Characterization of the microbiomic association between tonsil and saliva isolated from Korean children subjected to tonsillectomy using 16S rRNA sequencing: a pilot study

Oral microbes have the capacity to spread throughout the gastrointestinal system and are significantly associated with multiple diseases. Given that tonsils are located between the oral cavity and laryngoesophagus at the gateway of both alimentary and respiratory tracts, tonsillar tissue may also be affected by both oral (saliva) and alimentary tract microbiota. While several independent reports on oral microbiomes from saliva and tonsils are documented in the literature, limited studies have compared tonsil and saliva microbiota compositions. Here, we analyzed the distribution of the microbial communities in saliva and tonsils of healthy Korean children subjected to tonsillectomy. revealed that many bacterial communities are shared and show similarities in terms of diversity and composition, suggesting close interactions between the two microbial groups. Although further studies are clearly essential, we assume that the oral microbiome exerts significant effects on not only the tonsil itself but also tonsil-derived immune or stem cells through regulating the microbial community. Our study serves as a cornerstone of tonsillar microbiome research and the data obtained should facilitate the development of novel microbiome-based strategies to regulate tonsil-derived therapeutic platforms, such as immune or stem cell therapy.

microbiome. Based on the collective results, we propose that the identified microbes act as both intracellular and external environmental factors with a major impact on the functioning of tonsils and their primary cells.

Background
Human tonsils are lymphoid epithelial tissues of oral mucosa around the oropharynx and nasopharynx that protect the body from pathogen invasion through the mouth and nose. Tonsils are a component of immune system that develop from 2 years of age and grow until puberty, and gradually undergo atrophy thereafter. More frequently in children than in adults, the tonsils can become abnormally enlarged (adenotonsillar hyperplasia) or inflamed (tonsillitis), which can be surgically removed by tonsillectomy. Abnormally enlarged tonsillar tissue of pediatric patients is generally caused by hypertrophy rather than inflammation. Hyperplasia of tonsils is a cause of obstructive sleep apnea and recurrent infection of the upper airways. The most common pediatric operative procedure for tonsillar hyperplasia is tonsillectomy [1].
Oral bacteria travel throughout the body and are significantly associated with human diseases. Approximately 8 × 10 11 bacterial cells per day flow from the mouth to the alimentary tract, and several studies have reported associations of oral microbes between these regions [2][3][4]. Olsen et al. [5] reported that orally ingested Porphyromonas gingivalis affects intestinal microflora and a variety of oral microbes reach the gut microbiota, but only a subset of these bacteria colonize the gut under dysbiotic conditions. Microbiome analysis at 18 body sites of 200 individuals by Huse et al. [6] showed that core operational taxonomic units (OTUs) shared by more than 95% subjects were the most abundant in the oral cavity relative to the rest of the body.
Significant quantities of pathogenic bacteria are present in hyperplastic tonsils and adenoids, compared to their normal controls [7][8][9][10][11]. An investigation by Kielmovitch et al. [12] on the tonsillar surface and core of children with recurrent tonsillitis and tonsillar hypertrophy identified Haemophilus influenzae and Bacteroides melaninogenicus as the most prevalent bacteria. Jeong et al. [13] examined the differences between bacterial pathogen constituents of recurrent tonsillitis and tonsillar hypertrophy using the cultivation method in relation to age, season, and antibiotic sensitivity in patients subjected to tonsillectomy. Haemophilus influenzae, Streptococcus pyogenes, Staphylococcus aureus, and Streptococcus pneumoniae were isolated as representative abundant microbes in tonsils.
Until recently, microbiomes have been analyzed using cultivation and PCR methods. Brook et al. [14] performed a microbiological study on the core of tonsils from 50 children with recurrent tonsillitis using the cultivation method. The predominant isolated microbes were Staphylococcus aureus, Moraxella catarrhalis, Peptostreptococcus, pigmented Prevotella and Porphyromonas, and Fusobacterium species. However, culture-based studies present a potentially biased assessment of microbial diversity due to targeting of limited bacterial groups. More recently, 16S rRNA gene-based pyrosequencing and culture-independent techniques have been applied to survey bacterial communities.
Given that tonsils are located between the oral cavity and laryngoesophagus at the gateway of alimentary and respiratory tracts, the tonsil lymphoid system with antigen capture M cells similar to those of Peyer's patches of the gut is among the first handling sites for microbial agents and antigens in the human body. The microbial community of saliva can therefore affect that of tonsils and vice versa. To date, oral microbiome studies have mainly targeted saliva samples. Moreover, the majority of tonsil microbiota studies have focused on acute inflammatory disease and not normal conditions.
Since deeply branched crypts of the tonsillar epithelium provide optimal spaces for microbiota colonization, it is speculated that the tonsil microbiome has distinct properties from other oral microbiomes. However, no studies have examined this theory and further investigation of the correlation between the microbiomes of saliva and tonsils is warranted.
Here, we focused on the distribution and correlations of microbiota in saliva and tonsillar tissues of young tonsillectomy patients based on evaluation of the V3-V5 region of 16S rRNA genes to examine the hypothesis that the microbiome is associated with tonsillar hyperplasia in children. To our knowledge, this is the first comparative analysis of microbiomic portraits in tonsils and saliva of children subjected to tonsillectomy in South Korea. Elucidation of the microbiome constituents of saliva and tonsils in pediatric cases with tonsillar hypertrophy should aid in the treatment of oral Results General characteristics and biochemical indices of subjects A total of 29 healthy participants (23 boys and 6 girls) who had undergone tonsillectomy were enrolled. Table 1 presents the general and biochemical characteristics of the participants (left panel) along with the corresponding reference values for Korean children (right panel) [15,16]  The microbiome compositions of saliva and tonsillar tissues were determined based on average relative abundance assigned to the phylum and genus levels (Fig. 1). The saliva group showed significantly higher abundance of Firmicutes (45%), followed by Proteobacteria (23%), Bacteroidetes (15%), Actinobacteria (10%) and Fusobacteria (4%) at the phylum level, while the tonsil group contained more Proteobacteria (31%), followed by Bacteroidetes (22%), Firmicutes (21%), Proteobacteria (20%) and Actinobacteria (2%). At the genus level, Streptococcus (13.4%) was the predominant taxon in the saliva group and Haemophilus (23.5%) in the tonsil group.

