APS (purity, >98%; Shanghai Macklin Biochemical Co., Ltd.), ovalbumin (OVA) and aluminium hydroxide (Sigma-Aldrich; Merck KGaA) were used in the experiments in the present study.

A total of 36 healthy female BALB/c mice (weighing 18-22 g, 4-6 weeks old) were purchased from the Laboratory Animal Center of Hangzhou Medical College. All experimental protocols involving animals were approved by the Ethics Committee of the Zhejiang Center of Laboratory Animals (Hangzhou, China; approval no. ZJCLA-IACUC-20020187) and conducted conforming to the guidelines of the China Animal Protection Commission. In addition, all mice were raised in an environment with a temperature of 22±2˚C and with a relative humidity of 50±1%.

In a random manner, the 36 mice were classified into six groups, with 6 mice in each group, as follows: i) The normal control group sensitized with phosphate-buffered saline (PBS; HyClone; Cytiva); ii) the asthma group sensitized and challenged with OVA (grade V, Sigma-Aldrich; Merck KGaA); iii) the OVA + APS (2.5 g/kg) treatment group; iv) the OVA + APS (5 g/kg) treatment group; v) the OVA + APS (5 g/kg) treatment group; and vi) the OVA + dexamethasone (DXM, Sigma-Aldrich; Merck KGaA, 2 mg/kg) treatment group used as a positive control.

The establishment of the mouse model of asthma was as follows: i) The sensitization stage: The asthma model was constructed by the injection of 20 µg emulsifying OVA in 200 µl PBS on days 1, 8 and 15 of the experiment. ii) The excitation stage: Following initial sensitization, the mice were subjected to the aerosol inhalation of 1% OVA in a closed container for excitation every day from day 22, and each aerosol inhalation lasted for ~30 min for 1 week consecutively. In the negative control group, PBS was used instead of OVA during the sensitization and excitation stages. All drugs (APS and DXM) were dissolved in an equivalent amount of PBS and administered at a dose of 0.2 ml/mouse. APS was administered via intragastric administration every day for 1 week. DXM (2 mg/kg) served as the positive control and was administered in the same manner. On day 29, the enhancement suspension (Penh) was evaluated, and the mice were sacrificed on day 30. Serum, lung tissue and splenic cells were harvested for further analysis (Fig. 1). At the end of the animal experimental period, the experimental animals were subjected to cervical dislocation under anesthesia. The mice were anesthetized with 40 mg/kg pentobarbital sodium solution by intraperitoneal injection. All procedures were carried out following strict ethical guidelines, and all efforts were made to minimize animal suffering.

At 24 h following the final excitation, AHR was indirectly assessed using whole-body plethysmography. More specifically, each conscious mouse was stimulated with Mch at elevating doses (Mch aerosol containing 3.125, 6.25, 12.5, 25, 5 and 50 mg/ml saline) for 3 min. The 3-min Penh value following each Mch excitation was calculated.

Following collection, the lung tissue was fixed with a concentration of 4% paraformaldehyde (20˚C,48 h), paraffin-imbedded and subjected to histological staining. Briefly, the 4-µm-thick left lung sections were stained with H&E (Beijing Solarbio Science & Technology Co., Ltd.; 20˚C, 30 min) to evaluate eosinophil infiltration and inflammatory cell infiltration in the surrounding lung tissue.

DNA was extracted from the mouse feces, and cellular DNA was amplified using PCR. In brief, the amplification region was bacterial 16SV3+V4, and the primers used were 338F and 806R. The V3 upstream primer sequence was 5'-ACTCCTACGGGAGGCAGCA-3', and the V4 downstream primer sequence was 5'-GGACTACHVGGGTWTCTAAT-3'. The DNBseq sequencing process and method were applied in sequencing. In addition, the raw data were spliced with FLASH (version 1.2.11), and the spliced Tags were clustered into operational taxonomic units (OTUs) with the application of USEARCH software (v7.0.1090).

