Oral anaphylaxis to peanut in a mouse model is associated with gut permeability but not with Tlr4 or Dock8 mutations

Abstract

Background:

The etiology of food allergy is poorly understood; mouse models are powerful systems to discover immunologic pathways driving allergic disease. C3H/HeJ mice are a widely used model for the study of peanut allergy because, unlike C57BL/6 or BALB/c mice, they are highly susceptible to oral anaphylaxis; yet the immunologic mechanism of this strain’s susceptibility is not known.

Objective:

We aimed to determine the mechanism underlying the unique susceptibility to anaphylaxis in C3H/HeJ mice. We tested the role of deleterious toll-like receptor 4 (Tlr4) or dedicator of cytokinesis 8 (Dock8) mutations in this strain as both genes have been associated with food allergy.

Methods:

We generated C3H/HeJ mice with corrected Dock8 or Tlr4 alleles and sensitized and challenged with peanut. We then characterized the antibody response to sensitization, anaphylaxis response to both oral and systemic peanut challenge, gut microbiome and biomarkers of gut permeability.

Results:

In contrast to C3H/HeJ mice, C57BL/6 mice were resistant to anaphylaxis following oral peanut challenge; however, both strains undergo anaphylaxis with intraperitoneal challenge. Restoring TLR4 or DOCK8 function in C3H/HeJ mice did not protect from anaphylaxis. Instead, we discovered enhanced gut permeability resulting in ingested allergens in the bloodstream in C3H/HeJ mice when compared with C57BL/6 mice, which correlated with increased number of goblet cells in the small intestine.

Conclusions:

Our work highlights the potential importance of gut permeability in driving anaphylaxis to ingested food allergens and that genetic loci outside of Tlr4 and Dock8 are responsible for the oral anaphylactic susceptibility of C3H/HeJ mice.

Graphical Abstract

Introduction

Food allergy and resulting anaphylaxis is a serious and potentially life-threatening condition affecting 5% of adults and 8% of children in the United States, with increasing prevalence(1). Murine models of food allergy are vital to reveal mechanisms of anaphylaxis and allergic sensitization. To model human pathophysiology, it is imperative to identify the genetic, molecular, and cellular mechanisms operational in each model system. While most mouse strains can generate IgE antibodies to food allergens in the presence of adjuvants, including the common C57BL/6 strain, they are resistant to oral anaphylaxis(2). C3H/HeJ mice have become the archetypical strain used for studies of anaphylaxis following oral challenge as it is one of the few to react systemically to oral allergen challenge(3). However, the mechanism of anaphylactic predisposition of C3H/HeJ mice remains unclear.

C3H/HeJ mice are rarely used for immunologic studies because many immunologic tools do not exist on this background. One exception is food allergy models, which have focused on this strain because they demonstrate a heightened anaphylactic propensity after oral administration of food antigens(3). A spontaneous mutation in Toll-like receptor 4 (Tlr4) in the C3H/HeJ genome inactivates TLR4 signaling(4). TLR4 is a Toll-like receptor that recognizes lipopolysaccharide (LPS) from gram-negative bacteria. The role of TLR4 in food allergy models is unresolved. One study ascribed microbiota changes to TLR4 inactivation in C3H/HeJ mice and suggested that this enhances peanut-specific IgE production and susceptibility to anaphylaxis(5). Indeed, changes in the microbiota have recently been associated with food allergy development in both mice and humans(6–10). However, a study of TLR4-deficiency in C3H and BALB/c mouse strains did not find a definitive link between TLR4 and food allergy(11). These studies compared the C3H/HeJ strain with other related mouse strains such as C3H/HeOuJ, or with genetically disparate mice on the C57BL/6 or BALB/c background. No study has directly compared the effects of TLR4 sufficiency and deficiency on the C3H/HeJ background.

In addition to harboring a Tlr4 mutation, we recently discovered a point mutation in Dock8 (dedicator of cytokinesis 8) in C3H/HeJ mice that does not affect protein production but impairs dendritic cell (DC) migration(12). DOCK8 is an intracellular protein that regulates multiple aspects of immune cell function including regulation of the actin network and lymphocyte signaling(13,14). Patients with DOCK8 deficiency suffer a primary immunodeficiency associated with food allergy, though its role in murine models of food allergy is untested(15,16). We therefore hypothesized that the deleterious mutation in Dock8 on the C3H/HeJ background might drive oral anaphylactic susceptibility in this strain.

