Introduction

Vaccines are amongst the most significant public health advancements of the past century, effectively controlling or eradicating numerous infectious diseases. However, adjuvant options are extremely limited, and the mechanisms of action of currently licensed adjuvants are not entirely elucidated. For example, aluminum-based adjuvants (alum) have been the main immunostimulants used in vaccines for almost a century; however, the exact pathways through which alum triggers these immune responses remains incompletely understood1,2,3,4. Moreover, while vaccines are administered with the greatest frequency in the first year of life, the specific immune effects of vaccination and adjuvants during early life have not been well defined5.

The majority of mechanistic studies of vaccine adjuvants have been performed in adults and animal models of adult immune systems6. However, it has become increasingly clear that there are major differences between the infant and adult immune systems7. For example, at birth and during infancy, the immature immune system has an inherent Th2 bias8. More in-depth characterization of vaccination in the context of such differences in immune environments at these distinct developmental stages is thus of particular importance given the large percentage of vaccines that are given in infancy and the clear differences between the infant and adult immune systems, which likely play a role in differential responsiveness to vaccination in early life5.

As proposed by the hygiene hypothesis, exposure to T helper type 1(Th1)-polarizing stimuli drives the regulation of the immune system away from the inherently heavy T helper type 2 (Th2) bias present at birth and during infancy, ultimately resulting in a greater balance in Th1/Th2 polarization9,10. The Th2 bias presented in neonatal cells results from a reduced capacity for IFN-γ production as well as increased expression of inhibitory receptors that lead to enhanced immune regulation by pro-tolerogenic components that keep T cell responses in check to prevent pathology11,12,13,14,15,16. Perturbation of this shift away from Th2 immunity in early life may alter the normal progression of immune regulation and result in inappropriate skewing of immune responses later in life. Conversely, immunization with adjuvants that drive Th1-polarized immunity may prime the immune system for the development of Th1 or balanced Th1/Th2 immune responses. A majority of vaccines are administered during this early crucial phase of immune development, highlighting the potential to use vaccines to positively influence the maturation of the immune system during this window of opportunity. Furthermore, the neonatal period is of particular importance as trained immunity has the potential for more profound effects early in life when the immune system is the most susceptible to instruction17. As environmental exposures such as infections and pollutants are known to shape and train the immune system during development18,19, we hypothesize that vaccines and their accompanying adjuvants may also play a pivotal role in this process. As such, vaccines could be a tool by which to influence optimal immune responses throughout life.

Potential nonspecific protective effects of vaccines have been identified, with a growing amount of evidence suggesting that certain vaccines may exert heterologous effects beyond just their specific antigenic targets. For example, the Bacillus Calmette-Guérin (BCG) vaccine, which induces strong Th1 immune responses, has been shown to contribute to protection against heterologous pathogens and enhance antibody responses induced by other vaccines administered later on, with some studies demonstrating a reduction in all-cause mortality in infants that received the BCG vaccine20,21,22. Conversely, another study in Australia demonstrated that replacement of the Th1-polarizing whole-cell (wP) pertussis vaccine with the Th2-polarizing acellular antigen (aP) coincided with a significant increase in food allergy incidence23. Children that received the Th1-polarizing wP vaccine had a decreased risk of developing IgE-mediated food allergy compared to those receiving aP. Together, these findings suggest that similar to other environmental factors, vaccines may also have a role in training the immune system to influence subsequent immune responses in an antigen-agnostic manner.

Understanding the impact and mechanisms by which adjuvants influence immune development in early life is of basic and translational importance. Despite this, experimental models to address these mechanistic gaps are lacking. In this context, we focus on the major adjuvant currently used in early-life vaccinations: alum. Herein, we show that alum induces more strongly Th2 polarized immunity when administered to neonates. These studies interrogate the hypothesis that early-life immunization with Th2-polarized alum-adjuvanted vaccines may alter the maturation of the immune system, leading to a preferential priming of Th2 immune responses to other antigens encountered later in life and demonstrate that immunization with a Th1-polarizing adjuvant promoted more balanced Th1/Th2 immunity. Because the alum-adjuvanted hepatitis B (HB) vaccine is typically given within 24 h of birth24, it is crucial to understand how early-life alum immunization influences immune cell function to alter the immune environment to allow for rationale adjuvant design to promote optimal immune development.

Results

Model of early-life immunization

To determine the optimal dosage of alum in neonatal mice, a dose-response experiment was performed to assess immunogenicity of the HB-alum vaccine in neonates. Neonatal mice were immunized with 4 µg of HBsAg adsorbed on 1–100 µg of alum in 10 µl total vaccine volume. As expected, increasing the amount of alum adjuvant in the vaccine resulted in increased immunogenicity, as measured by HB-specific antibodies (Supplementary Fig. 1). The 25 µg dose was chosen to move forward for all subsequent studies in neonatal mice, as this was the lowest the dose that achieved maximal antibody induction. Because maintaining the same ratio of alum:immunogen is important for comparing immune responses, adult mice were immunized with 50 µl of the same vaccine as neonates, which delivers 20 µg HBsAg and 125 µg alum, consistent with other reports of alum-adjuvanted vaccines in adult mouse models25,26,27.

We next assessed the immunogenicity of the HB-alum vaccine at different developmentally relevant ages in mice. Mice were immunized at day of life (DOL) 7, 21, or 56. While it is not possible to exactly match mouse ages with corresponding developmental ages in humans, these ages are representative of neonatal, infant/child, and adult mice28,29. A single immunization with the HB-alum vaccine was sufficient to induce HB-specific IgG antibody titers (Fig. 1A). The age at immunization did not significantly affect the total HB-specific IgG titers [geometric mean titer (GMT) = 20,824; 84,472 and 12,306 for DOL7, 21, and 56, respectively]. HB-specific IgG titers were not significantly different in mice immunized at DOL 7 vs 21 (p = 0.1748) or DOL21 vs 56 (p = 0.0577) due to the titer spread after one immunization. However, the HB-specific IgG subclass repertoire was different, with mice that received the HB-alum vaccine at DOL7 and DOL21 having increased IgG1 and reduced IgG2a compared with mice immunized at DOL56 (Fig. 1B).

