Research ArticleVaccines

Vaccine priming is restricted to draining lymph nodes and controlled by adjuvant-mediated antigen uptake

See allHide authors and affiliations

Science Translational Medicine  07 Jun 2017:
Vol. 9, Issue 393, eaal2094
DOI: 10.1126/scitranslmed.aal2094

Moving beyond mice for vaccine studies

Vaccine enhancement by adjuvants has been known for decades, but the mechanistic differences in how specific adjuvants influence the immune response are just beginning to be elucidated. Liang et al. sought a model that closely mimics humans, so they intramuscularly immunized nonhuman primates with a prototypical HIV antigen in combination with various adjuvants. They then inspected the muscles and lymph nodes to characterize antigen-presenting cells and resulting adaptive immune responses. Their findings should provide valuable information on adjuvant selection for vaccine development in humans.


The innate immune mechanisms by which adjuvants enhance the potency and protection of vaccine-induced adaptive immunity are largely unknown. We introduce a model to delineate the steps of how adjuvant-driven innate immune activation leads to priming of vaccine responses using rhesus macaques. Fluorescently labeled HIV-1 envelope glycoprotein (Env) was administered together with the conventional aluminum salt (alum) adjuvant. This was compared to Env given with alum with preabsorbed Toll-like receptor 7 (TLR7) ligand (alum-TLR7) or the emulsion MF59 because they show superiority over alum for qualitatively and quantitatively improved vaccine responses. All adjuvants induced rapid and robust immune cell infiltration to the injection site in the muscle. This resulted in substantial uptake of Env by neutrophils, monocytes, and myeloid and plasmacytoid dendritic cells (DCs) and migration exclusively to the vaccine-draining lymph nodes (LNs). Although less proficient than monocytes and DCs, neutrophils were capable of presenting Env to memory CD4+ T cells. MF59 and alum-TLR7 showed more pronounced cell activation and overall higher numbers of Env+ cells compared to alum. This resulted in priming of higher numbers of Env-specific CD4+ T cells in the vaccine-draining LNs, which directly correlated with increased T follicular helper cell differentiation and germinal center formation. Thus, strong innate immune activation promoting efficient vaccine antigen delivery to infiltrating antigen-presenting cells in draining LNs is an important mechanism by which superior adjuvants enhance vaccine responses.


Effective vaccines against HIV/AIDS, malaria, and tuberculosis will require induction of high-magnitude antibodies (Abs), T cell responses, or both. Protein-based vaccines offer a safe, scalable, and flexible platform for inducing broad-based adaptive immunity but are usually poorly immunogenic without the addition of an immunostimulatory adjuvant (1). Adjuvants can be used to alter the magnitude and quality of the adaptive vaccine response, enhance the efficiency by lowering the antigen dose, and reduce the number of immunizations required to reach a protective threshold. The most commonly used adjuvant in humans is alum, consisting of precipitates of aluminum salt to which vaccine antigens are adsorbed. Although alum confers protective Ab responses when given with several vaccines (for example, hepatitis B virus and human papillomavirus), it is clear that a number of newer adjuvant formulations are more potent and will be important for improving vaccine efficacy against a variety of infections. In this regard, the introduction of oil-in-water emulsions, such as the clinically licensed MF59 and AS03, has significantly increased influenza vaccine efficacy (24). Another class of promising adjuvants is based on Toll-like receptor (TLR) ligands that can specifically activate antigen-presenting cells (APCs) and B cells. Among these, TLR ligands that activate intracellular TLR3, TLR7/8, or TLR9 have been shown to induce strong Ab and T cell responses. We have previously shown in vitro and in vivo that the use of TLR7/8 ligands can directly activate multiple subsets of human and nonhuman primate (NHP) dendritic cells (DCs) and B cells, resulting in enhanced antigen-presenting capacity and increased T cell and B cell responses (57).

Details on the in vivo mechanisms that dictate the priming of vaccine-induced immunity using models that more accurately translate to humans are lacking. Here, we established an NHP model to determine the innate immune mechanisms initiated by adjuvants in vivo. Outbred genetically diverse rhesus macaques have a high degree of similarity with the human immune system in terms of specific immune cells, such as DC subsets, repertoire of receptors including TLRs, and innate responses (5, 6, 8). Thus, they offer a powerful in vivo model for yielding data to optimize formulations and delivery approaches of human vaccines.