An appropriate amount of serum sample was supplemented with 200 µl water for homogenization, followed by vortexing for 60 sec. Thereafter, 800 µl methanol-acetonitrile solution (1:1, Merck, V:V) was supplemented for vortexing for a further 60 sec, followed by low-temperature ultrasonic treatment twice (30 min each). After being allowed to stand at -20˚C for 1 h to precipitate the protein, the sample was subject to centrifugation at 14,000 x g and 4˚C for 20 min to collect the supernatant. Subsequently, the sample was separated with the ACQUITY UPLC BEH Amide chromatographic column (100x2.1 mm, 1.7 µm) of the Agilent 1290 Infinity LC ultra-performance liquid chromatography system (UPLC), and mass spectrometry was carried out using the Agilent 6550 mass spectrometer (Agilent Technologies, Inc.). In addition, the metabolites were detected with the AB Triple TOF 6600 mass spectrometer, and the structural identification of metabolites was completed by collecting the first-order and second-order spectra of quality control (QC) samples. Following data preprocessing with Pareto scaling, unidimensional and multidimensional statistical analyses were performed, and the volcano plot was drawn using R software.

Based on the KEGG compound database, the metabolites were annotated. Furthermore, the annotated metabolites were then matched based on the KEGG pathway database. Significantly regulated metabolites were imported into the metabolite enrichment analysis MetaboAnalyst 5.0 for their metabolic pathway enrichment. The enrichment results were calculated for the significance (P-values) by hypergeometric tests and presented as bubble plots.

Statistical analysis was conducted using SPSS version 22.0 software (IBM Corp.). Measurement data are presented as the mean ± standard deviation (SD) and were compared using one-way analysis of variance for comparisons among multiple groups. In pairwise comparisons, the least significant difference (LSD) test was adopted in the case of homogeneity of variance. One-way ANOVA was used for comparison between multiple groups and pairwise comparisons were conducted using the Bonferroni method. Tamhane's T2 test was used for multiple comparisons in the case of heterogeneity of variance. A value of P<0.05 was considered to indicate a statistically significant difference.

With the purpose of evaluating the effects of APS on AHR, the Penh values induced with various doses of Mch (3.125, 6.25, 12.5, 25 and 50 mg/ml) were measured. The AHR level in each group increased with the increase in the inhaled Mch concentration. The AHR in the asthma group was markedly higher than that in the normal control group following induction with Mch (P<0.05). However, the Penh value decreased following treatment with APS and DXM (P<0.05; Fig. 2).

The H&E staining of mouse lung tissue revealed a large amount of inflammatory cell infiltration in the peribronchial area, such as eosinophils, neutrophils and lymphocytes (Fig. 3), with eosinophils being more prominent. Following treatment with APS, inflammatory cell infiltration markedly declined.

In the present study, the rarefaction curve tended to be flat, suggesting the reasonable sequencing data volume, which indirectly reflected the species richness. The Shannon curve also tended to be flat, suggesting a sufficiently large sequencing data volume (Fig. 4).

In the present study, the sequencing depth index coverage value of each sample was >0.996, suggesting that the experimental data could truly reflect the microbial communities of the experimental samples. Alpha diversity indicates the species evenness and richness, mainly including the ACE, Chao, Shannon and Simpson indexes. As shown in Table I, compared with the control group, the OTU number of gut microbiota in the asthma group was markedly decreased (P<0.05), and the ACE, Chao and Shannon indexes exhibited a significant decrease (P<0.05). Compared with the asthma group, the OTU numbers in the APS (5.0 and 10 g/kg) groups were notably increased (P<0.05), and the ACE, Chao and Shannon indexes were also significantly elevated (P<0.05), which tended to be similar to the levels in the control group. These results suggested a marked difference in the gut microbiota between the asthmatic and normal mice, following medium- and high-dose APS intervention. In addition, the OTU numbers and alpha diversity indexes in the gut microbiota of the asthmatic mice were increased, and the microbial community abundances were recovered.

As indicated by principal coordinates analysis (PCoA), the sample distance was the farthest between the model and control group, and these two groups were well distinguished, revealing the obvious difference in community structure between the two groups (Fig. 5). In other words, the gut microbial communities in the asthmatic mice were markedly altered. By contrast, the sample distribution in the high-dose APS group gradually approached the level in the normal group, and exhibited intersection with the latter. In particular, the high-dose APS group was the closest to the normal group, suggesting that APS regulated the gut microbial communities in asthmatic mice and alleviated asthma-induced dysbacteriosis.