Fig. 1: Early-life HB-alum immunization results in a more Th2 skewed HB-specific cellular immune response.
figure 1

Mice (n = 8–16) were immunized intramuscularly (IM) at DOL7, 21, or 56 with HB-alum or with alum alone at DOL7. HB-specific serum (A) IgG or (B) IgG1 and IgG2a subclasses was determined. C Splenocytes were harvested and re-stimulated with HB for 72 h, and cytokine secretion was measured. Antibody titers are presented as GMT ± 95% CI, and cytokine data are displayed as mean ± SEM. Statistical differences were determined using the Mann–Whitney test; *p < 0.05.

Full size image

HB-alum vaccine induces a more Th2 skewed cytokine response profile when administered in early life

While there were no age-dependent differences in the magnitude of the humoral immune response to the HB-alum vaccine, cellular immune responses were characterized to determine if there were changes in immune polarization. To determine if the cellular immune response was also impacted by age at immunization, splenocytes were harvested and restimulated with HB to determine antigen-specific cytokine secretion (Fig. 1C). Splenocytes from mice immunized on DOL7 or 21 produced greater Th2-type cytokines IL-5 and IL-13 compared to cells from mice immunized on DOL56. Immunization at DOL7 induced significantly more IL-5 than immunization at DOL56 (p = 0.0007). Immunization at DOL21 induced a trend for increased IL-5 compared with DOL56, but this was not statistically significant. IL-13 production correlated with age at immunization, as mice immunized at DOL7 produced significantly more IL-13 compared with DOL21 and 56 (p = 0.0014 and p = <0.0001). Mice immunized at DOL21 also produced significantly more IL-13 compared to mice immunized at DOL56 (p = 0.0002). Conversely, lymphocytes from mice immunized on DOL56 produced significantly more of the Th1-type cytokine IFN-γ, compared to cells from mice immunized on DOL7 and 21 (p = 0.0015 and p = 0.0266, respectively). HB-specific production of IFN-γ, IL-5, and IL-13 were all negligible for control groups given PBS, unadjuvanted HB, or alum only. These data indicate that the cellular response to the HB-alum vaccine is more Th-2 skewed when administered in early life as compared to adulthood.

Early-life HB-alum immunization results in the development of Th2-polarized immune responses to subsequent antigen exposure

We next wanted to investigate the effects of early-life HB-alum vaccination on the development of de novo immune responses to subsequent exposure to new antigens. Mice were immunized with HB-alum as described above and beginning 4 weeks later, received 4 weekly i.n. exposures to OVA to mimic mucosal antigen exposure (Fig. 2A). The mice were sacrificed 2 weeks after the final OVA exposure, and splenocytes were harvested and stimulated with OVA for 72 h in order to measure cytokine secretion profiles. IFN-γ, IL-4, IL-5, and IL-13 were quantified to determine the Th1 or Th2 polarization of the OVA-specific immune response induced by intranasal OVA exposure (Fig. 2B, C). Intranasal OVA exposure induced low but quantifiable cytokine responses in control mice treated with PBS at DOL7 or DOL56. Mice immunized at DOL7 or DOL56 with alum alone developed similar OVA-specific responses as PBS-immunized mice, demonstrating that immunization with alum alone without antigen did not result in changes to the induction of immune responses to intranasal OVA exposure. Immunization with HB alone without adjuvant also did not change the immune responses to intranasal OVA exposure (Supplementary Fig. 2). The OVA-specific cytokine production in these control groups was mostly limited to IFN-γ and IL-10, however mice immunized with PBS or alum alone at DOL7 also produced low levels of IL-5 which did not occur in mice immunized with these controls at DOL56. The immune response induced following intranasal OVA exposure was altered in mice immunized at DOL7 with HB-alum. IFN-γ and IL-10 production was significantly suppressed (p < 0.0001) compared to PBS immunized mice. Conversely, the production of OVA-specific Th2 cytokines, IL-4, IL-5, and IL-13, was significantly elevated in mice immunized with HB-alum at DOL7 compared with PBS immunized mice (p < 0.0001). The cytokine response induced following intranasal OVA administration was similar for mice immunized with HB-alum at DOL21 as was observed with mice immunized with HB-alum at DOL7, with very low secretion of IFN-γ and increased production of Th2 cytokines (Supplementary Fig. 3). However, Th2 cytokine production in the HB-alum group was lower overall than observed in mice immunized at DOL7. These alterations in the polarization of the immune response to intranasally administered OVA was unique to early-life immunization with HB-alum, as there was no suppression of IFN-γ or IL-10 or induction of Th2 cytokines in mice immunized at DOL56 (Fig. 2C). It has previously been reported in adult mice immunized with alum in allergic models that subsequent co-administration of the antigen included in the alum-based vaccine with a new antigen via intranasal or epicutaneous routes resulted in Th2-biased immune responses to the co-administered bystander antigen30,31. To test this in our model, adult (DOL56) mice were immunized with HB-alum followed by simultaneous co-administration of OVA with HB intranasally. Similar to previous observations, co-administration also resulted in reduced production of OVA-specific IFN-γ and increased IL-5 and IL-13 (Supplementary Fig. 4). However, only immunization with HB-alum in neonates (Fig. 2B) and not adults (Fig. 2C and Supplementary Fig. 3) resulted in alterations of the immune response to separately administered OVA, demonstrating these cross-priming effects are specific to immunization in early life.