The early immune responses determine the characteristics and magnitude of the adaptive responses that likely account for the differences in vaccine efficacy found with different adjuvants. One long-standing hypothesis was that certain adjuvants create a depot of the antigen at the site of injection. However, data indicate that the adjuvant effect is far more complex and that various aspects of innate immune activation are involved (1). The sequence of events occurring between the administration of a vaccine (most often an intramuscular injection) and the development of the adaptive response needs to be better defined. Skeletal muscle contains few immune cells (9, 10), and sufficient delivery of vaccine antigen therefore likely requires inflammation, including recruitment of such cells to the site of vaccine administration. It remains to be elucidated whether the augmented vaccine responses found with certain adjuvants are due to enhanced mobilization of immune cells, better antigen uptake, enhanced migration of antigen-bearing cells to lymph nodes (LNs), better antigen-presenting capacity, and/or preferential expansion and differentiation of desired T and B cell phenotypes (3).

Here, we used alum as a benchmark for comparing how a TLR7 ligand adsorbed to alum (alum-TLR7) and the clinically used emulsion adjuvant MF59 induce innate activation, leading to priming of adaptive vaccine response. These adjuvants were chosen because they are distinctly different in their composition and have demonstrated superiority over alum regarding enhancement of Ab production as well as T cell responses in humans and animal models (2, 1117). The HIV-1 envelope glycoprotein (Env) antigen was chosen because it is clinically used in HIV-1 vaccine efficacy trials, making it a relevant combination with these adjuvants.


Adjuvants induce a robust recruitment of neutrophils, monocytes, and DCs

Adjuvant-mediated enhancement of adaptive vaccine responses was first confirmed by comparing the neutralizing Ab responses at 26 weeks after four immunizations (0, 4, 12, and 24 weeks) of Env (12). Rhesus macaques immunized with Env in MF59 or alum-TLR7 showed significantly higher neutralization capacity of HIV-1 (tier 1 strain) compared to those immunized with Env in alum (P = 0.0141 and P = 0.009, respectively) or Env alone (P = 0.0085 and P = 0.0048, respectively) (n = 6 per group) (Fig. 1A). Although eliciting similar levels of vaccine-specific Abs, alum-TLR7 induced a T helper 1 cell (TH1) type of CD4+ T cell response defined by interferon-γ (IFN-γ) production, whereas MF59 induced a mixed TH1/TH2 response with both IFN-γ and interleukin-4 (IL-4) (Fig. 1B). Because innate immune responses dictate adaptive immunity, we hypothesized that such quantitative improvement and qualitative differences by MF59 and alum-TLR7 stem from the type of activation that these adjuvants induce directly after administration.

Fig. 1. Adjuvant-driven accumulation of APCs in muscle and draining LNs.

(A and B) Neutralization of HIV-1 (MW965.26 strain) (A) and IFN-γ+ or IL-4+ spots per million peripheral blood mononuclear cells (PBMCs) (B) 14 days after the fourth immunization with Env alone or with adjuvants (n = 6 per group). Bars represent the mean. ID50, amount of plasma inhibiting infection by 50%. Neutrophils and monocytes in muscle (C and D) and draining LNs (E and F) 24 hours after injection of PBS (P) or adjuvants with (+) or without (−) Env (MF59, n = 5; alum, n = 5; alum-TLR7, n = 6; Env alone, n = 3). (G) Staining of CD11c+ myeloid DCs (MDCs) and CD123+ plasmacytoid DCs (PDCs) in muscles receiving Env with indicated adjuvant (top) and control muscles (bottom). (H and I) Infiltrating MDCs (H) and PDCs (I) in the muscle 24 hours after injection. (J and K) DC subsets in the adjuvant-injected muscle (J) and draining LNs (K) at baseline (no injection), 24 or 72 hours (n = 3 per group) after Env and adjuvant injection. Means ± SEM. *P < 0.05, **P < 0.01 (Student’s t test).

To evaluate this, naïve rhesus macaques were divided into four groups that received Env with or without adjuvant to be analyzed after 24 hours (MF59, n = 5; alum, n = 5; alum-TLR7, n = 6; Env alone, n = 3) or 72 hours (n = 3 per group). Each animal was given intramuscular injections at four different sites, that is, the left and right deltoids and quadriceps, at the same time to study multiple sites of vaccine injections including controls using the same animals, thereby expanding our data collection and limiting the number of animals for experimentation (table S1).

