Biotin conjugated rat anti-mouse IgE paired antibodies, IgG1 and isotype standards were purchased from BD BioSciences (San Jose, CA, USA); Albumin from chicken egg white was obtained from Sigma (St Louis, MO, USA). All other reagents were obtained from Sigma (St. Louis, MO) unless otherwise noted.

All animal protocols are in compliance with the Guide for the Care and Use of Animals and were approved by the Institutional Animal Care and Use Committee (IACUC) at Michigan State University. Female 4-week-old BALB/c mice (The Jackson Laboratory, Bar Harbor, ME) were given control diet (AIN-93G special ordered with tBHQ omitted) (Dyets, Inc., Bethlehem, PA) or diet with 0.0014% tBHQ, 0.0014% butylated hydroxytoluene (BHT) or 0.0014% 3-hyrodxytyrosol (3-HT; AIN-93G, Dyets, Inc., Bethlehem, PA; where 0.0014% is the concentration of tBHQ that is normally included in the diet to prevent rancidification) one week prior to sensitization. The mice were then maintained on the diet for the duration of the study.

Mice were sensitized to sterile saline or ovalbumin as described below. Female BALB/c 3-week-old SCID (CBySmn.CB17-Prkdc^scid/J) immunodeficient mice were purchased from Jackson (The Jackson Laboratory, Bar Harbor, ME). Nrf2-null mice on a mixed C57Bl/6 and AKR background were generated by Dr. Jefferson Chan at the University of California San Francisco (10.1073/pnas.93.24.13943). The mice were backcrossed 8 generations on the Balb/c background and were found to be >98% congenic.

Dorsal skin was exposed by shaving the backs of the mice (n = 8/group) prior to transdermal sensitization to OVA (10mg/mouse) or sterile saline once per wk for 4 wks. Blood was collected from the saphenous vein into a lithium heparin pre-coated Microvette CB 300 Capillary Blood Collection Tube (Sarstedt, Newton, NC) on wks 2 and 4 after the first sensitization and the day of oral challenge ( Figure 1A ). Plasma was prepared from whole blood by centrifugation for 10 min at 2,000 x g at 4°C (Thermo Fisher Science Inc., Waltham, MA).

OVA-specific IgE (OVA-IgE) and IgG1 (OVA-IgG1) were measured using optimized ELISA-based methods described previously (25, 26). The OVA-IgE and OVA-IgG1 levels were measured by coating ELISA plates with OVA (5 mg/mL) followed by blocking, sample addition, and detection using a biotin-labeled anti-mouse IgE or IgG1 detection antibody (BD Biosciences) as described previously (25). Total IgE was quantified by a sandwich ELISA using paired antibodies (i.e., a capture anti-mouse IgE antibody and a biotin-labeled anti-mouse IgE detection antibody) and a mouse IgE standard (BD Biosciences). OVA-IgE level in the tBHQ dose response study was performed by using Mouse OVA specific IgE ELISA kit (Biolegend, San Diego, CA).

The OVA-sensitized mice and saline-sensitized mice were challenged by oral gavage with OVA (10 mg/mouse). Rectal temperature (RT) was measured before and every 10 min up to 30 min after challenge using a temperature probe (ThermoWorks Inc., American Fork, UT). Mice were assessed by a blinded observer for signs of systemic anaphylaxis for 30 min following oral challenge. Clinical signs were evaluated according to a scoring system (27). Clinical scoring (on a scale of zero to 5) was performed according to the method described here. Scores were: 0 = no symptoms; 1 = scratching and rubbing around the nose and head; 2 = puffiness around the eyes and mouth, diarrhea, pili erecti, reduced activity and/or decreased activity with increased respiratory rate; 3 = wheezing, labored respiration, cyanosis around the mouth and the tail; 4 = no activity after prodding, or tremor and convulsions; 5 = death.

Blood samples were collected 1 h after oral challenge. Plasma levels of mouse mucosal mast cell protease-1 (mMCP-1) were measured using Mouse mMCP-1 ELISA kit (eBioscience Inc., San Diego, CA), according to the manufacturer’s instructions.