Fig. 2: Early-life HB-alum immunization results in the development of Th2-polarized immune responses to bystander OVA exposure.
figure 2

A Mice (n = 8–10) were immunized at DOL7 or 56 with HB-alum and received 4 weekly i.n. exposures to OVA from weeks 4–7. Mice were sacrificed 2 weeks after the final OVA exposure, and splenocytes were harvested and stimulated with OVA for 72 h. B, C Cytokine secretion was measured in cell culture supernatants. D, E Mice were immunized at DOL7 with HB-alum and received 4 weekly i.n. exposures to OVA during weeks 11–14. Mice were sacrificed 2 weeks after the final OVA exposure, and splenocytes were harvested and stimulated with OVA for 72 h. E OVA-specific cytokine secretion was measured in cell culture supernatants. Data are displayed as mean ± SEM. Statistical differences were determined using the Mann–Whitney test; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Full size image

To demonstrate that alterations in the polarization of the immune response to intranasal OVA exposure were not dependent upon the age at which mice were exposed to OVA and to examine the longevity of the Th2 polarizing effects, a similar study was performed in which mice were immunized at DOL7 with HB-alum with intranasal OVA exposure delayed until 11 weeks following immunization when mice were 12 weeks old, which is the same age at which mice that begin the study at DOL56 received intranasal OVA (Fig. 2D). Even with delayed exposure to OVA, mice immunized at DOL7 with HB-alum still developed Th2-polarized responses to OVA. These results demonstrate that the alterations in the immune response that lead to the predisposition for developing Th2-polarized responses to intranasal OVA are durable and not dependent upon age at the time of the heterologous antigen exposure. Overall, these data demonstrate that immunization with HB-alum in early life can preferentially prime for the generation of Th2-biased immune responses to novel antigen exposures.

Mice immunized in early life with HB-alum have stronger induction of Th2- and reduced Th1-skewed responses to subsequent immunization with an OVA-MPLA vaccine

We next wanted to determine whether the effects of early-life immunization with HB-alum could be overcome by delivering the heterologous antigen with a Th1 skewing adjuvant. Neonatal (DOL7) mice were used for the early-life timepoint, as the data presented above demonstrate greater differences at this timepoint compared with DOL21, and the HB-alum vaccine is routinely administered to human newborns, making this a clinically relevant timepoint. Mice were immunized with HB-alum as described above and beginning 4 weeks later received a two-dose series of i.m. immunizations with OVA adjuvanted with monophosphoryl-lipid A (MPLA) (Fig. 3A). Regardless of timing or the type of the initial vaccination, all groups developed similar OVA-specific IgG in response to the two-dose OVA-MPLA vaccine series (Fig. 3B), demonstrating that the magnitude of the humoral immune response was not influenced by prior immunization. MPLA is an adjuvant that preferentially induces Th1-polarized immune responses in naïve animals. Indeed, mice immunized with PBS at either DOL7 or 56 developed Th1-polarized immune responses to OVA, characterized by more IFN-γ production and less production of Th2 cytokines, IL-5 and IL-13 (Fig. 3C). There were no differences in responses to the OVA-MPLA vaccine in mice that began the protocol at DOL7 compared with those that started on DOL56. Additionally, immunization with alum alone at DOL7 or 56 did not have any impact on the immune response to OVA-MPLA. Interestingly, mice that were first immunized with HB-alum at DOL7 developed a markedly different cellular immune response to the OVA-MPLA vaccine, with significantly increased Th2 cytokines (IL-5 and IL-13, p < 0.0001) and reduced IFN-γ compared with mice immunized with PBS at DOL7 (p < 0.0001). Serum OVA-specific IgG subclasses were also altered. Mice immunized at DOL7 with HB-alum had higher IgG1 antibodies and reduced IgG2a compared with mice immunized at DOL7 with PBS or alum alone, indicative of a more Th2 skewed immune response to the OVA-MPLA vaccine in mice immunized at DOL7 with HB-alum (Supplementary Fig. 5). These effects were only observed in mice immunized with HB-alum in early-life, as mice immunized with HB-alum on DOL56 had a similar pattern of cytokine secretion as mice immunized with PBS at either DOL7 or 56. Immunization with OVA-MPLA had no impact on the HB-specific immune responses, as mice that received HB-alum at DOL7 had similar titers of HB-specific IgG as the DOL56 immunized group, while the recall cytokine response to HB re-stimulation demonstrated significantly more IL-5 and IL-13 and reduced IFN-γ in mice that were immunized at DOL7 compared to those immunized at DOL56 (Supplementary Fig. 6). To determine if these results were specific to HB, we examined whether swapping the antigens used for early-life immunization gave similar results. In this experiment, OVA-alum was administered in early life followed by subsequent immunization with a HB-MPLA vaccine. Similar results were observed as described above, with reduced HB-specific IFNγ and increased IL-5 and IL-13 production in mice that were immunized at DOL7 with OVA-alum (Supplementary Fig. 7).

Fig. 3: Mice that are immunized in early-life with HB-alum have a more Th2 skewed response to subsequent immunization with OVA-MPLA.
figure 3

A Mice (n = 7–12) were immunized at DOL7 or 56 with HB-alum and received 2 i.m. immunizations with OVA-MPLA. B OVA-specific total IgG was measured in serum and (C) OVA-specific cytokine secretion was measured in splenocyte cultures. D Mice (n = 5–9) were immunized as shown in panel (A) with the OVA-MPLA vaccines administered in contralateral immunization sites. OVA-specific cytokine secretion was measured in splenocyte cultures. Data are displayed as mean ± SEM. Statistical differences were determined using the Mann–Whitney test; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Full size image

We next wanted to determine the effect of the site of immunization of the HB-alum vaccine on its potential to influence responses to subsequent immunization with OVA-MPLA. Alum has the potential to persist in low amounts at the site of immunization due to its depot promoting properties which contribute to its adjuvanticity32. The impact of immunization site on responses to subsequent immunizations has not been well-studied, but emerging evidence suggests that there may be differences in immune responses with booster immunizations given in contralateral versus ipsilateral arms; however this is likely dependent on the type of vaccine and adjuvant administered33,34,35. In the previous study, both vaccines were administered to the same location (right hind leg). To test if these effects were limited to sequential immunizations occurring at the same site, mice were administered HB-alum in the hind left leg at DOL7 followed by OVA-MPLA in the hind right leg, following the same schedule shown in Fig. 3A. Similar to the study with immunization at the same site, the response to the OVA-MPLA vaccine was similarly altered in mice immunized with HB-alum at DOL7 in the contralateral limb, with significantly reduced production of IFN-γ (p = 0.0002) and a concomitant increase in IL-5 (p = 0.0012) and IL-13 (p = 0.0033) compared to mice immunized with PBS at DOL7 (Fig. 3D). Immunization with HB-alum at DOL56 had no impact on the immune response to subsequent immunization with OVA-MPLA in the contralateral leg.