We first analyzed Env/adjuvant administration after 24 hours, which is a time point that has shown strong local innate immune activity in mice (14, 18). Consistent with previous observations, high numbers of CD66abce+ neutrophils and CD14+ monocytes infiltrated the adjuvant-injected muscle (Fig. 1, C and D, and fig. S1A). When compared to phosphate-buffered saline (PBS)–injected muscles, all the adjuvants induced significant infiltration of neutrophils (MF59, P = 0.0177; alum, P = 0.0111; alum-TLR7, P = 0.0059) and monocytes (MF59, P = 0.0034; alum, P = 0.0076; alum-TLR7, P = 0.0171) (Fig. 1, C and D). Env alone did not induce or induced only slight cell infiltration, and there was similar infiltration whether the adjuvant was given with Env or not, indicating that the recruitment of cells was adjuvant-driven (Fig. 1, C and D). There was no difference in monocyte and neutrophil recruitment into the muscle between the adjuvant groups. However, significantly higher numbers of neutrophils accumulated in the draining LNs in the MF59 group compared to the alum (P = 0.0181) or alum-TLR7 (P = 0.0233) group (Fig. 1E). Confocal microscopy confirmed the elevated levels of neutrophils in the LNs draining the sites of MF59 injection (fig. S1, B to D). Some but not all MF59-draining LNs had increased monocyte levels (Fig. 1F).

All adjuvants also induced robust infiltration of DCs to the injection site (Fig. 1G). The main DC subsets, CD11c+/low MDCs and CD123+ PDCs, were detected in the muscles injected with adjuvants at significantly higher levels compared to the animal-matched PBS-injected muscle (MDCs: MF59, P = 0.0087; alum, P = 0.012; alum-TLR7, P = 0.0089; PDCs: MF59, P = 0.0186; alum, P = 0.0220; alum-TLR7, P = 0.0029) (Fig. 1, H and I). MF59 recruited higher levels of MDCs compared to alum but not to alum-TLR7. In contrast, alum-TLR7 induced the highest infiltration of PDCs. Accumulation of DCs was rapid and transient at vaccine-injected sites and their draining LNs (Fig. 1, J and K).

Alum-TLR7 stimulates phenotypic maturation and IFN-α production

An important effect of adjuvants for enhancing antigen-presenting capacity may be their ability to induce phenotypic maturation of DCs (1921). Low or no up-regulation of costimulatory molecules, such as CD80 on DCs, monocytes, and neutrophils, was observed in muscles injected with MF59 and alum compared to the animal-matched PBS-injected muscles (Fig. 2A). In contrast, alum-TLR7–injected muscle showed strong phenotypic differentiation at 24 to 72 hours in all animals. Major histocompatibility complex (MHC) class II [human leukocyte antigen–DR (HLA-DR)] up-regulation was particularly evident on neutrophils and not noticeable on monocytes and DCs because their steady-state expression was high as expected (Fig. 2B). In addition, alum-TLR7, but not alum or MF59, induced strong maturation of primary human DCs after in vitro adjuvant stimulation (n = 5) (Fig. 2C).

Fig. 2. Alum-TLR7 induces strong DC activation involving type I IFN responses.

(A and B) Histograms of CD80 (A) and MHC II (HLA-DR) (B) expression on infiltrating cell subsets in the muscle 24 hours after Env plus adjuvant (open) versus PBS injection (filled) (MF59, n = 5; alum, n = 5; alum-TLR7, n = 6). (C) CD80 and CD40 expression by in vitro adjuvant-stimulated purified human primary DCs (n = 5). Sections of muscle 24 hours after injection with Env together with alum-TLR7 (D and E) or alum (F and G). Muscle bundles are visualized by wheat germ agglutinin (blue) (n = 3 per group). Red and green represent MxA+ and Env+ cells, respectively. MxA staining at 24 hours in draining LN of alum-TLR7 (H) and PBS-injection site (I) (n = 3 per group). DAPI+ cell nuclei are in gray. (J) IFN-α production by sorted human primary PDCs stimulated in vitro with the adjuvants or free TLR7 ligand (TLR7L) (n = 5). Bars represent the mean. ***P < 0.001 (Student’s t test).

Because PDCs are known producers of type I IFN in response to TLR7 signaling, we analyzed myxovirus resistance gene A (MxA) expression induced by type I IFNs. MxA+ cells were detected in the muscles receiving alum-TLR7 (Fig. 2, D and E) but not muscles receiving alum (Fig. 2, F and G) or MF59 (fig. S2A) (n = 3 per group). MxA expression was also detected in the alum-TLR7–draining LNs (Fig. 2H) but not in the animal-matched distal PBS-draining LNs (Fig. 2I) or in LNs of the alum or MF59 groups (fig. S2B). In addition, sorted human PDCs exposed to alum-TLR7 produced high levels of IFN-α in vitro (n = 5) (Fig. 2J). Together, this demonstrates that alum-TLR7 is unique compared to MF59 and alum at directly inducing phenotypic maturation and production of IFN-α.

