A particulate saponin/TLR agonist vaccine adjuvant alters lymph flow and modulates adaptive immunity

Abstract

Saponins are potent and safe vaccine adjuvants, but their mechanisms of action remain incompletely understood. Here, we explored the properties of several saponin formulations, including immunostimulatory complexes (ISCOMs) formed by the self-assembly of saponin and phospholipids in the absence or presence of the Toll like receptor (TLR) 4 agonist monophosphoryl lipid A (MPLA). We found that MPLA self-assembles with saponins to form particles physically resembling ISCOMs, which we termed saponin/MPLA nanoparticles (SMNP). Saponin-containing adjuvants exhibited distinctive mechanisms of action, altering lymph flow in a mast cell-dependent manner and promoting antigen entry into draining lymph nodes. SMNP was particularly effective, exhibiting even greater potency than the compositionally-related adjuvant AS01_B in mice, and primed robust germinal center B cell, T_FH, and HIV tier 2 neutralizing antibodies in non-human primates. Altogether, these findings shed new light on mechanisms by which saponin adjuvants act to promote the immune response and suggest SMNP may be a promising adjuvant in the setting of HIV, SARS-CoV-2, and other pathogens.

One Sentence Summary:

An adjuvant composed of saponin and a TLR4 agonist acts by enhancing lymph flow and antigen entry into lymph nodes.

INTRODUCTION

Adjuvants are important components of vaccines, promoting protective immune responses to poorly immunogenic antigens. Very few adjuvants have been approved as part of licensed human vaccines to date, and the development of new adjuvants that can safely augment adaptive immune responses is of great interest for vaccine efforts against infectious diseases for which billions of people are at risk, such as malaria, tuberculosis, human immunodeficiency virus (HIV) (1, 2), and now COVID-19 (3). Saponins are triterpene glycosides isolated from natural sources such as the Quillaja Saponaria tree, which have been under extensive study as vaccine adjuvants (4, 5). Although free saponins are toxic, formulation of saponins with lipids and cholesterol maintains their adjuvant activity in a non-toxic state, and even enables saponins to be safely combined with additional innate immune stimulators such as Toll-like receptor (TLR) agonists. Such formulation advances have led to the first licensed vaccines with saponin adjuvants, which employ liposomal saponin and MPLA (Glaxo Smith Kline’s AS01 adjuvant used in the Shingrix® and Mosquirix® vaccines) (6).

Another intensively investigated form of saponins are immune stimulating complexes (ISCOMs) – cage-like nanoparticles with diameters of approximately 40 nm formed by the self-assembly of saponins, cholesterol, and phospholipids. ISCOM-based vaccine formulations have been shown to induce adaptive immune responses in small animals (7, 8), non-human primates (NHPs) (9-13) and humans (14-17). Promising phase 3 clinical trial results for Novavax’s SARS-CoV-2 vaccine employing the saponin adjuvant Matrix M® may lead to the first FDA-approved vaccine incorporating an ISCOM-based adjuvant (18). ISCOMs were originally described as a multivalent antigen-delivery system with built-in adjuvant (i.e. saponins) (7). However, physical association with antigen was later found to be not important (16, 17, 19). Rather, “antigen-free” ISCOMs can be simply added to improve the immunogenicity of soluble antigens.

ISCOMs and other saponin-containing adjuvants are known to induce local proinflammatory cytokine and chemokine responses and activate dendritic cells (DCs) (20-23), and several mechanisms have been defined relating to enhanced CD8⁺ T cell priming through effects on DCs and promotion of antigen cross presentation (24-27). However, saponin adjuvants are also effective for inducing antibody responses, and how saponins promote antibody responses remains obscure. Saponins exhibit strong synergy with Toll-like receptor (TLR) 4 agonists (23, 28), but mechanisms underlying this synergy are also poorly defined. To gain further insight into these issues, we compared the immunological effects of ISCOMs versus a diverse panel of clinical and experimental adjuvants in mouse models enabling detailed dissection of antigen-specific B and T cell responses. Further, we designed ISCOMs incorporating the TLR4 agonist monophosphoryl lipid A (MPLA), forming saponin-MPLA nanoparticles (hereafter, SMNP), to evaluate potential synergistic effects of saponins with this TLR agonist.

As the majority of licensed vaccines are thought to function by inducing a protective antibody response (29, 30), we focused on analyzing adjuvant impacts on humoral immunity. Here, we demonstrate that saponin adjuvants exhibited potent activity in mice, eliciting robust germinal center (GC), T cell, and class-switched antibody responses superior to a range of alternative adjuvants, with SMNP exhibiting particularly strong potency. Mechanistically, we observed ISCOMs and SMNP altered lymph flow and lymph node permeability, leading to enhanced antigen acquisition by B cells in draining lymph nodes (dLNs), with maximal effects obtained for the combined saponin/TLR4 agonist adjuvant. In non-human primates, vaccination with HIV Env trimer and SMNP induced robust and durable GCs and autologous tier 2 neutralizing antibody responses. Altogether, these data provide insights into the mechanisms of action underlying saponin adjuvants and suggest that SMNP represents a promising new adjuvant.

RESULTS

Development and characterization of saponin/MPLA-based nanoparticle adjuvants

We first evaluated the ability of a panel of innate immune stimulators containing lipid moieties to co-assemble with Quil-A saponin, cholesterol, and phospholipids to form ISCOMs. Unlike lipid-linked NOD1, TLR2/7, TLR7, or TLR9 ligands (fig. S1A-B), the synthetic TLR4 agonist MPLA (PHAD®) could be incorporated into ISCOM particles at a 10:10:2.5:1 Quil-A:cholesterol:DPPC:MPLA (molar ratio) to yield homogenous ~40 nm nanocage structures (Fig. 1A-C). We focused on this particular formulation for further vaccine studies, and termed this adjuvant saponin/MPLA nanoparticles (hereafter, SMNP). Stimulation of TLR4 reporter cells confirmed that MPLA incorporated in SMNP had bioactivity comparable to free MPLA (Fig. 1D). In contrast to traditional ISCOMs, SMNP induced murine bone marrow-derived dendritic cells to secrete cytokines TNF-α and IL-6 in vitro (Fig. 1E), which are known to promote adaptive immune responses. We also assessed stimulation of human peripheral blood mononuclear cells (PBMCs), and observed robust production of IFN-γ, IL-12, TNF-α and other cytokines in response to SMNP in a dose-dependent manner (Fig. 1F). Importantly, SMNP was non-toxic to human PBMCs (fig. S1C).

Figure 1: Characterization of saponin-based nanoparticle adjuvants.

(A) Dynamic Light Scattering (DLS) size measurements of ISCOMs and SMNP.

(B) Size exclusion chromatogram of SMNP particles.

(C) Representative transmission electron microscope (TEM) images of ISCOM and SMNP particles. Scale bars, 50 nm.

(D) Activation analysis of Raw-Blue TLR reporter cell line after 18hr incubation with ISCOMS, SMNP, or free MPLA. ISCOMs were added at molar equivalent concentrations to SMNP as a control. Symbols represent mean values from duplicate wells.

(E) Analysis of cytokine secretion by bone marrow-derived DCs 6hrs after incubation with 1 μg/mL SMNP or ISCOMs. Each symbol represents a replicate well (n = 3) and the bar represents the mean.

(F) SMNP was added at indicated concentrations to human PBMCs in vitro for 24 hr and then cytokines and chemokines in the supernatant were analyzed by multiplexed ELISA. Shown are means ± SEM.

(G-H) Analysis of proinflammatory cytokine secretion in draining inguinal LNs of mice at 3, 6, and 18 hrs after adjuvant injection. Adjuvant was injected subcutaneously at the base of the tail in Balb/C mice. Representative data from 2 independent experiments (n=3 mice/group). Shown are the experiment timeline (G) and mean cytokine concentrations at each timepoint (H). Symbols represent mean values. Error bars are standard error of the mean (SEM).

Statistical analysis for (E) was performed by one-way ANOVA, followed by Tukey’s post-test. Statistical analysis for (H) is comparing responses to SMNP using two-way ANOVA, followed by Dunnett’s post test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

We next evaluated cytokine responses induced by SMNP in mice. A panel of proinflammatory cytokines and chemokines in dLNs were measured at 3h, 6h, or 18h after injection of ISCOMs, SMNP, MPLA, or ISCOMs mixed with free MPLA (Fig. 1G). Compared to traditional ISCOMs, SMNP rapidly induced 10-100-fold higher levels of the proinflammatory cytokines IFN-α, IFN-γ, IL-6, and TNF-α (Fig. 1H and fig. S1D). Free MPLA, administered in a dose equivalent to SMNP (0.5μg/mouse), failed to induce any response above background (Fig. 1G and fig. S1D). Notably, ISCOMs mixed with free MPLA at the same ratio used in SMNP did not show appreciable differences from ISCOMs alone (Fig. 1H and fig. S1D), indicating that physical incorporation of MPLA into the nanoparticle structure was essential for the activity of SMNP. Despite the high levels of inflammatory cytokines induced in LNs, cytokine and chemokine levels in the serum remained low following vaccination with the exception of IL-6, which showed a transient elevation at 6 hr that returned to baseline by 18 hr (fig. S1E). Together, these data show that incorporating MPLA into ISCOM nanoparticles synergistically enhances innate immune activation in dLNs.

