Clinical characteristics of participants
A total of 39 subjects (22 males and age 27 ± 8 years) participated in this study. Among them, 20 subjects were included in the AR group (10 males and 10 females, and age 30 ± 6 years), and 19 subjects were included in the HC group (12 males and 7 females, and age 25 ± 10 years). Eosinophils are the major effector cell of innate immunity, which are considered as a biomarker for AR . As shown in Table 1, the percentage of eosinophils was significantly increased in AR patients than that in HCs (P < 0.01). Moreover, IgE is the molecular component of atopy and the evaluation of total serum IgE levels has become a diagnostic criterion for AR . The result showed that patients with AR had higher total serum IgE levels as compared to HCs (P < 0.01). Dermatophagoides (D) allergen is a risk factor for the development of AR . The levels of serum specific IgE to D (serum sIgE-D1 and D2) were also dramatically higher in the AR patients than that in the HCs (P < 0.0001) (Table 1).
Phenotypic characterization of nasal EVs
The presence of nasal EVs from AR patients and HCs was investigated using TEM and nano-flow cytometry. TEM showed that nasal EVs presented a clear spheroid morphology from AR patients and HCs, however, the size of most of nasal EVs from AR patients were slightly larger than that in HCs (Fig. 1a). CD9 is a tetraspanin commonly used as a exosomal specific marker for the identification of EVs. Naon-flow cytometry showed an increase in CD9 from AR patients compared to HCs (Fig. 1b).
Alpha diversity of microbiome in nasal EVs
Alpha diversity is a pivotal indicator in the abundance and diversity of observed microbes, which is assessed by Chao1, Shannon, PD_whole_tree, Pielou, and Simpson indices. The Chao1 index is a qualitatively measure for species richness. As shown in Fig. 2a, there was no significant difference for the Chao1 index between AR patients and HCs (P = 0.1107). The Shannon index relates both species richness and evenness, and Pielou and Simpson give a greater weight to species evenness. The Shannon diversity of AR patients and HCs also did not differ significantly (P = 0.1354), while Pielou and Simpson indices were dramatically decreased in AR patients compared to those in HCs (P < 0.05) (Fig. 2b-d). The metric PD_whole_tree is Faith’s Phylogenetic Diversity, which is measured by adding up all the branch lengths based on the phylogenetic tree. There was no obvious difference in PD_whole_tree diversity between AR patients and HCs (P = 0.1417) (Fig. 2e).
Beta diversity of microbiome in nasal EVs
Beta diversity refers to microbial composition differences between different samples, presented by the principal coordinate analysis (PCoA) and hierarchical clustering based on UniFrac distance.According to the unweight UniFrac analysis, PCoA1 and PCoA2 accounted for 19.11% and 6.03% of total PCoA, respectively (Fig. 3a). For weighted UniFrac distances, PCoA1 and PCoA2 respectively accounted for 26.07% and 16.8% of total PCoA (Fig. 3b). Hierarchical clustering analysis showed that the microbial composition of AR patients and HCs presented relative differences based on both unweighted and weighted UniFrac distances (Fig. 3c, d).
