Diversity and Interactions of the Naso-Buccal Bacteriome in Individuals with Allergic Rhinitis, Asthma and Healthy Controls

Abstract

Allergic rhinitis and asthma are significant public health concerns worldwide. While previous studies have explored how nasal and buccal bacteriotas influence these conditions, few have directly compared their bacteriomes within the same cohort. To bridge this gap, I analyzed 16S rRNA next-generation sequencing data from 347 individuals, including participants with allergic rhinitis, asthma and healthy controls. The nasal and buccal bacteriomes shared all dominant bacterial taxa but differed significantly in their phylum- and genus-level relative abundances. Alpha-diversity was significantly higher in the buccal cavity, while beta-diversity varied significantly across all indices and clinical groups. Over 80% of the predicted metabolic pathways were differentially regulated between the two cavities, yet these functional differences remained fairly consistent across clinical groups. Naso-buccal bacterial networks exhibited striking differences in structure, complexity and hub nodes. Notably, the network of healthy controls showed a clear segregation between nasal and buccal bacteria, with 93.5% of the interactions occurring within each respective cavity, and contained few pathogenic keystone taxa. In contrast, bacterial networks from diseased individuals exhibited reduced ecological specialization and more pathogenic keystone taxa linked to airway disease. These findings, thus, demonstrate that the naso-buccal bacteriome plays distinct yet interconnected roles in allergic rhinitis and asthma.

1. Introduction

Allergic rhinitis and asthma are two of the most common chronic airway diseases in developed countries and a major public health concern [1,2,3]. Estimates indicate that approximately 400 million people worldwide suffer from allergic rhinitis [4]. Similarly, over 300 million individuals have been diagnosed with asthma, corresponding to more than 495 thousand deaths per year [5,6,7,8].

Allergic rhinitis is considered an inflammation of the nasal mucosa, characterized by sneezing, congestion, itching and rhinorrhea [9,10,11,12]. Similarly, asthma is also a condition of the airways characterized by obstruction, inflammation and mucous production [7,13,14]. Allergic rhinitis and asthma frequently coexist [15,16,17,18,19]; approximately 40% of individuals with allergic rhinitis also suffer from asthma, and up to 80% of asthma patients may experience symptoms of allergic rhinitis [20]. This suggests that they may represent a combined airway inflammatory disease with several pathophysiological, epidemiological and clinical connections [15,21,22,23,24]—i.e., the united airway disease concept [24].

Metagenomic analyses of bacterial genomes and 16S rRNA amplicons have shown that the nasal and buccal bacteriomes act as gatekeepers of respiratory health and play a significant role in the onset, development and severity of both allergic rhinitis [25,26,27,28,29,30,31,32] and asthma [31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49]. These same analyses have also demonstrated that the nasal cavity is a major reservoir for opportunistic bacterial pathogens, which can spread to other sections of the respiratory tract and potentially induce respiratory illnesses [26,27,28,29,32,34,38,39,40,41,42,44,50,51,52,53,54,55,56]. Most of these studies, however, have predominantly focused on a single niche or cavity, but it has been suggested that respiratory disease outcomes may be better explained by cross-niche microbial interactions [57,58,59,60,61]—the dynamic interplay between microbial communities across different body sites. Understanding these relationships may substantially increase our comprehension of airway inflammatory disease and our ability to formulate holistic hypotheses about its development, and also help to improve integrated diagnostic and treatment approaches [58,61,62,63].

Here, I have analyzed 16S rRNA high-throughput sequencing data from a cohort of 347 individuals, including those with allergic rhinitis (with and without asthma comorbidity), asthma and healthy controls, to compare their nasal and buccal bacteriomes. My goal was to identify similarities and differences in bacterial taxonomic and functional diversity, as well as community interactions, across health and respiratory disease states.

2. Materials and Methods

2.1. Studied Cohort

All participants in this study were enrolled in the ASMAPORT Project (PTDC/SAU-INF/27953/2017). This study was approved by the “Comissão de Ética para a Saúde” (Parecer_58-17, March 2017) of the Centro Hospitalar Universitário São João, Facultade de Medicina (Porto, Portugal). ASMAPORT was a cross-sectional study of adults and children from northern Portugal created in 2018 to investigate host–microbe interactions during asthma and rhinitis. Further details are provided in Pérez-Losada et al. [31,32].