In this study, we generated C3H/HeJ mice with wildtype Tlr4 or Dock8 sequences and studied the effects on the allergic response. We found that neither Dock8 nor Tlr4 rescue in C3H/HeJ mice alters IgE production to peanut or abrogates anaphylaxis following oral challenge. We further confirmed that gut microbial diversity and composition are comparable in Tlr4-deficient and -sufficient C3H/HeJ mice. Having discounted two strong genetic candidates for food allergy susceptibility, we compared the phenotype of C3H/HeJ mice to the most common inbred mouse strain, C57BL/6, which are resistant to anaphylaxis following oral peanut challenge, even when both strains are cohoused and IgE levels are comparable. In contrast to oral challenge, both C3H/HeJ and C57BL/6 strains have similar anaphylactic responses to systemic peanut challenge. As bypassing the gut barrier led to anaphylaxis in both strains, we tested the steady-state integrity of the gut barrier. We discovered that C3H/HeJ mice have enhanced gut permeability to both model dextran and protein antigens when compared with C57BL/6 mice. The enhanced gut permeability in C3H/HeJ mice was associated with increased goblet cell numbers in the small intestine when compared with C57BL/6 mice.

The loss of barrier integrity is thought to allow for sensitization through the damaged epithelium, promoting allergy to food(17,18) or other misdirected immune responses such as autoimmune reactions(19). Therefore, our work suggests that aberrant gut permeability may allow for more allergen to pass into the bloodstream, leading to anaphylaxis to ingested food allergens in C3H/HeJ mice but not in other strains with less permeable gut barriers. This finding likely explains why the C3H/HeJ strain is particularly suited as a model of oral anaphylaxis. It also emphasizes that the integrity of the intestinal barrier may have a central role in protection from food allergy and anaphylaxis.

Methods

Mice.

Age- and sex-matched 6-to-12 week old mice were kept on peanut- and egg- free Teklad Global 18% Protein Rodent Diet (2018S, Harlan laboratories). WT CD45.2 C57BL/6N mice were purchased from National Cancer Institute. C3H/HeJ mice were originally purchased from The Jackson Laboratory (JAX) and bred in Yale animal facilities for over 10 generations. C3H/HeJ mice from JAX were crossed with CD45.2 C57BL/6. Resulting F1 mice were then backcrossed over 10 generations with wildtype C3H/HeJ mice to generate C3H/HeJ mice with either two copies of C57BL/6-derived Dock8 or heterozygous for C3H/HeJ and C57BL/6 Tlr4. Genotypes of mice were ascertained by sanger sequencing using primers for Dock8 (F-5’ CTCTTTCCCTCCAGCTGTCA 3’) and (R-5’ CTCGCCCCTTTCCTGTAGTT 3’) and Tlr4 (F-5’ GGA CTG GGT GAG AAA TGA GC 3’) and (R-5’ CCA ACG GCT CTG AAT AAA GTG 3’). Littermates were used whenever possible. Prior to each experiment, mice were cohoused for at least 2 weeks such that at least one mouse of each genotype was present in each cage. All protocols used in this study were approved by the Institutional Animal Care and Use Committee at the Yale University School of Medicine.

Immunization and challenge.

Mice were immunized weekly for 6 weeks by oral gavage with 5mg ground blanched peanut (Western Mixers Produce & Nuts) and 10μg cholera toxin (CT) (List Biologicals) in 200μl of 0.2M sodium bicarbonate buffer per mouse. After 6 weeks of immunization, mice were rested for one week prior to challenge. Mice were fasted for four hours before being challenged orally with 800μl (25mg protein) defatted homemade crude peanut extract(20) or intraperitoneally with 500μg crude peanut extract (Greer) in 200μL PBS. Rectal temperature was measured every 15 minutes for up to 80 minutes, clinical symptoms scores were assessed at 45 minutes (Table E1), and sera were collected for analysis 1 hour post-challenge.

Serum collection.

Serum was collected prior to sensitization, 7 days after six weeks of peanut sensitization, and after oral challenge. Collected blood was incubated at RT for 1 hour and the resulting clot removed. Samples were centrifuged for 10 minutes at 1500 x g at RT and sera were collected and stored at −80 °C prior to analyses.

Passive cutaneous anaphylaxis (PCA).

15μL of sera from sensitized or naïve mice were injected intradermally into the left or right ear of a naïve mouse, respectively. 24 hours following injection, mice were challenged intravenously with 100 μg of crude peanut extract (Greer) in 1% Evans Blue solution in phosphate buffered saline (PBS). 30 minutes following challenge, mice were euthanized and ears collected for analysis. Ears were incubated at 56 °C in 500 μL formamide for 72 hours before being assayed for Evans Blue concentration at 650nm using a spectrophotometer.