Spleen cells (2 × 10⁶ cells/ml) in DMEM (Genesee Scientific, El Cajon, CA) containing 10% fetal bovine serum (Gibco), 25 mM HEPES (Sigma), 10 mM nonessential amino acids (Sigma), 100 u/mL penicillin-streptomycin solution (Sigma), and 55µM 2-mercaptoethanol (Invitrogen) were first stimulated with anti-CD3 (1.5 µg/ml) (eBioscience Inc., San Diego, CA)/anti-CD28 (1.5 µg/ml) (Invitrogen) and crosslinker AffiniPure F(ab’)₂ Fragment Goat Anti-Syrian Hamster IgG(H+L) (1.5 µg/ml) (Jackson ImmunoResearch, West Grove, PA, USA) for 72 hours in the presence of 2 mM monensin (Biolegend, San Diego, CA), and then stained with viability dye (eBioscience Inc., San Diego, CA) and antibodies to cell surface markers, CD3-AF488 (Biolegend, San Diego, CA), CD4⁺-PE/Cy7 (eBioscience Inc., San Diego, CA) and CD8a-PerCP/Cy5.5 (Biolegend, San Diego, CA). Cells were then permeabilized with Foxp3 Staining Buffer Set (eBioscience Inc., San Diego, CA) according to the manufacturer’s instructions and stained for 30 min at room temp with anti-GATA3-APC (Biolegend, San Diego, CA) and anti-IL-4-PE (Biolegend, San Diego, CA). Isotype controls (intracellular) and single-color controls (surface) were used to set gates. Fluorescence was detected by an Attune NxT flow cytometer (Thermo Fisher Science Inc., Waltham, MA).

Single-cell splenocyte suspensions were prepared in DMEM (Genesee Scientific, El Cajon, CA) containing 10% fetal bovine serum (Gibco), 25 mM HEPES (Sigma), 10 mM nonessential amino acids (Sigma), 100 u/mL penicillin-streptomycin solution (Sigma), and 55µM 2-mercaptoethanol (Invitrogen). The cells (5×10⁶ cells/ml) were cultured in a flat-bottom 96-well plate (200 ul/well) (Greiner Bio-One, Monroe, NC) in the presence or absence of OVA (1mg/well) for 96 h (37°C, 5% CO₂). IL-4, IL-5, and IL-13 were determined by the ELISA (eBioscience Inc., San Diego, CA) according to manufacturer’s instructions.

Formalin-fixed duodenum tissues were processed by the Michigan State University Histopathology and Cytology Service Laboratory using standard methods described previously (28). Paraffin sections (Reichert Jung 2030 rotary microtome) were stained with toluidine blue. The numbers of mast cells (intact and degranulated) in 40 high-power fields were counted, and the percent degranulation was calculated as described earlier (28).

CD4⁺ T cells were magnetically isolated from 6-week-old female wild-type or Nrf2-null mice on a BALB/c background (Milentyi Biotec). Likewise, B cells were isolated from 6-week-old female wild-type BALB/c mice. The cell isolations are performed in our laboratory routinely and typically result in >90% purity. 3-week-old female BALB/c SCID (CBySmn.CB17-Prkdc^scid/J) immunodeficient mice (The Jackson Laboratory, Bar Harbor, ME) were recipients. 0.2 ml (5×10⁶ cells/ml) CD4⁺ T cells from the donor mice (wild-type or Nrf2^-/-) were injected i.p. into the recipients. On the next day, 0.2 ml (5×10⁶ cells/ml) wild-type B cells were injected i.p. into the recipient mice. 8 wk after engraftment, mice were then sensitized for 6 wk and then challenged (29).

3-week-old female BALB/c SCID mice received B cells from wild-type BALB/c mice and CD4⁺ T cells from either wild-type (n=3) or Nrf2-null mice (n=3) on a BALB/c background following the protocol described in the adoptive transfer experiments and recipients were given on the control diet. 8 wk after engraftment, mice were sensitized for 6 wk. Mice were then sacrificed and total RNA was isolated from 4×10⁶ splenocytes using TRIzol Reagent per the manufacturer’s protocol (Thermo Fisher Science Inc., Waltham, MA). After isolation, reverse transcription was performed by using RT² First Strand kit (QIAGEN, Hilden, Germany). According to the manufacturer’s protocol, the cDNA was used on the real-time RT² Profile PCR Array Mouse Allergy & Asthma (QIAGEN, Hilden, Germany) in combination with RT² SYBR Green ROX qPCR Mastermix (QIAGEN, Hilden, Germany). Relative mRNA expression was calculated by ΔΔCt, normalized to housekeeping genes (Actb, B2m, Gadph, Gusb, and Hsp90ab1) using Ct values quantified by ABI Quant Studio 6 Flex (Thermo Fisher Science Inc., Waltham, MA).

Differentially expressed genes (DEGs) between wild-type vs Nrf2-null group were identified as fold change >2 and P-values<0.05. Kyoto Encyclopedia of Gene and Genome (KEGG) analyses were performed based on the ClueGO, Cytoscape software. Pathway groups were generated based on Kappa Score = 0.5 and pathways with pV ≤ 0.05 were shown.