2.2. Sample Collection and 16S rRNA Amplicon Sequencing

Below I present a short description of the molecular procedures and protocols carried out to sample bacterial communities from the upper airways and extract and sequence their DNA—further details can be found in [31,32]. A total of 347 children and adults (12.6 ± 5.2 years of age and 52.7% female) participated in this study from July 2018 to January 2020. Participants were not taking antibiotics at the time of sampling. They were classified in four clinical groups or phenotypes: allergic rhinitis (AR = 53 individuals), allergic rhinitis with asthma comorbidity (ARAS = 183), asthma (AS = 12) and healthy controls (HC = 99). Sample swabs were collected from the nasal and buccal cavities, rendering a total of 694 samples.

Total DNA was extracted from swabs using the ZymoBIOMICS™ DNA Miniprep Kit D4300. DNA extractions were prepared for sequencing using the Schloss’ MiSeq_WetLab_SOP protocol (09.2015) [64]. Each DNA sample, negative controls and mock samples were amplified for the V4 region (~250 bp) of the 16S rRNA gene and libraries were sequenced in a single run of the Illumina MiSeq sequencing platform at the University of Michigan Medical School. Sequence files and associated metadata and BioSample attributes have been deposited in the NCBI (PRJNA913468).

2.3. Bioinformatic and Statistical Analyses

16S rRNA–V4 amplicon sequence variants (ASV) in each sample were inferred using dada2 version 1.18 [65]. Reads were filtered using standard parameters, with no uncalled bases, maximum of 2 expected errors and truncating reads at a quality score of ≤2. Forward and reverse reads were truncated after 150 bases, merged, and chimeras were identified. Taxonomic assignment was performed against the Silva v138.1 database using the RDP naive Bayesian classifier [66,67]. ASV sequences (226 to 260 bp) were aligned using MAFFT [68] and a phylogenetic tree was built with FastTree [69]. The resulting ASV tables and tree were then imported into the R package phyloseq [70] for further analysis.

Samples were normalized using the negative binomial distribution as recommended by McMurdie and Holmes [71] and implemented in the Bioconductor package DESeq2 [72]. Taxonomic and phylogenetic alpha-diversity were estimated using Chao1 richness and Shannon, Simpson and phylogenetic (Faith’s) diversity (PD) indices. Beta-diversity was estimated using phylogenetic Unifrac (unweighted and weighted), Bray–Curtis and Jaccard distances and dissimilarity between samples was explored using principal coordinates analysis (PCoA).

Significant differences in alpha-diversity indices and microbial abundances (phyla and genera) between nasal and buccal bacteriomes in each clinical group (AR, ARAS, AS and HC) were assessed using the Wilcoxon test. Beta-diversity indices were compared using permutational multivariate analysis of variance (adonis) as implemented in the R package vegan [73]. I applied the Benjamini–Hochberg method at alpha = 0.05 to correct for multiple hypotheses testing [74,75]. All the analyses above were performed in R [76] and RStudio 2024.12.1 [77].

Bacterial function was predicted by coupling Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) [78] and the Integrated Microbial Genomes and Microbiomes (IMG/M) database [79]. ASVs abundances were normalized by 16S rRNA gene copy number, and gene abundances were estimated by multiplying the normalized ASV counts by the predicted gene copy numbers. Metabolic pathways were then estimated using the MetaCyc database [80,81] and PICRUSt2 default parameters. Differentially abundant pathways between nasal and buccal bacteriomes in each phenotypic group (AR, ARAS, AS and HC) were assessed using the DESEq2 Wald’s test with an adjusted p-value cutoff of 0.01.

