Female Balb/c mice (18–20 g) were obtained from the animal facility of the Institute of Biomedical Sciences, University of Sao Paulo. Animals were housed in groups of five per cage in a light- and temperature-controlled room (12 h light/dark cycles, 21 ± 2°C) with free access to food (AIN93-M) (Reeves et al., 1993) and water. The local Animal Care Committee of the University of Sao Paulo Institute of Biomedical Sciences approved the experiments.

Mice were anesthetized by an intraperitoneal (i.p.) injection of ketamine/xylazine (100 and 20 mg/kg, respectively). After an incision was made in the lower part of the abdomen, the ovaries were identified and removed from the adherent tissue. The effectiveness of the OVx procedure was assessed by analyzing the morphologic features of cells in vaginal smears and by quantifying the uterine weight. Mice subjected to similar manipulations except for the ovary removal were used as the sham-operated controls and labeled ‘sham’ animals. The same surgeon performed all surgical procedures.

Fecal samples were collected 1 day before and 10 days after surgery and fecal pellets were frozen in 2 mL tubes at -80°C until DNA extraction. Pellets were placed in MoBio PowerSoil bead tubes and incubated at 70°C for 10 min before proceeding with the protocol recommended by the manufacturer. Extracted DNA was quantified using a Qubit spectrophotometer (Thermo Scientific Technologies) and visualized in 2% agarose gels stained with ethidium bromide to evaluate DNA integrity. Microbiote analysis was performed essentially as described by Thomas et al. (2016). In summary, for PCR amplification and amplicon sequencing, the V4–V5 region of the 16S rRNA gene was amplified using the forward 5′-AYTGGGYDTAAAGNG-3′ and reverse primer 5′-CCGTCAATTCNTTTRAGTTT-3′, corresponding to the positions 562 and 906, respectively, of the Escherichia coli 16S rRNA gene. Three 20 μl amplification reactions were performed per sample, each containing: 2.5 μM of each primer; 10 μl of Kapa Hotstart High Fidelity Master Mix (Kapa Technologies); and 25 ng of genomic DNA (gDNA). Thermocycling conditions were: 95°C, 3 min; 98°C, 15 s; and 40°C, 30 s for 35 cycles. This was followed by a last extension step at 72°C for 5 min. Amplicons of the three reactions of each sample were pooled and purified using a MinElute PCR Purification Kit (Qiagen). The purified products were run on 1.5% agarose gels and bands within the expected amplicon range were excised using sterile and disposable scalpels and purified using the Qiaquick gel extraction kit (Qiagen) to remove artifacts, primer-dimers, and non-specific bands. Amplicons were end-repaired and Ion Torrent adaptors with barcodes were ligated using the Ion Plus Fragment Library Kit (Thermo Scientific). Equimolar amounts of barcoded-amplicons from each sample were pooled, using the Ion Torrent qPCR quantitation kit (Thermo Scientific), and used for emulsion PCR using the Ion PGM Template OT2 400 Kit (Thermo Scientific). All samples were sequenced on the Ion torrent PGM platform using a v2 chip and the Ion PGM Sequencing 400 Kit (Thermo Scientific). Sequences processed by the latest version of the Ion Torrent server (v3.6.2) were used as input into the Qiime (Quantitative insights into microbial ecology) software package (Version 1.6.0) (Caporaso et al., 2010). We removed sequences with an average quality score <20 using a 50 nt sliding window. Then, we identified barcodes used for fecal sample, allowing a maximum of two mismatches, and discarded sequences with no barcodes, and <200 nt or >500 nt after barcode removal. PCR primers identified at the start or at the end of the reads, allowing a maximum of 4 nt mismatches, were trimmed and sequences with no identifiable primers were discarded. After primer trimming, we removed all sequences below 200 nt and the remaining sequences were used as input for downstream analysis. Filtered sequences were clustered with 97% identity using UPARSE (implemented in USEARCH v7 – Edgar, 2013) and the seed sequence of each cluster was picked as a representative. Chimeric sequences (and clusters) were identified using UCHIME (Edgar et al., 2011) and the Broad Institute’s chimera slayer database (version microbiomeutil-r20110519) and excluded from further analysis. With the Qiime interface (default parameters), the RDP classifier (Wang et al., 2007) was used to assign a taxonomic rank to each sequence using a minimum confidence value of 80% and, subsequently, to each operational taxonomic unit (OTU). Unless otherwise stated, OTUs that occurred in fewer than 25% of all samples and with less than 3 reads were not considered. Using Qiime, we rarefied the OTU table to a depth of 13,851 sequences in order to calculate species richness and phylogenetic diversity, using the total number of observed OTUs and Faith’s phylogenetic diversity (Faith et al., 2009), respectively. Rarefaction analysis was performed using Qiime and a cutoff of 13,851 for species richness. For beta-diversity analysis, OTU-representative sequences were aligned using PyNAST (Caporaso et al., 2010) against the aligned green genes core set (DeSantis et al., 2006) using Qiime default parameters, and the alignments were lane-mask filtered to build a phylogenetic tree using FastTree (Price et al., 2009). Distance matrices were generated for four metrics, unweighted and weighted UniFrac (Lozupone and Knight, 2005) Bray–Curtis and Euclidean, to compare pairwise distances between the four groups using ADONIS (Oksanen et al., 2016). Due to unequal sequence sampling depth across samples, we normalized the raw number of reads for each OTU using metagenome Seq’s (Paulson et al., 2013) cumulative sum scaling function. We then used normalized counts to investigate differences between the groups at the OTU, genera, and family level.

