The 6- to 12-week-old female C57BL/6 or 129 Sv mice were used. C57BL/6 WT, MyD88-KO, IL-10-KO, IL-12p40-KO, IFNγ-KO, and RAG-KO mice on C57BL/6 background were originally purchased from Jackson Laboratories (Bar Harbor, ME, USA). Double KOs IL-10/IL-12p40-KO or IL-12p40/IFNγ-KO mice were generated in our breeding unit. The 129 Sv WT, IFNα/β receptor (IFNα/βR)-KO, and IFNγ receptor (IFNγR)-KO mice on 129 Sv background (9) were provided by Dr. Luiz Fernando Reis (Ludwig Institute for Cancer Research, São Paulo, Brazil).

Mice were kept at specific pathogen-free breeding unit, Institute of Biomedical Sciences (ICB IV-USP), in five animals per cage in a ventilated caging system with filter tops under a laminar flow hood. Twelve-hour light/dark cycle, temperature-controlled rooms, food and water ad libitum, cardboard tubes, and towel paper were used for environment enrichment. Mice were treated according to animal welfare guidelines of ICB (Ethic Protocol 081/09) under National Legislation-11.794 Law.

RAG-KO mice received 20 × 10⁶ spleen cells diluted in sterile phosphate buffer saline (PBS) via intraperitoneal (i.p.) route from spleen macerates of WT or MyD88-KO mice. Reconstituted RAG-KO mice were used 14 days later.

Mice were subcutaneously (s.c.) sensitized on days 0 and 7 with 4 µg of OVA adsorbed to 1.6 mg of Al(OH)₃ (Alum) gel in 0.2 mL of PBS, containing or not TLR agonists. TLR4 agonist (LPS from Escherichia coli 055:B5) was purchased from Sigma-Aldrich (St. Louis, MO, USA) and TLR9 agonist (CpG-ODN 2395 Class C) from Invivogen (San Diego, CA, USA). The standard dose of TLR ligands was 10 µg. On days 14 and 21, mice were intra-nasally challenged with OVA (10 µg) in 40 µL of PBS. Control mice consisted of naïve non-manipulated animals. Sensitization and challenge were done under anesthesia with ketamine (50 mg/kg) and xylazine (20 mg/kg). Animals were euthanized by inhaled halothane 24 h after the last challenge; samples were collected, unaware numbered, and decoded after analyses.

Endotoxin (LPS) removal from OVA and Alum preparation was performed as previously described (7). Briefly, chicken OVA (Sigma-Aldrich, St. Louis, MO, USA) was diluted in PBS (2 mg/mL) and depleted of the endotoxin activity (measured by Limulus amoebocyte lysate QCL-1000 kit from BioWhittaker, Walkersville, MD, USA), using two to four cycles of Triton X-114 extractions, as described by Aida and Pabst (10). The endotoxin level of purified OVA (2 mg/mL) was below the limit of detection (less than 0.1 EU).

Blood samples were collected by cardiac puncture, centrifuged, and serum stored was at −20°C. BAL was acquired after lung washing with 1 mL of cold PBS via the trachea. Total and differential cell counts of BAL fluids were determined by haemocytometer and cytospin preparation stained with Instant-Prov (Newprov, Brazil) or with Panótico-Rápido (Laborclin, Brazil), both stains based on Romanowsky formulation.

Total mouse IgE was determined by sandwich-ELISA using kit BD OptEIA ELISA Set (BD, San Diego, CA, USA). OVA-specific IgE was determined by adding serum at multiple dilutions to plates with anti-IgE (SouthernBiotech, Birmingham AL, USA). After washing, biotin-labeled OVA was added and revealed with avidin-HRP plus substrate. Internal sample arbitrarily assigned as 1,000 U was used as standard. OVA-specific IgG2a/c were measured by coating the plates with 20 µg/mL of OVA. Serum samples were added at multiple dilutions and revealed with goat anti-mouse IgG2a conjugated to HRP (Invitrogen, San Diego, CA, USA), which also reacts with IgG2c isotype. Purified mouse IgG2a (Southern Biotech) was used as standard. All ELISA were performed in 96-well maxisorp plates (Nunc, NY, USA). Cytokines levels were assayed by sandwich kit ELISA according to the manufacturer’s recommendation (BD Biosciences, PharMingen, San Diego, CA, USA) as previously described (5). Values were expressed as picograms per milliliter deduced from a standards curve of recombinant cytokines ran in parallel. The limits of detection ranged from 5 to 30 pg/mL.