As illustrated in Fig. 6, at the phylum level, the gut microbiota in the mice primarily consisted of Firmicutes, Proteobacteria, Bacteroidetes and Actinoidetes, among which, Firmicutes and Bacteroidetes were the dominant bacterial communities. Compared with the control group, the relative abundance of Firmicutes in the model group was elevated by 16.5%; compared with the model group, the relative abundances of Firmicutes in the three APS groups (2.5, 5.0 and 10 g/kg) decreased by 0.21, 7.9 and 14.5%, respectively. Compared with the normal control group, the relative abundance of Bacteroidetes was reduced by 8.9% in the model group. Compared with the model group, the relative abundance of Bacteroidetes in the APS (2.5 g/kg) group decreased by 3.2%, while the abundance in the other APS (5.0 and 10 g/kg) groups was elevated by 8.4 and 12.3%, respectively. The Firmicutes/Bacteroidetes (F/B) ratio is usually used as a marker to evaluate gut microbial disturbance (7). In asthmatic mice, the gut microbial imbalance mainly manifested as a low abundance of Bacteroidetes and a high abundance of Firmicutes, yielding an increased F/B ratio. When compared with the control group, the F/B ratio in the model group tended to significantly increase. Compared with the model group, the F/B ratios in the APS (5.0 and 10 g/kg) groups exhibited a decreasing trend. In the high-dose APS group, the abundance of Firmicutes decreased, and that of Bacteroidetes increased, thereby decreasing the F/B ratio and correcting the gut microbial imbalance (Table SI).

Analysis of gut microbial species structure at the genus level. As demonstrated in Fig. 7, compared with the normal control group, the relative abundances of Lactobacillus, Prevotella, Helicobacter and Barnesiella were decreased in the model group, while those of Alistipes and Clostridium_XlVa were increased. Compared with the model group, the abundances of Lactobacillus, Barnesiella and Prevotella in the fecal samples of mice in the APS (5.0 and 10 g/kg) groups were increased, while the abundance of Clostridium_XlVa was decreased (Table SII). These findings indicated the positive regulatory effects of APS on gut microbial communities in asthmatic mice at the genus level. More specifically, APS increased the abundances of probiotics and suppressed the growth of pernicious bacteria.

As shown in Fig. 8, in the partial least squares-discriminant analysis (PLS-DA) model, the model group was notably distinguished from the other groups, indicating that the occurrence of asthma and the interventions of low-, medium- and high-dose APS treatment markedly altered the intestinal metabolite concentrations in the mice. Combined with fold change (FC) analysis and the t-test, the significance in the changes of metabolites between two samples was analyzed and visualized in the form of a volcano plot (Fig. 8). In the volcano plot, red and blue points indicated metabolites of FC>1.5 and P<0.05, namely, the differential metabolites between two groups, with red points indicating significantly upregulated, whereas blue points represented significantly downregulated metabolites. In the present study, significantly up- and downregulated differential metabolites were found between the model and control groups (Fig. 8A), the model group and low-dose APS group (Fig. 8B), the model group and medium-dose APS group (Fig. 8C), the model group and high-dose APS group (Fig. 8D), as well as between the model group and DXM group (Fig. 8E). Further hierarchical clustering suggested that there were 145 differential metabolites between the model group and APS (5.0 g/kg) group, including the significantly upregulated 3-hydroxyhexanoylcarnitine, N-palmitolenoyltaurine, 1-palmitoleoylglycero and 3-hydroxybutyrylcarnitine, as well as the significantly downregulated glycylvaline, S-adenosylhomocysteine and 5-methyl-2'-deoxycytidine (Fig. 9A).

Upon further hierarchical clustering, 105 significant differential metabolites were detected between the model group and APS (10 g/kg) group, including the significantly upregulated cholate, 7-ketodeoxycholate and ursocholate, together with the significantly downregulated glycitein sulfate, pinitol, glycylvaline, dihomo-linolenoyl-choline, catechol glucuronide, 3-(3-hydroxyphenyl)propionate sulfate and ursodeoxycholate. These results revealed that APS could alter the blood metabolites in mice (Fig. 9B).

Analysis of related metabolic pathways. As presented in Fig. 10, KEGG pathway enrichment analysis suggested that the most significantly affected metabolic pathways in the model group and APS (5.0 g/kg) group were Biosynthesis of unsaturated fatty acids and Arginine biosynthesis (Fig. 10A), while those in the model group and APS (10 g/kg) group were Biosynthesis of unsaturated fatty acids and Pyrimidine metabolism (Fig. 10B).