2.1 Animal

The experimental procedures were approved by the Animal Ethical Committee. BALB/c mice aged 3 weeks (SPF) were obtained and housed in the Animal Research Center of the Ruijin Hospital, Shanghai Jiao Tong University School of Medicine with a 12-hr light/12-hr dark cycle in 23 ± 2°C.

2.2 Ethics statement

The procedure of animal was approved by the Ruijin Hospital, School of Medicine, Shanghai Jiao Tong University. Animal study was carried out in accordance with the Guide for Care and Use of Laboratory Animals of National Research Council (SYXK [Shanghai 2011–0113]).

2.3 Sensitization and challenge

The FA were induced by OVA in mice, as previously described (Kim et al., 2016), with some modification. Briefly, to determine the optimal OVA concentration inducing FA, the mice were sensitized with 10, 20, and 50 μg/0.2 ml OVA (grade V; Sigma-Aldrich, St. Louis, MO) by intraperitoneal injection adsorbed on aluminum potassium sulfate adjuvant (A-7210, Sigma-Aldrich) on Days 0 and 14, respectively. From Day 28, 500 mg/L OVA in saline was administered orally to challenge every 3 days, five times altogether. The mice were killed on the first day after the final challenge (Figure 2a).

2.4 Probiotic supplements

Bb M-16V (Dipro; AB-BIOTICS, S.A, SPAIN) were provided by MIHC Science Laboratory, DERBY CARE MEDICAL TECHNOLOGY CO., LIMITED. Mice in the probiotic group received Bb M-16V orally (1 × 10⁷ CFU/g formula per mouse daily; Figure 2a).

2.5 Histology

The intestinal tissue was fixed by 4% paraformaldehyde, embedded in paraffin, and sections of paraffin were stained with chloroacetate esterase. Tissue sections were counted by investigators blinded to the sample identity in 10 high power fields (HPFs) at 400 times magnification.

2.6 ELISA experiments

Cytokine and immunoglobulin levels in serum were assayed by ELISA, using a kit (Cusabio Biotech, Wuhan, China) following the manufacturer's use instructions. Quantitative sandwich enzyme immunoassay was used in this assay. This assay employs the qualitative enzyme immunoassay technique. This experiment has high sensitivity and sensitivity for the detection of serum samples in mice. No significant cross-reaction or interference was observed in the relationship between detection index and their analogs.

2.7 Quantitative polymerase chain reaction

Gene expression was detected by quantitative polymerase chain reaction (qPCR). In brief, RNA lysis buffer homogenized the samples of small intestine tissue. Total RNA was extracted using TRIzol reagent (Qiagen, Hilden, Germany). Complementary DNA (cDNA) was prepared using the kit of iScript™ cDNA synthesis (Promega, Madison, WI). qPCR was done with ABI prism 7900 sequence detection system (Application Binary Interface). Total IL-33 and ST2 messenger RNA levels were measured by qPCR using appropriate primers. IL-33 (F: 5′-GGTGTGGATGGGAAGAAGCTG-3-G R: 5′-GAGGACTTTTTGTGAAGGACG-3′) and ST2 (F: 5′-CAAGTAGGACCTGTGTGCCC-3′ and R: 5′-CGTGTCCAACAATTGACCTG-3′). The gene expression levels were standardized and calculated with the $$ method for relative quantification normalized by glyceraldehyde-3-phosphate dehydrogenase gene expression.

2.8 16S rRNA gene amplification and sequencing

The fecal samples of mouse were collected and stored at −80°C. The DNA extraction of samples was performed using the kit of MagAttract PowerSoil DNA Isolation (Qiagen, Hilden, Germany). The V4-V5 variable region of the 16S rRNA gene was amplified, standardized, pooled, and sequenced by the Illumina MiSeq platform using 2 × 250 cycles.