Gut Permeability PCA.

PCA was performed as described earlier with modification(21). Briefly, ~20 μL of pooled serum from peanut- and CT-sensitized C57BL/6 and C3H/HeJ mice (or serum from naïve control mice) were injected into ear pinnae of naïve C57BL/6 or C3H/HeJ recipients. 16 hours later, the recipient mice were deprived of food for 4 hours. They were then injected intravenously with 100μL 2% Evans blue dye and immediately challenged with 800μL of homemade crude peanut extract (25mg protein) through oral gavage. 45 minutes later, mice were euthanized, and ear pinnae were harvested and incubated in formamide before assaying for Evans Blue concentration as above.

16S rRNA sequence data availability.

All sequence read data have been deposited with links to BioProject accession number PRJNA689615 in the NCBI BioProject database (https://www.ncbi.nlm.nih.gov/bioproject/)

Statistical analysis.

Except for 16S data, all statistical analyses were performed using Prism software (Graphpad Software). Exact tests used are denoted in figure legends. Statistical significance is defined as *P< 0.05, **P< 0.01, ***P< 0.001, ****P<0.0001. Complete methods are available in this article’s supplementary materials.

Results

C3H/HeJ but not C57BL/6 mice are susceptible to anaphylaxis following oral sensitization and challenge.

C57BL/6 mice were cohoused at weaning with C3H/HeJ mice for 3–4 weeks prior to experiments. Following a 6-week sensitization with peanut and CT via gavage, mice were challenged orally with peanut (Fig 1, A). We observed a decrease in temperature in C3H/HeJ mice following oral challenge, indicative of an anaphylactic response (Fig 1, B and E1, A)(22). In contrast, C57BL/6 mice did not have a drop in temperature, indicating a lack of systemic anaphylactic response. This was also reflected in the clinical presentation of the mice: mean anaphylactic score (Table E1) in C3H/HeJ mice was significantly higher than in C57BL/6 mice (Fig 1, C). As mucosal mast cell protease 1 (mMCP1) is a factor released by degranulating mast cells(23,24), it is a proxy for mast cell degranulation during anaphylaxis. Serum mMCP1 was detected in all C3H/HeJ mice following oral challenge, but not in C57BL/6 mice (Fig 1, D). The clinical response difference between these groups was not due to differences in sensitization as peanut (PN)-specific immunoglobulin E (IgE) present in pre-challenge serum samples was similar between C3H/HeJ and C57BL/6 mice (Fig 1, E). Total PN-specific IgG antibodies were moderately but not statistically higher in C3H/HeJ mice, indicating that possible protection afforded by PN-specific IgG in C57BL/6 mice was not a primary explanation for the observed phenotypic difference (Fig E1, B).

Figure 1. C3H/HeJ mice, but not C57BL/6 mice are susceptible to anaphylaxis following oral sensitization and challenge.

Cohoused C3H/HeJ mice and C57BL/6 mice were sensitized with peanut and CT for six weeks before being orally challenged with defatted peanut protein. (A) Sensitization model. Peanuts represent an oral immunization with peanut and CT. (B) Temperature change following oral challenge. (C) Anaphylactic scores 45 minutes following oral challenge. ELISA quantification of (D) Serum-mucosal mast cell protease 1 (mMCP1) one hour following oral challenge and (E) peanut-specific IgE in the serum following sensitization. For each experiment, each group was assessed for normality via normal quantile plots before being compared using (B) one-way ANOVA or (C,D,E) Student’s t-test. Dotted lines represent limits of detection. AU denotes arbitrary units assigned to ELISA standards. Error bars represent SEM for 4–5 mice per group. *P<0.05, ***P<0.001; ns, not significant, n.d., not detected. 1 representative experiment out of 3 (B,C,E) or 2 (D) independent experiments.

Rescue of TLR4 function does not alter the microbiome.

As defects in TLR4 signaling have previously been suggested to contribute to the susceptibility of C3H/HeJ mice to orally-induced anaphylaxis when compared with similar mouse strains(5), we sought to correct the Tlr4 gene on the C3H/HeJ background to definitively determine its contribution to the anaphylactic phenotype. We crossed C3H/HeJ mice with C57BL/6 mice, generating mice with a functional Tlr4^WT allele derived from C57BL/6. We subsequently backcrossed mice with a functional Tlr4 allele onto the C3H/HeJ background for ten generations. This resulted in C3H/HeJ mice which had one functional TLR4 gene (indicated as Tlr4^Het) as well as littermate C3H/HeJ mice that were homozygous for the original non-functional Tlr4 gene (indicated as Tlr4^C3H/HeJ) (Fig 2, A). To determine that the rescue was successful, we generated bone marrow-derived dendritic cells (BMDCs) from each of these genotypes and stimulated with LPS, which is the ligand for TLR4(4). Following LPS stimulation, Tlr4^Het but not Tlr4^C3H/HeJ BMDCs upregulated CD86, indicating activation of DCs with a restored WT copy of Tlr4 (Fig 2, B), confirming intact function of the TLR4 protein in Tlr4^Het mice.