SMNP supports potent antibody responses in mice

Saponin particulate adjuvants have been reported to elicit potent immune responses in mice, but we first sought to benchmark them against a panel of alternative experimental and clinical adjuvants. For these studies we assessed the capacity of ISCOMs or SMNP to stimulate humoral responses in Balb/c mice against a poorly immunogenic HIV antigen, a stabilized HIV Env BG505 SOSIP trimer (termed MD39) (31), representative of SOSIP Env trimers currently in clinical trials (Fig. 2A). Mice were immunized subcutaneously at the base of the tail, and draining inguinal lymph nodes (iLN) were harvested for analysis. SMNP elicited GC B cell (B_GC) responses greater than traditional ISCOMs, and outperformed other adjuvants, including AddaVax (an MF59-like adjuvant), Ribi (an oil-in-water emulsion adjuvant containing MPLA), AS01_B (a clinical liposomal saponin/MPLA adjuvant (6)), and alum (Fig. 2B, fig. S2A). Follicular helper CD4⁺ T cells (T_FH) that drive GC responses were also increased by SMNP 2-3 fold over all other adjuvants tested (Fig. 2C and fig. S2B), suggesting a significant impact of combining saponin with MPLA in an ISCOM particle. We also analyzed B_GC and T_FH responses in the axillary lymph nodes (axLN), which drain efferent lymph from the iLNs. Interestingly, we saw that ISCOMs and SMNP elicited robust GC responses at these distal dLNs, in contrast to all other adjuvants tested, including AS01_B (Fig. 2D). Hence, SMNP and ISCOMs elicit robust GC and T_FH responses that extend beyond the proximal dLNs in mice.

Figure 2. SMNP is a potent adjuvant for humoral immunity in mice.

(A) Schematic of the experiment for (A-F). Balb/C mice were immunized s.c. with specified adjuvants and 2 μg of MD39 HIV Env trimer, and were boosted with identical formulations 4 weeks later. Germinal center B cell (B_GC) and T follicular helper (T_FH) responses were analyzed 12 days post prime immunization and antibody titers were measured 2, 4, 5, and 7 weeks post prime. Pooled data from 2 independent experiments. Each symbol represents a mouse and the bar represents the mean. n = 9 or 10 mice per group.

(B) Representative flow plots and quantification of B_GC cells in draining iLN on day 12. Live, single cells were gated on B220⁺, CD4⁻, CD38^lo, GL7^hi

(C) Representative flow plots and quantification of T_FH cells in draining iLN on day 12. Live, single cells were gated on B220⁻, CD4⁺, CXCR5^hi, PD1^hi.

(D) Enumeration of B_GC and T_FH cells in axillary LNs on day 12.

(E) Total HIV MD39 Env-specific IgG Ab titers.

(F) HIV MD39 Env-specific IgG1 Ab titers.

(G) HIV MD39 Env-specific IgG2a Ab titers.

Statistical analysis (B-D) was performed by One-way Anova, followed by Dunnett’s post-test. Statistical analysis (E-G) was performed by Two-way Anova, followed by Benjamin, Krieger and Yekitieli’s post-test. Log-distributed datasets (E-G) were log-transformed before statistical analysis. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

Serum Ig responses were measured longitudinally after prime and booster immunization with MD39 HIV Env trimer and the same panel of diverse adjuvants. Analysis of Env-specific IgG responses revealed that the saponin adjuvants elicited the strongest IgG responses following the priming immunization, with a hierarchy in potency of SMNP > AS01_B ~ ISCOMs (Fig. 2E). Following boosting, the saponin adjuvants, Ribi, and Addavax all elicited high total IgG titers, but IgG titers with SMNP were 7-fold higher than the next best comparator. SMNP was also superior over the other adjuvants for priming an IgG2a response (Fig. 2F-G). Comparing two most potent adjuvants for generating high IgG titers, SMNP and ISCOMs, the strongest plasmablast responses were elicited by SMNP (fig. S2C-D). SMNP additionally induced superior humoral responses to other vaccine antigens in separate experiments, including diphtheria toxin (DTx) and influenza hemagglutinin (fig. S2E-G).

We also compared SMNP formulations prepared using Quil-A versus QS-21, a purified sub-fraction of Quil-A saponin used in the licensed AS01_B adjuvant. Both SMNP formulations elicited comparable B_GC, T_FH and MD39 Env-specific antibody responses (fig. S2H-J). We compared serum cytokine profiles and animal weight loss following immunization with QS-21 SMNP, ISCOMs, AS01_B, or alum to gain insights into potential differences in toxicity. QS-21 SMNP induced transient elevations in proinflammatory cytokine and chemokine levels in circulation that were comparable to or less than AS01_B (fig. S2K-L). AS01_B also induced transient weight loss 24hrs after injection, which was not observed with SMNP (p<0.001, fig. S2M). Overall, these experiments suggest that saponin adjuvants are highly effective, with saponin/MPLA combination adjuvants promoting a broader range of antibody effector functions, and SMNP in particular capable of enhancing multiple facets of the humoral response elicited by subunit vaccination while maintaining a safety profile that appears comparable to an approved saponin adjuvant.

SMNP promotes antigen capture by cognate B cells

We next sought to test potential mechanisms of action underlying SMNP adjuvanticity. Based on the rapid induction of cytokines in dLNs, we suspected that SMNP might influence early B cell responses. To investigate this possibility, we made use of an adoptive transfer model in which B cells expressing the inferred germline-reverted VRC01 B cell receptor (VRC01^gHL B cells) are seeded in C57BL/6 recipient mice at defined precursor frequencies, enabling assessment of antigen-specific B cell responses in vivo under challenging conditions designed to be relevant for vaccine regimens intended to induce HIV broadly neutralizing antibodies in humans (32-34). Mice adoptively transferred with VRC01^gHL B cells were immunized with a 60mer form of the HIV gp120 immunogen eOD-GT5 (engineered outer domain germline-targeting, version 5, eOD-GT5 60mer) labeled with Alexa Fluor 647 (eOD-GT5 60mer-AF647) to assess antigen uptake and early activation of antigen-specific B cells in dLNs (Fig. 3A, B). This experiment models germline-targeting immunizations for HIV that are currently undergoing initial clinical trials (35). We compared the panel of adjuvants evaluated earlier in wild type mice, adding CpG as an additional comparator targeting a distinct Toll-like receptor (TLR9).

Figure 3. SMNP enhances antigen uptake, activation, and proliferation of B cells in draining LNs.

(A-G) 10⁶ CTV⁺ VRC01^gHL B cells were adoptively transferred into B6 mice prior to vaccination with eOD-60mer-AF647 with various adjuvants. Antigen uptake and activation of VRC01^gHL B cells in draining iLNs was assessed at 24h post-immunization.

(A) Schematic of the experiment.

(B) Gating strategies to identify endogenous B cells (Endo B) and VRC01^gHL B cells. Cells were stained in vitro using AF488-labeled eOD-60mer to identify VRC01^gHL B cells. AF647 fluorescence indicates in vivo antigen uptake. Representative plots from B6 mice immunized with ISCOMs + eOD-60mer-AF647 are shown.

(C) Representative histograms showing VRC01^gHL B cell phenotype in LN. The uptake of eOD-60mer was inferred from AF647 fluorescence (left column). Surface expression of early activation markers, CCR7 (middle) and CD86 (right), was assessed. Dotted line represents endogenous B cells.

(D) Quantification of AF647 uptake (geometric MFI) in LN.

(E) Quantification of CCR7 expression (geometric MFI normalized to Ag alone group) in iLNs.

(F) Quantification of CD86 expression (geometric MFI normalized to Ag alone group) in iLNs.

(G) Quantification of cell size (mean FSC-A normalized to Ag alone group) in iLNs.

(D-G) Each symbol represents a mouse and the bar represents the mean (e-g) or geometric mean (d). Pooled data from 2-4 independent experiments. n = 8 mice (unprimed; 4 experiments), n = 12 mice (Ag alone, ISCOM; 4 experiments), n = 9 mice (SMNP, MPLA; 3 experiments), n = 6 mice (AddaVax, Ribi, amph-CpG, Alum; 2 experiments).

(H-K) Antigen uptake and activation of VRC01^gHL B cells in draining iLNs was assessed 24h after immunization with eOD-60mer-AF647 and SMNP, AS01_B or no adjuvant.

(H) Quantification of AF647 uptake (geometric MFI).

(I) Quantification of CCR7 expression (geometric MFI).

(J) Quantification of CD86 expression (geometric MFI).

(K) Quantification of cell size (mean FSC-A).

(H-K) Each symbol represents a mouse and the bar represents the mean (K) or geometric mean (H-J). Pooled data from 2 independent experiments. N = 4 mice (unprimed), n = 7 mice (Ag alone), n = 9 mice (SMNP, AS01_B).

(L-N) 10⁶ CTV⁺ VRC01^gHL B cells were adoptively transferred into B6 mice prior to vaccination with eOD-GT5 60mer with various adjuvants. VRC01^gHL B cell proliferative responses in draining iLNs were assessed at d3.

(L) Representative histograms (tinted) showing VRC01^gHL B cell proliferation following different immunizations. Unprimed VRC01^gHL cells (dotted histograms) are overlaid. The vertical line divides CTV^hi (undivided) and CTV^lo (divided) populations.

(M) Normalized frequency of undivided VRC01^gHL B cells in proximal draining (inguinal) LNs. Frequency of undivided VRC01^gHL B cells of the total B cells were calculated for each mouse. Normalization was performed for each experiment by dividing the frequencies with the average (mean) in unimmunized mice, and the values were expressed as percentages.

(N) Quantification of divided VRC01^gHL B cell numbers in draining inguinal LNs.

(O) Quantification of divided VRC01^gHL B cell numbers in distal draining axillary LNs.

(P) Ratio of divided cells in the distal axillary LN vs the proximal inguinal LN.

(L-P) Each symbol represents a mouse and the bar represents the mean. (L-N) Pooled data from 2 independent experiments. n = 6 mice (unprimed; 4 experiments), n = 6 mice (Ag alone, alum; 2 experiments), n = 12 mice (ISCOM, SMNP; 4 experiments). (O, P) Pooled data from 2 independent experiments, n = 6 mice per group.

Statistical analysis was performed by One-way Anova, followed by Tukey’s post-test (D, H) or Dunnett’s post-test (E-G, I-P). **, p < 0.01; ***, p < 0.001; ****, p < 0.0001. Log-distributed datasets (D, H) were log-transformed before statistical analysis.