Infiltrating cell subsets efficiently take up the vaccine antigen

After establishing that both MF59 and alum-TLR7 generally induced a more powerful cell migration and activation compared to alum, the impact on antigen uptake was assessed. Env gp120 used in our studies was labeled with Alexa Fluor 680 to enable tracking of antigen after delivery. Env+ cells were readily detected in the muscle and draining LNs in all adjuvant groups (Fig. 3, A to C). No Env+ cells were found in the PBS-injected muscle and draining LNs. The restricted deposit of the vaccine to the LNs draining the vaccine injection site was also illustrated by the localization of MF59 by confocal microscopy (n = 3) (Fig. 3, D and E).

Fig. 3. Env+ cells in muscle and draining LNs.

(A to C) Alexa Fluor 680–labeled Env uptake by MDCs in the muscle and draining LNs 24 hours after injection in the MF59 [dioactadecyloxacarbocyanine (DiO)-labeled] (A), alum (B), and alum-TLR7 (C) groups (MF59, n = 5; alum, n = 5; alum-TLR7, n = 6). SSC-A, side scatter area. (D and E) Draining LNs at 24 hours after MF59 (red) (D) or PBS (E) injection. CD20+ B cells and CD3+ T cells are in green and blue, respectively. DAPI+ cell nuclei are in gray. (F) Env+ cells in the muscles and draining LNs at 24 or 72 hours (n ≥ 3 per adjuvant group; n ≥ 1 Env alone group) after injection. (G) Proportions of Env+ cell subsets in muscle and draining LNs at 24 hours. Means ± SEM. *P < 0.05 (Student’s t test).

Neutrophils, monocytes, and MDCs showed substantial Env uptake in the muscle (Fig. 3F). The highest numbers of Env+ cells in the muscle were detected at 24 hours after injection, followed by a rapid decline at 72 hours. Although the alum group tended to show the overall lowest frequencies of Env+ cells, there were no statistical differences between the groups (Fig. 3F). Because neutrophils and monocytes were frequent cell types infiltrating the muscle, they also represented the largest proportions of Env+ cells in all adjuvant groups (Fig. 3G). However, in the draining LNs, a large proportion of Env+ MDCs and also PDCs were found in the alum and alum-TLR7 groups compared to the MF59 group where Env+ monocytes and neutrophils were the predominant populations in line with the migration pattern (Fig. 3G). Env uptake was found to be similar at the injection sites and draining LNs in animals naïve to Env (n = 4 per group) versus those previously immunized (n = 3 per group) (fig. S3).

Antigen+ DCs, monocytes, and neutrophils in vaccine-draining LNs present to CD4+ T cells

DCs are essential for priming of T cells, but because Env was also efficiently taken up by monocytes and neutrophils, we investigated whether they present Env to T cells after vaccination. To address this in the NHP model, we used animals previously immunized with Env that had generated strong Env-specific CD4+ T cell responses at the time (>1.5 years earlier) (n = 3). The animals received a new injection with Alexa Fluor 680–labeled Env together with either of the adjuvants and an injection with PBS in the opposite deltoid. Draining LNs were collected at 24 hours, and the Env+ fraction of MDCs, monocytes, and neutrophils were isolated by flow cytometry (Fig. 4A). The isolated Env+ cells were cocultured, without the addition of any exogenous Env, with CD4+ T cells from cryopreserved PBMCs collected at peak response. Antigen presentation was demonstrated by proliferating Env-specific memory T cells [carboxyfluorescein diacetate succinimidyl ester (CFSE) low] after 5 days of coculture. As expected, MDCs induced the highest proliferation of responding T cells followed by monocytes (Fig. 4B). Neutrophils also showed T cell stimulatory capacity in line with their up-regulated MHC II and costimulatory molecule expression. The specificity of the proliferative response to Env was confirmed by the lack of proliferation stimulated by MDCs, monocytes, and neutrophils sorted from the PBS-draining LNs in the same animal (Fig. 4, B and C). Furthermore, Env+ cells sorted from Env/adjuvant–draining LNs in naïve animals were unable to induce T cell proliferation (n = 2) (Fig. 4D). The hierarchy of presentation capacity of MDCs, monocytes, and, last, neutrophils was the same with any of the adjuvants. Consistent with this, cells directly isolated from vaccine-draining LNs in all the adjuvant groups showed higher CD80 and HLA-DR expression on MDCs and monocytes compared to neutrophils (MF59, n = 4; alum, n = 3; alum-TLR7, n = 4; PBS, n = 3) (Fig. 4E).

Fig. 4. Env+ cells in draining LNs present Env with different efficiency.