For comparison of two groups (control diet and tBHQ diet), Student’s t-test was used (SigmaPlot 14 software). For studies with two factors (wild-type vs. Nrf2-null; control vs tBHQ diet), two-way ANOVA followed by Bonferroni’s or Holm-Sidak post hoc analysis was used (SigmaPlot 14 software). For comparison of multiple groups (tBHQ dose response), one-way ANOVA followed by Holm-Sidak post hoc analysis was used (SigmaPlot 14 software). Pearson correlation analysis was used to test the relationship between mMCP-1 levels and the drop in body temperature. An online software service was used in Pearson correlation analysis (http://www.socscistatistics.com/tests/). For comparison of clinical reactivity, statistical significance between two groups was analyzed by using the Wilcoxon rank-sum test and statistical analysis between multiple groups was performed using a Kruskal-Wallis test with Dunn’s post hoc test (SigmaPlot 14 software). Data are shown as mean ± SEM. For all analyses, statistical significance level was set at p < 0.05. DEGs and KEGG analysis is described separately (above).

To determine the effect of the food additive tBHQ, in the present studies, mice were transdermally exposed to OVA (10 mg/mouse) weekly for a total of 4 weeks ( Figure 1A ). Transdermal OVA sensitization induced a robust OVA-IgE antibody response by week 4 which was greater in the mice on the tBHQ diet ( Figures 1B–D ) ( Table 1 ). Consistent with its effects on IgE, tBHQ also caused an increase in plasma OVA-IgG1 levels ( Figures 1E, F ) ( Table 1 ).

One week after the last sensitization, the mice were challenged orally to OVA (10 mg/mouse) and observed for clinical symptoms of anaphylaxis. Oral challenge of OVA-sensitized mice resulted in clinical signs of immediate hypersensitivity reaction, including inactivity, isolation, scratching/rubbing and increased respiratory rate with onset occurring almost immediately following challenge. Notably, mice on the tBHQ diet demonstrated more severe symptoms with an average clinical score around 2.5 ( Figure 1G ).

We also evaluated the hypothermia shock response (HSR) by measuring the drop in rectal temperature after oral allergen challenge. OVA-sensitized mice on both the control diet and the tBHQ diet showed a significant drop in rectal temperature with maximal drop at 30 min post oral challenge. The decrease in temperature was significantly more pronounced in mice on the tBHQ diet (2.3°C drop in tBHQ diet group vs. 1.5°C drop in control diet group at 10 min) ( Figure 1H ). There was no difference in baseline body temperatures between the groups ( Supplementary Figure 1 ). Overall, these data indicate that the mice on a tBHQ diet show exacerbated clinical signs of food allergen-induced anaphylaxis as well as HSR upon oral allergen challenge.

Because splenic immune responses play a role in the development of allergic diseases and spleen is a site where mast cells are induced in the development of food allergy, we next investigated mast cells in the spleen of allergic mice after oral challenge (30). The percentage of FcϵRI⁺/c-kit⁺ cells (mast cell population) was significantly higher in mice on the tBHQ diet compared to those on control diet, consistent with an increase in the number of mast cells ( Supplementary Figure 2 ).

IgE-mediated mast cell degranulation is a critical event to initiate anaphylaxis. To quantify the extent of IgE-mediated mast cell degranulation, we measured plasma levels of mucosal mast cell protease-1 (mMCP-1). Oral challenge with OVA elicited a robust elevation of mMCP-1. Notably, mMCP-1 response to OVA challenge was 3-fold greater in mice on the tBHQ diet ( Figure 2A ). A Pearson analysis 0.973, p<0.05) suggests a correlation between mMCP-1 plasma levels and a drop in body temperature at 20 min post challenge ( Figure 2B ), indicating that mast cell degranulation is likely correlated to HSR. In addition, duodenum tissue sections were collected and stained with toluidine blue for detection of mast cells. Histopathology studies demonstrated increased numbers of degranulated mast cells in the duodenums of mice in the tBHQ group ( Figure 2C , Supplementary Figure 1 ). Taken together, these data show that increased mast cell degranulation and elevated mMCP-1 levels in mice in the tBHQ group strongly correlates with significantly higher HSR and clinical symptom scores.

Skewed Th2 polarization can promote IgE production and is associated with the development of food allergy. Moreover, our previously published studies demonstrate that tBHQ promotes Th2 polarization in vitro (23). Thus, we quantified the percentage of splenic Th2 cells in this study. Notably, following challenge, the percentage of splenic Th2 cells (CD4⁺ GATA-3⁺IL-4⁺ cells) was significantly increased in the tBHQ group ( Figures 2D–F ). Taken together, our previous and present studies combined suggest that tBHQ promotes Th2 polarization.