2.9 Quantification and statistical analysis

All experiments were conducted by using randomly assigned mice. Data analysis was performed with GraphPad Prism 7.0. One-way analysis of variance was used to analyze the results. The sparse canonical correlation analysis was applied to identify the relationship between immunological variables and gut microbiota taxa. All data are presented as mean ±  standard error of the mean. The p values of lower than .05 were considered significant: ^∗p ≤ .05, ^∗∗p ≤ .01, ^∗∗∗p ≤ .001, ^∗∗∗∗p ≤ .0001.

3.1 The determination of optimal OVA concentration inducing FA

The avirulent food allergens, such as OVA, may cause oral sensitization in mice with anaphylaxis that occurs in IgE-dependent and T-helper 1/T-helper 2 (Th1/Th2) imbalance manner (Noah et al., 2019; O'Konek et al., 2018).

The results showed that the intestinal tissue structure was intact and there was no obvious inflammatory change in the control group. In 10 and 50 µg OVA groups, a small amount of inflammatory cell infiltration was observed. In 20 µg OVA group, intestinal mucosa was destroyed, villous epithelial cells were not neatly arranged or exfoliated, and a large number of inflammatory cells such as lymphocytes and eosinophils were observed to infiltrate (Figure 2b).

SlgE was significantly increased in all the OVA-sensitized groups, suggesting the occurrence of anaphylaxis in mice (Figure 2c). The upregulation of IL-4 and nonsignificant change of interferon-γ (IFN-γ) resulted in the increase of IL-4/IFN-γ ratio in the 20 µg OVA group (Figure 2d-f).

Twenty microgram OVA-sensitized, but not 10 and 50 µg OVA-sensitized, mice had anaphylaxis on the oral challenge with OVA, suggesting that the FA model by 20 µg OVA is more suitable for probiotic therapy studies.

3.2 The immune effect of Bb M-16V16V on the OVA-induced FA mouse model

While different doses of OVA were used to induce FA, corresponding Bb M-16V treatment groups were also assigned. We selected OVA (20 µg)-sensitized and probiotic (Bb M-16V)-intervened mice as research objects to further explore the therapeutic role of probiotics on FA.

M-16V significantly improved the inflammation of small intestine promoted by OVA and reduced the infiltration of inflammatory cells (Figure 3a). The increase of IL-4 and IL-4/IFN-γ induced by OVA was significantly inhibited, but the impact on IL-5 and IL-13 was not obvious (Figure 3b-f). Moreover, Bb M-16V had no significant inhibitory effect on the sIgE and sIgG induced by OVA. It indicated that Bb M-16V may have an immunomodulatory effect on FA (Figure 3g,h).

3.3 The effect of oral Bb M-16V treatment on IL-33 and ST2 expression in OVA-sensitized mice

IL-33/ST2 may be an alarm in cytokine with important roles in allergic diseases, such as asthma and atopic dermatitis (AD) (Cayrol et al., 2018; Galand et al., 2016), regardless of association with Type 2 immunity. We used OVA-sensitized mice to verify the role of IL-33/ST2 in this model. Compared with the control group, the IL-33 expression increased in the OVA-sensitized (OVA) group and the same results were observed in the probiotic-intervened (OVA + Bb) group (Figure 4a). Interestingly, ST2 expression had not significantly changed in the OVA group and was significantly inhibited in the OVA + Bb group (Figure 4b).

These results suggest that IL-33 plays a crucial part in the FA elicited by OVA, accompanied by the production of sIgE antibody and antigen-driven Th2 cytokines in sensitized mice. Whereas administration of Bb M-16V during FA showed no significant inhibitory effect on IL-33. Meanwhile, The ST2 blockade induced by Bb M-16V not only affected the systemic Th2 response but also significantly reduced the severity of allergic reactions. However, the alleviation of FA by Bb M-16V is not associated with IgE-mediated immune response.

