement of antigen-specific responses by AIM assay demonstrated induction of Env and Gag-specific cTfh cells with a trend for higher responses elicited by DNA_IP10+IL6 (median, 0.42%, IQR 0.3–1.7%) relative to D_IP10 (median, 0.12%; IQR: 0.04–1.4%, Figure 7—figure supplement 1).

Based on the transient accumulation of cTfh cells in the blood, we further explored GC T_fh responses. Out of 12 collected FNAs, 8 were successful (four per vaccine group), and demonstrated significant increase in GC T_fh frequencies following vaccination (Figure 7D). Gating on Ki-67 + cells encompassing both CXCR5 + PD-1+and PD-1 ++ subsets, an increase in activated T_fh cells within the LN was observed (Figure 7E), accompanied by a corresponding increase in the T_fh1 subset (Figure 7F). Altogether, these data demonstrate induction of GC T_fh1 responses during the priming phase post immunization with DNA_IP10 or DNA_IP10+IL6.

Measurement of C.1086 Env antibodies in serum revealed induction of antibody responses after the 3rd prime in 8/12 animals, with comparable kinetics across vaccine groups. Notably, antibody concentrations 4 weeks post 3rd DNA prime strongly correlated with both the frequencies of proliferating T_fh cells (Ki-67 +PD-1+CXCR5+) within the LN and with the proliferating T_fh1 (Ki-67 +PD-1+CXCR5+CXCR3+) subset at 2 weeks post 3rd DNA prime (Figure 7G–H). Moreover, the strong association of antibody concentrations after the 3rd DNA prime with peak antibodies following the first protein boost (Figure 7I) indicated efficient recall of Env-specific memory B cells induced during the prime. Overall, the data show that stimulation of T_fh1 cells during the DNA priming stage is closely linked with robust antibodies post boost.

In summary, our results strongly support maximizing GC T_fh1 responses during both the prime and boost phases as a strategy to generate potent and durable humoral immunity. These findings have significant implications for the development of more effective vaccination strategies aimed at eliciting robust and long-lasting immune responses.

For LAMV studies, sixteen female colony-bred Indian origin rhesus macaques (Macaca mulatta) were utilized. Animals were dichotomized by age into two groups; young [n=8, mean = 4 years] and aged [n=8, mean = 16 years]. Young adults with a higher baseline weight [mean = 5.9 kg] were selected for the cohort to closer match the aged animals in size [mean = 8.6 kg]. For the HIV vaccine study, 12 adult (10 males and 2 females) Indian origin rhesus macaques (Macaca mulatta) were utilized. At study initiation, animals were 3.4–5.6 years of age with a median weight of 6.8 kg. Animals in both studies were SIV negative (SIV-), simian T-cell leukemia virus negative (STLV-), and simian retrovirus negative (SRV-); and had no history of dietary, pharmacological, or surgical manipulation. All animals were bred and housed at the California National Primate Research Center (CNPRC) in accordance with the American Association for Accreditation of Laboratory Animal Care (AAALAC) guidelines. All studies were approved by the University of California, Davis Institutional Animal Care and Use Committee (IACUC).

The animals received an IM booster vaccination of canine distemper-measles vaccine (Vanguard).

Animals were immunized with DNA vaccines intradermally in both thighs at weeks 0, 4, and 8. At each prime, animals received 4 mg of the pGA2/JS2 plasmid DNA encoding either SHIV C.1086 T/F Env +rhesus interferon-induced protein (IP)–10 (Group 1; n=6) or SHIV C.1086 T/F Env +rhesus interferon-induced protein (IP)–10+rhesus interleukin (IL)–6 (Group 2; n=6). Subsequent protein boosters were administered at weeks 12 and 20 with Group 1 animals receiving 50 μg C.ZA 1197 MB gp140 protein adjuvanted with MPLA +QS-21 and Group 2 animals receiving 50 μg C.ZA 1197 MB gp140 protein adjuvanted with CAF01 delivered in a 175 μL volume subcutaneously into each thigh.

CAF01 was generously provided by Statens Serum Institut, Denmark. Admixing with protein was performed by addition of 25 μg of C.ZA 1197 MB gp140 protein (Immune Technology, USA) to 250 µL of CAF01 (625 μg DDA +125 μg TDB) adjuvant formulation. The solution was vortexed vigorously for 15–30 s followed by visual inspection. Remaining protein was added in 10 μg increments with intermittent vortexing until a final amount of 100 μg gp140 was achieved.