Figure 2. Rescue of TLR4 function does not alter the microbiota.

The Tlr4 point mutation in C3H/HeJ mice was rescued by crossing to C57BL/6 mice with a WT Tlr4 gene. Mice were selectively crossed back onto the C3H/HeJ background for 10 generations, preserving the functional Tlr4 gene. (A) Chromatogram of the Tlr4 point mutation in C3H/HeJ mice (Tlr4^C3H/HeJ) and heterozygous with the WT C57BL/6 Tlr4 gene (Tlr4^Het). (B) Representative histogram and quantification of surface CD86 expression on LPS-stimulated bone marrow-derived dendritic cells. (C) Relative bacterial abundance in stool between pre-immunization (“Naïve”) and post-immunization (“Immunized”) Tlr4 sufficient (Tlr4^Het) and deficient (Tlr4^C3H/HeJ) C3H/HeJ cohoused littermates. Each bar represents pooled stool samples from a single mouse. Samples are color-coded by bacterial composition at class level (D) Principal coordinate analysis (PCoA), of fecal bacteria from Tlr4^Het and Tlr4^C3H/HeJ cohoused littermates, using the proportionally normalized Bray-Curtis dissimilarity matrix. (E) Shannon diversity indices for measurement of alpha diversity in bacterial fractions isolated from stool samples of cohoused Tlr4^Het and Tlr4^C3H/HeJ littermates pre- and post-peanut sensitization. Each circle represents pooled stool samples from a single mouse in (D) and (E). Tlr4^Het and Tlr4^C3H/HeJ mice were sensitized and challenged orally with peanut as in (Fig 1). Error bars in (B) represent SEM for 4 mice per group (***P<0.001). 1 representative experiment out of 2 (B) or pooled data from two independent experiments (C-E).

Since TLR4 is the innate sensor for lipopolysaccharide found on gram-negative bacteria and its loss-of-function could alter microbiota composition, we performed 16S sequencing of fecal bacterial DNA from littermate Tlr4^Het and Tlr4^C3H/HeJ mice pre- and post- peanut sensitization to compare their bacterial composition. Similar to a previous study(25) in TLR4-deficient and -sufficient C57BL/6 mice, microbial composition was analogous betweenTlr4^Het and Tlr4^C3H/HeJ mice as shown by the relative abundance of different bacterial classes found in the stool samples from cohoused mice (Fig 2, C and E2, A). This similarity was maintained even after peanut sensitization (Fig 2, C and E2, A). Using both principle coordinate analyses (PCoA) (Fig 2, D) and non-metric multidimensional scaling (nMDS) ordination (Fig E2, B), we further demonstrate that beta diversity of fecal bacteria was not different between Tlr4^Het and Tlr4^C3H/HeJ littermates. Statistical analysis (PERMANOVA) confirmed no significant differences in community composition by genotype. Lastly, Shannon indices revealed that alpha diversity of fecal bacteria was also similar in Tlr4^Het and Tlr4^C3H/HeJ littermates (Fig 2, E). Statistical analysis (ANOVA) confirmed that genotype did not significantly impact microbial diversity. Collectively, these data suggest that the loss-of-function mutation in TLR4 did not lead to major alterations in gut microbiota of C3H/HeJ mice at steady-state or post sensitization, allowing us to directly test the role of TLR4 on susceptibility to oral anaphylaxis without confounding effects of disparate microbiomes.

Rescue of TLR4 function does affect susceptibility to orally-induced anaphylaxis.