Comparison Of Alpha Diversities
We further assessed alpha diversity within each group indicative of species abundance. The alpha diversities of saliva and tonsils were measured based on the Chao1 index and species richness compared. Rarefaction curves for the sequences per sample are shown in Fig. 2a. Chao 1 and Shannon indices were employed to calculate species diversity of samples between saliva and tonsillar groups (Fig. 2b). The Chao1 index values representing microbiome richness in saliva and tonsils showed no significant differences (p = 0.7). On the other hand, the Shannon index value representing richness and evenness of the microbial community in saliva was significantly higher than that in tonsil tissue (p < 0.001). Our findings suggest that the bacterial community in saliva is more evenly distributed, compared to that in tonsils.

Principal-coordinates Analysis (pcoa) Biplots
Microbe communities from saliva and tonsillar tissues were discriminated based on principal coordinates analysis (PCoA). PCoA was used to compare community phylogenetic compositions and revealed microbial diversity between the groups. Based on PCo2 scores, tonsillar tissues were clearly distinguishable from saliva samples (Fig. 3a). A Venn diagram illustrated 334 genera in the saliva group and 168 in the tonsil group (Fig. 3b). Detailed information on common genera between the groups (saliva-specific and tonsil-specific) is listed in Additional file 1. In total, 147 common genera were identified in both sample groups, corresponding to 44.1% and 87.5% of the 334 and 187 microbial genera determined in saliva and tonsils, respectively. Overall, 87.5% of the tonsil microbiome was identical to the oral saliva microbiome at the genus level, clearly signifying high similarities of the microbial communities of the two groups.
To determine bacterial distribution from the Venn diagram, a bar graph of all genera at the phylum level was generated (Fig. 3c).

Correlation Plots Of Individual Microbes In Saliva And Tonsils
Spearman correlation analysis was applied to compare microbiomes within saliva and tonsil tissues as well as between saliva and tonsils for the top 10 ranked taxa (Fig. 4). In general, the top 10 ranked microbiota within saliva showed high positive correlations with each other (Fig. 4a) while the top 10 ranked microbes within tonsils displayed both positive and negative correlations (Fig. 4b). Haemophilus

Taxonomic-level Comparison Of Microbiota In Saliva And Tonsils
Linear discriminant analysis (LDA) effect size (LEfSe) analysis was used to identify significantly different saliva and tonsil microbiota compositions for the top 10 ranked taxa (Fig. 5).