Synthetic MPLA was purchased from InvivoGen, USA. 100 μg C.ZA 1197 MB gp140 protein was dissolved in a solution of 100 μg MPLA with 50 μg QS-21 saponin [Desert King International, USA] (2:2:1) in PBS.

Lymph node (LN) biopsies were obtained at baseline (week 0), week 14 (week 2 post first protein boost) and week 23 (week 3 post second protein boost) and processed as described previously (Verma et al., 2019). Isolated cells were washed in complete media, counted, and cryopreserved until subsequent analysis. Two weeks after the 3rd DNA immunization, fine needle aspirates of LN were obtained using a 22-gauge needle, as previously described (Verma et al., 2019). PBMCs were isolated from whole blood collected in CPT vacutainer tubes at weeks 0, 4, 9, 10, 13, 21, and 32. Serum was collected from animals at weeks 0, 4, 8, 8+day 3, 9,10, 12, 13, 14, 16, 20, 20+day 3, 21, 22, 23, 28, 32, 50 timepoints and stored at –80 °C until subsequent analysis. Rectal secretions were collected for assessment of mucosal antibody production using premoistened Weck-Cel sponges as previously described (Verma et al., 2019).

Detection of antigen specific CD4 T cells detection was quantified by activation induced marker (AIM) and intracellular cytokine staining (ICS). PBMCs/LN cells were stimulated with overlapping peptide pools of HIV consensus C and HIV-1 C.1086 Env gp140C protein (NIH AIDS Reagent Program) in AIM/R10 media in the presence of 0.2 μg CD28/49d co-stimulatory antibodies (BD) per test. As a positive control, cells were stimulated with 1 X Cell Stimulation Cocktail (PMA and ionomycin) (eBioscience, USA). Unstimulated controls were treated with volume-controlled DMSO (Sigma-Aldrich). Tubes were incubated in 5% CO₂ at 37 °C overnight for AIM assay. For ICS assay, after 1 hr of stimulations, protein transport inhibitors 2 μl/mL GolgiPlug (Brefeldin A) and 1.3 μl/mL GolgiStop (Monensin) (BD, Biosciences, USA) were added to the tubes for 8 hr at 37 °C, 5% CO₂. Following stimulation, the cells were stained for AIM and ICS surface markers (see Table 1). Cells were then fixed with cytofix/cytoperm for 10 min at 4 °C, permeabilized with 1 X Perm wash buffer (BD, Biosciences, USA), and stained for intracellular markers (see Table 1) for 45 min. Cells were then washed and acquired the same day on a BD FACS Symphony.

T_fh cell staining was performed on whole blood and LN cells as previously described (Verma et al., 2019). Samples were acquired on BD FACS Symphony with FACS Diva version 8.0.1 software and data were analyzed using FlowJo (Versions 9 and 10). For cell sorting, cryopreserved cells were enriched by NHP CD4 isolation kit (Miltenyi Biotec,USA) and stained with CD3, CD4, CXCR5, CD95, and live/dead in complete media (incubated for 1 hr at 4 °C on a shaker). Stained cells were washed twice with 5 mL RPMI plain media (Gibco, USA) and resuspended in 0.5 mL of sorting buffer containing 2% FBS +PBS. The cells were then filtered through a 40 μM strainer into 5 mL sterile FACS tube (Blue cap tube). Cell sorting was performed using a BD FACSAria III. Naive, and CD95 +CD4 T cell populations were collected in complete media supplemented with 20% FBS. Sorted cells were stimulated with overlapping peptide pools of HIV consensus C and HIV-1 C.1086 Env gp140C protein (NIH AIDS Reagent Program) in R10 media in the presence of 0.2 μg/mL CD28/49d co-stimulatory antibodies (BD) for 14 hr. The culture supernatant and stimulated cells were collected and stored at –80 °C for subsequent analysis.