Three lobules (superior, middle, and post caval) of right lungs were digested with collagenase IV (2 mg/mL) and DNAse I (1 mg/mL) (Sigma-Aldrich), at 37°C for 30 min. Cell suspension was obtained after erythrocyte lysis. Extra-cellular staining to identify neutrophils and eosinophils was done using fluorochrome-conjugated antibodies (BD, San Diego CA, USA): anti-CD45 (C363.16A), -Gr-1 (RB6-8C5), -siglecF (E50-2440), and anti-major histocompatibility complex class II (MHC-II) (107630) (Biolegend, San Diego, CA, USA). For each sample, at least 1,000,000 events were collected. Cytokine production by lung cells was determined after stimulation of cells in vitro with 10 ng/mL PMA (phorbol 12-myristate 13-acetate) and 750 ng/mL ionomicin (Sigma-Aldrich, St. Louis, MO, USA) incubated with GolgiStop (BD-Biosciences, San Diego, CA, USA) for 12 h at 37°C in RPMI-1640 medium. Cell suspensions were stained with fluorochrome-conjugated antibodies against surface markers CD45 (C363.16A), CD3 (145-2C11), and CD4 (RM4-5). Cells were permeabilized and subsequently fixed with Cytofix/Cytoperm kit with GolgiPlug (BD-Biosciences, San Diego, CA, USA). For intracellular cytokine staining, we used antibodies against IL-4, IFNγ, and IL-10 (BD, San Diego, CA, USA). Only viable and non-doublet cells were considered. Cell acquisition was performed on a FACs CANTO II instrument using FACSDiva software (BD, San Diego, CA, USA). For each sample, at least 300,000 events were collected. Pools of three to five samples from the same group were used. Data were analyzed using FlowJo software (Tree Star). The number of cytokine-producing CD4⁺ T cells was calculated based on the total number of viable cells (trypan blue exclusion test) recovered from digested lung lobules. FACS analysis was performed with pooled samples from three to five mice.

Lungs were perfused with 10 mL of cold PBS through the right ventricle, fixed in 10% PBS-formalin for 24 h and then in 70% ethanol until embedding in paraffin. Five-micrometer sections were stained with hematoxylin/eosin or periodic acid-Schiff for analysis of cell infiltrates and mucus production, respectively. Lung inflammation score was performed in coded samples by two researchers scoring peribronchial and perivascular cellular infiltrates as: 0 (not present); 1 (<20% of the airways affected); 2 (20–40%), 3 (40–60%), 4 (60–80%), and 5 (>80% of the airways affected) arbitrary units (AU). Mucus score was calculated as the percentage (%) of mucus positive bronchi throughout the lung sections. In experiments shown in Figure 7, lung inflammation was scored by measuring the area of peribronchial infiltrate per length of bronchial basal membrane.

1-Methyltryptophan was used to inhibit IDO activity. Three different treatments were performed on OVA/Alum/CpG sensitized animals: (a) 30-day slow release 1-MT polymer pellet (150 mg/pellet) (Innovative Research of America) was inserted under the dorsal skin of mice on day 0, as previously described; (b) 1-MT (10 mg) (Sigma-Aldrich) diluted in PBS (500 mL) was daily injected by i.p. route from day 0 to day 14; or (c) 1-MT (35 mg) was absorbed to OVA/Alum/CpG formulation. To estimate IDO enzymatic activity in lung, the concentration of kynurenine, an IDO metabolite, was measured using a modified spectrophotometric assay (11).

Statistical analyses were performed using GraphpadPrism (V.5; GraphPad Software, USA). One-way ANOVA followed by Tukey post-test was performed, as appropriate. Differences were considered statistically significant when p value ≤ 0.05. Data are presented as mean ± SE.