Microbial interactions were inferred for each naso-buccal bacteriome (AR, ARAS, AS and HC). First, bacterial ASVs were classified as nasal or buccal using the niche indicator species analysis in the R package indicspec (function multipatt) [82]. This function calculates an indicator value for each ASV at each niche (mouth and nose), taking its total abundance per niche into account. An ASV was deemed a niche indicator only when it had a p-value <0.05 for one specific niche. Bacterial networks were then constructed using the SPIEC-EASI (SParse InversE Covariance Estimation for Ecological Association Inference) R package [83]. ASVs not observed more than 1 time in at least 10% of the samples were discarded. Meinshausen–Buhlmann estimation with nlambda = 20, lambda.min.ratio = 0.01 and rep.num = 50 was used [83]. This method is based on the concept of conditional independence rather than correlation, making it less likely to detect spurious connections between taxa that are indirectly connected but not directly connected. Further, it uses centered log-ratio transformation of the data to overcome the compositionality of microbiome data. Keystone taxa (taxa considered key players in the network) were selected using three criteria: degree of centrality (nodes with the highest number of connections), betweenness centrality (nodes that frequently serve as bridges in the network) and closeness centrality (nodes that can quickly reach other nodes). Only nodes in the top 95% percentile for those three indices were considered hub nodes (i.e., keystone taxa). The following properties were also estimated for each network: node degree (number of edges a node has with other nodes within the network), neighborhood connectivity (degree of interconnectivity among immediate neighbors in the network) and cohesion (how tightly linked are the nodes of a network).

3. Results

I analyzed a total of 694 buccal and nasal swabs from 347 participants grouped in four phenotypes: AR = 53, ARAS = 183, AS = 12 and HC = 99 (for more detail see [31,32]). Singletons and six samples with <1771 reads were eliminated. The naso-buccal bacteriomes (688 samples) comprised 13,971,293 clean reads, ranging from 1771 to 82,430 sequences per sample (mean = 20,307.1), and 7363 ASVs. In previous studies [31,32] I described and compared the bacteriomes of those four clinical phenotypes within each cavity (nose and mouth); here, I have assessed similarities and differences in taxonomic and functional diversity and interactions between cavities for each clinical group (AR, ARAS, AS and HC).

3.1. The Nasal and Buccal Bacteriomes Differ in Taxonomic Composition and Diversity

The nasal and buccal bacteriomes shared the same dominant phyla and genera, but in different proportions (Table S1 and Figure 1). More information about their unique composition can be found in [31,32]. Of the five dominant bacterial phyla comprising the naso-buccal bacteriomes, three to five showed significant differences (≤0.015) in their mean relative proportions between cavities across phenotypes (AR, ARAS, AS and HC) (Table S1). Similarly, all of the 22 dominant bacterial genera showed significant differences (≤0.023) in mean relative abundance across phenotypes, except Haemophilus in the AS group (Table S1).

As a whole (all phenotypes combined), the nasal and buccal bacteriomes had 5191 and 1167 unique ASVs, respectively, and shared 1005 ASVs. Then, within each clinical group, AR, ARAS, AS and HC shared 7, 106, 2 and 43, respectively, between cavities (Figure S1).

Alpha-diversity indices [Chao1, Shannon, Simpson and phylogenetic diversity (PD)] of community richness and evenness varied between nasal and buccal bacteriomes for each phenotype (Figure 2 and Table S2). The buccal bacteriomes showed higher diversity than the nasal bacteriomes for all the indices and phenotypes except for PD. All of the alpha-diversity pairwise comparisons between cavities across phenotypes yielded significant results (p < 0.01), except for Chao1 in AR and AS, and PD in AS.

To characterize the structure of the bacteriomes (beta-diversity), I applied principal coordinates analysis (PCoAs) to Unifrac (unweighted and weighted), Bray–Curtis and Jaccard distances matrices. The PCoAs of the samples grouped either by cavity alone (Figure S2) or by cavity and clinical phenotype (Figure 3) separated the buccal and nasal bacteriomes. Group segregation was confirmed by the adonis test, which output significant differences (p < 0.0001) in beta-diversity for all indices and pairwise comparisons performed.

3.2. The Nasal and Buccal Bacteriomes Differ in Functional Diversity

Predicted functional profiles varied greatly between nasal and buccal bacteriomes for the four clinical groups (Table S3). Most distinct pathway [log2FoldChange (FC) > 2 and adjusted p-value < 0.01] tallies in each group were as follows: AR = 79 pathways (32 upregulated and 47 downregulated), ARAS = 93 pathways (56 up and 37 down), AS = 88 pathways (55 up and 33 down) and HC = 92 pathways (55 up and 37 down) (Figures S3–S6). Eight to sixteen of the top pathways above showed log2FC > +/−10 (=1024) in each clinical group. There was considerable overlap in up and downregulated pathways across phenotypic groups, with 17 and 13 of the top pathways shared among the four naso-buccal bacteriomes, respectively. Similarly, the diseased phenotypes alone (AR, ARAS and AS) shared 14 and 6 up- and downregulated top pathways, respectively. Overall, functional profiles in the naso-buccal bacteriomes were dominated by biosynthesis and degradation pathways, followed by fermentation.