The species of bifidobacteria (B. longum 5^1A) used in this study was isolated and characterized at the Laboratory of Ecology and Physiology of Microorganisms, Institute of Biological Sciences, Federal University of Minas Gerais (Souza et al., 2013). The bifidobacteria were isolated from the feces of healthy children up to 5 years old in the city of Salvador (Bahia, Brazil) and identified by morfotintorial, respiratory, and biochemical tests, followed by multiplex PCR (Kwon et al., 2005). The bacteria were replicated in MRS broth medium (Difco) and grown under anaerobic conditions in an anaerobic jar at 37°C without stirring for 48 h. For the administration of probiotic bacteria, the mice received a daily inoculum by gavage of 0.1 mL containing 10⁸ bacterial cells, beginning 15 days before the first sensitization (when animals were 4 weeks old) and until ovariectomy for prophylactic administration and 1 day after ovariectomy for the treatment protocol.

Mice were sensitized intraperitoneally with 10 μg of chicken egg OVALBUMIN (OVA) grade V (Sigma Chemical Co., Saint Louis, MO, United States) dissolved in 0.2 mL of Imject alum suspension (1 mg). At days 14, 15, and 16, the mice were exposed to aerosolized OVA (1% in phosphate-buffered saline PBS) for 15 min using an ultrasonic nebulizer device (Respira Max^®, SP, Brazil) coupled to a plastic inhalation chamber (18.5 cm × 18.5 cm × 13.5 cm). Ten days after the last challenge, the animals were subjected to ovariectomy (OVx), and 10 days later, they were re-challenged over 3 days as described above. The experimental asthma protocols were started when mice were 6 weeks old. All measurements of inflammatory parameters were performed 24 h after the last aerosol re-challenge.