We have previously shown that sensitization to OVA performed with TLR4 agonist adsorbed to Alum prevented the development of asthma-like responses without shifting the lung inflammation toward a Th1 pattern (7). Here, we extended our study and compared the effect of sensitizations with agonists of TLR4 (LPS) or TLR9 (CpG) adsorbed to Alum on the development of OVA-induced airway allergic disease in WT C57BL/6 mice. We found that absorption of LPS or CpG onto OVA/Alum prevented airway allergic inflammation. Decreased number of total cells and eosinophils were observed in BAL fluid (Figures 1A,B) when comparing CpG to allergic group (OVA/Alum). A slight, but significant increase in the number of lymphocytes or macrophages was observed, respectively, in LPS and CpG groups (Figure 1B). The number of infiltrating neutrophils was low and not significantly affected by TLR agonists. The levels IL-5 were significantly decreased in LPS or CpG groups, but the levels of IFNγ production in BAL were not affected (Figures 1C,D). Particularly, significant inhibition of mucus production was only observed in CpG group when compared to allergic group (Figure 1E). Finally, LPS or CpG suppressed total and OVA-specific IgE production when compared to allergic group, being CpG superior to LPS in suppressing IgE production (Figures 1F,G). We conclude that CpG is more effective than LPS in dampening OVA-induced Th2-mediated allergic responses.

We next evaluated the effect of different doses of CpG on the development of OVA-induced airway allergic disease. WT C57BL/6 mice were sensitized to OVA adsorbed to Alum with or without CpG, at doses of 0.1, 1, or 10 µg. Figure 2 shows that the inhibitory effect of CpG on Th2 responses was dose-dependent and even at low CpG dose (0.1 µg), the number of total cells and eosinophils in the BAL were reduced (Figures 2A,B) as well as lung inflammation and mucus production scores (Figures 2C,D). Regarding IgE antibody production, we found that the higher the dose of CpG the better the inhibition of total and OVA-specific IgE production (Figures 2E,F). OVA-specific IgG2c, a Th1-associated isotype, was significantly increased at high dose of CpG (Figure 2G). Based on the number of total and differential cells (Figures 2A,B) and lung inflammatory score (Figure 2C), we could not detect any deviation toward Th1-type lung inflammation. Overall, adsorption of CpG to Alum was effective in dampening the development of Th2-mediated responses and modifying qualitatively humoral immunity without inducing a noticeable lung inflammation.

To evaluate whether the CpG effect on allergic sensitization depends on adsorption to Alum, we compared the effect of CpG in animals sensitized with CpG adsorbed to OVA/Alum to animals that received separately CpG by subcutaneous (on the opposite site of OVA/Alum injection) or i.p. routes. PBS (allergic) group was used as a positive control. We found that optimal inhibition of allergic sensitization by CpG when compared to PBS group requires its absorption to Alum (Figure 3). Specifically, the number of total cells and eosinophils in BAL of mice sensitized with CpG adsorbed to Alum (CpG group) was more reduced than in animals that received CpG separately by s.c. or i.p. routes (CpG s.c. and CpG i.p. groups) when compared to PBS group (Figures 3A,B). Notably, IgE production and OVA-specific IgE antibodies were only significantly reduced in animals sensitized with CpG absorbed to OVA/Alum (Figures 3C,D). FACS analysis of infiltrating lung CD4⁺ T cells producing-cytokines showed that CpG administration dampened Th2 responses in all CpG groups, as revealed by the decreased percentage of IL-4-producing T cells when compared to PBS group (Figure 3E) and confirmed the superior efficacy of CpG absorbed to Alum in inhibiting Th2 sensitization since the CpG group showed the lowest value IL-4-producing T cells when compared to other groups (Figure 3E). Importantly and in line with previous results, the percentage of IFNγ-producing T cells in all CpG groups did not increase when compared to PBS group (Figure 3E) while the percentage of IL-10-producing T cells was lower in all CpG groups when compared to allergic (PBS) group (Figure 3E). Overall, these results clearly indicate that CpG adsorbed to Alum is more effective in dampening the development of allergic responses than when given separately (in a non-absorbed form) during OVA/Alum sensitization. In addition, the results obtained with lung-infiltrating CD4⁺ cytokine-producing T cells (IL-4 versus IFNγ or IL-10) reinforce the view that OVA/Alum/CpG sensitization inhibits Th2 responses without inducing immune deviation toward Th1 pattern or regulatory mechanisms.