(A) Isolation of the Env+ fraction of CD14+ monocytes, CD66abce+ neutrophils, and CD11c+ MDCs from vaccine-draining LNs 24 hours after injection. (B) Env-specific autologous CD4+ T cell responses evaluated by CFSE dilution assay after 5 days of coculture with Env+ neutrophils, monocytes, or MDCs, sorted from vaccine-draining LNs (top) and same cell subsets from PBS-draining LNs (Env) (bottom). Compiled data showing percent proliferating CD4+ T cells in animals with preexisting Env immunity (n = 3) (C) and naïve animals (n = 2) (D) after coculture with Env+ or Env cell subsets. (E) Expression of CD80 and MHC II (HLA-DR) by the indicated cells in the draining LNs at 24 hours (MF59, n = 4; alum, n = 3; alum-TLR7, n = 4; PBS, n = 3). MFI, mean fluorescence intensity. Means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 (Student’s t test).

Priming of T cell responses is restricted to the vaccine-draining LNs

Because the location of Env+ antigen-presenting DCs, monocytes, and neutrophils was exclusively in the vaccine-draining LNs, we analyzed whether the kinetics and magnitude of priming of Env-specific CD4+ T cell responses in the LNs were influenced by the adjuvant given. Naïve animals were immunized in separate muscles at different time points, and the draining LNs were collected (n = 3 per group). Cells from the LNs were restimulated with Env protein or peptides in vitro to evaluate whether Env-specific responses had been generated. No or very few Env-specific proliferating CD4+ T cells could be expanded from the draining LNs after 2 or 24 hours after immunization (Fig. 5A). At 72 hours, low yet detectable levels of responding T cells were found, whereas strong responses were found after 10 days. There was no proliferation detected at 10 days in LNs not draining the vaccine injection site. This again demonstrates that priming of responses occurs exclusively in the vaccine-draining LNs and that Env-specific T cells are not disseminated to other LNs at a detectable level during the first 10 days of priming. At day 10 after immunization, the animals receiving MF59 or alum-TLR7 showed higher proliferation compared to the animals in the alum group (Fig. 5, A and B). Low levels of Env-specific CD8+ T cells were also detected in the vaccine-draining LNs at 10 days in some of the animals in the alum-TLR7 and MF59 groups. This suggests that the enhanced activation and mobilization of Env+ MDCs, monocytes, and neutrophils with MF59 or alum-TLR7 translate into more efficient and/or rapid onset of antigen presentation, potentially even some cross-presentation, and generation of antigen-specific T cells. The same pattern was observed after boost immunization (MF59, n = 5; alum, n = 5; alum-TLR7, n = 3) (Fig. 5, C and D). However, at boost immunization, notable responses were found in the PBS-draining LNs, presumably representing circulating Env-specific memory T cells responding to in vitro recall stimulation. Nevertheless, there were substantially higher levels of proliferating T cells in the vaccine-draining LNs, demonstrating that the location for antigen presentation and stimulation of T cell responses is exclusively the vaccine-draining LNs.

Fig. 5. MF59 and alum-TLR7 enhance priming of Env-specific CD4+ T cells.

Animals received injections with Env plus indicated adjuvant at different sites before collecting the draining LNs at either 2 hours, 24 hours, 72 hours, or 10 days. (A) CD4+ T cell proliferation in CFSE-labeled LN suspensions in the presence of Env protein. (B) Compiled background (medium only)–subtracted data of CD4+ T cell responses at indicated time points in the absence or presence of Env antigen (n = 3 per group). (C) CD4+ T cell proliferation in response to Env protein in vaccine-draining LNs 10 days after fifth immunization (MF59, n = 5; alum, n = 5; alum-TLR7, n = 3). (D) Compiled background-subtracted data of Env-specific CD4+ T cell proliferation after boost in the absence or presence of Env antigens. (E) Pearson correlation analysis of Env-specific CD4+ T cell responses in vaccine-draining LNs at day 10 and Env+ cells in the draining LNs detected at 2 hours to 10 days after immunization. Means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 (Student’s t test).

The number of Env+ cells in the vaccine-draining LNs at 2 hours to 10 days correlated with the level of proliferating T cells at 10 days after vaccination (Fig. 5E). The availability of Env+ APCs may therefore directly regulate the priming efficiency and generation of Env-specific T cells.