Injection of eOD-GT5 60mer-AF647 alone resulted in a moderate shift of AF647 fluorescence on VRC01^gHL B cells, indicating that in vivo uptake of antigen within dLNs is inefficient in the absence of adjuvant (Fig. 3C left column, D). Most adjuvants did not alter the in vivo uptake of antigen by specific B cells (Fig. 3C, D). Notably, saponin-containing ISCOMs and SMNP promoted substantial antigen acquisition by VRC01^gHL B cells in LNs (Fig. 3C, D), which was largely internalized (fig. S3A-D). SMNP resulted in a 10-fold increase in antigen uptake compared to the non-adjuvanted control. Free MPLA, or Ribi (containing 50 μg MPLA), did not impact antigen uptake by the B cells, in contrast to SMNP (Fig. 3C, D). VRC01^gHL B cells robustly upregulated activation markers including CCR7, CD86, CD69, and CD44 in response to ISCOMs or SMNP vaccination, with SMNP triggering maximal rapid B cell activation compared to other adjuvants (Fig. 3C, E-F, fig. S3E-G). In addition, ISCOMs and SMNP exhibited a rapid induction of metabolic activity by the antigen-specific B cells, based on increased cell size, which was significantly greater than any of the other adjuvants tested (Fig. 3G, fig. S3E). SMNP had much weaker effects on polyclonal B cells (e.g., no cell size increase, Fig. 3C, fig. S3E, H-M), indicating the adjuvant predominantly enhanced cognate BCR-dependent activation. To rule out the possibility that other adjuvants might be as effective as SMNP at triggering B cell activation but with slower kinetics, we carried out 48 hr time course experiments of antigen acquisition and B cell activation. SMNP elicited higher overall antigen uptake and B cell activation signatures over time compared to non-saponin-containing adjuvants (fig. S3N-Q). Given the striking impact of SMNP on B cell antigen acquisition, we repeated these experiments including AS01_B from the commercial Shingrix® vaccine, to assess if a liposomal form of saponin and MPLA would have similar effects. AS01_B also increased early antigen-specific B cell uptake of antigen, activation, and increased metabolic activity (Fig. 3H-K). These data suggest that the capacity to promote antigen uptake by cognate B cells is a distinctive attribute of saponin-containing particulate adjuvants, and integration of MPLA amplifies this process in ways that are not accomplished by saponin or MPLA individually.

To explore adjuvant effects on B cell proliferation in vivo, CellTrace Violet (CTV)-labelled VRC01^gHL B cells were tracked for three days post-immunization. Approximately 50% of CTV⁺ VRC01^gHL B cells in dLNs remained undivided at d3 after immunization with eOD-60mer without adjuvant (Fig. 3L-M), and there was minimal accumulation of CTV^lo (divided) VRC01^gHL B cells (Fig. 3N). Alum did not promote early B cell proliferation (Fig. 3L-N), and 50% of the B cells did not proliferate in response to Ribi (Fig. 3L, M). In contrast, nearly all antigen-specific B cells proliferated in the presence of SMNP or ISCOMs, in an antigen-dependent manner (Fig. 3L-N, fig. S3R-T). SMNP also led to substantial proliferation of VRC01^gHL B cells in distal axillary LNs (axLNs) (Fig. 3O). Interestingly, with SMNP, VRC01^gHL B cell proliferation was similar in distal and proximal LNs (Fig. 3P). In sum, enhanced cognate B cell activation induced by saponin formulations subsequently leads to robust B cell proliferation in vivo, explaining an additional important mechanism of action of these adjuvants.

Saponin/MPLA combination adjuvants induce an antiviral T_FH functionality profile

GCs and the development of most neutralizing antibodies depend on T_FH cells. To explore effects of SMNP on CD4⁺ T cell responses, we used a similar adoptive transfer model approach, now employing T cell receptor (TCR) transgenic CD4⁺ T cells to track antigen-specific T cell responses. SMARTA CD4⁺ T cells specific for the lymphocytic choriomeningitis virus (LCMV) gp_66–77 antigen were transferred into C57BL/6 recipient mice prior to immunization with eOD-GT5 60mer containing the LCMV gp_66–77 epitope (eOD-GT5_gp61 60mer) in the presence of SMNP or other adjuvants (Fig. 4A). Immunization with SMNP led to the greatest early (24h) upregulation of activation markers on SMARTA CD4⁺ T cells, including ICOS, IL-2Rα (CD25) and CD69, without affecting non-cognate T cells (Fig. 4B-C, fig. S4A-D). To assess T cell differentiation, we analyzed dLNs 4 days after immunization (Fig. 4D). Accumulation of SMARTA T cells at d4 was equivalent when immunizing with SMNP, ISCOMs, AS01_B, or Ribi (Fig. 4E-F, fig. S4E). However, production of the cytokines IL-21 and IFN-γ by SMARTA non-T_FH and T_FH cells was notably increased by SMNP compared to the other adjuvants, including relative to the other saponin adjuvants ISCOMs and AS01_B (Fig. 4G-J). IL-21 was particularly notable, as IL-21 is the signature cytokine of T_FH cells, and detection of IL-21 production by antigen-specific T_FH cells is typically very challenging (36, 37). In sum, SMNP potently enhances the quality of CD4⁺ T cell help provided to B cells.

Figure 4. SMNP enhances T_FH differentiation and IL-21 production by CD4⁺ T cells in draining LN.

(A–C) 10⁶ CTV⁺ SMARTA CD4⁺ T cells were adoptively transferred into B6 mice prior to vaccination with eOD_gp61-60mer with various adjuvants. Activation of SMARTA CD4⁺ T cells in draining iLN was assessed at 24h post-immunization.

(A) Schematic of the experiment.

(B) Representative histograms showing surface expression of ICOS (left) and CD25 (right) by SMARTA (solid line) and endogenous CD4⁺ T cells (dotted line).

(C) Quantification (geometric MFI) of ICOS (left) and CD25 (right). Each symbol represents a mouse, bars indicate the mean.

(B, C) Pooled data from two independent experiments. n = 3 mice (unimmunized) or n = 6 mice (no adjuvant, ISCOM, SMNP, Ribi).

(D–J) 2×10⁴ SMARTA CD4⁺ T cells were adoptively transferred into B6 mice prior to vaccination with eOD_gp61-60mer with various adjuvants. SMARTA responses in iLNs were analyzed at d4 post-immunization.

(D) Schematic of the experiment.

(E) Quantification of total SMARTA CD4⁺ T cells.

(F) Quantification of CXCR5^hi PD1^hi SMARTA T_FH.

(G–J) Cells were re-stimulated ex vivo with LCMV gp_66–77 or irrelevant OVA_323–339 to assess cytokine responses.

(G) SMARTA CD4⁺ T cells stimulated with LCMV gp_66–77. Numbers in quadrants indicate the percentages.

(H) Quantification of IL-21⁺ cells as a percentage of SMARTA CD4⁺ T cells.

(I) Quantification of IFNγ ⁺ cells as a percentage of SMARTA CD4⁺ T cells.

(J) Quantification of IL-21⁺ cells as a percentage of SMARTA T_FH.

(K) Quantification of IFNγ ⁺ cells as a percentage of SMARTA T_FH.

(E-K) Pooled data from 3-5 independent experiments. Ag alone, n = 6 (Ag alone; 2 experiments), n = 15 (ISCOMs, SMNP; 5 experiments), n = 11 (Ribi; 4 experiments), n = 9 (AS01_B; 3 experiments). Each symbol represents a mouse, bars represent the geometric mean (C, E, F) or the mean (H-K). **, p < 0.01; ***, p < 0.001; ****, p < 0.0001 by One-way Anova with Tukey’s test. Log-distributed data sets (E-F) were log-transformed before statistical analysis.

Saponin adjuvants promote antigen entry in lymph nodes and enhance lymph flow

Motivated by the findings of enhanced antigen acquisition, rapid B cell activation, and robust responses in distal LNs, we hypothesized that saponin adjuvants might impact antigen delivery and/or access into lymph nodes. We first examined antigen distribution within dLNs. Four hours post-injection, fluorescent 70 kDa dextran injected without adjuvant or with AddaVax was observed primarily in the subcapsular sinus and collagen conduits (Fig. 5A, fig. S5A). By contrast, dextran administered with SMNP exhibited widespread distribution deep into the LN parenchyma (Fig. 5A-C). Flow cytometric analysis revealed that SMNP was captured by numerous LN cell types, including CD169⁺ subcapsular sinus macrophages (SSM, Fig. 5D-E, fig. S5B). SMNP rapidly induced a >50% loss of CD169⁺ SSM in the dLN within 4h of injection (Fig. 5F). ISCOMs also induced SSM depletion, although to a lower extent, whereas other adjuvants had no effect on CD169⁺ SSM at this timepoint (Fig. 5F). Immunofluorescence imaging of dLN sections confirmed that injection of antigen (eOD-GT5 60mer) with SMNP was accompanied by loss of CD169⁺ cells in the subcapsular sinus (Fig. 5G-H). Injection of AS01_B also caused a loss of CD169⁺ SSMs (fig. S5C), similar to previous reports showing that liposomal saponins can deplete this population (21). Prior studies have suggested that saponin adjuvant effects may involve inflammasome activation. However, neither the NLRP3 inflammasome nor non-canonical inflammasomes regulated by caspase-1 or caspase-11 were required for SMNP-mediated macrophage depletion (Fig. 5I).

Figure 5. SMNP adjuvant enhances lymph flow and promotes antigen delivery to draining lymph nodes.

(A) Histological images of fluorescent 70kDa dextran drainage to iLN 4h after subcutaneous base-of-tail injection in balb/C mice. Scale bar = 200 um.

(B) Quantification of fluorescent 70kDa dextran diffusion into the LN parenchyma 4h after injection in balb/C mice. Dextran intensity values were measured every 10um from LN surface. n = 6 mice per group. Each symbol represents an analyzed LN section. 9 LN sections were evaluated for Naïve, 7 for SMNP, and 13 for AddaVax. Data representative of 2 independent experiments.