We asked whether rescuing TLR4 function in the C3H/HeJ strain impacted allergic antibody production. We quantified serum PN-specific IgG1 and IgE(26) in Tlr4^Het and Tlr4^C3H/HeJ mice after six weeks of oral sensitization. We found no difference in PN-specific IgG1 or IgE between the two genotypes (Fig 3, A and B). We next tested the anaphylactic capacity of the IgE generated in each group through passive cutaneous anaphylaxis (PCA) challenge(27–30). PCA challenge revealed similar IgE functionality between Tlr4^Het and Tlr4^C3H/HeJ mice (Fig 3, C). Therefore, there was no detectable effect of TLR4 signaling rescue on the oral induction of allergic antibodies in C3H/HeJ mice. We next challenged cohoused 6-week sensitized Tlr4^Het and Tlr4^C3H/HeJ mice orally with peanut. Following challenge, both sensitized groups showed significant drops in temperature when compared to challenged naïve mice, but there was no statistically significant difference between sensitized Tlr4^Het and Tlr4^C3H/HeJ mice (Fig 3, D). The two groups also demonstrated similar anaphylactic scores (Fig 3, E) and serum mMCP1 concentrations (Fig 3, F). Thus, there was no detectable effect of TLR4 signaling on anaphylactic susceptibility in C3H/HeJ mice.

Figure 3. Rescue of TLR4 function does not affect susceptibility to orally-induced anaphylaxis.

Representative ELISA quantification of peanut-specific (A) IgG1 or (B) IgE in serum one week after the final immunization with peanut and CT. Data from one representative experiment of three. Dotted lines represent limit of detection. (C) Evans blue dye extravasation after passive cutaneous anaphylaxis (PCA) assay performed by transferring day seven post-6^th immunization sera from Tlr4^Het or Tlr4^C3H/HeJ mice into naïve C3H/HeJ mice and challenging intravenously with peanut and 1% Evans blue in PBS. Data from 1 representative experiment out of 2. The dotted line represents reading from a naïve mouse after challenge. (D) Temperature change following oral peanut challenge from 1 representative experiment out of 3 independent experiments. (E) Pooled anaphylactic scores 45 minutes following oral peanut challenge from 3 independent experiments. ELISA quantification of (F) serum-mucosal mast cell protease 1 (mMCP1) one hour post-challenge pooled from 2 independent experiments. For each experiment, groups were assessed for normality via normal quantile plots before being compared with either (D) 2-way ANOVA with Tukey-corrected multiple comparisons by timepoint or (A-C,E,F) Student’s t-test. Error bars represent (D) SD or (A-C,E,F). SEM for 4–15 mice per group. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001; ns, not significant.

Rescue of DOCK8 function does not impact allergic sensitization to peanut or susceptibility to orally-induced anaphylaxis.

We previously discovered that C3H/HeJ mice harbor a point mutation in Dock8(12). Given the allergic predisposition of patients with mutations in DOCK8, we tested whether impaired DOCK8 function in C3H/HeJ mice impacted the susceptibility of this strain to food allergy. We generated C3H/HeJ mice with a functional C57BL/6 Dock8 allele (Fig E3, A). We confirmed that DOCK8 function was restored by immunizing mice homozygous for mutated Dock8 (Dock8^C3H/HeJ) or homozygous for C57BL/6-derived rescued Dock8 (Dock8^WT) intranasally with the TLR2 ligand Pam2CysK4. Consistent with our previous work(31), we found that Dock8^C3H/HeJ mice had a selective defect in CD11b⁺ but not CD103⁺ DC migration to draining lymph nodes (LNs) when compared with mice containing WT Dock8 sequence (Fig E3, B–D).

We then investigated the effect of Dock8 rescue on oral allergic sensitization and challenge. In 6-week peanut-sensitized Dock8^WT and Dock8^C3H/HeJ mice we detected similar levels of PN-specific IgG1 and IgE (Fig 4, A and B). PCA also indicated that the IgE was similarly functional (Fig 4, C). Therefore, rescue of DOCK8 function did not alter oral peanut sensitization in C3H/HeJ mice. We then challenged cohoused sensitized mice orally with peanut. Both Dock8^WT and Dock8^C3H/HeJ groups had significant temperature drops relative to naïve mice following challenge (Fig 4, D), but were not statistically different from each other. Further, neither the mean anaphylactic scores (Fig 4, E) nor serum mMCP1 (Fig 4, F) between the two groups were significantly different. Therefore, the rescue of DOCK8 function did not affect anaphylactic susceptibility in C3H/HeJ mice.

Figure 4. Rescue of DOCK8 function does not impact allergic sensitization to peanut or susceptibility to orally-induced anaphylaxis.