Addition of a Th1-polarizing adjuvant in an early-life vaccine results in a predisposition for Th1 polarization of subsequent heterologous immune responses

Given that HB-alum induces significantly stronger Th2 immunity in neonatal mice, we hypothesized that reducing the Th2 bias of the initial early-life vaccination could prevent the subsequent preference for driving Th2 immunity. In order to determine this, the Th1-polarizing adjuvant CpG was added to the HB-alum vaccine. The addition of CpG to the HB-alum vaccine resulted in a HB-specific immune response that was Th1-polarized in mice that were immunized at DOL7 or DOL56 (Supplementary Fig. 8). Production of Th2 cytokines IL-4, IL-5, and IL-13 was significantly higher in mice that had been immunized as neonates. However, the production of these cytokines was greater than 10-fold lower than IFN-γ. There was not a significant difference in IFN-γ production. Mice were immunized with HB-alum+CpG at DOL7 and subsequently immunized with OVA-MPLA as described in Fig. 4A, and OVA-specific cytokine production was quantified to determine the effects of early-life immunization on subsequent immunity to the heterologous immunization with OVA. Interestingly, the addition of CpG was sufficient to prevent the Th2 polarization that occurred in mice immunized at DOL7 with HB-alum (Fig. 4B). Instead, in mice immunized with HB-alum+CpG at DOL7, OVA-MPLA immunization induced a cellular immune response towards OVA with significantly increased IFN-γ (p = 0.0006) and IL-10 (p = 0.0023) with significantly less IL-4, IL-5, and IL-13 (p = 0.0006). These results suggest that adjuvants that induce Th1-polarizing immune responses can prevent the immune changes leading to Th2 polarization of subsequent immune responses to heterologous antigens.

Fig. 4: Neonatal immunization with a CpG adjuvanted HB vaccine primes for the generation of Th1-polarized immune responses to subsequent immunization with OVA-MPLA or OVA-alum.
figure 4

A, B Mice (n = 8) were immunized with PBS or HB-alum or HB-alum+CpG at DOL7 and subsequently immunized with OVA-MPLA. OVA-specific cytokine secretion was assessed in splenocytes. C–F Mice (n = 5–7) were immunized at DOL7 with HB-CpG, HB alone, CpG alone or PBS and subsequently immunized with OVA-alum. OVA-specific cytokine secretion was assessed in splenocytes. Data are displayed as mean ± SEM. Statistical differences were determined using the Mann–Whitney test; *p < 0.05, **p < 0.01.

Full size image

We next investigated if a CpG-adjuvanted HB vaccine could result in preferential priming of Th1-polarized immune responses to subsequent immunization with OVA. Similar to the previous experiments with early-life HB-alum immunization, mice were immunized at DOL7 with HB-CpG. Mice were subsequently immunized with an OVA vaccine adjuvanted with alum and a recall assay was performed in splenocytes to characterize OVA-specific cytokine production (Fig. 4C). Control mice that were sham immunized at DOL7 with PBS had a balanced Th1/Th2 response to the OVA-alum vaccine (Fig. 4D–F). Conversely, mice that were immunized at DOL7 with HB-CpG produced significantly more IFN-γ (p = 0.0025) and significantly less IL-5 and IL-13 (p = 0.0177 and p = 0.0025, respectively). These data suggest that early-life immunization with HB-CpG influences the response to a subsequent alum-adjuvanted vaccine to promote stronger Th1 polarization.

Differential responses of BMDCs from mice immunized in early life to activation of innate immune receptors

To determine if early-life immunization resulted in systemic immune changes that could be found at the level of bone marrow immune progenitor cells, the responsiveness of bone marrow-derived DCs (BMDCs) to various TLR ligands was investigated. Mice were immunized with PBS at DOL7 or HB-alum or HB-CpG at DOL7 or DOL56. Bone marrow was harvested 4 weeks after immunization, and BMDCs were generated by culturing in vitro. BMDCs were stimulated with TLR ligands Pam3CSK4 (TLR2), LPS (TLR4), and CpG (TLR9), and secreted cytokines were measured in cell culture supernatants to characterize activation of BMDCs. IL-6 production in response to stimulation of TLR2 (Pam3CSK4) or TLR9 (CpG) was not impacted by early-life immunization (Fig. 5A, E), however these TLR ligands induced significantly less TNFα in BMDCs from mice immunized with HB-alum at DOL7 compared to unvaccinated controls (Fig. 5B). BMDCs from mice immunized at DOL7 with HB-alum had the most significant alterations in their response to TLR4 stimulation with LPS, with significant reductions in both IL-6 and TNFα. IL-12 production was almost completely absent in BMDCs from mice immunized at DOL7 with HB-alum in response to stimulation with all three TLR ligands (Fig. 5C). Conversely, IL-1β production was significantly greater upon stimulation of TLR2, TLR4 and TLR9 in BMDCs from mice immunized with HB-alum at DOL7, with the greatest increase following TLR4 stimulation (LPS) (Fig. 5D). As these results suggested that HB-alum immunization at DOL7 altered DCs, we assessed the phenotype of DCs in the draining iLNs (gating strategy shown in Supplementary Fig. 9). Mice were immunized at DOL7 with PBS or HB-alum and immunized with OVA-MPLA 4 weeks later as shown in Fig. 3, and lymphocytes were harvested from iLNs 24 h following immunization with OVA-MPLA. While the overall frequency of DCs and expression of the activation marker CD86 were not changed in mice immunized at DOL7 with HB-alum, these mice had an increased frequency of type 2 conventional DCs (cDC2s) compared with mice immunized at DOL7 with PBS (Fig. 6A–D). Additionally, MHCII expression was increased on DCs following OVA-MPLA immunization in mice previously immunized with PBS, however this increase in MHCII was not observed in mice immunized with HB-alum at DOL7 (Fig. 6E–G). These results confirm the presence of alterations of DCs in vivo, demonstrating changes in DCs in the draining lymph node as well as in DC precursors in the bone marrow.