Acetic acid (99.7%) and citric acid were purchased from Sigma–Aldrich (St. Louis, MO, United States), butanol from Carlo Erba (Cornaredo, Italy), acetonitrile from Merck (Darmstadt, Germany), and tetrahydrofuran from Acros Organics (Fair Lawn, NJ, United States). Chromatographic analyses were performed using an Agilent 6850 system with EzChrom software, equipped with a 7683B automatic liquid sampler, a flame ionization detector (FID) (Agilent Technologies, United States), and a fused-silica capillary DB-23 column (Agilent Technologies, United States) with dimensions of 60 m × 0.25 mm internal diameter (i.d.) coated with a 0.15 μm thick layer of 80.2% 1-methylnaphatalene. The initial oven temperature was 100°C (hold 7 min), which was then increased to 200°C at a rate of 25°C/min (hold 5 min). The FID temperature was maintained at 260°C, and the flow rates of H2, air, and the make-up gas N2 were 30, 350, and 25 mL/min, respectively. Sample volumes of 5 μL were injected at 250°C using a split ratio of approximately 25:1. Nitrogen was used as the carrier gas at 1 mL/min (hold 4 min), reduced to 0.8 mL/min (hold 1 min) and then 0.6 mL/min (hold 1 min), and finally increased to 1 mL/min. The runtime for each analysis was 16.5 min. Acetic acid (HAc) was diluted to 1.0 mg/mL in butanol:tetrahydrofuran:acetonitrile (5:3:2), and further dilutions were made using the same solvent to generate a series of standard solutions. Blank human plasma was spiked with 1.0 mg/mL HAc. This solution was further diluted with blank plasma to prepare plasma standards of 0.015–1 mg/mL. To each tube, 40 mg of sodium chloride, 20 mg of citric acid, 40 μL of 0.1 M hydrochloric acid, and 200 μL of butanol:tetrahydrofuran:acetonitrile (5:3:2) were added. To quantify the acids, a calibration curve for the concentration range of 0.015–1 mg/mL was constructed. Probiotic culture medium (100 μL) was used to measure the acetate concentration. The concentration of plasma acetate was expressed in mM. Fecal samples were weighed into 1.5 mL tubes, crushed and homogenized in 100 mL of distilled water. Subsequently, 40 mg of sodium chloride, 20 mg of citric acid, 40 μL of 0.1 M hydrochloric acid, and 200 μL of butanol: tetrahydrofuran: acetonitrile (5:3:2) were added. The tubes were vortexed for 1 min and centrifuged at 10,000 rpm for 10 min. The supernatant was transferred to microtubes, and 5 μL was injected in triplicate into the gas chromatograph. The retention time was 7.2 min. The precision and accuracy of the assay were evaluated by analyzing samples (0.62, 0.25, and 0.75 mg/mL) in five replicates. The precision, expressed as relative standard deviation (RSD), was less than 14.34%, and the accuracy was between 87.51 and 113.56%. The standard acids, HAc, HPr, and HBu, were successfully separated in the gas-liquid chromatogram and eluted between 7 and 10 min in the temperature and gas flow program used here. The retention time was 7.2, 8.2, and 9.2 min for HAc, HPr, and HBu, respectively. The standard calibration curves showed linear relationships for all acids (r = 0.999). There was no interference present in the blanks at the retention time of the acids analyzed, and the analyte peaks were identifiable, discrete, and reproducible. The precision and accuracy of the assay were evaluated by analyzing samples (0.62, 0.25, and 0.75 mg/mL) in five replicates. The precision, expressed as RSD, was less than 14.34, 13.93, and 12.68%, for HAc, HPr, and HBu, respectively. The accuracy was between 87.51 and 113.56% for HAc, 91.44 and 111.60% for HPr, and 86.71 and 98.56% for HBu. The limit of quantification (LOQ) was 0.031 mg/mL for both acids analyzed, with a precision lower than 15% and accuracy within ±15%. The limit of detection (LOD) for both acids was 0.015 mg/mL, where the mean coefficient of variation (CV) value was under 20% for precision and under ±20% for accuracy. The feces extraction efficiency was 97.35 ± 5.23%. Therefore, the developed methods were shown to be reliable (Shah et al., 1991) for quantifying the SCFAs HAc, HPr, and HBu in mouse feces.

Female mice were anesthetized with ketamine/xylazine (100 and 20 mg/kg, respectively), the trachea was exposed, and a cannula was inserted. The lungs were washed three times with aliquots (each 0.8 mL) of saline injected through the cannula. From the BAL fluid (BALF), the total number of cells was counted microscopically using a Neubauer chamber. Differential cell counts were performed by cytospin analysis and prepared from aliquots of BALF (200 μL) centrifuged at 300 g for 1 min using a Citospin^® (Fanem). The cells were stained with Instant Prov (Newprov^@, São Paulo, Brazil), and a total of 300 cells were counted to determine the proportion of neutrophils, eosinophils, and mononuclear cells using standard morphological criteria.

All animals were anesthetized with ketamine (100 mg/kg, i.p.) and xylazine (20 mg/kg) and paralyzed with pancuronium bromide, and a stable depth of anesthesia was maintained (Ferreira et al., 2012). After tracheostomy, the trachea was cannulated with a blunt 18-gauge metal tube, and the mouse was ventilated with a computer-controlled small-animal ventilator (flexiVent; Scireq, Montreal, QC, Canada) using a tidal volume of 10 mL/kg and a respiratory frequency of 150 breaths/min. A positive end-expiratory pressure (PEEP) of 2 cm H₂O was applied throughout. An external jugular vein was isolated for an intravenous (i.v.) infusion of methacholine (MCh). At the outset, 6 μg of MCh was provided intravenously to ensure that the animal was indeed responsive to MCh and that airway resistance returned to the baseline value after its MCh-induced rise, which indicated that the mouse was in stable physiological condition. To obtain a dose-response curve, a bolus of MCh was then injected starting at a dose of 4 μg (200 μg/mL solution in PBS; i.v. boluses of 10–40 μL). Prior to each MCh dose, the expiratory path was obstructed for 15 s to produce a deep inflation, after which exhalation was immediately allowed. Ventilation was continued for approximately 2 min between consecutive MCh doses. Airway responsiveness was equal to the Newtonian resistance (Rn) (Ferreira et al., 2012).