It is well established that CpG signals through MyD88 pathway (12), although it was recently shown that TRIF pathway could also be involved (13). In vivo and in vitro models have shown that CpG induce the release of type 1 cytokines, notably IL-12 and IFNγ (14), that in turn, downregulate type-2 immune responses (15, 16). To determine whether MyD88 pathway and type 1 cytokines were involved in the CpG-induced inhibition of allergic sensitization, we studied the effect of CpG in MyD88-KO or IL-12p40-KO or IL-12p40/IFNγ double-KO mice. We first found that OVA/Alum sensitization induced significant airway allergic inflammation in MyD88-KO, IL-12-KO, or IL-12/IFNγ double KOs mice, as shown by increased number of total cells, eosinophils in the BAL, lung mucus production, and total and OVA-specific IgE antibody production (Figures 4A–E). Control mice did not show these alterations (data not shown). Absorption of CpG to Alum (OVA/Alum/CpG group) inhibited the development of allergic responses in all KO mice, except MyD88-KO mice (Figures 4A–E). These findings indicate that the expression of MyD88, but not IL-12/IFNγ molecules, is essential for the inhibition of allergic sensitization by CpG. Besides IFNγ, type I interferons (IFNs) have also been proposed as mediators of the immune modulatory effects of CpG (17). Studies on murine models of allergic airway inflammation have shown that IFN-α and IFN-β downregulate eosinophil and CD4⁺ T cell recruitment into airways (18, 19). Therefore, to extend our results we also included in our study IFNα/βR-KO and IFNγR-KO mice. We found that addition of CpG to OVA/Alum reduced airway eosinophilic inflammation, IL-5, and IgE production (Figures S1A–E in Supplementary Material), indicating that IFNαβ or IFNγ are not critical in the CpG-induced attenuation of allergic responses. The induction of IDO enzyme activity is another pathway that has been proposed for the inhibition of experimental asthma by CpG (13, 20). Therefore, we treated OVA/Alum/CpG sensitized animals with 1-MT, an inhibitor of IDO activity, and also monitored IDO activity by measuring kynurenine levels in lung tissue. As shown in Figure S2 in Supplementary Material, the inhibitory effect of CpG on Th2 immunity was preserved in animals treated with 1-MT when compared to untreated animals. In addition, no augmented IDO activity was detected in the lung of OVA/Alum/CpG group (Figure S2 in Supplementary Material). Altogether, these results indicate that the role of CpG in inhibiting allergic process does not involve type I or type II IFN receptors or IDO activity.

Since MyD88 molecule was critical for the inhibitory effect of CpG, we next determined the role of MyD88 expression on myeloid or adaptive lymphoid cells. For this, we first reconstituted RAG-1-KO mice, which express MyD88 molecule on myeloid cells but lack adaptive mature B and T lymphocytes (21) with spleen cells of WT or MyD88-KO mice as depicted in Figures 5A,F. Reconstituted RAG-KO mice with WT spleen cells and sensitized with OVA/Alum developed allergic eosinophilic inflammation, showing increased total and eosinophils cell counts in BAL (Figures 5B,C) as well as increased levels of total and specific IgE, when compared to non-reconstituted (RAG^−/−) or reconstituted but non-manipulated (Control) groups (Figures 5D,E). By contrast, mice sensitized with OVA/Alum/CpG did not develop airway allergic inflammation (Figures 5B,C) and showed decreased IgE production when compared to OVA/Alum group (Figures 5D,E). As expected, non-reconstituted (RAG^−/−) mice have no serum IgE (Figure 5D). RAG-KO mice reconstituted with MyD88-KO lymphoid cells and sensitized with OVA/Alum also developed allergic eosinophilic inflammation (Figures 5G,H) as well as increased levels of total and OVA-specific IgE antibodies when compared to control groups (Figures 5I,J). Importantly, OVA/Alum/CpG group did not develop airway allergic inflammation and IgE production was decreased when compared to OVA/Alum group (Figures 5G–J). Lung mucus and inflammation scores confirmed the inhibitory effect of CpG on lung allergic inflammation in reconstituted RAG-KO mice with MyD88-KO lymphoid cells (Figures 5K,L). Therefore, CpG signaling through MyD88 pathway on myeloid cells was sufficient to inhibit Alum Th2 sensitization.