The Dock8 point mutation was corrected to produce C3H/HeJ mice which were homozygous for the WT C57BL/6 Dock8 allele or homozygous for the deficient C3H/HeJ Dock8 allele. ELISA quantification of peanut-specific (A) IgG1 or (B) IgE in serum one week after final immunization with peanut and CT. Dotted line represents limit of detection. (C) Evans blue dye extravasation after passive cutaneous anaphylaxis (PCA) assay. The dotted line represents reading from a naïve mouse after challenge. Mice were sensitized then challenged orally with peanut as in (Fig 1). (D) Temperature change following oral peanut challenge of Dock8^C3H/HeJ and Dock8^WT C3H/HeJ mice. (E) Anaphylactic scores from mice 45 minutes post-challenge. (F) Serum-mucosal mast cell protease 1 (mMCP1) levels one hour following challenge. For each experiment, groups were assessed for normality via normal quantile plots before being compared with either (D) 2-way ANOVA with Tukey-corrected multiple comparisons per timepoint, or (A,B,C,E,F) Student’s t-test. Error bars represent (D) SD or (A,B,C,E,F) SEM for 4–15 mice per group. *P<0.05,; ns, not significant. 1 representative experiment of 3 (A, B, D) or 2 (C) independent experiments. (E, F) Pooled data from 2 of 3 independent experiments.

Both C3H/HeJ and C57BL/6 mice are susceptible to intraperitoneal challenge.

Having demonstrated a difference in oral anaphylactic susceptibility between C3H/HeJ and C57BL/6 mouse strains (Fig 1), we asked if this was due to a gut-specific defect or if it was indicative of a systemic effect. To investigate this, we cohoused C3H/HeJ and C57BL/6 mice to equalize microbiota before sensitizing with peanut and CT. We then challenged these mice orally or interperitoneally with peanut antigen (Fig 5, A). As we had observed before, C3H/HeJ but not C57BL/6 mice demonstrated evidence of anaphylaxis with oral peanut challenge (Fig 5, B). However, when the same mice were challenged interperitoneally (bypassing the gut barrier) we observed temperature drop, positive anaphylactic scores and detectable serum mMCP1 levels in both C3H/HeJ and C57BL/6 strains (Fig 5, B–D and E4, A). Together these data indicate that both sensitized C3H/HeJ and C57BL/6 mice are capable of anaphylactic responses to peanut when antigen is introduced systemically, but only C3H/HeJ mice respond to oral peanut challenge. Therefore, the gut mucosa might block systemic exposure to peanut in C57BL/6 mice.

Figure 5. C57BL/6 mice are susceptible to intraperitoneally-induced anaphylaxis, but not orally-induced anaphylaxis.

Wildtype C3H/HeJ and C57BL/6 mice were cohoused for 2 weeks prior to experimentation. Mice were orally sensitized for six weeks with peanut and CT. (A) Sensitization and challenge model. Peanuts represent an oral immunization with peanut and CT. (B) Temperature change 45 minutes following either 500μg intraperitoneal (IP) or 25mg oral (PO) peanut challenge of sensitized C3H/HeJ or C57BL/6 mice; temperature change was taken at 15 minutes for mice that succumbed during challenge (marked with an open symbol). (C) Anaphylactic scores from mice 45 minutes post-IP challenge; open symbols are mice that died during challenge. (D) Serum mMCP1 levels one hour following IP challenge were assayed by ELISA. For each experiment, groups were assessed for normality via normal quantile plots before being compared with Student’s t-test. Dotted lines represent limits of detection. Error bars represent SEM for 3–7 mice per group. **P<0.01, ***P<0.001; ns, not significant. 1 representative experiment of 3 (B and C) or pooled data from two (D) independent experiments of mice that survived challenge.

C3H/HeJ mice demonstrate enhanced translocation of luminal antigens into the circulation compared with C57BL/6 mice.

We hypothesized that the difference between susceptibility to oral anaphylaxis in between C3H/HeJ and C57BL/6 mouse strains may be due to differences in gut barrier integrity. To explore this, we directly tested the intestinal permeability at steady-state of cohoused C3H/HeJ and C57BL/6 mice. As the permeability of the gut might increase following sensitization and oral challenge coincident with mMCP1 release(32), we utilized naïve mice. We orally administered FITC-conjugated dextran and screened for FITC fluorescence in the serum after two hours. We found that there was markedly increased FITC fluorescence in the serum of C3H/HeJ mice compared with C57BL/6 mice following challenge with 4kDa FITC-Dextran conjugate (Fig 6, A). 4kDa FITC-dextran is capable of traversing the epithelial barrier via multiple routes including paracellular transport, whereas larger antigens cross epithelia via transcellular pathways in the absence of overt epithelial damage(33). Therefore, the increased gut permeability we observed in C3H/HeJ mice could be due to differences in either transcellular or paracellular transport.