(C) Area under curve (AUC) analysis for (B).

(D-E) C57Bl/6 mice were subcutaneously injected with Cy-5 labeled SMNP (SMNP-Cy5). Cellular uptake of Cy5 was assessed at 2h post-injection.

(D) Representative flow cytometry histograms of uptake of Cy5-labeled SMNP by the indicated cell type (tinted red histogram) in the draining iLN 2h after subcutaneous injection. The vertical line divides Cy5⁺ and Cy5⁻ cells. The number indicates the frequency (percentage) of Cy5⁺ cells from a representative sample. The gate was set based on background signals from mice immunized with non-fluorescent SMNP (dotted histogram). Subcapsular sinus macrophage (SSM), medullary macrophage (MSM).

(E) Quantification of percent Cy5⁺ cells for each indicated cell type. Each symbol represents a mouse and bars indicate the mean. n = 6 mice for SMNP-Cy5, n = 3 mice for SMNP. Pooled data from two independent experiments. Gating strategy is depicted in fig. S5B.

(F) Flow cytometry analysis of CD169⁺ SSM remaining in the draining iLN 4hrs after subcutaneous adjuvant injection. n = 6 mice per group. Live single cells were gated on B220⁻, CD3⁻, Ly6G⁻, Ly6C⁻, CD11b⁺, CD169^hi. Pooled data from 2 independent experiment. Each symbol represents a mouse and the bar represents the mean.

(G) Representative confocal images of CD169⁺ SSM in the draining LN at 24h after injection of adjuvants with eOD-GT5 60mer. CD169 staining is shown in red, IgD staining is shown in blue. Scale bar = 100μm.

(H) Quantification of loss of SSM from (G). Each symbol represents a mouse and bars represent the mean. n = 6 mice per group. Pooled data from two independent experiments.

(I) Casp-1/Casp-11 KO, NLRP3 KO, or WT C57Bl/6 mice were injected with 5ug of SMNP. 24hrs later, the draining LNs were harvested for flow cytometry analysis of SSMs. Each symbol represents a mouse. Bars represent the mean. n = 3 or 4 mice per group.

(J-K) B6 mice were pre-treated with clodronate liposome (CLL) to deplete CD169⁺ macrophages or with control liposome. 10⁶ CTV⁺ VRC01^gHL B cells were then adoptively transferred one day before immunization with eOD-60mer:AF647 alone or with SMNP.

(J) Quantification of antigen uptake by VRC01^gHL B cells in draining iLNs at 24h post-immunization. Each symbol represents a mouse. Bars represent the geometric mean. Results from two independent experiments, n = 6 mice per group.

(K) Activation of VRC01^gHL B cells as assessed by CCR7 expression (geometric MFI). Each symbol represents a mouse. Bars represent the mean. Results from two independent experiments, n = 6 per group.

(L-P) B6 mice were pre-treated with CLL or control liposome. 2×10⁴ SMARTA CD4⁺ T cells were then adoptively transferred prior to vaccination with eOD-GT5_gp61 60mer with SMNP.

(L) SMARTA CD4⁺ T cell numbers in iLNs at d4 post-vaccination.

(M) CXCR5^hi PD1^hi SMARTA T_FH cell numbers in iLNs at d4 post-vaccination.

(N) Percentage of IL-21⁺ SMARTA CD4⁺ T cells upon ex vivo gp_66-77 re-stimulation.

(O) Percentage of IFNγ ⁺ SMARTA CD4⁺ T cells upon ex vivo gp_66-77 re-stimulation.

(P) Percentage of IL-21⁺ SMARTA T_FH cells upon ex vivo gp_66-77 re-stimulation.

For (C, E, F, H, I, J and K) statistical analysis was performed by One-way Anova, followed by Tukey’s post-test. Datasets in (K) were log-transformed before statistical analysis. For (L-P), statistical analysis was performed using Mann Whitney U test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

To assess the impact of SSM depletion on the function of SMNP, we tested the effects of pre-depleting SSMs using clodronate liposomes (CLL) in the VRC01^gHL adoptive transfer model. Pre-depletion of macrophages prior to injection of eOD-GT5 60mer alone enhanced antigen uptake and showed a trend toward enhanced early activation of VRC01^gHL B cells in dLNs (Fig. 5J, K), consistent with the idea that CD169⁺ SSMs limit antigen access to naïve B cells in dLNs. By contrast, injection of eOD-GT5 60mer together with SMNP led to the same level of antigen uptake whether or not CLL was used to pre-deplete macrophages (Fig. 5J). Early VRC01^gHL B cell activation was blunted when antigen and SMNP were administered after CLL (Fig. 5K), but subsequent serum antibody titers were not impacted (fig. S5D). We similarly assessed the impact of macrophage pre-depletion on responses to eOD-GT5 60mer and SMNP immunization in the SMARTA T cell adoptive transfer model. Overall SMARTA T cell proliferation and T_FH expansion in response to SMNP immunization were not impacted by macrophage pre-depletion with CLL (Fig. 5L, M), but cytokine polyfunctionality of antigen-specific T cells was significantly reduced (Fig. 5N-P, fig. S5E). Thus, SMNP acting on lymph node macrophages does not have an obvious impact on early antibody responses, but does impact T cell polyfunctionality.

CLL treatment followed by eOD-GT5 60mer administered alone led to 2.6-fold lower antigen uptake by VRC01^gHL B cells compared to eOD-GT5 60mer administered with SMNP (Fig. 5J, +CLL No Adj vs. −CLL SMNP), demonstrating that depletion of SSMs alone cannot explain the enhanced antigen acquisition observed with SMNP adjuvant. We thus next examined lymph flow and antigen accumulation in dLNs following injection of SMNP or other adjuvants. SMNP increased the rate of fluorescent antigen (DTx) clearance from the injection site 24h after immunization (Fig. 6A), implicating enhanced antigen drainage. Strikingly, transport of the lymphatic tracer indocyanine green (ICG) or fluorescent DTx to dLNs was increased by SMNP and ISCOMs, but not other adjuvants (Fig. 6B-D). Intravital imaging revealed that SMNP increased the diameter of lymphatic vessels draining the injection site within 1h after injection (Fig. 6E-F). Hence, uniquely among a panel of common vaccine adjuvants, saponin-containing adjuvants appear to alter lymphatic drainage. These data show that enhanced lymphatic drainage and disruption of LN CD169⁺ SSMs are two mechanisms of action of saponin adjuvants that contribute to enhanced antigen capture and B cell activation, with saponin/MPLA particulate adjuvants having the greatest impact on these events.

Figure 6. SMNP adjuvant enhances lymph flow in a mast cell-dependent manner.

(A) Fluorescent antigen (DTx) remaining at injection site 24hrs after injection. Wistar rats were injected with 20 μg of fluorescent antigen intradermally in 20 or 40 μl of PBS. Representative data from 2 independent experiments. (n=4 animals, 4 separate injection sites were evaluated in each animal per group).

(B-C). Balb/C mice were injected with 5 μg Indocyanine Green (ICG) dye in the footpad and the draining popliteal LN (pLN) was monitored. Shown are intravital images showing ICG drainage to popliteal lymph-node in mice 1hr following injection (B) and quantification of ICG in pLN (C). Representative data from 2 separate experiments. n=4 mice (no adj) or n=6 mice (adjuvant groups). Each symbol represents a mouse and the bar represents the mean.

(D) Quantification of DTx trafficking to iLN in balb/C mice 4hr after subcutaneous base-of-tail injection (one on each side) with 20ug of fluorescent DTx. Representative data from 2 separate experiments. n = 4 mice per group. Each symbol represents an individual iLN measurement (2 per mouse) and the bar represents the mean.

(E) Representative images of lymphatic vessels 1hr after subcutaneous injection. Scale bar = 50 μm.

(F) Quantification of lymphatic vessel diameter in (E). n = 6 mice per group. Each symbol represents one mouse and the bar represents the mean. Shown are results from 2 independent experiments.

(G) Quantification of ICG drainage into inguinal LNs of Balb/C mice 4 hrs after base of tail injection. A subset of mice were treated with antihistamine (diphenhydramine) just prior to ICG injection. Representative data from 2 separate experiments. n=4 mice (naive) or n=10 mice (ICG groups). Each symbol represents a mouse and the bar represents the mean.

(H) Quantification of ICG drainage into inguinal LNs of C57/b6 or cKitw-sh (mast cell-deficient) mice 4 hrs after base of tail injection. Representative data from 2 separate experiments. n=10 mice (C57b/6) or n=8 mice (cKitw-sh). Each symbol represents a mouse and the bar represents the mean.

(J-L) 10⁶ CTV⁺ VRC01^gHL B cells were adoptively transferred into B6 or cKitw-sh mice prior to vaccination with eOD-60mer-AF488 with SMNP. Antigen uptake and activation of VRC01^gHL B cells in draining iLNs was assessed at 24h post-immunization.

(I) Quantification of eOD-60mer AF488 uptake (geometric MFI).

(J) Quantification of CCR7 expression (geometric MFI).

(K) Quantification of CD86 expression (geometric MFI normalized).

(I-K) Pooled data from 2 independent experiments. n = 7 (B6) or n = 8 (cKitw-sh). Each symbol represents a mouse, bars represent the geometric mean (I) or the mean (J and K).

(L-Q) 2×10⁴ SMARTA CD4⁺ T cells were adoptively transferred into B6 or cKitw-sh mice prior to vaccination with eOD_gp61-60mer with SMNP. SMARTA responses in iLNs were analyzed at d4 post-immunization.

(L) Quantification of total SMARTA CD4⁺ T cells.

(M) Quantification of CXCR5^hi PD1^hi SMARTA T_FH

(N–Q) Cells were re-stimulated ex vivo with LCMV gp_66–77 or irrelevant OVA_323–339 to assess cytokine responses.