3.3. Bacterial Interactions in the Nasal and Buccal Cavities

I inferred bacterial interactions in the naso-buccal bacteriome for each phenotypic group (AR, ARAS, AS and HT) using SPIEC-EASI networks (Figure 4). All nodes (ASVs) in each network were statistically assigned to a cavity (M = mouth, N = nose or NM = undetermined or mixed), as indicated by niche indicator analysis. ASVs in N and M represent specialists strongly associated with each cavity, while ASVs in NM represent generalist occurring in both cavities (low-prevalence ASVs were filtered out) (Table S4).

All naso-buccal networks looked different in their structure, complexity and keystone nodes (Table S4). The AR network included 219 nodes (M = 92, N = 37, NM = 90) and 464 edges, with 46.1% of them connecting intra-niche ASVs (N-N or M-M) and 53.9% connecting inter-niche ASVs (N-M, N-NM, M-NM and NM-NM). The AR network showed a mean node degree of 4.24, a mean neighborhood connectivity of 5.21 and a mean cohesion of 0.12. It included 13 keystone taxa (ASVs) (M = 4, N = 4 and NM = 5) belonging to bacterial genera Lautropia, Fusobacterium, Abiotrophia, Gemella, Staphylococcus, Chryseobacterium, Micrococcus, Enhydrobacter, Veillonella, Neisseria, Prevotella_7, Granulicatella and Haemophilus.

The ARAS network included 231 nodes (M = 11, N = 15, NM = 205) and 374 edges with only 0.5% of them connecting intra-niche ASVs (99.5% inter-niche) and many disconnected nodes. It showed a mean node degree of 3.24, a mean neighborhood connectivity of 5.06 and a mean cohesion of 0.2. The ARAS network included 16 keystone taxa (M = 3, N = 1 and NM = 12) of the genera Corynebacterium, Abiotrophia, Staphylococcus, Porphyromonas, Granulicatella, Finegoldia, Fusobacterium, Lautropia, Dolosigranulum, Fretibacterium, Streptococcus, Gemella and Enhydrobacter.

The AS network included 205 nodes (M = 68, N = 14 and NM = 123) and 374 edges with 21.7% of them connecting intra-niche ASVs (78.3% inter-niche). It showed a mean node degree of 3.65, a mean neighborhood connectivity of 4.2 and a mean cohesion of 0.07. The AS network included 17 keystone taxa (M = 7, N = 0 and NM = 10) of the genera Alloprevotella, Oribacterium, Cardiobacterium, Peptostreptococcus, Fusobacterium, Capnocytophaga, Leptotrichia, Campylobacter, Parvimonas, Treponema, Selenomonadaceae_sp, Prevotella, Fretibacterium, and Streptococcus.

Finally, the HC network included 177 nodes (M = 141, N = 35 and NM = 1) and 458 edges with 93.5% of them connecting intra-niche ASVs (6.5% inter-niche). It showed a mean node degree of 5.18, a mean neighborhood connectivity of 6.75 and a mean cohesion of 0.23. The HC network included 10 keystone taxa (M = 8, N = 2 and NM = 0) of the genera Fusobacterium, Atopobium, Actinomyces, Corynebacterium, Haemophilus, Streptococcus, Gemella, Staphylococcus and Enhydrobacter.

None of the keystone bacterial ASVs was shared across all or just the three networks of diseased individuals (AR, ARAS and AS), but some ASVs were shared across network pairs (Table S4). Similarly, at the genus level, several keystone ASVs were shared across network pairs, but only Fusobacterium was found in the four naso-buccal networks (Table S4).