Our results indicated that type 1 cytokines or respective receptors were not essential for the inhibition of allergic sensitization induced by CpG. Since CpG signaling through MyD88 also triggers the production of IL-10, a molecule that has been shown to suppress both Th1 and Th2 responses (22), we turned our focus to the role of IL-10 in our model. For this, animals were sensitized with OVA/Alum or OVA/Alum/CpG and the results obtained in IL-10 KO mice were compared with those obtained in WT mice. Both mouse strains developed allergic inflammation when sensitized with OVA/Alum as revealed by total and eosinophil cell counts in BAL (Figures 6A,B). However, an apparent difference was found between WT and IL-10-KO mice sensitized with OVA/Alum/CpG, since WT mice did not develop airway inflammation while IL-10-KO mice developed an intense recruitment of inflammatory cells in BAL that was predominantly constituted of neutrophils (Figures 6A,B). All CpG groups showed lower IL-4 production than OVA/Alum groups (Figure 6C) while the levels of IL-10 were increased in allergic (OVA/Alum) group but not in OVA/Alum/CpG groups and as expected, they were bellow detection in IL-10-KO mice (Figure 6D). Importantly, in OVA/Alum/CpG groups, high levels of IFNγ in BAL were only found of IL-10-KO but not in WT mice (Figure 6E). IL-17 levels in BAL were bellow detection in all groups (Figure 6F).

These results indicate that, in absence of IL-10, sensitization with OVA/Alum/CpG instead of inducing non-polarized immunity results in the development of polarized IFNγ-dominated lung inflammation. Because IL-12 is pivotal to induce Th1 immunity, we also determined the inflammatory phenotype of IL-10/IL-12p40 double-KO mice. Double-KO animals sensitized with OVA/Alum also developed allergic airway inflammation as revealed by total and eosinophil cell counts in BAL (Figures 6A,B). Eosinophil cell counts were decreased in CpG group when compared to OVA group (Figure 6B), but CpG group also showed increased number of neutrophils in BAL (Figure 6B). Notably IL-17, but not IFNγ production was significantly increased only in CpG group of IL-10/12-KO mice when compared to OVA/Alum groups (Figures 6E,F). These results indicate that in IL-10 KO mice IFNγ predominate while in IL-10/12-KO mice IL-17 predominate in BAL of CpG groups. As shown before, IgE production was inhibited while IgG2c production was increased in CpG groups when compared to allergic (OVA/Alum) group (Figures 6G,H) and in IL-10-KO mice, the production of OVA-specific IgG2c antibodies reached the highest levels (Figure 6I).

The above results indicate that phenotype of adaptive immunity to OVA resulting from sensitization with OVA/Alum/CpG varies depending on the presence of IL-10 and IL-12 cytokines. In WT mice, intranasal OVA challenge does not results in lung inflammation while in IL-10 KO or IL-10/IL-12-KO mice intranasal OVA challenge induces airway neutrophilic inflammation and lung inflammation associated with IFNγ or IL-17 production. In order to reveal and quantify lung inflammation, we performed histological analysis of lung sections of Control, OVA/Alum, and OVA/Alum/CpG groups in WT, IL-10-KO, and IL-10/12-KO mice. We found that all OVA/Alum sensitized strains developed peribronchial inflammatory infiltrates when compared to respective Control groups as shown by representative micrographs of lung section and lung inflammation scores (Figures 7A,B). In WT mice, lung inflammation of mice sensitized with OVA/Alum/CpG was absent and similar to Control group (Figures 7A,B) while in IL-10-KO or IL-10/12-KO mice sensitized with OVA/Alum/CpG the intensity of lung inflammation was significantly higher when compared to Control group (Figures 7A,B). Therefore, we conclude that besides MyD88, IL-10 is another molecule that plays a critical role in shaping Alum/CpG-adjuvant activity.