Th2 cytokines, such as IL-4, IL-5 and IL-13, have a number of effects associated with allergy, including IgE class-switching by B cells, eosinophil activation and recruitment, and mucus production (31). Thus, we quantified the ex vivo recall response to OVA in splenocytes derived from OVA-sensitized and challenged mice. We found that splenocytes from mice on the tBHQ diet produced 3-4 fold more IL-4 and IL-13 than splenocytes from mice on control diet ( Figures 3A, C ). Though IL-5 from OVA-restimulated splenocytes was not different between groups, there was a trend toward an increase in basal levels in the tBHQ group ( Figure 3B ). Likewise, we found an increased percentage of Th2 cells in the tBHQ group (7%) as compared to those from the control group (4.5%) ( Figure 3D ) but not in the saline control group ( Supplementary Figure 1 ). These data further support the hypothesis that tBHQ promotes type 2 immune responses.

To determine the dose response of tBHQ on allergic sensitization and anaphylaxis, mice were fed AIN-93G with 0.0014% tBHQ (standard), 0.0007% tBHQ (50% of the standard), and 0.00028% tBHQ (20% of the standard) or control diet and exposed to OVA (10 mg/mouse) weekly for a total of 4 weeks. Sensitization with OVA elicited a more robust OVA-specific IgE antibody response in mice on the tBHQ diet at the highest concentration (0.0014%, the concentration found in standard chow, Figure 4A ). One week after the last sensitization, the mice were challenged orally to OVA (10 mg/mouse) and body temperature and plasma levels of mMCP-1 were measured. Compared to the mice on the control diet, mice on the 0.0014% tBHQ diet exhibited a greater drop in body temperature and significantly elevated levels of mMCP-1, while 0.0007% tBHQ and 0.00028% tBHQ groups showed a non-significant trend toward an increase in mMCP-1 ( Figures 4B, C ). Overall, the data point to a dose-dependent effect of tBHQ on allergic sensitization and anaphylaxis.

It is established that oxidative burden impacts immune cell polarization and many different disease states. To determine whether other antioxidants would have a similar effect to that of tBHQ, mice were exposed to additional antioxidants through diet. Butylated hydroxytoluene (BHT) is a synthetic antioxidant that is widely used as a food preservative. 3-hydroxytoluene (3-HT) is a naturally-occurring antioxidant found in olive oil. The effects of BHT and 3-HT were not similar to those of tBHQ. Whereas tBHQ enhanced IgE sensitization and exacerbated anaphylaxis, BHT and 3-HT did not ( Figure 5 ). Indeed, 3-HT and BHT caused a significant decrease in IgE sensitization, which correlated with a trend toward diminished change in body temperature and mMCP-1 levels. Taken together, the results indicate that the effect of tBHQ on allergic sensitization and anaphylaxis is selective and not observed with other similar antioxidants.

Our previously published studies suggest that tBHQ promotes Th2 polarization through activation of the stress-activated transcription factor, Nrf2 (23). To investigate the role of Nrf2 in the effect of tBHQ on OVA-induced anaphylaxis, we implemented adoptive transfers of CD4⁺ T cells from wild-type or Nrf2-null mice and B cells from wild-type mice into SCID mice. The expansion of T cells in SCID recipients is no different between WT and Nrf2 KO ( Supplementary Figure 1 ). The study revealed marked genotype differences. Of importance, recipients of Nrf2-deficient CD4⁺ T cells had markedly decreased plasma levels of OVA-IgE and OVA-IgG1 ( Figures 6A, B ). Strikingly, mice receiving Nrf2-deficient T cells did not become anaphylactic following OVA challenge (i.e., there was no drop in body temperature and no increase in clinical scores) ( Figures 6C–E ). Likewise, tBHQ caused an increase in OVA-IgE plasma levels and a greater drop in body temperature in mice receiving wild-type, but not Nrf2-deficient, CD4⁺ T cells, suggesting these effects are dependent on Nrf2. There were also notable genotype differences in Th2 cells and mast cells in which mice adoptively transferred with CD4⁺ T cells from Nrf2-null mice showed greatly attenuated IL-4 and GATA3 expression and decreased mMCP-1 release following OVA challenge ( Figures 6F–H ). Overall, these data suggest a critical role for Nrf2 during food allergen-induced sensitization and anaphylaxis.

Wild-type and Nrf2-null splenocytes were analyzed to identify DEGs by using RT² Profiler PCR Arrays Test. Compared with Nrf2-null group, DEGs were identified by setting the threshold for fold change > 2 and P-values < 0.05. Among the DEGs, expression of genes associated with Th2, mast cell, IgE receptor, and eosinophils were upregulated in the WT group, while the Th1-associated gene IFNγ and the Treg-associated genes Foxp3 was downregulated ( Figure 7A , Table 2 ). To estimate the functions of identified DEGs, Cytoscape pathway and KEGG enrichment analysis was conducted, which revealed enrichment in pathways associated with asthma, inflammatory bowel disease, Th1 and Th2 differentiation, and Fc epsilon RI-mediated signaling ( Figures 7B, C ).