Lungs were removed from mice after BAL and fixed by immersion in 4% paraformaldehyde. The lobes were sagittally sectioned, embedded in paraffin, cut into 5-μm sections, and stained with periodic acid-Schiff (PAS); mucus production was measured as previously described (Tong et al., 2006).

Mouse Treg cells were collected from the BALF and analyzed for CD4⁺ CD25⁺ Foxp3⁺ expression using a mouse Treg cell staining kit containing APC-labeled anti-CD4, PE-labeled anti-CD25, and FITC-labeled anti-Foxp3 (eBioscience) according to the manufacturer’s instructions. Briefly, prepared cells (1 × 10⁶) were washed by centrifugation with cold PBS, resuspended in 1 mL of fixation/permeabilization solution, and incubated in the dark at 4°C for 30–60 min. The cells were washed once with 2 mL of permeabilization buffer, collected by centrifugation, resuspended in 20 mL of blocking agent with 2% (2 mL) normal rat serum in permeabilization buffer, and incubated at 4°C for 15 min. Next, 20 mL of a fluorochrome-conjugated antibody or isotype control in permeabilization buffer was added, followed by incubation in the dark at 4°C for 30 min. Finally, the cells were washed with 2 mL of permeabilization buffer, resuspended in flow cytometry buffer (PBS with 2% FBS), and analyzed using a FACSCanto II cytometer (BD Bioscience, San Diego, CA, United States). The data were analyzed using FlowJo^® software.

Commercial preparations of paired antibodies and protein standards for measurements of mouse IL-4, IL-5, IL-10, and IFN-γ (BD Bioscience) in the BAL were used to develop ELISAs according to the manufacturer’s instructions. The determinations were performed in duplicate. Serum estradiol and progesterone concentrations were determined using an enzyme immunoassay kit according to the manufacturer’s instructions (Cayman Chemical, Ann Arbor, MI, United States).

The data are presented as the mean ± standard error of mean, Student’s t-test and ANOVA followed by a Bonferroni post-test were used for comparisons between 2 and 3 or more groups, respectively. Values of p ≤ 0.05 were considered significant.

Ten days after ovariectomy, we measured the body and uterus weight of female mice and analyzed the vaginal cell morphology. The ovariectomized (OVx) animals had significantly higher body weights and lower uterus weights than animals subjected to false surgery (Sham). In addition, the vaginal cell morphological changes observed after ovariectomy were representative of diestrus phase and were corroborated by decreased plasma concentrations of estrogen and progesterone (Table 1). Table 1 shows the parameters measured to analyze ovariectomy efficacy of animals studied in the Figure 1.

To analyze the effect of ovariectomy on gut microbial diversity and composition, we performed 16S rRNA next-generation sequencing. We obtained a total of 1,246,075 QC-passed sequences with an average length of 329 ± 2.9 nt and an average of 56,639 sequences/sample. A total of 258 operational taxonomic units (OTUs) were obtained, which was reduced to 248 after filtering to include only those OTUs present in at least 25% of samples and with at least three reads. When analyzing species richness, we found no significant differences between groups for any of the comparisons (Figure 1A) but did find increased phylogenetic diversity when comparing OVx POS (post-ovariectomy) mice with Sham PRE (pre-ovariectomy) mice and OVx PRE mice with Sham PRE mice. We found significant differences in pairwise distances using the weighted UniFrac metric when we compared OVx POS mice with Sham POS mice and all groups simultaneously (Table 2). Multidimensional scaling (MDS) revealed that the OVx POS samples clustered farther away from the other samples (Figure 1B). Bacterial community composition was also investigated, revealing that the most abundant bacterial phyla were (in decreasing order of abundance) Bacteroidetes, Firmicutes, and Proteobacteria. Despite seeing increases in Bacteroidetes abundance and decreases in Firmicutes abundance in OVx POS mice, these were non-significant. At the family level, the most abundant bacterial families among the groups were Porphyromonadaceae, unclassified Clostridiales, Bacteroidaceae, Lachnospiraceae, and Ruminococcaceae. Hierarchical clustering of rarefied OTU abundances clustered all samples from the OVx POS group together in one main tree branch, whereas samples from the other groups were scattered across various branches (Figure 1C). When we evaluated differences between the groups at different taxonomic levels, we found a significant increase of Bifidobacteriaceae, but not any other bacterial taxon, in OVx POS samples (Figure 1D).