Serum IL-21 cytokine was quantified using an IL-21 ELISA Development kit (Novus Bio, USA) in accordance with manufacturer’s protocol. Briefly, the capture mAb (MT216G) was diluted to 2 μg/mL in PBS, pH 7.4 and added (100 μL/well) to 96-well microtiter plates with high binding capacity (Thermo Fisher, USA) and incubated overnight at 4 °C. The next day, plates were aspirated, and wells blocked with 200 μL/well of PBS with 0.05% Tween 20 and 0.1% BSA (incubation buffer). After 1 hr of incubation, plates were washed five times with PBS containing 0.05% Tween 20 (300μL/well). Samples or working standards were added at 100 µL/well and incubated for 2 hr at room temperature. After washing, plates were incubated for 1 hr with 100 μL/well of detection mAb diluted to 1 μg/mL in incubation buffer. Streptavidin-HRP conjugates were diluted 1: 1000 in incubation buffer and added to the plates at 100 µL/well for 1 hr followed by washing. Plates were washed and then developed with TMB substrate (Thermo Fisher, USA), and the reaction quenched with 0.2 M H₂SO₄ (Sigma, USA). Absorbance was recorded using a Spectramax 5 plate reader (Molecular Devices) at 450 nm with a reference filter at 570 nm within 15 min. The concentration of IL-21 in serum was calculated based on an IL-21 standard curve using SoftMax Pro.

RNA was extracted from sorted naive, and CD95 +CD4 T cells using a RNeasy plus mini kit (QIAGEN) (Table 2). Isolated RNA sample quality was assessed using a BioAnalyzer RNA pico assay (Agilent Technologies Inc, California, USA) and quantified by Qubit 2.0 RNA HS assay (Thermo Fisher, Massachusetts, USA). Library construction was performed based on manufacturer’s recommendation for the SMART-Seq v4 Ultra Low Input RNA Kit (Takara Bio USA Inc, California, USA) followed by the Nextera XT DNA Library Prep Kit (Illumina, California, USA). Final library quantity was measured using the KAPA SYBR FAST qPCR and library quality evaluated using a TapeStation D1000 ScreenTape (Agilent Technologies, CA, USA). Final library size was about 450 bp with an insert size of about 300 bp. Illumina 8-nt dual-indices were used. Equimolar pooling of libraries was performed based on QC values and sequenced on an Illumina NovaSeq S4 (Illumina, California, USA) with a read length configuration of 150 PE for 40 M PE reads per sample (20 M in each direction). Reference rhesus macaque genome (Macaca_mulatta _GCF_003339765.1_Mmul_10) and gene model annotation files were downloaded directly from the genome website. Index of the reference genome was built using Hisat2 v2.0.5 and paired-end clean reads were aligned to the reference genome using Hisat2 v2.0.5. The mapped reads of each sample were assembled by StringTie (v1.3.3b) using a reference-based approach.

The quality of the raw RNA-seq data was assessed using FastQC and poor-quality ends were trimmed (Trimgalore). High-quality sequences were aligned against the Macaca mulatta reference genome using STAR aligner v2.7.9, and counts were generated using featureCount (Dobin et al., 2013). A list of differentially expressed genes (DEGs) were generated using DESeq2 based on the negative binomial distribution (Love et al., 2014). The resulting DEGs between the groups were defined at cut-off criteria of |log₂ fold-change| ≥ 1.5 and p-value < 0.05 adjusted using the Benjamini and Hochberg’s approach for controlling the false discovery rate (Liao et al., 2014). Gene set enrichment analysis (Subramanian et al., 2005) was used to assess the statistical enrichment of gene ontologies and pathways, and visualized using Clusterprofiler v4.8.1 (2012). All statistical analyses were performed using R 4.2.0 (Yu et al., 2012).