T follicular helper cell differentiation and germinal center formation correlate with the level of T cell priming

An important step after T cell priming is the differentiation of CD4+ T follicular helper (Tfh) cells that are central for the formation of germinal centers (GCs) critical for the selection, affinity maturation, and class switching of vaccine-specific B cells (22). We found that the MF59 and alum-TLR7 groups showed higher numbers of Tfh cells, identified as CD3+ CD4+ CXCR5high and PD-1high among the central memory T cells (23), in the vaccine-draining LNs 10 days after priming compared to alum (n = 3 per group) (Fig. 6, A and B). The GCs, defined by their polarized location of PD-1+ Tfh cells and Ki67+ B cells in B cell follicles (Fig. 6, C to H, and fig. S4, A to D), were also found to be more frequent in the MF59 group (fig. S4E) and of larger size in both the MF59 and alum-TLR7 groups compared to alum (Fig. 6, I and J). In addition, the frequencies of PD-1+ Tfh cells as well as proliferating Ki67+ cells within the GCs were higher in the MF59 and alum-TLR7 groups (Fig. 6, K and L) and were closely linked to the size of GCs (fig. S4, F and G).

Fig. 6. Tfh cells and GCs in draining LNs after priming.

(A) Tfh cells within CD4+ central memory T cells. (B) Percent Tfh cells in vaccine-draining LNs versus non–vaccine-draining (distal) LNs at day 10 (n = 3 per group). Means ± SEM. (C to H) GCs at day 10 after Env plus MF59 (C and D), alum (E and F), or alum-TLR7 (G and H) immunization. CD3 (blue), Ki67 (red), and PD-1 (programmed cell death protein 1) (green). (I and J) Individual GC areas at day 10 after priming (I) and when normalized by LN section area (J) (n = 3 per group). (K and L) PD-1+ (K) and Ki67+ (L) cells within GCs. Bars represent mean. (M and N) Correlation of normalized total GC area at day 10 (two sections per LN) versus Env+ cells in draining LNs (M) or percent Env-specific CD4 T cells (N) (n = 3 per group). *P < 0.05, **P < 0.01 (Student’s t test).

The differentiation of Tfh cells and B cells in the GC reaction has been proposed to be driven by the availability of antigen (24). Consistent with this, both the levels of Env+ cells and primed Env-specific CD4+ T cells correlated with the GC area (Fig. 6, M and N). This indicates that increased mobilization of vaccine antigen+ cells and priming result in enhanced Tfh and B cell activity, which, in turn, determines the induction of GCs and production of vaccine-specific Abs (fig. S5).


Extreme antigen variability and the stringent requirements for potent and durable Ab and T cell responses for protection are major hurdles in the development of next-generation vaccines against infections like HIV-1, malaria, and tuberculosis. A much better understanding of how strong vaccine responses can be elicited, tailored, and sustained over time with potent adjuvants would have a significant impact on the design of new vaccines. Although alum is the most common clinically used adjuvant, there are a number of new adjuvant candidates that induce superior vaccine responses. For influenza vaccines, the approved oil-in-water adjuvants MF59 and AS03 were shown to induce higher virus-specific Ab seroconversion rates and heterosubtypic neutralization against diverse H5N1 types in addition to allowing antigen dose sparing (2, 4). This demonstrates that existing vaccines can be improved by replacing alum with a more potent adjuvant and that the choice of adjuvant may be critical for the success of new vaccines to induce high Ab titer and breadth for protection.

The few in vivo reports that characterize the early mechanisms dictating vaccine responses were performed in mice, which differ substantially from humans in their distribution of immune cell populations and innate immune responses. Volume and doses of antigen and adjuvants used in mice are often not equivalent to those given in humans. We therefore used rhesus macaques to characterize how the innate immune events differ after administration of alum compared to two more potent, clinically relevant adjuvants, alum-TLR7 and MF59, shown in clinical as well as preclinical studies to induce superior vaccine responses over alum (2, 11, 12). However, although rhesus macaques likely offer more translational value for human vaccine responses than inbred mice, there are limitations in exploring certain mechanistic aspects because of the lack of genetically modified primate models. The ethical considerations using minimal numbers of primates in combination with the natural variation in their responses can make it challenging to obtain conclusive results especially when precise sampling procedures and kinetics are critical.

The main function of a vaccine adjuvant, such as alum, was originally proposed to be the creation of a depot of the vaccine antigen at the site of injection, which led to a gradual antigen release (1). We found that substantial frequencies of immune cells infiltrated the vaccination site and internalized the antigen/adjuvant. This peaked at 24 hours and thereafter rapidly declined in all groups. Similar kinetics were observed in the vaccine-draining LNs. This suggests that immune activation, antigen uptake, and cell mobilization occurs quickly after adjuvant administration regardless of the adjuvant formulation. Creating a vaccine depot in the injected muscle is unlikely central to any of the three adjuvants studied here. A consistent finding was that the vaccine distribution after delivery was restricted to the site of infection and the draining LNs, which is in line with earlier mouse studies (14, 25, 26). This was also supported by our data showing that antigen presentation and stimulation of Env-specific T cell responses were exclusively taking place in the vaccine-draining LNs, both at prime and boost immunizations.