Fig. 5: Dendritic cells from mice immunized in early-life display differential activation with TLR ligands.
figure 5

BMDCs (n = 4–8) generated from mice 4 weeks after immunization with PBS at DOL7 or (A–D) HB-alum at DOL7 or 56 or (E–H) HB-CpG at DOL7 or 56 were stimulated with Pam3CSK4, LPS, or CpG for 3 days. Secreted cytokines were measured in cell culture supernatants. Data are displayed as mean ± SEM. Statistical differences were determined using the Mann–Whitney test; *p < 0.05.

Full size image
Fig. 6: Increased MHCII expression on DCs from draining lymph nodes of mice immunized as neonates with HB-alum.
figure 6

Mice were immunized at DOL7 with PBS or HB-alum. Mice were immunized 4 weeks later with OVA-MPLA as shown in Fig. 3. Lymphocytes were isolated from draining iLNs 24 h after OVA-MPLA immunization and DCs were phenotypically characterized by flow cytometry. A total DC (B) cDC1 and (C) cDC2 frequencies were determined as well as (D) the mean fluorescence intensity (MFI) of CD86 on DCs. E–G MFI of MHCII on DCs, cDC1, and cDC2 populations. Data are displayed as mean ± SEM. Statistical differences were determined using the Mann–Whitney test; *p < 0.05,**p < 0.01, ns (not significant).

Full size image

Early-life immunization with HB-CpG also altered responsiveness of BMDCs to TLR stimulation. There were no significant differences in IL-6 production (Fig. 5E), demonstrating the ability of BMDCs to respond to stimulation. However, BMDCs generated from mice immunized at DOL7 with HB-CpG produced significantly more TNFα and IL-12, compared with BMDCs from mice immunized with PBS or with HB-CpG at DOL65 (Fig. 5F, G). There were no significant differences in production of IL-1β following TLR stimulation of BMDCs from PBS-immunized mice and CpG compared with DOL56 CpG (Fig. 5H). Moreover, there were no differences in cytokine production in BMDCs harvested from mice treated with PBS at DOL7 or DOL56 (data not shown). Finally, there were also no differences in cytokine production in BMDCs harvested from mice immunized with HB-alum or HB-CpG at DOL56 compared with PBS treated mice, demonstrating that both the HB-alum and HB-CpG vaccines did not alter bone marrow-resident cells when administered to adult mice.

BMDCs from mice immunized as neonates promote altered polarization of CD4+T cells

The strong reduction of IL-12 production by DCs generated from mice immunized with HB-alum as neonates and increase in IL-12 following early-life immunization with HB-CpG suggested an altered capacity of these cells to induce differentiation of Th1 polarized CD4+ T cells. To investigate this, we established an in vitro co-culture model in which BMDCs differentiated from mice 4 weeks following HB immunization were co-cultured with OVA-specific CD4+T cells from 6 to 8 week old DO11.10 transgenic mice (Fig. 7A). BMDCs were harvested 4 weeks after immunization to allow for sufficient time for cell tracking and changes back to the bone marrow. Cells were stimulated with OVA, and cytokine secretion was evaluated. T cells co-cultured with BMDCs from mice immunized with HB-alum at DOL7 produced significantly lower levels of IFN-γ (p = 0.0022) and significantly higher levels of IL-4 (p = 0.0022), IL-5 (p = 0.0.0022) and IL-13 (p = 0.0152) compared with co-cultures with BMDCs from PBS-immunized mice (Fig. 7B). Conversely, T cells co-cultured with BMDCs from mice immunized with HB-CpG at DOL7 produced significantly higher levels of IFN-γ (p = 0.0087) compared with co-cultures with BMDCs from PBS-immunized mice (Fig. 7C). HB-CpG immunization did not significantly alter the production of Th2 cytokines. Consistent with previous data, BMDCs from mice immunized with HB-alum or HB-CpG at DOL56 displayed similar responses as control mice immunized with PBS. Taken together, these data demonstrate that the effects of early-life immunization are imprinted on bone-marrow progenitors and that these effects can be transmitted through DCs to preferentially prime CD4+T cells for Th1 or Th2 responses.

Fig. 7: Co-culture of dendritic cells from mice immunized in early-life with HB-alum prime more Th2 biased T cell responses.
figure 7

A Bone marrow was harvested (n = 6) from mice immunized with (B) HB-alum or (C) HB-CpG at DOL7 or DOL56. BMDCs were co-cultured with OVA-specific CD4+ T cells from DO11.10 mice. Cells were stimulated ex vivo with OVA, and secreted cytokines were measured. Data are displayed as mean ± SEM. Statistical differences were determined using the Mann–Whitney test; *p < 0.05.

Full size image

Discussion

Understanding differential immune effects of adjuvants in early life compared with adulthood is of utmost importance. Currently, there is a significant focus on developing new vaccine adjuvants to improve the effectiveness of both existing vaccines and those in development, so the potential exists to tailor vaccines for use in early life to promote both optimal immune responses and influence immune development36. Here we demonstrate that while HB-alum induces comparable humoral immunity when administered during early life versus adulthood, the polarization of the induced cellular immune response is markedly distinct at these different life stages of development. Additionally, we demonstrate that early-life immunization with an alum or CpG-adjuvanted HB vaccine alters subsequent responses to heterologous antigens. While there have been numerous studies investigating efficacy and safety of vaccines administered to neonates and infants, the broader and more long-term immune effects of early delivery of these vaccines and their influence on the developing immune system remains to be fully explored37.