(N) Quantification of IL-21⁺ cells as a percentage of SMARTA CD4⁺ T cells.

(O) Quantification of IFNγ ⁺ cells as a percentage of SMARTA CD4⁺ T cells.

(P) Quantification of IL-21⁺ cells as a percentage of SMARTA T_FH cells.

(Q) Quantification of IFNγ ⁺ cells as a percentage of SMARTA T_FH cells.

(N-Q) Pooled data from 2 independent experiments. n = 8 per group. Each symbol represents a mouse, bars represent the geometric mean (L, M) or the mean (N-Q). For (A, H-Q) statistical analysis performed by Mann-Whitney test. For (C -D, F-G) statistical analysis was One-way Anova, followed by Dunnett’s post-test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

Given the unexpected finding of lymph flow alterations, we assessed candidate pathways that might be triggered by SMNP to impact lymphatic vessels. Nitric oxide (NO), VEGF-C, and histamine have been reported to impact lymph flow in certain conditions (38-41). We tested the effect of administering inhibitors of NO or histamine or blocking antibodies against VEGF-R together with SMNP and the lymphatic tracer dye ICG. While NO or VEGF inhibition had no effect on dye accumulation in dLNs, histamine inhibitors modestly reduced ICG uptake in the dLNs (Fig. 6G). As mast cells are key cellular producers of histamine, we next tested whether the lymph flow effects of SMNP were preserved in cKit^W-sh mice, which have a mutation in the 5’ regulatory region of the Kit locus that leads to mast cell abolishment shortly after birth. The enhanced ICG accumulation in dLNs triggered by SMNP was ablated in mast cell-deficient animals (Fig. 6H). To determine if mast cells contributed to the enhanced antigen capture by cognate B cells in dLNs seen with SMNP, we evaluated cognate B cell responses in the VRC01^gHL B cell adoptive transfer model. Absence of mast cells reduced uptake of eOD-GT5 60mer by ~2-fold (Fig. 6I), correlating with weaker VRC01^gHL B cell activation (Fig. 6J, K). While expansion and T_FH differentiation of CD4⁺ T cells was not impacted by mast cell deficiency (Fig. 6L, M), the cytokine polyfunctionality of antigen-specific T_FH cells was reduced (Fig. N-Q). Together, these data suggest that mast cells are one cell population targeted by SMNP for initiating enhanced vaccine immune responses.

SMNP is a highly effective adjuvant in non-human primates

Given the effectiveness and distinctive mechanisms of action observed for SMNP in mice, we next sought to evaluate the potency of this new adjuvant in rhesus monkeys. We previously found that NHPs immunized with Quil-A ISCOMs plus a stabilized SOSIP-type HIV Env trimer based on MD39 (BG505 Olio6 (42)) made robust GC responses and anti-Env IgG binding titers, but little or no autologous tier 2 neutralizing antibodies (10), consistent with HIV Env being a difficult target for eliciting neutralizing antibodies (43). We therefore immunized NHPs with SMNP combined with the MD39 HIV Env trimer. Autologous tier 2 neutralizing antibody titers after SMNP + BG505 MD39 were elicited in all animals and were much higher compared to the previous study (p = 0.0014, Fig. 7A). These responses are comparable to the best BG505 autologous tier 2 neutralizing antibody titers reported to date following 3 immunizations in NHPs by a conventional (not slow delivery (9, 10)) immunization regimen (44-46). High quality neutralizing antibodies are generally the result of successful GC responses. Impressively robust Env-specific B_GC responses were observed in lateral and central axillary LNs proximal to the immunization site (Fig. 7B-D, fig. S6A-D). Notably, Env-specific B_GC were also induced in more distal pectoral and apical LNs, indicating broad recruitment of LNs by SMNP (Fig. 7B-D, fig. S6A-D), and consistent with the enhanced lymphatic drainage observed in mice. As a measure of the safety of SMNP, we immunized animals and collected serum for analysis of systemic cytokines/chemokines over 48 hr post injection. This data revealed transient elevations in IL-6, G-CSF, IL-10, and the chemokines MCP-1 and MIP-1β, and low levels of IFN-γ; however all of these responses returned to baseline by 48 hr (fig. S6E). TNF-α, IL-12, and MIP-1α did not rise above baseline levels (data not shown). Thus, SMNP is a potent and likely safe adjuvant in NHPs.

Figure 7. SMNP is a potent adjuvant in non-human primates.

(A–D) RMs were subcutaneously immunized in the deltoid area with soluble native-like Env trimer BG505 MD39 with SMNP at week 0, 10 and 24. LN necropsies were performed at week 30.

(A) Autologous BG505 pseudovirus ID50 neutralization titers assessed at week 26. Each symbol represents an animal, bars indicate the geometric mean. Previously published data (ref. (10)) of RMs subcutaneously immunized in the mid-thighs with soluble native-like Env trimer BG505 Olio6_CD4ko with ISCOMs in the same schedule are shown as a reference. n = 6 animals per group. Data from one experiment.

(B) Schematic showing the cluster of axillary LNs (lateral, central, pectoral and apical LNs) draining the deltoid region (triangle) in NHPs with approximate distances from the injection site.

(C) Top panel, Representative FACS plots pre-gated on B cells, showing CD38⁻ CD71⁺ total B_GC responses. Bottom panel, MD39 Env-binding cells within the B_GC gate are identified. Numbers indicate percentages.

(D) Quantification of MD39 Env⁺ B_GC cells in draining LNs identified as in (C), expressed as number per 10⁶ lymphocytes. n = 6 animals. Left and right LNs were separately examined. Each symbol represents one side of an animal. Data from one experiment.

(E) Schematic of experimental design for (F-M) to compare the adjuvanticity of CDN, amph-CpG, and SMNP in NHPs. RMs were immunized subcutaneously in each inner thigh with MD39-ferritin NPs along with the specified adjuvant at week 0, 10 and 24. n = 5 (no antigen), n = 6 (adjuvant groups).

(F) HIV MD39 Env binding titers at various time points post prime and boost immunization. Symbols represent mean values. Error bars are SEM.

(G) Antibody dependent cellular phagocytosis, neutrophil mediated phagocytosis, and complement deposition from purified Ab 25 wks post immunization. Data is displayed in arbitrary units (AU).

(H) ID50 autologous BG505 pseudovirus neutralizing titers analyzed 25wks post immunization.

(I, J) B_GC responses of biopsied dLNs at (I) 6 and (J) 14 wks post immunization. Data is shown as percent of total B cells.

(K) HIV MD39 Env binding B_GC cells 6 wks post immunization. Data is shown as percent of total B_GC cells.

(L-M) T_FH responses of biopsied dLNs at 6 and 14 wks post immunization. Data is shown as percent of total CD4⁺ T cells.

For (A, G-M), each symbol represents one animal and the bar represents the mean. Statistical analysis was performed by Mann-Whitney test (A), or One-way Anova, followed by Dunnett’s post-test (G-M). *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

To gain further insight into the nature of immune responses elicited by SMNP in NHPs, we compared SMNP with two other adjuvants capable of promoting type I IFN responses, cyclic dinucleotides (CDNs) (47) and CpG (48). CDNs and CpG are potent adjuvants in mice and act on distinct innate pathways (STING or TLR9, respectively). An appropriate CDN capable of effectively stimulating NHP STING alleles was selected based on rhesus PBMC activation by CDN variants in vitro. 2’3’-c-di-AM(PS)₂(Rp,Rp) was selected as the most potent inducer of cytokine release (fig. S6F). NHPs were vaccinated with a ferritin-based nanoparticle presenting 8 copies of MD39 Env trimer (MD39-NP) (49) combined with SMNP, CDN or CpG at weeks 0, 10, and 24 (Fig. 7E). Adjuvant alone controls were also included. Immunization with MD39-NP and CDNs or MD39-NP and CpG elicited no Env trimer-specific IgG until after the week 10 booster immunization (Fig. 7F). By contrast, immunization with MD39-NP and SMNP elicited substantial IgG titers following a single immunization, with all (6/6) animals seroconverted by week 8 (Fig. 7F). Env IgG titers remained ~100 to ~1000-times higher in animals that received SMNP following the second and third immunizations (Fig. 7F). SMNP also induced more polyfunctional humoral responses, evident by higher antibody-dependent monocyte and neutrophil phagocytosis, and complement deposition 25 weeks post-immunization (Fig. 7G). Additionally, SMNP induced significantly higher autologous neutralization titers compared to the other adjuvants (p < 0.01, Fig. 7H). Flow cytometry analysis of dLNs revealed robust B_GC responses at week 6 following immunization with MD39-NP and SMNP, while B_GC responses were near baseline in other conditions (Fig. 7I); SMNP-induced GC responses were also stronger than the other groups at week 14 (Fig. 7J). Similarly, vaccination with SMNP induced the most robust Env trimer-specific B_GC cells (Fig. 7K). T_FH responses were also significantly greater in animals that received SMNP compared to other adjuvants (Fig 7L-M). Taken together, these data demonstrate that SMNP is a potent adjuvant for humoral immunity in NHPs, consistent with outcomes and mechanisms of action in mice.

DISCUSSION

Saponins incorporated into particulate lipid carriers are safe for use in humans, and are now part of licensed vaccines and ongoing clinical development campaigns. Despite their clinical importance, their means of promoting humoral immunity remain poorly understood. Here, we analyzed immunological mechanisms of action underlying the potency of saponin-based adjuvants, and developed a new ISCOM-type adjuvant physically incorporating the TLR4 agonist MPLA. We found that compared to a broad panel of experimental and clinical alternatives, saponin-containing adjuvants exhibit superior potency, with the saponin/MPLA combination adjuvants SMNP and AS01_B exhibiting the most robust humoral immune responses in mice. HIV Env trimers are poorly immunogenic, and single immunizations with various adjuvants were previously found to induce little or no Env-specific IgG in NHPs (9, 46, 50). Here, we found that vaccination with Env trimers combined with the saponin/MPLA adjuvant SMNP led to seroconversion of all animals following a single immunization and had the capacity to elicit excellent HIV neutralizing antibody titers. Mechanistically, we found that saponins are distinctive in their capability of inducing increases in lymphatic flow, antigen access to cognate B cells, secondary LN engagement, and improved T_FH cell quality.