The availability of vaccine antigen and suitable cells to present antigen likely plays a major role for determining the magnitude of vaccine responses generated. The intracellular rate of antigen decay has been shown to be slower in cells exposed to antigen adsorbed to alum than antigen alone, which resulted in prolonged duration of antigen presentation by DCs (27). We found that the kinetics of priming of Env-specific CD4+ T cell responses in vaccine-draining LNs were similar in all adjuvant groups, indicating that MF59 (where the antigen is not adsorbed to the adjuvant as with alum and alum-TLR7) had sufficient antigen retention and presentation to generate good responses. The alum group showed lower levels of primed T cells in the vaccine-draining LNs compared to MF59 and alum-TLR7. This could be explained by the lower number of infiltrating Env+ cells, particularly of Env+ neutrophils and monocytes, in the vaccine-draining LNs in the alum group.

The distinct characteristics of the innate immune responses induced by MF59 versus alum-TLR7 defined by neutrophil mobilization and type I IFN response, respectively, likely imprint the type of adaptive response elicited. We found that alum-TLR7 induced a strict TH1 type of CD4+ T cell response, whereas MF59 induced a mixed TH1/TH2 response. IFN-α production by alum-TLR7 could account for the polarized TH1 response (28). The induction of an early IFN-α response is also associated with creating an innate antiviral state that can play an important role for preventing viral transmission to some degree before adaptive vaccine immunity has been established. This could be particularly important for HIV-1 vaccines where inflammatory innate responses by MF59 may not have this effect (29). The TH1-polarized response with alum-TLR7 may also be suitable for cancer vaccine strategies. Selected TLR agonists alone or in combination with emulsion (MONTANIDE) have shown promise for cancer vaccines by enhancing CD8+ T cell immunity (30, 31). We have reported earlier that the TLR7/8 agonist in MONTANIDE amplified the boost effect of an adenovirus vector–based vaccine in NHPs (5).

In contrast to the IFN-α and TH1 response, the neutrophil activation induced by MF59 may contribute to the mixed TH1/TH2 response. Neutrophils have been demonstrated to influence the recruitment and differentiation of certain TH types via the production of selected chemokines (32, 33). Env+ neutrophils isolated from the vaccine-draining LNs in the current study were able to present antigen to T cells. Mouse and human neutrophils have been shown earlier to present antigen to memory T cells (32, 34, 35). Although the levels of MHC II and costimulatory molecules on neutrophils are lower than those on MDCs and monocytes, it is plausible that neutrophils contribute to presenting vaccine antigen to T cells at least in a boost situation. Neutrophils were also reported to exert suppressive effects on the stimulation of T cell responses in a mouse model (36). However, deletion of neutrophils in mice did not affect the adjuvant effect of MF59 (14). Neutrophils can also promote B cell survival and differentiation (37). The differentiation of Tfh cells and GC formation may therefore be supported by the mobilization of neutrophils induced by MF59 and lead to enhanced Ab responses. Instead, IFN-α production, known to support B cell differentiation and Ab production, may be an important contributor to the improved Ab responses by alum-TLR7 (6, 28). Despite the differential innate profiles by alum-TLR7 and MF59, they induce similar quantity and quality of Ab responses, as we reported recently (12). Further studies on the qualitative aspects of the Abs induced by MF59 versus alum-TLR7, such as the isotypes and the glycosylation pattern associated with distinct functions (38), are needed to increase the understanding on when alum-TLR7 versus MF59 would be the more suitable adjuvant.

Frequencies of Tfh, GC B cells in LNs, and GC structures have been associated with successful vaccine immunity in several mouse studies (15, 16, 39, 40). Our finding that GC size correlates with the level of Env+ cells mobilized to or residing in the vaccine-draining LNs suggests that the more robust innate activation as a whole with MF59 and alum-TLR7 compared to alum leads to efficient antigen delivery to APCs and is a critical mechanism for enhancing vaccine responses. In conclusion, in this study, we demonstrate that the magnitude of vaccine immunity is determined early and locally after vaccine administration. The in vivo model introduced here can easily be adapted for other vaccine formulations to decipher vaccine persistence, biodistribution, targeting of specific immune cell subsets, and other functions central for priming the vaccine responses. Information of this kind will guide future vaccine design and rational selection of the most appropriate formulation to a given pathogen.