Discussion
To our knowledge, this is the first study to compare the core microbiomes of saliva and tonsils in Korean pediatric patients without tonsillitis. By employing 16 s rRNA gene sequencing, we determined the relative abundance of the microbial community in saliva and tonsils, with a view to establishing the association between the oral saliva microbial community and tonsillar microbiome profile in Korean children (Fig. 6). Our data suggest that the microbiome profiles of saliva and tonsils are largely similar (87.5%), suggestive of interactions between their microbial components. To date, most studies on pediatric tonsils have reported the tonsil microbiomes under inflammatory conditions such as tonsillitis. Our study is distinct from these previous studies, in that we examined saliva and tonsil microbiomes in the absence of tonsillitis. Nevertheless, our study could not elucidate the microbiome of perfectly healthy tonsils, because the tonsil microbiota samples used in this study were obtained from patients diagnosed with tonsillar hypertrophy. However, because children with normal tonsils do not need to undergo tonsillectomy, it is almost impossible to obtain perfectly normal tonsil tissues.
Since tonsils have emerged as an good source of adult stem cells in tissue engineering and regenerative medicine with the added advantages of noninvasiveness of tissue collection [17], relatively high proliferation rate and low allogenicity, our research may facilitate the development of microbiome-based tools to regulate tonsilderived therapeutic platforms. Further studies on larger groups are warranted to confirm initial findings.
Recent advances in sequencing technology and metagenomics have expanded our knowledge of the composition and association of the oral microbiome with human health and disease. The oral cavity microbiome has generally been examined by collecting oral rinse samples, including saliva, but it remains to be established whether saliva provides an accurate representation of the microbiome of the oral cavity [18]. Given that the human oral cavity is composed of various subsites, including teeth, gingival sulcus, tongue, hard and soft palates, and tonsils, which provide appropriate space for colonization but slightly different environments for microorganisms, the salivary microbiome profile may not be correlated with those of other subsites of the oral cavity.
Although a number of differences in the microbial profiles between recurrent tonsillitis and tonsillar hypertrophy pediatric groups have been determined by Jeong et al. [13] using culture-based analysis, these experiments were limited in that they only included aerobic bacteria existing in the tonsillar core and many human associated microorganisms are not cultivable under laboratory settings. Accordingly, we aimed to elucidate and compare the core microbiome profiles from paired saliva-tonsil samples using metagenome-wide analysis.
We obtained an average of 12,745 raw read counts, which was markedly lower than the 80,829 raw read count average reported by Jensen et al. [19]. However, Caporaso and co-workers [20] demonstrated that 2,000 reads are sufficient to resolve relationships among samples observed with the full dataset. Accordingly, we considered our data adequate to determine microbiome correlations between saliva and tonsils. Our dataset identified from 12,745 reads included 1,678 OTUs in saliva and 1,461 OTUs in tonsils, which were reasonable counts for further analysis.
Alpha-diversity indices, including Chao1 and Shannon-Weaver diversity indicies, are generally used to compare the differences between microbial communities [21][22][23]. Both saliva and tonsils showed similar richness in microbial communities but the saliva microbiome presented higher evenness than tonsil tissue. The microbiome profile of saliva appears more stable relative to that of tonsils. Saliva is constantly secreted from salivary glands into the oral cavity containing a diverse bacterial population while the microbial community in tonsils is mainly colonized in the recessed epithelium of deeply branched crypts [18,24]. Therefore, saliva continuously secreted from various oral sites contains diverse bacteria but the range in tonsils is lower since the microbial populations stagnate in the crypts and remain limited.
Haemophilus and Staphylococcus are the representative microbiota in both tonsillectomized and nontonsillectomized children with recurrent and tonsillar hypertrophy [27,28]. Haemophilus was identified as the most common bacterial species in both saliva and tonsils in our study (Table 3). However, Staphylococcus, which is frequently detected in tonsillitis, was less abundant in our study and weakly correlated between the groups (data not shown). Specifically, Staphylococcus was ranked as 22 in terms of abundance in tonsils and 38 in saliva.
The low abundance of Staphylococcus may be explained by the fact that the bacterium is mainly associated with inflammation and we selected only hyperplasia cases for analysis (except tonsillitis).
According to Kuhn et al. [10], the predominant aerobic and facultative organisms were Haemophilus influenzae, followed by Neisseria and Staphylococcus aureus, in pediatric patients subjected to tonsillectomy, while the predominant anaerobic bacteria were Fusobacterium, Bacteroides and Prevotella melaninogenica. These earlier results were identical to our finding that Haemophilus, Fusobacterium, Prevotella, and Neisseria are among the top 10 ranked genera in tonsil and saliva samples of Korean children (Table 3). Among the top 10 microbiota identified in the current study, the dominant bacteria in both saliva and tonsils were facultative anaerobes.
The top 10 ranked microbes in our study mostly showed a strong positive correlation within the saliva microbiome while both positive and negative correlations were detected within the tonsil microbiome. These differences could be explained as follows: since saliva is continuously secreted and renewed in the oral cavity, the microbial community is relatively constant. Compared with saliva, microbiota in tonsils are exposed to an environment with insufficient supply of oxygen and nutrients due to stagnation in the crypt, potentially leading to competition between two microbes and, consequently, negative correlation.
Treponema 2 showed the most significant positive correlation between saliva and tonsil environments (Fig. 4), although its abundance was low (1.3% in saliva, 3.2% in tonsils; Table 3). The genus Treponema contains both pathogenic (treponematoses) and non-pathogenic species. Non-pathogenic treponemes form part of the normal microbial flora of the oral cavity, intestinal tract or genital tract. A number of oral treponemes have been associated with gingivitis and periodontal disease [30]. Prevotella 7 in tonsils was positively correlated with Streptococcus, Veillonella, and Alloprevotella in saliva. Prevotella 7 is one of the major periodontal diseasecausing microbiota, such as periodontitis and gingivitis. Furthermore, Streptococcus in saliva, which affects Prevotella 7 in tonsils, triggers airway infections, such as pharyngitis, tonsillitis, and tympanitis. This strong positive correlation supports a link between dental inflammatory and oropharyngeal inflammatory diseases, further indicating a significant connection between the oral cavity (saliva) and oropharynx (tonsil).
Based on the collective results, we have expanded our understanding of the interactions between microbiomes of saliva and tonsils. Several factors are responsible for loss of microbial diversity and homeostatic function, including inflammation, diet, xenobiotics, and altered host cell function [31]. Given its critical roles in multiple diseases, the microbiome has become an extremely attractive target for therapeutic interventions [32]. One of the biggest hurdles in microbiome research is identification of cause-effect relationships and designing of microbiome-based therapies that are able to achieve predictable outcomes on the microbial community and host cell function. Further studies are required to develop strategies aimed at modulating the microbiome profile for improvement of host function and health.