Herein, we developed a mouse model of early-life immunization to study differences in immune effects of vaccination over the course of development at day of life (DOL) 7 (neonates), 21 (weaning), and 56 (adults). Murine and human immune maturation as well as responses to a variety of vaccines administered to infants are remarkably conserved across species, including both antibody and T cell responses29. Because mice are less developed at birth than humans, the immune system of DOL7 mice corresponds to those of human neonates, while DOL21 mice are similar to human children13,28,29. The cellular immune response was more strongly Th2 polarized, with a dramatic reduction in IFN-γ production, in mice immunized at early-life timepoints compared with adults38. The differences in immune responsiveness were more pronounced following immunization at DOL7 compared to at DOL21. DOL7 was chosen for subsequent experiments because it more closely aligns with the immunological environment of a human newborn. Although DOL21 is still considered young in murine models, this age begins to approach the immune characteristics of adulthood. Further investigation is warranted to pinpoint the developmental transition from DOL21 to DOL56, when adult immune responses are established. The work presented here was performed in BALB/c mice, which have been reported to be more Th2 skewed than other mouse strains39,40. While BALB/c mice may be inherently predisposed to generate more strongly Th2 polarized immunity, we also observed that early-life immunization of these mice with HB-CpG resulted in a stronger induction of Th1-polarized immune responses, suggesting that even in a more Th2-skewed environment, early-life immunization may be able to alter immune polarization. Future studies in other mouse strains are warranted to confirm these findings.

The lack of IFN-γ and pronounced Th2 biased cellular responses to the HB-alum vaccine in neonates may be attributed to the overall polarization and immunomodulation of the neonatal cellular immune response. In utero, the immune system is skewed away from Th1 immunity to avoid harmful fetal-maternal interactions41. Recent work has also demonstrated that neonatal T cells are uniquely regulated by inhibitory receptors and pro-tolerogenic cells which may augment their responsiveness as well as suppress the ability to induce IFN- γ14,15,16. This results in a strongly Th2-biased neonatal immune system that matures following birth in response to environmental and microbial stimuli to induce Th1-type immune responses, resulting in a balanced Th1/Th2 profile9,10. For this reason, vaccines administered in early life may generate more Th2-skewed immune responses than the same vaccines given later in life, consistent with the results obtained herein in our mouse model of HB-alum vaccination42,43,44. This has mostly been attributed to hypermethylation of the IFN-γ locus in neonates, leading to decreases in gene expression; however mechanistic studies are largely lacking11. This Th2 bias observed in our studies of early-life HB-alum immunization in mice as well as the relative lack of IFN-γ correlates with what has been previously reported for the HB-alum vaccine in human infants43. However, it is important to note that these neonatal mice are not entirely incapable of generating IFN-γ. In our model, introduction of the Th1-polarizing adjuvant CpG effectively reinstates IFN-γ production (Fig. 4 & Supplementary Fig. 8). Additionally, other adjuvants have been shown to induce Th1 immune responses in neonatal mice and human cord blood monocytes, highlighting the potential for modifying immune responses through strategic adjuvant use45,46,47. This is of specific importance as HEPLISAV-B is an FDA approved HB vaccine adjuvated with CpG48. While it is currently only approved for use in adults, it has been shown to induce strong Th1 responses, consistent with the data presented here. The current alum-adjuvanted HB vaccine approved for use in infants and children has demonstrated remarkable efficacy and good safety and has been a critical vaccination that has reduced mortality and morbidity from HB for decades. Exploring the safety and efficacy of HEPLISAV-B in infants may be worthwhile to pursue in order to promote Th1 polarized immune responses.

Here we demonstrate that early-life immunization with HB-alum leads to the generation of Th2-polarized immune responses following exposure to a heterologous antigen (OVA) (Figs. 2, 3) while early-life immunization with HB-CpG leads to a preference for Th1 immunity to heterologous antigen exposure to immunization with OVA-alum (Fig. 4). These findings are novel, as virtually all characterization of vaccines has only addressed immune responses specific to the antigen included in the vaccine, while only few studies have interrogated bystander immune responses influencing other heterologous antigens. Previous work has demonstrated in the context of asthma models sensitized with one antigen (antigen A) and alum, co-administration of antigen A with a second antigen (antigen B) leads to Th2-biased immune responses to antigen B30,31. These studies were done with alum immunization only in adult mice and with subsequent epicutaneous or intranasal antigen co-administration. Like these previous publications, we also demonstrated that alum immunization in adults could lead to development of Th2 polarized responses to a second antigen when they were co-delivered. (Supplementary Fig. 4). However, the work presented here demonstrates that the induction of Th2 polarized responses to a second antigen can occur without requiring co-administration only when the initial immunization was performed in early life. These data clearly demonstrate that there are distinct effects of alum immunization unique to early life that can result in a prolonged predisposition for developing Th2 immune responses to new antigen exposures.

The impact of the HB-alum vaccine on immune training in humans has not been thoroughly elucidated. A clinical trial comparing HB vaccines adjuvanted with AS01 or alum demonstrated that in adults, the AS01 adjuvanted vaccine but not the alum-adjuvanted vaccine induced epigenetic changes in innate immune cells49. The absence of immune imprinting by the alum-adjuvanted HB vaccine in adults aligns with our findings that only early-life immunization led to changes in subsequent heterologous immune responses. Future work elucidating this following early-life immunization is required to validate our work in humans. One limitation of this study is that the dose of alum necessary for immunogenicity in mouse models, including this study, is higher by weight ratio compared to that used in humans. Therefore, conducting human studies will be critical to validate these findings. Such studies will not only confirm the broader relevance of our results but also advance our understanding of alum-adjuvanted vaccine mechanisms in diverse age groups.

The durability of the early-life imprinting of immune polarization is another critical aspect of our findings. Notably, the influence of early-life immunization remains robust even when OVA-MPLA or intranasal OVA were administered at DOL84, 11 weeks after the initial early-life immunization. This suggests that the immunological imprint established during early life is durable and has further reaching impact beyond the immediate immune effects. This durability may be linked to changes within the bone marrow, where DCs derived from early-life immunized mice exhibit altered behavior to promote Th2 immunity and reduce Th1 polarization. We also found that the heterologous immune effects still occurred when the HB-alum and OVA-MPLA vaccines were administered at contralateral sites, and when the HB-alum vaccine was administered parenterally followed by exposure to OVA via the intranasal route. These results further support our findings that early-life immunization results in significant crosstalk between immunological compartments as well as systemic changes that extend to the bone marrow.