Figure 6. Increased oral anaphylaxis susceptibility of C3H/HeJ mice is associated with increased gut permeability.

Cohoused naïve C3H/HeJ and C57BL/6 mice were administered FITC-dextran orally. (A) FITC fluorescence detected in serum 2 hours following 4 kDa and (B) 40 kDa FITC-dextran administration to cohoused naïve mice. Dotted line represents the average naïve reading (autofluorescence). (C) ELISA quantification of serum OVA concentration 30 minutes following oral gavage with solubilized OVA. Dotted line represents the reading in naive mice (no OVA administration), which is at the limit of detection of the assay. (D) Quantitation of Evans blue dye extravasation in ear after passive cutaneous anaphylaxis (PCA) assay in mice orally challenge with 25mg peanut protein. The dotted line represents the post-challenge reading from mice receiving naïve serum (without antigen-specific IgE). (E) Representative image of Periodic acid-Schiff stain for goblet cells (dark purple cells) in C57BL/6 and C3H/HeJ jejunum and quantification. Scale bar represents 200μm. For each experiment, groups were assessed for normality via normal quantile plots before being compared with Student’s t-test. Error bars = SEM. *P<0.05, **P<0.01. (A-E) Pooled data from 2 independent experiments with 3–5 mice per group.

We next tested whether there was a difference in intestinal permeability to a larger 40kDa FITC-Dextran molecule that could only reach the systemic circulation via the transcellular route. Similar to the low molecular weight dextran, we found that only C3H/HeJ but not C57BL/6 mice transported high molecular weight dextran into the blood (Fig 6, B). As these experiments used a model sugar-fluorophore conjugate, we then investigated the gut barrier with the food protein antigen ovalbumin (OVA). We gavaged cohoused C3H/HeJ and C57BL/6 mice with OVA in solution before measuring OVA concentration in the serum 30 minutes later by ELISA. We saw a marked increase in mean OVA serum concentration in C3H/HeJ mice compared with C57BL/6 mice (Fig 6, C), demonstrating that a food protein also passages more efficiently through the gut barrier in C3H/HeJ mice than in C57BL/6 mice.

A positive correlation between the amount of antigen necessary to induce a PCA reaction in the ear and the quantity of antigen present has been previously observed(30). Therefore, we injected serum from immunized mice into the pinnae of both naïve C3H/HeJ and naïve C57BL/6 mice. We then injected Evans Blue dye intravenously into these mice before challenging orally with peanut. C3H/HeJ but not C57BL/6 mice demonstrated a local edematous response in the ear after oral challenge, demonstrating that sufficient antigen could pass through the gut into the serum of C3H/HeJ but not C57BL/6 mice (Fig 6, D). Together, these data indicate greater absorption of both carbohydrate and protein antigens from the gut lumen in C3H/HeJ than C57BL/6 mice. Therefore, the susceptibility of C3H/HeJ mice to anaphylaxis following oral antigen challenge is associated with a more permeable gut than in mice that are resistant to oral challenge such as C57BL/6 mice.

C3H/HeJ mice have increased jejunal goblet cell numbers compared to C57BL/6 mice.

One potential explanation for increased antigen entry into the serum from the gut is increased antigen passaging by secretory cells such as goblet cells. As goblet cells are known to form secretory antigen passages (SAPs) that traffic food antigen across the intestinal barrier(34), we performed histological analysis by periodic acid-Schiff staining to quantitate goblet cell numbers in the small intestine. We found that C3H/HeJ mice have increased number of goblet cells as compared with C57BL/6 (Fig 6, E). These surprising findings suggest a potential cellular mechanism for gut permeability to food allergens in C3H/HeJ mice, thereby predisposing this mouse stain to oral anaphylaxis.

Discussion

C3H/HeJ mice are a common model of oral food allergy and anaphylaxis(3). In this study, we have demonstrated an association between enhanced gut permeability to oral antigens and susceptibility to anaphylaxis. By challenging both anaphylaxis-prone C3H/HeJ and anaphylaxis-resistant C57BL/6 mice in parallel we demonstrated that sensitized C3H/HeJ and C57BL/6 mice are both susceptible to anaphylaxis following intraperitoneal challenge, but only C3H/HeJ mice are susceptible to anaphylaxis following oral challenge. This is associated with increased gut permeability, which may allow for more antigen to pass into the systemic circulation following challenge, thus triggering tissue mast cells, leading to an anaphylactic phenotype in C3H/HeJ mice when challenged orally. Further, we demonstrated that this phenotype may be due in part to increased goblet cell presence in the gut of C3H/HeJ mice, which warrants further study.