We found that saponin exhibited substantial synergy when combined with the TLR4 agonist MPLA. TLR agonists have long been viewed as important candidate vaccine adjuvants. Both MPLA and TLR9 agonists (CpG DNA) have reached approval as adjuvant components for licensed vaccines, and there remains high interest in TLR7/8 ligands as candidate adjuvants (51). Although some studies have suggested an advantage of TLR7/8 ligands over other TLR stimuli in NHPs (52, 53), this now seems less clear. For example, a recent study (54) found that a nanoemulsion formulation of TLR4 agonist outperformed alum/TLR7 agonist or TLR7 agonist in the same nanoemulsion, and alum/TLR4 agonist formulations were found to promote greater antibody titer durability than alum/TLR7 agonist (55). Further, a recent NHP study with a candidate COVID-19 subunit vaccine (3) showed that an alum/TLR7 agonist combination was worse than traditional alum in promoting neutralizing responses to SARS-CoV-2. TLR agonists have also been combined with ISCOMs in preclinical studies, and shown the potential for synergy with saponins (11, 12, 27, 56). However, unformulated TLR agonists, especially low molar mass compounds such as imidazoquinoline TLR7 ligands or CpG DNA, exhibit problematic pharmacokinetics with rapid dissemination into the bloodstream from injection sites that can trigger unacceptable systemic inflammation (48, 57). This has motivated a number of approaches to formulate TLR agonists in carriers that can help retain the adjuvants at the injection site and/or dLNs, including alum, oil-in-water emulsions, or formulation in nanoparticles (48, 52, 54, 55, 58-60). Such formulation strategies are capable of retaining the potent immunostimulatory capacity of the TLR agonist while achieving an acceptable safety profile in NHPs. This likely contributes to the success of liposomal AS01 adjuvants (6). SMNP’s nanoparticle form both promotes efficient lymphatic trafficking and enables synergy between saponin and MPLA at a very low dose of the TLR4 agonist; together these factors likely lead to the limited and transient systemic inflammation following parenteral immunization in mice or NHPs.

Vaccine adjuvants have often been reported to recruit phagocytic cells to injection sites, which subsequently acquire and transport antigen to draining lymph nodes (61-63). However, an effect of adjuvants on direct lymphatic transport of antigens has not been previously described. Using B cell adoptive transfers, we unambiguously show that SMNP strongly improves access of cognate B cells to intact antigen in dLNs. We found that AddaVax only had marginal effects on cognate B cell antigen uptake and activation, despite earlier studies implicating this adjuvant class in enhancing antigen accumulation in dLNs (62, 63). Despite the technical challenges involved, this finding underscores the value of analyzing antigen-specific B cells.

Disruption of CD169⁺ SSM may also contribute to enhanced antigen accumulation in dLNs by SMNP. CD169⁺ SSM capture lymph-born particulate antigens (64-66), which can be relayed to follicular B cells (64, 67-70). However, SSM can also limit lymph-born particulate antigens access to follicular B cells (71). Previous studies have shown adjuvants such as LPS and CpG can disrupt CD169⁺ SSM over the course of 4-7 days (68), but the only adjuvants tested herein capable of inducing substantial loss of CD169⁺ SSM within 24h—when antigen trafficking is most concentrated—were saponin-based adjuvants including SMNP and ISCOMs. This finding is consistent with prior work reported similarly rapid SSM loss in response to liposomal saponin (21).

SMNP had profound effects on cognate CD4⁺ T cell responses. CD4⁺ T cells primed in the presence of SMNP produced high levels of IL-21, a key T_FH cytokine. IL-21 derived from T_FH cells is important for B cell proliferation, isotype-switching and differentiation (72, 73). Robust early expression of IFN-α and IL-6 in dLN induced by SMNP likely drive this phenotype (74-77). Although compositionally-related AS01_B also led to early expression of IFN-α and IL-6, SMNP induced substantially more IL-21⁺ T_FH cells. The exact cell types induced to express IFN-α and/or IL-6 by SMNP and AS01_B may underlie the difference. Notably, we found that SMNP also primed a substantial IFN-γ⁺IL-21⁺ T_H1 population, which is important for controlling viral infections. High levels of ICOS and CD25 induced rapidly on cognate CD4⁺ T cells by SMNP likely facilitated efficient differentiation of highly polarized T_FH and T_H1 (73).

We have reported that immunization regimens enabling extended vaccine kinetics can also be used to substantially enhance humoral response to Env trimers (9, 10). Notably, these “extended delivery” methods are based on concepts related to the kinetics of adaptive immune responses and, as such, are complementary to mechanisms of action of adjuvants such as SMNP. Extended delivery immunizations still rely on the use of adjuvants, and we expect that more potent adjuvants such as SMNP will further enhance the efficacy of such approaches.

A limitation of these studies is the inherent inability to predict the safety profile of SMNP in humans without clinical testing. The licensed AS01_B adjuvant provides a useful comparator, however. AS01_B is comprised of saponin and MPLA inserted in 100 nm diameter liposomes at a ~1:1:20 saponin:MPLA:lipid mass ratio. The ~40 nm diameter ISCOMs “honeycomb cages” of SMNP are formed by a 1:0.1:0.1 mass ratio of the same components; both AS01_B and SMNP are dosed in mice at 5 μg saponin. SMNP elicited transient serum cytokine elevations similar to or lesser than AS01_B, and elicited no weight loss in mice. No injection site reactions or gross effects on animal behavior have been reported across more than 74 individual immunizations of NHPs with SMNP to date (including unpublished ongoing studies), suggesting no obvious safety issues. Hence, while safety must ultimately be determined by human testing, these preclinical studies are encouraging. A second limitation is that TLR4 expression exhibits some differences between mice and humans, with TLR4 expressed by naive mouse B cells but not human naive B cells (78). However, as we observed little or no detectable SMNP uptake by B cells in mouse lymph nodes (Fig. 5E), that species difference is not expected to be relevant for SMNP mechanisms of action. Further, the potent cytokine induction in human PBMCs in vitro and humoral responses elicited by SMNP in NHPs—where TLR expression is similar to humans—suggests this adjuvant is promising for human translation.

Overall, we have shown that SMNP is a potent and safe vaccine adjuvant that strongly promotes humoral immunity in mice and NHPs. SMNP augments the immune response via multiple mechanisms of action that synergize to enhance high quality humoral immune responses. These findings suggest SMNP is a promising candidate vaccine adjuvant for clinical translation.

Methods

Study design

The major objective of this study was to evaluate the adjuvant activity of ISCOMs-like saponin nanoparticles in the presence of absence of co-incorporated TLR4 agonist, comparing against well-known experimental and clinical adjuvants in small and large animal models, and to defined mechanisms of action underlying the action of these saponin-based adjuvants. Mice and non-human primates were immunized with clinically-relevant subunit protein immunogens combined with the saponin or alternate adjuvants, and both early (antigen uptake, T_FH and germinal center induction) and later (serum antibody, plasmablast) responses were assessed over time. Mechanistic studies employed adoptive transfer models to allow a focused analysis of antigen acquisition and activation of antigen-specific B cells and T cells, as well as analysis of the immune response in mice genetically modified to lack certain key genes/cell populations or chemically treated to remove target cell subsets such as macrophages.

Mice and Rats

Mouse and rat experiments were performed at MIT or the La Jolla Institute for Immunology (LJI). All experimental procedures were approved by institutional IACUC committees. Experiments were done using sex- and age-matched female mice between 6-12 weeks of age. Balb/C mice (strain: 000671) and C57B/6J mice (strain: 000664) were purchased form the Jackson Laboratory. VRC01^gHL mice, BCR transgenic knock-in mice carrying inferred germline reverted VRC01 IgH and IgL (34), were maintained on the B6.GFP (strain 004353; Jackson Laboratory), C57BL/6J or B6.SJL-Ptprc^apepc^b/BoyJ (B6.CD45.1) (strain 002014; Jackson Laboratory) backgrounds as homozygous lines at LJI. SMARTA mice were maintained on the B6.CD45.1 background at LJI. Mast cell deficient c-Kitw-sh mice (strain 030764; Jackson Laboratory) were used for lymphatic drainage mechanism studies. Female 12-16 wk old Wistar rats (the Jackson Laboratory) were used for imaging antigen drainage from intradermal injections.

Antigens

eOD-GT5 60mer, a self-assembling protein nanoparticle formed by the fusion of the eOD-GT5 antigen with lumazine synthase, was generated as previously described(34, 79). eOD-GT5 has a monovalent binding affinity for VRC01^gHL Fab of K_D ~250 nM, which is near the high end of the spectrum of affinities of eOD-GT8 for naive VRC01-class precursors in HIV-uninfected humans(34). eOD-GT5 60mer was fluorescently labeled with Alexa Fluor 647 (ThermoFisher; A20173) according to the manufacturer’s protocol. In some experiments, we used eOD-GT5 60mer fused with the LCMV gp_61-80 helper epitope (GLKGPDIYKGVYQFKSVEFD) to track CD4⁺ T cell responses. For this immunogen, a GGS linker (GGSGGSGGSGGSGGG) between the N-terminal lumazine synthase and C-terminal eOD was replaced with the LCMV gp_61-80 sequence. All eOD constructs were analyzed by SEC-MALS and SDS-PAGE gels. 60mers were injected at a dose of 5-10 μg per mouse. MD39 HIV Env trimer was generated as previously described(31) and was administered at a dose of 2 μg per animal. Diphtheria toxin CRM197 was purchased from Scarab Genomics LLC (CRM197-10) and was administered at 2 ug per animal. Trimeric hemagglutinin (HA) ectodomain from A/New Caledonia/20/1999 (NC99) was expressed in 293F cells and purified as described previously(80, 81). Mice were immunized with 3.5 μg of HA trimer.