Formalin-fixed paraffin-embedded (FFPE) blocks were prepared using lymph node tissue from biopsies at baseline, week 2 post protein1, and week 3 post protein 2 immunization timepoints to analyze the spatial protein profiling using the Nanostring GeoMx Digital Spatial Profiler (DSP). Tissue sections of 5 μm thickness were cut from FFPE blocks and mounted on GeoMx-NGS BOND RX slide as per manufacturer’s recommendations. Sections were baked at 60 °C for 30 min, deparaffinized, rehydrated in CitriSolv, ethanol and washed in water. For antigen retrieval, slides were placed in a staining jar containing 1 X Citrate Buffer (pH 6) for 15 min using a preheated pressure cooker (high pressure and high temp). The tissue slides were washed with 1 X TBS-T and blocked with Buffer W for 1 hr at room temperature in a closed humidity chamber. Tissue slides were stained with a cocktail of fluorescent morphological markers SYTO13 (nuclear stain; AF532), CD20 (Clone: L26; AF488), CD3 (Clone: CD3-12; AF594), and Ki-67 (Clone: B56; AF647) at 1:10,000, 1:250, 1:100, and 1:1000 dilutions, respectively. For protein detection, a multiplex cocktail of primary antibodies was used from core panels; GeoMx immune cell profiling panel, GeoMx IO drug target module, and GeoMx immune activation status module (See Table 2). In total, 60 regions of interest (ROIs) were selected within GCs based on co-localization of CD3 with CD20 +Ki-67+GC B cells. Indexing oligos were released from each ROI by exposure to UV light as described (Merritt et al., 2020), and 10 μl of liquid from above the ROI was collected by a microcapillary tip and deposited in a 96-well plate.

Indexing oligos from each ROI were PCR amplified using GeoMx Seq Code primers. PCR products were pooled and purified twice with AMPure XP beads (Beckman Coulter, Brea, CA). Library concentration and purity were measured using a high-sensitivity DNA Bioanalyzer chip (Agilent Technologies, Santa Clara, CA). Paired-end sequencing was performed on an Illumina HiSeq 2000 instrument (Illumina, San Diego, CA). After sequencing, fastq files were run through the GeoMx NGS Pipeline where reads were trimmed, merged, and aligned to a list of indexing oligos to identify the source probe. Analysis of filtered normalized gene expression data was performed in R with Bioconductor. We profiled 32 proteins along with core protein in GeoMx Human Protein array. We first calculated the signal-to-noise ratio by dividing raw count values by the geometric mean of the negative IgG probes. The data were normalized using three negative control IgG probes with the GeomxTools package. Normalized expression values were used for downstream analyses. Dimensionality reduction analysis was performed with principal component analysis using the facomineR package (Lê et al., 2008). Differential expression analysis was conducted with the mixedModelDE function from the GeomxTools package. We tested for differences between time points and/or vaccine groups using linear mixed effect models that incorporated animal ID as a random effect term to account for non-independence of the multiple ROIs sampled per animal. Differential expression results were visualized in heatmap plots generated using the ComplexHeatmap R package (Gu et al., 2016).

Serum IgG titers against HIV-1 C.1086 Env gp140 were determined by ELISA as described previously (Verma et al., 2019). In brief, 96-well microtiter plates with high-binding capacity (Thermo Fisher, USA) were coated overnight at 4 °C with C.1086 Env gp140 protein from the NIH AIDS Reagent Program (ARP) diluted in coating buffer. Plates were washed and blocked with nonfat dry milk in PBS for two hours at room temperature and washed five times with 1 X PBS-0.05% tween-20 (PBST). Diluted standard and serum samples were added to plate wells and incubated overnight at 4 °C. Plates were washed, and HRP conjugated goat anti-monkey IgG (Nordic MUbio, Netherlands) was added to plate wells and incubated at room temperature for 1 hr. Plates were then washed and developed with the addition of TMB substrate (Thermo Fisher, USA). Absorbance was recorded at 450 nm with a reference filter at 570 nm using a Spectramax 5 plate reader (Molecular Devices, USA). Baseline sera from each animal served as negative control and optical density (OD) values twofold above baseline were considered positive and extrapolated using in-plate standards to determine anti-Env antibody concentrations.