4. Discussion

Allergic rhinitis and asthma are healthcare burdens that cause major distress worldwide [84,85,86]. The bacteriomes of the nose and mouth have been closely related to the onset and development of both conditions [26,30,31,32,48,53,87,88,89,90,91,92,93,94,95,96]. However, very seldomly have these two cavities been compared in the same cohort to assess their similarities and differences and potential interactions during rhinitis or asthma [96,97,98]. To address this gap, I have here analyzed 16S rRNA sequence data from 347 individuals (694 nasal and buccal samples) with allergic rhinitis (with and without comorbid asthma), asthma and healthy controls.

The nasal and buccal bacteriomes were comprised of the same dominant phyla and genera across all studied phenotypes (AR, ARAS, AS and HC) (Table S1 and Figure 1). All these bacterial taxa are natural residents of the two cavities [25,26,27,29,30,38,40,41,99,100,101,102,103,104,105,106,107], including some pathogenic groups (e.g., Moraxella, Streptococcus, Haemophilus, Neisseria and Staphylococcus) associated with rhinitis, asthma and allergy in both the mouth [31,93,94,95] and nose [32,37,42,48,108]. Relative mean abundances varied greatly between cavities for most dominant phyla and genera (Table S1). Previous comparisons of the nasal and buccal bacteriomes have also reported large differences in bacterial composition during both health [96,97,98,109,110,111,112] and airway disease (allergic rhinitis and asthma) [96,97,98].

The buccal bacteriomes showed higher richness and evenness than the nasal bacteriomes for almost all alpha-diversity indices and phenotypes (Figure 2 and Table S2), with most of the comparisons yielding significant results (p < 0.01). Similar results were observed in previous studies of asthmatic children and adults [95,96,97], but the opposite trend was described in a study of Chinese children with allergic rhinitis [98]. Similarly, when comparing the intra-sample diversity of the buccal and nasal bacteriomes in healthy individuals, several studies have shown that the former have significantly higher alpha-diversity than the latter [96,97,109,110], but some have suggested the opposite [98,110]. Frequent exposure to food and drinks, saliva exchange and high topographical diversity may increase bacterial diversity in the mouth [113,114,115]. The nasal microbiome, however, although usually less diverse due to lower airflow, environmental exposure and nutrient availability, seems more stable due to mucosal immunity influence [116,117,118].

Naso-buccal bacterial communities were structured differently (p < 0.0001) across cavities and clinical phenotypes for all indices compared, with the former being the mayor driver of divergence (Figure S2 and Figure 3). Similar results were observed in previous studies of naso-buccal bacteriomes in both healthy [96,97,98,109,110,111,112] and rhinitic and asthmatic individuals [96,97,98]. As indicated above, ecological and physiological differences between both cavities are likely to shape the structure of these two bacteriomes and drive differences in beta-diversity. Moreover, these results also highlight the potential of beta-diversity indices as indicators of heterogeneity/stochasticity associated with dysbiosis in the naso-buccal bacteriome [31,32,119,120,121,122].

Predicted functional profiles varied greatly between nasal and buccal bacteriomes for all clinical phenotypes with >80% of the pathways tested showing significant differences (p < 0.01) in relative abundance (Table S3). These differences may result from variation in bacterial composition, oxygen availability (the mouth has more anaerobic zones than the nose), nutrient availability (the mouth is exposed to dietary sugars, starches, and protein) or differential host immune pressure [109,123]. Consequently, for example, fermentation pathways are upregulated in the mouth, while fatty acid and amino acid metabolism pathways are upregulated in the nose, as seen in Figures S3–S6. Interestingly, these differences in naso-buccal bacterial regulation remained fairly consistent for many pathways across clinical groups and health status, suggesting again that niche (cavity) composition is the main driver of functional diversity.