Bacterial composition at different taxonomic levels
At the phylum level, the top 10 bacterial phyla in nasal EVs from both AR patients and HCs were listed in Supplementary Table S1. Taxonomic classification at phylum level showed that Proteobacteria was the dominant bacterial phylum in both AR patients and HCs (Fig. 4a). However, the abundance of Proteobacteria was not significantly different between AR patients and HCs (90.564% in AR patients vs. 87.303% in HCs, P = 0.888). Among the top 10 bacterial phyla, the Tenericutes (0.314% in AR patients vs. 0.041% in HCs, P = 0.002, Kruskal-Wallis test) and Verrucomicrobia (0.067% in AR patients vs. 0.006% in HCs, P = 0.012, Kruskal-Wallis test) were significantly more abundant in AR patients than that in HCs (Supplementary Table S1, Fig. 4a). At the class level, the top 10 bacteria in nasal EVs from AR patients and HCs were listed in Supplementary Table S2. The dominant bacterial classes were Alphaproteobacteria, Gammaproteobacteria, and Betaproteobacteria in nasal EVs from both AR patients and HCs (Fig. 4b). The abundance of these three bacterial classes all did not show significant differences between AR patients and HCs (P = 0.273 for Alphaproteobacteria, P = 0.086 for Gammaproteobacteria, and P = 0.226 for Betaproteobacteria). Among the top 10 bacterial classes, Mollicutes was significantly more abundant in AR patients than that in HCs (0.314% in AR patients vs. 0.041% in HCs, P = 0.002, Kruskal-Wallis test) (Supplementary Table S2, Fig. 4b). Within the order taxonomic rank, top 12 bacterial orders in nasal EVs from AR patients and HCs were listed in Supplementary Table S3. The most abundant orders presenting in both AR patients and HCs were Pseudomonadales, Burkholderiales, Rhizobiales, and Sphingomonadales (Fig. 4c). These four dominant orders all did not show significant differences between AR patients and HCs (P = 0.123 for Pseudomonadales, P = 0.555 for Burkholderiales, P = 0.129 for Rhizobiales, and P = 0.778 for Sphingomonadales). In other orders, Rhodocyclales was dramatically less abundant in AR patients than that in HCs (1.077% in AR patients vs. 5.147% in HCs, P = 0.009, Kruskal-Wallis test) (Supplementary Table S3, Fig. 4c). At the family level, the top 10 bacterial families in nasal EVs from AR patients and HCs were listed in Supplementary Table S4. The most abundant families in both AR patients and HCs were Moraxellaceae, Burkholderiaceae, Methylobacteriaceae, and Sphingomonadaceae (Fig. 4d). Among the top 10 families, Zoogloeaceae showed significantly less abundant in AR patients than that in HCs (1.060% in AR patients vs. 5.118% in HCs, P = 0.028, Kruskal-Wallis test) (Supplementary Table S4, Fig. 4d). Within the genus taxonomic rank, top 10 genera in nasal EVs from AR patients and HCs were listed in Supplementary Table S5. Among them, Acinetobacter, Ralstonia, Methylobacterium, and Sphingomonas were the dominant bacterial genera in both AR patients and HCs (Fig. 4e).The genus Zoogloea was dramatically less abundant in AR patients than that in HCs (1.054% in AR patients vs. 5.098% in HCs, P = 0.002, Kruskal-Wallis test) (Supplementary Table S5, Fig. 4e).
Differential bacterial communities in nasal EVs
To unveil the differential bacterial communities in nasal EVs from AR patients and HCs, LEfSe analysis was performed. The bacterial class that differed markedly between AR patients and HCs was Mollicutes. At the order level, Mycoplasmatales was increased in AR patients and Rhodocyclales was increased in HCs. Analysis at the family level showed ascending levels of Halomonadaceae and Mycoplasmataceae for AR patients, and ascending levels of Streptococcaceae, Zoogloeaceae, and Pseudomonadaceae for HCs. At the genus level, Acetobacter, Escherichia, Halomonas, and Mycoplasma were abundant in AR patients, and Streptococcus, Burkholderia, Zoogloea, and Pseudomonas were abundant in HCs (Fig. 5a). The 17 taxa mentioned above presented significant differences between AR patients and HCs (the logarithmic LDA score > 3.0) (Fig. 5b).
Enrichment of microbial metabolic pathways
To analyze the microbial metabolic function in nasal EVs of AR, the PICRUSt based on the KEGG database was conducted to predict corresponding microbial metabolic pathways. Total 35 metabolic pathways with significant differences between AR patients and HCs were identified (P < 0.05). Among them, 25 microbial metabolic pathways were unregulated in AR, including valine, leucine and isoleucine degradation, tryptophan metabolism, toluene degradation, propanoate metabolism, and primary bile acid biosynthesis, etc. The rest 10 pathways presented significantly more abundant in HCs than that in AR patients, including selenocompound metabolism, protein kinases, prenyltransferases, porphyrin and chlorophyll metabolism, nitrogen metabolism, glycosyltransferases, glycosylphosphatidylinositol (GPI)-anchor biosynthesis, folate biosynthesis, D-arginine and D-ornithine metabolism, and C5-branched dibasic acid metabolism (Fig. 6). Accordingly, our data suggest that metabolism functions were changed by alteration of bacterial composition.