ELISA was used to measure concentrations of gp120-specific IgG1-4 antibodies. Ten rows of a 96-well Immulon 4 microtiter plate (VWR) were coated overnight at 4 °C with 50 ng C.1086 gp120Δ7 (HIV Reagent Program) per well in PBS, pH 7.2. To generate a standard curve with a known amount of IgG1, IgG2, IgG3 or IgG4 antibody, 2 rows of the plate were coated with 50 ng per well recombinant CD40 (NHP Reagent Resource). The following day, the plate was washed with PBST and blocked with PBST containing 0.1% BSA (ELISA buffer: EB). Serum samples diluted in EB were then added to gp120 wells. Duplicate dilutions of anti-CD40 rhesus IgG1, IgG2, IgG3, or IgG4 antibody (NHP Reagent Resource) were added to the CD40 wells. Following overnight incubation at 4 °C, the plate was washed and treated for 1 hr at 37 °C with 0.5 µg/ml of the appropriate biotinylated monoclonal antibody (all from NHP Reagent Resource) diluted in EB: anti-rhesus IgG1 clone 7H11 (ena), anti-rhesus IgG2 clone 8D11 (dio), anti-rhesus IgG3 clone 6F5 (tria) or anti-rhesus IgG4 clone 7A8 (tessera). Plates were washed, treated with 1/4,000 diluted neutralite avidin peroxidase (SouthernBiotech) for 30 min at room temperature, developed with TMB substrate and quenched with a H₂SO₄ stop solution. After recording absorbance at 450 nm, a standard curve constructed from the anti-CD40 IgG subclass antibody was used to interpolate the concentration of anti-gp120 IgG of the same subclass in samples.

A binding antibody multiplex assay avidity index (BAMA-AI) method (Buchbinder et al., 2017; Seaton et al., 2021) was used to measure the strength of IgG antibody-antigen interactions in serum collected at baseline and 0-, 2-, 8-, and 12 weeks post second protein boost. Baseline sera were diluted at 1:80, and post-immunization sera were diluted at 1:80 and titrated 5-fold for 6 dilutions. Diluted serum samples were incubated with a mixture of magnetic bead sets coupled to one of five HIV-1 antigens C.1086 gp140, C.1086_gp120, C.1086_V1V2, Con C (clade C consensus) gp140, Con S (group M consensus) gp140 for 30 min. Beads were then washed and treated with dissociative sodium citrate buffer (pH 4.0, Teknova) or PBS for 15 min, prior to addition of a goat anti-monkey IgG-biotin secondary detection antibody (4 µg/mL; Rockland) for 30 min followed by phycoerythrin (PE) streptavidin (1:100 dilution; BD Pharmingen) for 30 minutes. Beads were acquired on a Bio-Plex 200 instrument (Bio-Rad), with antibody binding expressed as mean fluorescence intensity (MFI). Positive controls included HIV IgG immunoglobulin (HIVIG) and CH58_4 A IgG (HIV-1 V2-specific) monoclonal antibody titrations. Negative controls included blank (uncoupled) beads, normal human reference serum (Sigma-Aldrich), and pooled seronegative monkey plasma (BioIVT). Titer of HIV-1-specific binding antibodies was reported as AUC, calculated across the titrated PBS-treated wells. Avidity of antibody binding was reported as avidity index (a percentage), defined as MFI in the citrate treated well divided by MFI in PBS-treated well x 100. AI was calculated only if the response in the PBS-treated well was considered positive and in the linear range of the assay. Positivity criteria were as follows: (1) MFI >100 (2) MFI >antigen-specific cut-off (95th percentile of all baseline sample binding per antigen after high baseline exclusion), (3) MFI >threefold of matched baseline sample binding before and after blank bead subtraction. Positivity calls were made at the 1:80 dilution for samples tested in PBS. AI values were confirmed to be within 10% across the sample dilutions, with the lowest dilution meeting this criterion selected for reporting. Samples that did not meet the above-mentioned criteria were reported as indeterminate for AI.

Serum antibodies were evaluated for ability to enhance phagocytosis of gp120 expressing fluorescent beads by THP-1 monocytes as previously described (Verma et al., 2019). Briefly, 1 µm avidin-coated fluorospheres (Invitrogen) were labeled with biotinylated anti-His tag monoclonal antibody (ThermoScientific) and used to capture His-tagged clade C gp120 Du151 protein (Immune Technologies). The gp120-expressing beads were then incubated with triplicate fivefold dilutions of heat-inactivated serum samples in V-bottom plates for 1 hr at 37 °C. THP-1 cells (2x10⁴ per well) were then added and incubated at 37 °C in 5% CO₂. After 5 h, cells were washed with Ca^+2/Mg⁺²-free DPBS, then treated with 0.12% trypsin-EDTA for 5 min at 37 °C. The cells were washed and resuspended in 1% paraformaldehyde. Fluorescence was measured using a FACSCanto (BD Biosciences) and FlowJo 9 software. Phagocytosis was quantified by multiplying the percentage of fluorescent cells by their median fluorescence intensity. A phagocytic score was then calculated by dividing the phagocytosis measured in each test sample by the phagocytosis observed for pooled pre-immune serum at the same dilution.