Co-occurrence network analyses revealed distinct patterns of connectivity (i.e., interactions) for each naso-buccal bacteriome (Figure 4). The bacterial network of healthy individuals (HC) showed the highest connectivity (i.e., complexity) compared to the networks of diseased individuals (AR, ARAS and AS). This contrasts with previously published intra-niche (nose or mouth) networks [31,32,96], which have shown more complex interactions in the disease groups. Other intra-niche studies, however, have revealed the opposite trend, i.e., more complex networks in asthmatics than in healthy controls [41,124,125,126]. The HC-network showed two very distinct clusters of taxa (nodes) composed almost exclusively of bacterial specialists (N or M). All the other networks included different proportions of specialists and generalists (NM), with the ARAS-network dominated by NM taxa and the AR-network containing a cluster of almost exclusively N nodes. A higher number of NM taxa also confirmed the lower stability and more transient nature of the co-occurrences in networks of disease phenotypes. Concomitantly, most of the interactions (93.5%) in the HC-network involved intra-niche ASVs, while the other three naso-buccal networks showed lower variable rates of intra-niche connections (0.5% to 46.1%). All this evidence, therefore, highlights that loss of ecological specialization in the naso-buccal region (or bacterial translocation via the pharynx) may lead to respiratory disease, as indicated before in previous studies of the oral-nasopharyngeal bacteriome during respiratory infection. It also suggests that the preservation of ecological topography may be important to reduce disease susceptibility [58,127,128,129]. Similarly, network fragmentation, which is indicative of potential dysbiosis and loss of stability, as seen in the ARAS group, could make those individuals more vulnerable to airway disease, as indicated before in respiratory tract infections [58].

The four naso-buccal networks were composed of unique ASV keystone nodes—key players in their respective networks that influence community structure, stability and function (Table S4). The HC-network included more putative commensals and a lower number of pathogenic or opportunistic keystone taxa than the other three networks of diseased participants (Table S4). Some of these pathogens (e.g., Moraxella, Streptococcus, Haemophilus, Neisseria and Staphylococcus) have been associated to rhinitis, asthma and other respiratory diseases [26,27,28,29,31,32,34,38,39,40,41,42,44,50,51,52,53,54,55,56].

Several keystone ASVs of the genera Veillonella, Capnocytophaga, Alloprevotella, Haemophilus, Neisseria, Prevotella, Moraxella, Gemella, Dolosigranulum, Leptotrichia, Porphyromonas, Granulicatella, Actinomyces, Corynebacterium, Fusobacterium, Streptococcus and Staphylococcus also showed a relatively high abundance in the nasal or oral cavities, while others (Abiotrophia, Atopobium, Oribacterium, Cardiobacterium, Peptostreptococcus, Finegoldia, Fretibacterium, Lautropia, Enhydrobacter, Chryseobacterium, Micrococcus, Campylobacter, Parvimonas and Treponema) showed low abundances, despite being highly connected (Table S4). This supports the “rare taxa” hypothesis, which postulates that the abundance of a species is not necessarily the best indicator of its importance within a microbial community [130,131]. Studying the airway microbiome in a more system-centric context may provide further insights into the importance and roles of lesser-known microbes as indicators of disease and potential targets for intervention [58,124,132,133,134]. Further research is still needed to assess the role of microbial networks in the pathogenesis of respiratory illnesses [31,32,87,96,97,98,124].

This study has some limitations. 16S rRNA amplicon sequence data have limited taxonomic resolution (genus or above) and may experience PCR biases [135,136]. Metabolic pathways were predicted from 16S rRNA abundances. This is a cross-sectional study, hence, potential longitudinal variation in bacterial profiles or causality cannot be determined. Similarly, potential confounding factors (antibiotic or probiotic use, clinical severity or allergens) not considered here could also influence the outcomes of the study. Finally, this study focuses on a Portuguese cohort, limiting generalizability to other populations with different environmental exposures and genetics.

5. Conclusions

I compared the nasal and buccal bacteriomes of a cohort of 347 individuals with allergic rhinitis (with and without comorbid asthma), asthma and healthy controls. Bacterial communities differed significantly in taxonomic composition and structure and functionality across all clinical groups. Naso-buccal bacterial networks also displayed remarkable differences in connectivity and keystone taxa. Notably, the networks of diseased individuals showed less complexity and ecological specialization and more pathogenic hub nodes. Moreover, multiple keystone taxa of variable relative abundance were detected in each network. Altogether, these findings demonstrate that the naso-buccal bacteriome plays distinct yet interconnected roles in allergic rhinitis and asthma. They also confirm that, despite the pathophysiological and clinical similarities between allergic rhinitis and asthma (the united airway disease concept [15]), these two conditions may represent two distinct diseases with different allergen sensitization and onset [137], severity [138] and treatment response [139], as suggested by this and other omic studies [31,32,121,122,140,141]. Further research is needed to explore the role of the naso-buccal bacteriome in chronic inflammation, particularly in allergic individuals.