We propose that the influence of early-life immunization with HB-alum on maintaining Th2 predisposition is indicative of immune training, as the BMDCs show distinct responses to stimulation with TLR ligands. While DCs from mice immunized as neonates with HB-alum respond with similar overall magnitude to ex vivo stimulation, they produce dramatically lower amounts of type 1 cytokines with increased production of type 2 cytokines (Fig. 5). Specifically, we found that BMDCs from DOL7-immunized mice produced reduced levels of IL-6 and TNF-α, both of which play critical roles in shaping T cell differentiation. IL-6 can modulate both Th1 and Th2 pathways, while TNF-α is essential for Th1 development50. Previous work has demonstrated that neonatal DCs make significantly lower amounts of IL-12, which is required for Th1 responses51,52,53. Conversely, activation of innate immunity in infants leads to increased IL-23 and IL-6 which enhance Th2 responses54. The BMDCs used in this study were isolated from mice 4 weeks after immunization, at which point the mice were no longer neonates, however the altered phenotype of these DCs suggests that early-life immunization with HB-alum trained the DCs to maintain a reduced capacity for IL-12 production similar to responses reported for neonatal DCs. This altered cytokine milieu may contribute to a reduced capacity to skew T cells towards Th1 responses, alongside an increased bias toward Th2. Additionally, analysis of DCs isolated from the LNs revealed that early-life immunization with HB-alum resulted in an increased frequency of cDC2s in the LN and also limited the upregulation of MHCII on DCs following a subsequent immunization with OVA-MPLA (Fig. 6). The increased ratio of cDC2s to cDC1s correlates with enhanced Th2 skewing in mice immunized with HB-alum in early life, and cDC2s play critical roles in T cell skewing55. MHCII is crucial for antigen presentation to CD4+ T cells, and increased MHCII expression may result in more effective antigen presentation and a strengthened MHCII-TCR signal, which could enhance T cell activation or influence T cell skewing56,57,58. Future work is required to more fully investigate the mechanistic determinants of changes to DC subsets and phenotypes. However, these results demonstrate fundamental changes to DC populations, both in bone marrow precursors and DCs isolated directly from the LN that as associated with a shift in the ability to prime Th1 and Th2 immune responses.

IL-1β production was also enhanced in BMDCs from mice immunized with HB-alum at DOL7 (Fig. 5D) but not following immunization with HB-CpG (Fig. 5H). The enhanced IL-1β production in BMDCs only from mice with enhanced Th2 predisposition may suggest a role for inflammasome activation. Others have identified a role for IL-1β and inflammasome activation in Th2 immunopathology induced by early-life infection with respiratory syncytial virus (RSV)59,60. Future studies are necessary to identify a role for inflammasome activation in this mechanism.

Our data also demonstrate that early-life immunization alters the function of BMDCs affecting their ability to skew polarization of naïve T cells (Fig. 7). DCs from early-life HB-alum immunized mice alone were sufficient to alter the polarization of co-cultured DO11.10 adult T cells to promote the production of Th2 cytokines with significantly reduced secretion of IFN-γ. This ex vivo data supports the in vivo findings that early-life immunization alters DCs to promote Th2 polarized immunity. Expanding these earlylife studies to other adjuvants will be important to harness the ability of vaccines to optimally prime the polarization of the immune system and tailor vaccines specifically to the immunologically distinct populations to which they are most often administered.

Our results align with existing literature on the implications of vaccine composition and immune polarization. For instance, differences between acellular (aP) and whole-cell (wP) pertussis vaccines have been shown to influence allergic outcomes23. Additionally, prior studies have linked RSV infection in early life to an increased Th2-biased predisposition for asthma, and recent work has suggested RSV infection may promote Th2-biased immune responses to subsequent vaccines, reinforcing the notion that early immune experiences can shape long-term health outcomes61,62,63. Conversely, there is evidence that early-life immunization with the Bacillus Calmette-Guérin (BCG) vaccine alters the developing immune system to predispose for the induction of heterologous Th1 immune responses20,21,64. Importantly, this has been associated with protection from other unrelated infections as well as sparked clinical trials to investigate an impact of these strong Th1 immune responses on immune-mediated diseases64,65,66. This highlights the potential positive impact for bystander immune effects induced by early-life immunization. Taken together, these studies suggest the potential for both Th1 and Th2 polarizing immune events in early life to influence subsequent immune polarization and responsiveness in later life with different outcomes.

Taken together, these findings suggest that like other environmental factors, vaccines can also train the immune system to influence subsequent immune responses in an antigen-agnostic manner. While further research is needed to identify links between vaccination and antigen-independent immune effects, a strong focus on early-life administration is required to fully understand how vaccine adjuvants influence immune ontogeny, polarization, and memory. Data surrounding BCG, RSV, and pertussis vaccines imply that a more Th1-polarized immune response in early life may foster better immune outcomes. While inducing Th1 immunity may present challenges, our results demonstrate that the addition of CpG can effectively prevent the subsequent Th2 bias, paving the way for enhanced and optimally primed immune development. Defining the immune imprinting effects in early-life compared to later in life is important for developing targeted age-specific adjuvants which maintain the public health benefit of vaccination while shaping broad effects on the immune system to favor optimal immune development.

Methods

Animals

All mice used in this study were bred in-house at the University of Michigan Unit for Laboratory Animal Medicine. Six to eight-week-old male and female BALB/c and DO11.10 mice were ordered from Jackson Laboratory (Bar Harbor, ME) to be used as breeders. Mice were housed in microisolation cages on corn cob bedding in ventilated racks under constant environmental conditions (68 to 72 °F [20–22 °C], 30–50% relative humidity, 12:12 h light/dark cycle). Mice were provided with rodent chow (Laboratory Rodent Diet 5001, PMI Lab Diet, St. Louis, MO) and reverse osmosis water ad libitum. All procedures were performed at an AAALAC International-accredited facility and with the approval of the University of Michigan Institutional Animal Care and Use Committee in accordance with the Guide for the Care and Use of Laboratory Animals.