Serum neutralizing activity was quantified using a previously described pseudovirus infectivity assay where TZM-bl cells were co-incubated with tier 1 clade C pseudovirus MW965.26, tier 2 clade C virus Ce1086_B2, tier 2 clade C Ce1176_A3, or a control MLV-pseudotyped virus (to measure non-HIV-specific activity). Neutralization titers were measured in serum samples collected at week 0 and week 2 following the second protein boost. Samples neutralizing antibody activity against SIV pseudoviruses was measured relative to MLV pseudovirus negative controls.

A custom binding antibody multiplex assay (BAMA) assay with C.1086 gp140 K160N-conjugated Bio-Rad Bio-plex Pro magnetic COOH beads was used to quantify Env-specific IgG in rectal secretions and specific IgA in IgG-depleted sera and rectal secretions as previously described (Verma et al., 2019). Briefly, 50 µg gp140 was conjugated to 10x10⁶ beads as described (Iyer et al., 2015). The gp140-labeled beads (3500 per well) were shaken overnight at 1100 rpm and 4 °C with 10-fold dilutions of serum or secretion and standard Verma et al., 2019. Beads were then washed with PBST and reacted for 30 minutes at room temperature with biotinylated goat anti-human IgG (Southern Biotech) or goat whole IgG containing anti-monkey IgA (Novus) followed by phycoerythrin-labeled neutralite avidin (Southern Biotech). Fluorescence was measured with a Bio-Rad Bioplex 200. Antibody concentrations were determined using standard curves prepared with Bioplex Manager software (Bio-Rad). Concentrations of gp120-specific IgG in secretions were divided by the concentration of total IgG or IgA measured in the sample by ELISA to obtain the specific activity (ng of gp140-specific IgG or IgA antibody per µg of total IgG or IgA, respectively).

Statistical analysis was performed using Prism v.9.5.1 (Graph Pad). Gaussian distribution was not assumed, thus comparisons across time, between groups, and correlations were analyzed using non-parametric tests. For within group comparisons, Wilcoxon matched pairs signed rank test was used. For between group comparisons, Mann-Whitney rank comparisons were used. Correlation coefficients were computed using Spearman correlation. p-values <0.05 were considered significant.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

The authors are grateful to Dennis Christensen, previously at the SSI, for provision of CAF01 adjuvant for the studies. The authors are grateful to Brian Schmidt for technical assistance with the LAMV study; Wilhelm Von Morgenland and Miles Christensen and the CNPRC SAIDS team for coordination of macaque studies; Stephenie Liu at the UC Davis Genomics Resource for optimization of morphological markers for spatial profiling, and Daniel Newhouse at the Spatial Data Analysis Service (sDAS) for analysis of spatial proteomics data. We thank Robert L Wilson for excellent technical assistance in processing secretions and measuring mucosal antibodies. The authors thank Aloki Patel for technical assistance with BAMA-AI assays, Kelly Seaton and Angelina Sharak for BAMA-AI data selection/consultation, and Lu Zhang for data analysis and visualization (Center for Human Systems Immunology, Duke University). The authors acknowledge funding support from NIAID, NIH R56AI150409 (SSI) and contract # HHSN272201800004C (XS); Duke CFAR Grant P30 AI064518 (GDT). Reagents for measurement of rhesus IgG subclasses were obtained from the NIH Nonhuman Primate Reagent Resource supported by AI126683 and OD10976.

All animals were bred and housed at the California National Primate Research Center (CNPRC) in accordance with the American Association for Accreditation of Laboratory Animal Care (AAALAC) guidelines. All studies were approved by the University of California, Davis Institutional Animal Care and Use Committee (IACUC).

1. Sent for peer review: June 28, 2023
2. Preprint posted: July 19, 2023
3. Reviewed Preprint version 1: October 5, 2023
4. Reviewed Preprint version 2: January 8, 2024
5. Version of Record published: February 22, 2024

You can cite all versions using the DOI https://doi.org/10.7554/eLife.89395. This DOI represents all versions, and will always resolve to the latest one.

© 2023, Verma, Hawes et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.