Adjuvants

ISCOMs adjuvant was prepared as previously described(9). Briefly, 20 mg/ml solutions of cholesterol (Avanti Polar Lipids Cat# 700000) and DPPC (Avanti Polar Lipids Cat# 850355) were prepared in Milli-Q water containing 20% w/vol MEGA-10 (Sigma D6277) detergent. Quil-A saponin (InvivoGen; vac-quil) was dissolved in Milli-Q water at a final concentration of 100 mg/ml. All components were mixed at a molar ratio of 10:10:5 (Quil-A:chol:DPPC) followed by dilution with PBS to a final concentration of 1 mg/ml cholesterol. The solution was allowed to equilibrate overnight at room temperature, followed by dialysis against PBS using a 10k MWCO membrane. The adjuvant solution was then sterile filtered, concentrated using 50k MWCO centricon spin filters, and further purified by FPLC using a Sephacryl S-500 HR size exclusion column. SMNP was formulated in a similar manner but included MPLA (Avanti Polar Lipids 699800P) at a molar ratio of 10:10:2.5:1 (Quil-A:chol:DPPC:MPLA). SMNP QS-21 was prepared in a similar manner, substituting QS-21 for Quil-A. QS-21 was purchased from DesertKing.com. SMNP Cy5 was produced in a similar manner to SMNP, substituting the lipid DPPC for DSPC-Cy5. Doses of ISCOMs and SMNP are reported in terms of the amount of saponin administered, calculated by measuring the concentration of cholesterol (Cholesterol Quantitation kit; Millipore Sigma; Cat# MAK043) in the preparation and assuming quantitative incorporation of the saponin during synthesis.

All adjuvant and antigens were prepared in sterile PBS, mixed, and injected bilaterally at the base of tail in 50-100 μL volume per injection site (100-200 μL per mouse). For ISCOMs and SMNP, each mouse received a dose corresponding to 5 μg Quil-A (corresponding to 0.5 μg MPLA for SMNP). AS01_B was taken from licensed Shingrix vaccine doses; the adjuvant is packaged as pure AS01b in a separate vial from the antigen. Each mouse received 1/10^th the human dose of AS01_B, containing 5 μg QS-21 and 5 μg MPLA. This dose has been previously reported and is in line with FDA animals-to-human dosing conversion (82, 83). Ribi (Sigma Adjuvant System, Sigma-Aldrich; Cat# S6322), AddaVax (InvivoGen; Cat# vac-adx-10), or Alum (Alhydrogel adjuvant 2%, InvivoGen; Cat# vac-alu-250) were mixed 1:1 vol:vol with antigen prepared in sterile PBS. At this dose, Ribi contains 25-50 μg MPLA per mouse. In some experiments, 0.5-5 μg free MPLA (PHAD, Avanti Polar Lipids, Cat # 699800) or 1 μg LPS-B5 from E. coli 055:B5 (InvivoGen; Cat# tlrl-pb5lps) was injected in control mice. Amph-CpG 1826 was produced as previously described (48) and used at 1.24 nmol per mouse. CpG 1826 (Invivogen; Cat# vac-1826-1) was used at 50 ug per mouse.

Biophysical characterization of SMNP

Dynamic light Scattering (DLS) images were acquired using a DynaPro Dynamic Light Scatterer. Nanoparticle images were acquired using Transmission Electron Microscopy (TEM) on a JEOL 2100F microscope. The microscope was operated at 200 kV with a magnification range of 10,000-60,000X. All images were recorded on a Gatan 2kx2k UltraScan CCD camera. Size exclusion chromatography (SEC) was performed with an AKTA Pure FPLC system (GE Healthcare) using a Sephacryl S-500 HR column.

Adoptive transfer of T and B cells

B cells were purified from spleens using CD43 MACS MicroBeads (Miltenyi Biotec, Cat# 130-049-801) following the manufacturer’s protocol. Purity of VRC01^gHL B cells (eOD-binding, B220⁺) purified from homozygous VRC01^gHL mice was ~90%. Cell numbers for adoptive transfer were adjusted according to the purity of VRC01^gHL B cells. SMARTA CD4⁺ T cells were purified from the spleens of SMARTA mice using the EasyStep Mouse CD4⁺ T Cell Isolation Kit (STEMCELL, Cat# 19852) according to the manufacturer’s protocol. Purity of SMARTA CD4⁺ T cells (CD4⁺ TCR Vα2⁺) typically exceeded 95%. Purified B or T cells were adoptively transferred via retro-orbital injection in 200 μL RPMI supplemented with 1% FCS 1-2d before vaccination. To examine VRC01^gHL B cell responses in vivo, 10⁶ VRC01^gHL B cells from homozygous VRC01^gHL mice were adoptively transferred. To examine SMARTA CD4⁺ T cells, 2×10⁴ or 10⁶ SMARTA CD4⁺ T cells were adoptively transferred.

In some experiments, purified B or T cells were labeled with CellTrace™ Violet (CTV) (ThermoFisher). For CTV labeling, B cells resuspended at 10⁷ cells/mL in DPBS supplemented with 0.1% bovine serum albumin (Sigma-Aldrich) were incubated with 5 μM CTV for 9.5 minutes at 37°C. The reaction was quenched using 3x volume of ice-cold fetal calf serum (FCS), and washed once in RPMI supplemented with 5% FCS (RP5).

Mouse flow cytometry analyses

Single cell suspensions were generated by mechanical dissociation of spleens and lymph nodes. Erythrocytes in spleens were removed by washing cells in Gibco™ Ammonium-Chloride-Potassium (ACK) lysing buffer (ThermoFisher) at 4°C. To analyze CD169⁺ macrophages or dendritic cells, splenocytes were digested with liberase TM (Roche, Cat# 5401127001) and DNase I (Roche, Cat# 10104159001) at 20 μg/mL and 70 μg/mL, respectively, for 20 min at room temperature. Antibody staining was performed in FACS buffer (PBS, 1% BSA, 0.02% NaN₃, 2mM EDTA). Fc-mediated binding was blocked using 1% normal rat serum (STEMCELL) and 2.5 μg/mL purified anti-CD16/32 (93; Biolegend) at 4°C for 15 min, prior to performing staining with primary antibodies. Staining panels that included CXCR5 or CCR7 was performed at room temperature for 45 min. All other surface staining was performed at 4°C for 30 min. To assess B_GC cells, cells were stained with GL7 PerCPCy5.5 (GL7; Biolegend), CD38 AF488 (90; Biolegend), B220 PECy7 (RA3-6B2; Biolegend) and CD4 BV711(GK1.5; Biolegend). To assess T_FH cells, cells were stained with CD4-BV711 (GK1.5; Biolegend), B220-PECy7 (RA3-6B2; Biolegend), CXCR5-PE (L138D7; Biolegend), PD1-BV421 (29F.1A12; Biolegend) To assess early B cell activation, cells were stained with B220-BV785 (RA3-6B2; Biolegend), CCR7-PE (4B12; Biolegend), CD86 BV605 (GL-1; Biolegend), CD69-PECy7 (H1.2F3; Biolegend), CD44-PerCPCy5.5 (IM7; Biolegend), and CD4 AF700 (RM4.4; Biolegend). eOD-GT8 60mer was generated in-house and conjugated to Alexa Fluor 488 (ThermoFisher; A10235) to detect VRC01^gHL B cells. To assess early T cell responses, cells were stained with CD4 AF700, B220 BV785, CD69 PECy7, CD62L BV711 (Mel-14; Biolegend), CD25 PerCPCy5.5 (3C7; Biolegend), ICOS BB515 (7E.17G9; BD), CD44 BV605 (IM7; Biolegend) and CD45.1 BV510 (A20; Biolegend). Dead cells were stained using Fixable Viability Dye eFluor™ 780 (ThermoFisher). To analyze SMARTA CD4⁺ T cell cytokine responses, single cell suspension was prepared in sterile D10 (DMEM, 10% FCS, Gibco GlutaMAX, Gibco Penicillin-Streptomycin). 2-5×10⁶ cells were restimulated with either LCMV gp66-77 or non-specific OVA323-339 at 2 μg/mL in the presence of GolgiPlug and BME in 300 μL D10 using F-bottom 96-well plates for 5h at 37C. Cells were then surface stained with CXCR5-biotin, CD4-AF700, CD45.1 BV510, B220 BV785, SA-PECy7, PD1 BV605 (29F.1A12; Biolegend), and Viability Dye e780. Cells were then fixed using BDCytofix/Cytoperm for 15min at 4C, then intracellularly stained with recombinant mouse IL-21R-Fc Chimera Protein (R&D System) for 60min at room temperature. After washing, cells were intracellularly stained with F(ab’)₂ goat anti-human IgG PE (Jackson) and IFN-γ FITC. Background cytokine responses (OVA323-339 stimulation) were subtracted from LCMV gp66-77 stimulated samples Data acquisition was performed using FACSCelesta™ (BD Bioscience) and analyzed using FlowJo 10 (Tree Star).

Measurement of lymphatic trafficking and antigen distribution in LNs

Noninvasive fluorescence imaging of lymphatic drainage with Indocyanine Green (ICG): 10 μL aqueous ICG solution (0.5 mg/mL) was injected subcutaneously in the hind footpad of mice with or without the specified adjuvant. 1hr after the injection, the lymphatic clearance of ICG from the injection site into the popliteal lymph node was visualized noninvasively through intact mouse skin. An 808-nm continuous wave laser (CNI lasers, MDL-N-808) was used to illuminate the imaging area with a 50 mW/m² power density. Fluorescence images were captured by an InGaAs SWIR camera (Princeton Instrument OMA V) with a 1,100-nm long-pass filter at 200 ms exposure time. ImageJ was used to calculate intensity values in the popliteal LN region of interest.