3.4 The oral administration of Bb M-16V improved the structure of gut microbiota during OVA sensitization

After discussing the immune effects of Bb M-16V, we used 16S rRNA gene amplification and sequencing technology to detect the effects of Bb M-16V on the gut microbiota of OVA-induced FA mice.

The gut microbiota was analyzed from different taxonomic levels. At the phylum, class, order, and family level, Actinobacteria, Epsilonproteobacteria, Campylobacterales, and Bacteroidaceae composition of OVA and OVA + Bb groups showed significant differences, respectively (Figure 5a-d). In addition, at the genus level, the average abundance of Bacteroides in the OVA group was higher, while Bb M-16V significantly reduced the composition of Epsilonproteobacteria (Figure 5e).

3.5 Search for the biomarkers of microbiota in the feces of mice with FA and Bb M-16V intervention

The results and differences in taxa caused by Bb M-16V are seen in the LEfSe analysis (Figure 6a). After the Bb M-16V treatment, the abundance of Actinobacteria was higher than before the treatment at the phylum level. At the class level, Bb M-16V increased Actinobacteria and Erysipelotrichia. The abundances of Bifidobacteriales, Coriobacteriales, and Erysipelotrichales were enriched by Bb M-16V at the order level. Bb M-16V enhanced the abundances of seven families, including Corynebacteriaceae, Bifdobacteriaceae, Coriobacteriaceae, Porphyromonadaceae, Christensenellaceae, Eubacteriaceae, and Erysipelotrichaceae, while it inhibited Peptostreptococcaceae.

Besides, the comparison of LDA is exhibited (Figure 6b). Compared with the baseline composition of the microbiota in the control group, OVA sensitization had upregulated seven bacteria proportion, including, Bacteroides, Peptostreptococcaceae, Romboutsia, Bilophila, Parabacteroides, Vampirovibrio, and Anaerotruncus, but bacteria like Parabacteroides, Gemella, Bacillales, Clostridium, Odoribacter, Desulfovibrio, Proteus, and Alistipes were significantly downregulated. Simultaneously, Bb M-16V intervention had increased seven bacteria proportion, including Porphyromonadaceae, Turicibacter, Barnesiella, Corynebacteriaceae, Corynebacterium, Christensenella, and Christensenellaceae, reducing the abundance of bacteria increased by OVA.

3.6 The oral administration of Bb M-16V regulated the functional gene composition of gut microbiota

To explore the impact of Bb M-16V intervention on physiological function, we used PICRUSt to analyze and predict the functional gene composition in the metabolic pathways. The result showed that Bb M-16V significantly regulated the 12 subfunctional gene composition of gut microbiota (Figure 7). Before Bb M-16V treatment, OVA sensitization upregulated the functional genes, including Pentose and glucuronate interconversions, Proteasome, Carbon fixation in photosynthetic organisms, Prostate cancer, and NOD−like receptor signaling pathway, whereas after Bb M-16V treatment, six other functional genes, for instance, Styrene degradation, Butanoate metabolism, Fatty acid metabolism, Aminobenzoate degradation, and Benzoate degradation. The results indicated that Bb M-16V treatment significantly influenced the functional gene composition of gut microbiota.

3.7 The gut microbiota taxa in the feces associated with immune factors and immunoglobulin in serum

Next sparse canonical correlation analysis was used to identify nine immunological variables and gut microbiota taxa that associated with each other. The gut microbiota with the highest positive and negative loading indicated that most closely related to immunological variables. The results displayed that the concentration of Epsilonproteobacteria and Campylobacterales were positively correlated with the IL-4/IFN-γ (Figure 8a-e), indicating that these bacteria can induce Th2-mediated allergic reaction. On the other hand, Bacteroidaceae and Bacteroides were negatively correlated with IL-33 and positively correlated with ST2, while Actinobacteria was opposite to these bacteria, which corresponded with the analysis results of immune factors and gut microbiota (Figures 3d, 4, and 5).