Study design

The overall objective of this study was to investigate the type of early in vivo immune responses that would account for the enhanced potency of vaccine adjuvants MF59 and alum-TLR7 when compared to benchmark adjuvant alum. In-depth characterization of innate activities in muscles injected with adjuvanted HIV-1 Env and the corresponding draining LNs as well as priming of vaccine immunity were assessed using an NHP model for definite translation of such responses in humans. No randomization or blinding was performed. In most of the experiments, four immunizations were performed (left and right deltoid and quadriceps muscles) to maximize data collection using fewer animals and to be able to include internal controls such as PBS injection and adjuvant alone for comparison. With this study design, a minimum of three animals per group was used similar to previous studies (9, 14, 18). Number of animals analyzed is stated in figure legends. Primary data are in table S3.

Adjuvants and antigens

The adjuvants and Env were provided by Novartis Vaccines. The HIV-1 Env gp120 monomer derived from the TV1 or SF162 strains was fluorescently labeled with Alexa Fluor 680 (Molecular Probes). The labeled Env was adsorbed to either aluminum hydroxide (alum) or alum preadsorbed with the TLR7 agonist LDH 153 (100 μg/ml) (11). Alternatively, labeled Env was mixed with DiO-labeled MF59 (5% squalene, 0.5% Tween 80, and 0.5% Span 85 in citrate buffer), as described earlier (14). Responses were assessed by neutralization of HIV-1 (heterologous tier 1A strain MW965.26) plus T cell ELISpot for IFN-γ and IL-4, as described previously (12).


Approval for this study was granted by the Animal Care and Use Committees of the Vaccine Research Center, National Institutes of Health (NIH). Indian rhesus macaques were housed at BIOQUAL or the Division of Veterinary Resources, NIH, and handled according to the standards of the Association for Assessment and Accreditation of Laboratory Animal Care. Each Env-naïve animal received intramuscular injections at four different sites, as mentioned above (table S1) and as described earlier (9). All injection volumes were 1 ml. Two sites were injected with 100 μg of labeled Env alone, mixed with MF59/DiO at a 1:1 ratio, or adsorbed to either 1 mg of alum or alum-TLR7. The remaining two injection sites received adjuvant alone and PBS.

Statistical analysis

Two-sided paired or unpaired t test and Pearson correlation analysis were performed using GraphPad Prism version 5.0c software and considered significant at *P < 0.05, **P < 0.01, and ***P < 0.001.


Materials and Methods

Fig. S1. Accumulation of neutrophils in response to adjuvant injection.

Fig. S2. MxA expression after alum or MF59 immunization.

Fig. S3. Env uptake by Env-naïve or Env-boosted animals.

Fig. S4. In situ analysis of GCs.

Fig. S5. Schematic overview on adjuvant-induced innate and adaptive immune responses.

Table S1. Intramuscular immunization schema.

Table S2. Abs for flow cytometry.

Table S3. Primary data.

References (41, 42)


  1. Acknowledgments: We thank S. Rao, J. P. Todd, D. Ambrozak, K. McKee, J. D. Hernandez, A. Ploquin, A. S. Martinez, N. Björkström, M. Friess, and M. Vono for assistance and advice and T. Sandberg for the illustration. Funding: K.L. was supported by grants from Vetenskapsrådet and the Swedish Governmental Agency for Innovation Systems (Vinnova). F.L. and K.J.S. are recipients of scholarships from the Swedish Society of Medicine, the Fernström Foundation, and the Swedish Society for Medical Research. N.J.S., R.A.K., and R.A.S. were supported by intramural funds as U.S. NIH investigators. The HIV-1 Env protein production was funded by the National Institute of Allergy and Infectious Diseases–NIH HIV Vaccine Research and Design grant #5P01 AI066287 to former Novartis. Author contributions: F.L., N.J.S., R.A.K., R.A.S., and K.L. designed the research. F.L., G.L., K.J.S., E.A.T., J.R.F., and K.L. performed the experiments. F.L., G.L., K.J.S., J.R.F., R.A.S., N.J.S., R.A.K., and K.L. analyzed the data. D.T.O., A.S., E.D.G., and S.B. designed the research and contributed with the adjuvants and antigens. F.L., G.L., K.J.S., R.A.S., and K.L. wrote the paper. Competing interests: D.T.O., A.S., and E.D.G. are inventors on patents owned by the GSK group of companies, which includes alum-TLR7 #US20110053893A1. D.T.O., A.S., and E.D.G. were previous employees of Novartis Vaccines and Diagnostics and are now employed by GSK Vaccines. S.B. was a previous employee of Novartis Vaccines and Diagnostics and is now employed by Bill and Melinda Gates Foundation. All other authors declare that they have no competing interests. Data and materials availability: The adjuvants and Env antigen were available from Novartis at the time under a Materials Cooperative Research and Development Agreement with the Vaccine Research Center, NIH.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article