Vaccine preparation and immunization

Vaccine adjuvants aluminum hydroxide gel (Alyhydrogel, alum), monophosphoryl-lipid A (MPLA) and class B CpG oligonucleotide (ODN 1826) were purchased from Invivogen (San Diego, CA). Hepatitis B surface antigen subtype adw was ordered from Biosynth (Staad, Switzerland) and endotoxin-free ovalbumin (OVA) was purchased from Lionex (Braunschweig, Germany). Vaccines administered to neonatal [day of life (DOL) 7] mice contained 4 µg of HB and/or 25 µg alum and/or 4 µg CpG in phosphate buffered saline with a total injection volume of 10 µl. Vaccines administered at DOL 28 and 56 contained 20 µg HB and/or 125 µg alum and/or 20 µg CpG in phosphate buffered saline (PBS) with a total injection volume of 50 µl. OVA-MPLA vaccines contained 20 µg OVA and 5 µg MPLA in 50 µl volume per animal. OVA-alum vaccines contained 20 µg OVA and 125 µg alum in 50 µl volume per animal67,68. All immunizations were performed under isoflurane anesthesia using an IMPAC6 vaporizer. Unless indicated otherwise, all intramuscular immunizations were administered to the quadriceps of the right hind limb. Intranasal OVA exposure: 20 µg OVA was administered to mice 12 µl/mouse (6 µl/nare) under isoflurane anesthesia once per week for 4 weeks. Methods of euthanasia: at the end of the study, mice were euthanized via isoflurane overdose in a drop jar followed by thoracotomy, exsanguination via cardiac puncture, and tissue harvest.

Measurement of antigen-specific IgG

Sera were obtained from mice at the end of the study, as specified in figure legends. Concentrations of HB-specific or OVA-specific IgG antibodies were measured by ELISA. Immunograde 96-well ELISA plates were coated with 2 µg/ml HB or 20 µg/ml OVA overnight at 4 °C and blocked with 1% non-fat dry milk in PBS for 2 h at 37 °C. Serum samples were serially diluted in PBS + 0.1% BSA and added to the plates and incubated overnight at 4 °C. IgG was probed with alkaline phosphatase-conjugated mouse IgG specific antibodies (Jackson ImmunoResearch Laboratories) for 1 h at 37 °C and developed by incubation with p-nitrophenyl phosphate substrate in diethanolamine buffer (Thermo Fisher). Absorbance was measured at 405 nm, and titers were calculated based on a cutoff defined by the sum of the average absorbance for naïve sera and two times the standard deviation.

Analysis of antigen-specific recall response

The cellular recall response to HB or OVA was evaluated in splenocytes. Spleens were mechanically disrupted, and red blood cells were lysed with ACK lysis buffer. Cells were resuspended in T-cell media (DMEM 1x, 5% fetal bovine serum, 1% penicillin/streptomycin, 1% L-glutamine, 1% sodium pyruvate, 1% MOPS and 50 M 2-mercaptoethanol) and filtered through 70 µm cell strainers to generate single-cell suspensions. Cells were plated in 96 well plates at a density of 4 × 106 cells in 200 µl/well and were cultured ex vivo ± HB (2 µg/well) or OVA (5 µg/well) for 72 h, cytokine secretion (IFNγ, IL-4, IL-5, IL-10 and IL-13) was measured in cell culture supernatants using a Luminex Multiplex detection system (Millipore).

Flow cytometric characterization of LN DCs

Draining iLNs were isolated 24 h following OVA-MPLA immunization. Lymphocytes were isolated by mechanical disruption followed by digestion with Collagenase D (Roche) to obtain single-cell suspensions69. Cells were incubated with Fc block (Biolegend) in FACS buffer (PBS, 1% FBS, 2 mM EDTA) for 30 min at 4 °C followed by staining for surface markers CD11c (clone N418), CD11b (clone M1/70), MHCII (I-A/I-E clone M5/114.15.2), XCR1 (clone ZET) and CD86 (clone GL-1) in FACS buffer for 30 min at 4 °C, followed by staining with LIVE/DEAD Fixable Aqua for 20 min at 4 °C. FACS was performed on a Novocyte3000 flow cytometer and analyzed using FlowJo software.

Generation of bone marrow derived dendritic cells (BMDCs)

Bone marrow was aspirated from the femurs and tibias of neonatal and adult BALB/c mice. Aspirated bone marrow cells were washed with PBS and filtered through a 70 µm cell strainer and resuspended in RPMI1640 supplemented with 10% HI-FBS, 1 mM sodium pyruvate, 1x non-essential amino acids, 10 mM HEPES buffer, 50 µM 2-mercaptoethanol, 100 IU penicillin, 100 µg/mL streptomycin, and 20 ng/mL GM-CSF. Bone marrow cells were seeded at 2 × 107 cells in 20 ml of complete medium in a T-150 flask. 10 ml of fresh medium was added on day 3 of culture and hemidepletion was performed on days 5 and 7. On day 8, BMDCs were analyzed for maturity by determining the cell surface expression of CD11c+CD11b+ cells to verify that >95% of cultured cells were DCs.

In vitro stimulation of BMDCs

BMDCs were plated at 5 × 105 cells per well in a 96-well plate and stimulated with 10 ng/ml Pam3CSK4 (Invivogen), 100 ng/ml LPS (Invivogen) or 4 μg/ml CpG for 3 days. Cytokine secretion was measured in cell culture supernatants using the Luminex Multiplex detection system as described above.

DC:T cell co-culture

BMDCs were harvested from neonatal and adult mice as described above. OVA-specific CD4+T cells were harvested from spleens of adult DO11.10 mice using an EasySep mouse CD4+ T cell isolation kit (Stemcell Technologies). Cells were resuspended in complete media. BMDCs were seeded at 2.5 × 104 cells per well in a 96-well plate in 100 μl of complete media and were incubated with 1 μg/ml OVA MHC II peptide (Invivogen, OVA323–339) for 4 h at 37 °C. The excess peptide was then washed off and D011.10 CD4+ T cells (2 × 105 cells per well) were added. Cells were co-cultured for 3 days, and cytokine secretion was measured in cell culture supernatants using the Luminex Multiplex detection system as described above.

Statistics

Statistical analyses were performed using GraphPad Prism (La Jolla, CA) using the Mann–Whitney nonparametric test. For all comparisons p < 0.05 was considered the threshold for significance. Data are shown as mean ± SEM.