Conclusions
Analysis of the microbiomes of paired tonsil and saliva samples from Korean pediatric patients diagnosed with hyperplasia without tonsillitis revealed that many bacterial communities are shared and show similarities in terms of diversity and composition, suggesting close interactions between the two microbial groups. Although further studies are clearly essential, we assume that the oral microbiome exerts significant effects on not only the tonsil itself but also tonsil-derived immune or stem cells through regulating the microbial community. Our study serves as a cornerstone of tonsillar microbiome research and the data obtained should facilitate the development of novel microbiome-based strategies to regulate tonsil-derived therapeutic platforms, such as immune or stem cell therapy. Oral examination and sample collection were conducted at the Department of Otolaryngology-Head and Neck Surgery of CNUH. All participants remained fasted after midnight the day before until completion of surgery according to the regular preoperative protocol.

Methods
Saliva was collected from mouths of patients on the morning of surgery. Prior to sample collection, participants were instructed not to brush their teeth from the previous night to the time of sampling and prohibited from eating or drinking for at least 9 hours. In brief, participants were asked to chew on sterile gauze (4 × 4 inch) for 1 minute to stimulate salivation before placing the gauze with absorbed saliva back into a sterile Eppendorf conical tube (50 ml). Tubes were centrifuged for 15 minutes at 1,763 g with a 100 µm cell strainer and used to collect clear saliva samples that were (immediately) stored at − 70 °C until DNA extraction.
After general anesthesia, en-bloc resection of both palatine tonsils via the peroral approach was performed according to the traditional tonsillectomy procedure using a cold knife and electrocautery. About half the tissue around the upper pole of palatine tonsils was sent to a pathological laboratory for clinical diagnosis and the remainder used for microbiome analysis.

Anthropometric Measurements And Serum Biochemical Indices Of Subjects
We collected anthropometric data from all subjects, including height and weight of all patients, which were measured the day before surgery. Body mass index (BMI) was calculated according to a previous report [33]. In addition, with reference to the 2017 Korean National Growth Chart, subjects were divided into two groups based on BMI 85th percentile to determine obesity status. primers. Libraries were prepared using PCR products and quantified with the aid of the QIAxpert kit (QIAGEN).

Analysis Of Bacterial Compositions Of Microbiomes
Paired-end reads matching the adapter sequences were trimmed using cutadapt (version 1.1.6) [34]. The resulting FASTQ files containing paired-end reads were merged with CASPER and quality-filtered with Phred (Q) score of 20 [35]. Any reads < 350 bp or > 550 bp after merging were additionally excluded. To identify the chimeric sequences, a reference-based chimera detection step was conducted with VSEARCH against the SILVA gold database [36,37]. Next, sequence reads were clustered into OTUs using VSEARCH with a de novo clustering algorithm under 97% sequence similarity. The representative sequences of OTUs were finally classified using SILVA 128 database with UCLUST (script on QIIME version 1.9.1) under default parameters [38].
interest to disclose.         LDA effect size (LEfSe) analysis of tonsils compared to saliva for the top 10 ranked genera. LEfSe analysis used to select microbiota significantly more abundant in one group, while the bar size represents the size of the effect. Green bars represent microbiota associated with saliva while red bars represent tonsil-associated microbiota. LDA effect size (LEfSe) analysis of tonsils compared to saliva for the top 10 ranked genera. LEfSe analysis used to select microbiota significantly more abundant in one group, while the bar size represents the size of the effect. Green bars represent microbiota associated with saliva while red bars represent tonsil-associated microbiota.

Figure 6
Overview of the experimental framework.

Figure 6
Overview of the experimental framework.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.