Imaging of antigen drainage to LNs using ICG or fluorescent proteins: DTx was labeled with Alexa-fluor 750 (ThermoFisher: A20111) according to the manufacturer’s instructions. 20 μg of fluorescent protein was injected subcutaneously with or without the specified adjuvant at the tail base of Balb/C mice in 50 ul volume. 5 hrs post-injection the draining inguinal LN was harvested and imaged using an IVIS imaging system (PerkinElmer: 124262) with imaging parameters optimized for Alexa-fluor 750 (Ex: 745, Em: 820). Living Image (PerkinElmer) was used to calculate total radiant efficiency values at each LN. IVIS imaging of draining lymph nodes was also performed on mice receiving adjuvants combined with 10 μL aqueous ICG solution (0.5 mg/mL). For some experiments, mice received pharmacological blockade of NO (1400W, 10mg/kg intraperitoneal injection; Tocris), VEGF (anti-VEGF receptor Ab, 20mg/kg intravenous injection; BioXcel DC101) or histamine (diphenhydramine hydrochloride, 10mg/kg intraperitoneal injection; Tocris) prior to ICG/SMNP injections.

Imaging of antigen drainage from injection site: Female 12-16 wk old Wistar rats (the Jackson Laboratory) were shaved then injected intradermally (ID) in the dorsal paraspinal region with 20 μg of DT-750 in 20 μL PBS with or without 5 μg of SMNP. Rats were anesthetized with isoflurane and imaged (Ex: 745, Em: 820) immediately and 24h post injection using an IVIS Spectrum bioluminescent and fluorescent imaging system (Perkin Elmer). Total radiant efficiency per injected area was analyzed with Living Image software (Perkin Elmer).

Measurement of lymphatic diameter

Lymphatic vessel diameter measurements were performed as previously described(84, 85). Mice were anesthetized with 250 μl of 100 mg/ml Ketamine/xylazine. A 5 μg injection of SMNP was administered in the right footpad followed by a 2 μl injection of 2% w/v FITC-dextran (2 million molecular weight; Sigma Aldrich cat # FD2000S) 1hr later in the same site. The afferent collecting lymphatic vessels draining to the popliteal lymph node (PLV) were surgically exposed and imaged. Intravital images of the exposed lymphatic vessels were captured with an inverted fluorescence microscope (Olympus IX70 Inverted Epi-Fluorescence Microscope).

ELISA

Serum anti-MD39 Env antibody ELISAs were performed as previously described (86). For anti-DTx and NC99 Influenza ELISAs, antigen was coated on MaxiSorp plates (ThermoFisher) at 2 ug/ml. Mouse sera was diluted in 2% BSA block starting at 1:100 with 3x serial dilutions and incubated for 1 hr, followed by detection with 1:5000 goat anti-mouse IgG-HRP, IgG1-HRP, or IgG2a-HRP conjugate in block. Titers are reported as inverse dilutions giving an 0.2 HRP absorbance.

Cytokine analysis

Draining LN cytokine measurements were performed using a LEGENDplex anti-virus response panel (Biolegend: 740622) following the manufacturer’s instructions. Serum cytokine samples were submitted for analysis by Eve Technologies using a 31-plex cytokine/chemokine array (MD31).

Subcapsular sinus macrophage depletion

For SSM depletion, clodronate or control liposomes (Encapsula NanoSciences) were injected subcutaneously 6 days prior to immunization according to the manufacturer’s instructions.

NHP studies

For studies in Fig. 7A-D, male and female Indian rhesus monkeys (RMs, Macaca mulatta) were housed at the Yerkes National Primate Research Center and were maintained in accordance with NIH guidelines. This study was approved by the Emory University Institutional Animal Care and Use Committee (IACUC). Animals were between 4 – 6.5 years of age at time of 1st immunization. Animals were immunized at 3 time points: week 0, week 10, and week 24. All immunizations were given subcutaneously in the left and right deltoid with 50μg of MD39 and 375ug of SMNP adjuvant per side.

At the end of the 30-week study, animals were sacrificed. Clusters of draining axillary lymph nodes were harvested and defined based on their location: lateral, central, pectoral or apical. For each animal, 1 lateral, 1 central, 1 pectoral, and 2 apical/subclavicular lymph nodes were collected from each side. To generate fluorescent antigen probes, biotinylated MD39-avi was tetramerized with fluorescently labeled streptavidin (Alexa Fluor 647 and BV421) at RT for 20 minutes. For the B_GC panel, cells were incubated with the probes for 30 minutes at 4°C and then with surface antibodies [CD4 (APCe780; SK3; ThermoFisher), CD8α (APCe780; RPA-T8; ThermoFisher), CD16 (APCe780; eBioCB16; ThermoFisher), CD20 (Alexa Fluor 488; 2H7, Biolegend), IgG (PE-Cy7; G18-145; BD Biosciences), IgM (PerCPCy5.5; G20-127; BD Biosciences), CD38 (PE; OKT10; NHP Reagents), CD71 (PE-CF594; L01.1; BD Biosciences)] and Fixable Viability Dye e780 (ThermoFisher) for an additional 30 minutes. Validation of CD38 and CD71 as surface markers for B_GC cells were previously described(10). For the GC-T_FH panel, cells were incubated with surface antibodies [CD4 (BV650; OKT4; Biolegend), CD8 (Qdot705; 3B5; ThermoFisher), CD20 (PE-TexasRed; B9E9; Beckman), CXCR5 (PE; MU5UBEE ThermoFisher), PD1 (BV605; EH12.2H7; Biolegend), CD3 (BV786; SP34-2; BD Biosciences)] and Fixable Viability Dye e506 (ThermoFisher) for 30 minutes at 4°C, washed twice, and fixed with 1% paraformaldehyde. The samples stained for B_GC were acquired and bulk sorted on a FACSAria II (BD Biosciences). The samples stained for GC T_FH were acquired on LSR Fortessa (BD Biosciences). B_GC and GC T_FH data reported are normalized to the number of lymphocytes gated to account for the varying sizes and numbers of harvested lymph nodes.

For studies in Fig. 7E-M, male Indian rhesus monkeys (RMs, Macaca mulatta) were housed at the Southwest National Primate Research Center and were maintained in accordance with NIH guidelines. This study was approved by the Texas Biomedical Research Institutional Animal Care and Use Committee (IACUC). Animals were between 6.8-10.1 years of age at time of 1st immunization. Animals were immunized at 3 time points: week 0, week 10, and week 24. All immunizations were given subcutaneously in the left and right inner thigh (1mL per injection site).

MD39 Ferritin nanoparticles were produced as previously described(49) and were administered at a dose of 50 μg s.c. in the left and right inner thigh (100 ug per animal). Adjuvants used were amph-CpG (250 ug per site, 500 μg total per animal), cyclic dinucleotide (2’3’-c-di-AM(PS)₂, 50 μg per site, 100 ug total per animal) or SMNP adjuvant (187.5 μg per site, 375 ug total per animal). Inguinal lymph node biopsies were collected at weeks 6 and 14 for flow cytometry analysis.

Fluorescent MD39 or Ferritin probes were generated by combining biotinylated antigen with fluorescent streptavidin (APC and BV605) at RT for 20 min. For the B_GC panel, cells were incubated with fixable viability dye Live/Dead Acqua (ThermoFisher) for 20 minutes, followed by the probes for 30 minutes at 4°C and then with surface antibodies [CD3 (AF700; SP34-2; BD Bioscience), CD20 (BUV737; 2H7, Biolegend), IgG (PE-Cy5; G18-145; BD Bioscience), IgM (FITC; G20-127; BD Bioscience)]. Subsequently cells were fixed with True-Nuclear transcription factor buffer set (Biolegend) following manufacturer instructions, and stained with intracellular antibodies [Bcl-6 (PE; K112-91; BD Bioscience), Ki67 (BV421; B56, BD Bioscience)]. For the GC-T_FH panel, cells were incubated with surface antibodies [CD3 (AF700; SP34-2; BD Bioscience), CD4 (APC-Cy7; OKT4; Biolegend), CD8a (BV711; RPA-T8; Biolegend), CXCR5 (PE-Cy7; MU5UBEE, Biolegend), PD1 (BV785; EH12.2H7; Biolegend)] and fixable viability dye Live/Dead Aqua (ThermoFisher) for 30 minutes at 4°C, washed twice, and fixed with 1% paraformaldehyde. All samples were acquired on LSR Fortessa (BD Biosciences).

Serum samples were collected at weeks 0, 6, 8, 10, 11, 12, 14, 24, 25 and 26. Antibody titers were measured by ELISA as follows: MAXIsorp 96 well plates (ThermoFisher) were coated with 1 ug/mL of rabbit anti-his tag antibody (GenScript) in PBS, followed by his-tagged MD39 trimer at 1 ug/mL. Sera was diluted in 2% BSA block starting at 1:50 with 4x serial dilutions and incubated for 2 hr, followed by detection with 1:5000 goat anti-human IgG-HRP (Jackson ImmunoReserach). Alternatively, samples were detected using anti anti-Rhesus IgG1, IgG2 or IgG3 (NHP Reagent Resource), followed by anti-mouse IgG-HRP in block. Titers are reported as inverse dilutions giving an 0.2 HRP absorbance.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism v8. Bars plotted in linear scale represent the mean. Bars plotted in log scale represent the geometric mean. Symbols in bar graphs represent individual mice unless otherwise stated in figure legends. Data sets were tested for normality log-distributed datasets were transformed to linear scale before performing statistical analysis. Multiple group comparisons were performed by one-way ANOVA followed by a post hoc test specified in figure legends. Data containing two independent variables were analyzed using two-way Anova, followed by Benjamin, Krieger and Yekitieli’s post-test. Data were considered to be statistically significant at * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001.