We next investigated whether re-exposure of OVx allergic mice to antigen (OVA) exacerbates lung inflammation (Figure 2A). We found a significant increase in total cells recovered in the BALF from Sham and OVx allergic mice that was exacerbated by re-exposure of OVx allergic mice to OVA (Figure 2C). Moreover, the number of eosinophils in the BAL from re-challenged OVx allergic mice significantly increased compared with Sham allergic mice. The total cells in the BAL recovered from non-sensitized or OVx mice did not differ.

Next, considering that airway hyperresponsiveness is a hallmark of asthma and that its evolution is often associated with lung inflammation (Maddox and Schwartz, 2002; Tong et al., 2006), we investigated whether the ovariectomy of allergic mice would also lead to exacerbated hyperresponsiveness. We quantified the respiratory Newtonian resistance (Rn) changes in response to intravenous administration of a cholinergic agent, methacholine (MCh), after the last re-exposure to antigen challenge. Re-challenged OVx allergic mice exhibited significantly increased airway hyperresponsiveness to MCh compared with Sham OVx allergic mice (Figure 2B).

We also determined the effect of ovariectomy on the Th2 cytokine profile. Our data revealed a significant increase in IL-5 levels in the BAL recovered from OVx allergic mice re-exposed to antigen challenge compared with that of Sham OVx allergic mice (Figure 2D). Although a low level of IL-4 was found in the BALF, no significant difference between these two groups was observed (unpublished data). In contrast, IFN-γ and IL-10 were reduced in the BAL of re-challenged OVx allergic mice compared with Sham OVx allergic mice (Figure 2D).

Excessive mucus production is a characteristic of asthmatic airways; therefore, we evaluated the presence of PAS-stained cells in the lungs of re-challenged mice. As seen in Figure 2E, a significant increase in PAS-stained cells was observed in the lungs of re-challenged OVx allergic mice compared with re-challenged Sham OVx allergic mice. The lungs of non-sensitized mice did not show PAS-stained cells (Figure 2E).

The role of probiotics in regulating the immune system and allergic diseases has been extensively debated (Abrahamsson et al., 2007). An oral probiotic (B. longum 5^1A) was administered daily by gavage for 15 days before the first OVA sensitization of mice and continued until 4 h before the mouse ovaries were removed (Figure 3A). The preventive effect of B. longum 5^1A supplementation on the exacerbation of lung inflammation and airway hyperresponsiveness in re-challenged OVx allergic mice was investigated. Oral supplementation with B. longum 5^1A attenuated the exacerbation of the MCh-induced airway hyperreactivity of re-challenged OVx allergic mice. The airway hyperresponsiveness of supplemented OVx-OVA mice was significantly reduced compared to the OVx-OVA group. Additionally, B. longum 5^1A treatment significantly reduced the number of eosinophils recovered from the BAL of re-challenged OVx allergic mice (Figure 3B). B. longum 5^1A administration also resulted in a significant reduction of IL-5 levels in the BAL of re-challenged OVx allergic mice relative to non-treated mice (Figure 3C). By contrast, B. longum 5^1A treatment significantly increased the levels of IFN-γ in re-challenged OVx allergic mice. The BAL levels of IL-4 and IL-10 were not changed (Figure 3D). The probiotic treatment also significantly reduced the mucus production in the lungs of re-challenged OVx allergic mice compared with non-treated re-challenged OVx allergic mice (Figure 3E).

Because SCFAs, especially acetate, have anti-inflammatory properties, we hypothesized that SCFA production by B. longum 5^1A could modulate allergic lung inflammation. Thus, we measured SCFAs in the probiotic culture medium and in mouse feces. Our data indicated that acetate was the SCFA quantified in the bacterial medium (Figure 4A), whereas we did not find measurable levels of propionate or butyrate. In addition, B. longum 5^1A treatment significantly increased the levels of acetate in the feces of re-challenged OVA allergic mice compared to non-treated mice (Figure 4B). Butyrate was also measured but did not differ between groups (unpublished data). Since Treg cells play a role in the control of asthma, and SCFAs interfere with the functional activity of Tregs, we examined the levels of Tregs in the BALF of the re-challenged OVx allergic mice previously treated with B. longum. The re-challenged OVx allergic mice supplemented with the probiotic exhibited a significant increased percentage of Tregs in BALF compared with non-treated mice (Figure 4C).

In this set of experiments, we investigated the effect of probiotic supplementation on exacerbated lung inflammation and airway hyperresponsiveness after ovariectomy, when lung inflammation is established. The mice were supplemented with B. longum 5^1A after ovary removal. As shown in Figure 5, when the probiotic was given after ovary removal, the profile of total inflammatory cells and eosinophils recovered in the BAL was not altered in the re-challenged allergic mice relative to the non-treated re-challenged OVx allergic mice (Figure 5).