As the anaphylactic susceptibility of C3H/HeJ mice could have been due to a strain differences in microbiome, which have been previously observed in C3H/HeJ mice when compared with other non-cohoused strains and were associated with mutations in Tlr4(5), we characterized the gut microbial composition. The microbiome has also been implicated in other mouse and human studies of allergy(6–9,35). In mice, changes in the microbiome have been directly linked to changes in antigen passage through the epithelium during sensitization as well(36). We did not find any significant changes in microbiome composition or subsequent protection between either functional Tlr4^Het or inactive Tlr4^C3H/HeJ genotypes. Changes in gut permeability may instead be due to genes outside of the Tlr4 loci that regulate the gut barrier, including specialized epithelial cells that transport antigen across from the gut lumen(34).

DOCK8 deficiency is associated with severe immunodeficiency in humans, as well as high rates of food allergy(14). As we have previously found a deleterious mutation in the Dock8 gene in the C3H/HeJ strain(12), we sought to determine whether this was responsible for the strain’s oral anaphylactic susceptibility. After correcting the Dock8 gene on the C3H/HeJ background, we did not see demonstrable protection against anaphylaxis. The deleterious mutation in the C3H/HeJ strain is a single point mutation that hampers DOCK8 function, but does not eliminate DOCK8 protein(12), as opposed to a variety of mutations found in human cases of DOCK8 deficiency that often result in complete DOCK8 loss(37). Therefore, although the mutation in Dock8 in C3H/HeJ mice is not responsible for the strain’s susceptibility to anaphylaxis, total loss of DOCK8 function in human patients predisposes an individual to other aspects of food allergy through loss of tolerance and aberrant IgE production (14,16,21,38).

Food allergy is driven by complex interactions of multiple environmental and genetic risk factors(1). Different environments may lead to drastically different results. Indeed, C3H/HeJ mice in some facilities, but not all, demonstrate this unique susceptibility to oral anaphylaxis(5,11,39–43). Recent studies of outbred mice to identify genetic risk factors for anaphylaxis to peanut observed that C3H/HeJ mice did not mount an anaphylactic response, while an outbred strain, CC027, did(10,40). Although the genes underlying this phenotype have not yet been identified, the differences between this study and our own in the same inbred strain of C3H/HeJ mice highlight the impact of environment on susceptibility to food allergy. It also highlights that food allergy is multi-factorial and that multiple different possible perturbations of the immune system can result in disease.

Secretory cells such as goblet cells have been implicated as conduits for passage of food antigens in the gut(34). The observation that C3H/HeJ have more goblet cells and the association between oral anaphylactic susceptibility in C3H/HeJ mice and increased gut permeability to model food antigens such as OVA, invites further study towards mechanistic understanding of gut barrier defects or alterations in luminal antigen sampling. This has the promise of identifying causative genetic pathways as well as cellular mechanistic aspects underlying the risk for food allergy. In addition, other aspects of mucosal immunity that could impact gut permeability such as IgA levels, tight junction integrity and cellular passages remain to be investigated(18,44). Further studies of both murine models and clinical cases should be mindful of physical alterations in gut barrier integrity that could lead to increased susceptibility to oral anaphylaxis.

Key Messages.

• C3H/HeJ mice are predisposed to oral anaphylaxis to peanut, but not due to mutations in Tlr4 or Dock8.

• Gut permeability is increased in C3H/HeJ mice and is associated with oral anaphylactic susceptibility.

• Future therapy for food allergy may address changes in gut permeability.

Capsule Summary.

C3H/HeJ mice are prone to food allergy and anaphylaxis, and this is associated with increased gut permeability. Future treatment or prophylaxis of food allergy may utilize therapies targeting gut permeability.

Abbreviations used

DOCK8

Dedicator of Cytokinesis 8

TLR4

Toll-like receptor 4

Tlr4 ^C3H/HeJ

C3H/HeJ mice with two mutant, non-functional copies of Tlr4

Tlr4 ^Het

C3H/HeJ mice with one mutant, non-functional and one wild type, functional copy of Tlr4 resulting in mice with the ability to respond through the TLR4 receptor

Dock8 ^C3H/HeJ

C3H/HeJ mice with two mutant copies of Dock8

Dock8 ^WT

C3H/HeJ mice with two wild type copies of Dock8

BMDC

Bone marrow-derived dendritic cell

IgE

Immunoglobulin E

mMCP1

Mucosal mast cell protease 1

PCA

Passive cutaneous anaphylaxis