Background
The human oral microbiome has been relatively well characterized [
1‐
4], and ongoing efforts are focused on studying compositional and spatial organization of oral communities as well as similarities and differences across populations [
5]. Another widely appreciated aspect is the role of oral microbial communities in oral and systemic diseases [
6,
7]. One common condition- the inflammation of tonsils (termed tonsillitis), is also regarded to have a microbial aetiology. Tonsils are part of the lymphatic system and can become inflamed when infected by bacteria or viruses. While tonsillitis is usually a self-limiting disease, abscesses can form in some cases and require surgical intervention to drain accumulated pus. Culture-based characterization of microorganisms in tonsil tissues of patients with tonsillitis have reported a polymicrobial association commonly involving group A streptococci,
Staphylococcus aureus,
Streptococcus pneumoniae,
Haemophilus influenzae, and members of the
Prevotella,
Bacteroides,
Fusobacterium,
Porphyromonas and
Veillonella genera [
8‐
10]. Since many human-associated microorganisms are not cultivable under laboratory conditions, recent studies have applied culture-independent DNA sequencing methods to identify the kinds of microorganisms present and quantify their abundances in the human microbiome [
11]. By sequencing the microbial small subunit ribosomal RNA gene (16S), a study of microorganisms enriched in tonsillar tissue from adult subjects with recurrent tonsillitis identified enrichment of members of the
Treponema,
Fusobacterium,
Streptococcus,
Selenomonas,
Gemella,
Tannerella and
Prevotella genera compared to healthy individuals [
12].
While it has been reported that tonsil tissue-associated microbial communities are altered in subjects with tonsillitis, there are no studies describing whether the total oral cavity microbiome is also influenced by this condition. The oral cavity microbiome is commonly studied by collecting oral rinse samples [
11,
13,
14], and its composition has been shown to correlate with diseases such as oral cavity and oropharyngeal cancers [
13,
15]. Here, we collected oral rinse samples from patients showing symptoms of acute tonsillitis and compared their oral cavity microbial community composition to healthy individuals without oral disease. We hypothesized that the oral microbiome composition in tonsillitis patients differed from healthy individuals, and that disease state (i.e. tonsillitis vs no disease) was the primary factor attributable to differences in community composition. We also investigated the influence of prescription antibiotics, the presence of oral abscesses and smoking habits on oral rinse community compositions.
Discussion
Tonsillitis is traditionally associated with an overgrowth of bacterial taxa such as
Bacteroides,
Fusobacterium,
Veillonella,
Prevotella,
Streptococcus,
Staphylococcus and
Haemophilus based on laboratory cultures of tonsillar tissue and material from abscesses [
8,
22]. In addition to supporting these observations, culture-independent microbial community surveys based on DNA sequencing have also uncovered additional associations with other taxa such as
Treponema,
Fusobacterium, Gemella and
Tannerella [
12]. Our data extended these findings to a cohort of tonsillitis patients in Hong Kong in which we observed a subtle but significant difference in oral rinse microbial community composition compared to healthy individuals, in line with previous reports that composition of microbial communities in oral rinse samples is predictive of oral disease [
13]. An even smaller effect of gender and age was identified, although these two factors were only significantly associated with oral community composition in healthy subjects when the tonsillitis and healthy cohorts were analysed separately. We postulate that the lack of association in tonsillitis subjects could be due to a larger influence of tonsillitis overriding effects linked to age and/or gender, and also partly because of reduced power in detecting age/gender differences in community composition in the smaller tonsillitis cohort compared to healthy individuals (Table
1).
Although compositional differences in oral community composition linked to tonsillitis were subtle as indicated by PERMANOVA and the lack of statistical difference in alpha diversity indexes between cohorts, five of the top 10 ESVs associated with tonsillitis consistently classified as
Prevotella suggested that multiple members of this genus may be involved with oral disease. Indeed,
Prevotella sequence variants have been implicated in various oral conditions such as periodontitis [
23‐
25] and endodontic abscesses [
26], and are possibly enriched through their preferential utilization of proteins and production of cytotoxic end products [
27,
28]. We also identified several
Streptococcus,
Veillonella,
Lactobacillus and
Atopobium variants associated with the tonsillitis cohort. While these tonsillitis-associated taxa were mostly distinct from those associated with healthy samples,
Streptococcus ESVs were linked to both tonsillitis and healthy cohorts (Fig.
2). A limitation of using 16S amplicon sequencing is the insufficient resolution in classifying species level differences [
29,
30], which is compounded by microbial genera in oral communities encompassing multiple species and/or polyphyly in their phylogenetic classifications [
3,
31]. Ideally, full-length 16S sequences should be obtained to confirm exact species identities of any implicated microorganisms.
While the main aim of this study was to describe differences in oral community composition between patients with tonsillitis and healthy individuals, we also observed secondary differences in community composition linked to smoking in both healthy and tonsillitis cohorts. Firstly, we wish to point out that smokers and former smokers were underrepresented in numbers compared to non-smokers (29 smokers, 19 former smokers and 160 non-smokers in total). Nevertheless, we identified associations of multiple
Fusobacterium ESVs with smokers and
Neisseria with non-smokers in both the tonsillitis (Fig.
3) and healthy cohorts (Additional file
1: Figure S3) consistent with other studies comparing oral community composition between smokers and non-smokers. Previous oral community surveys have consistently reported reductions of
Neisseria detected in mucosal surfaces [
32], oropharynx [
33] and oral rinses [
34‐
36] of smokers compared to non-smokers. Conversely, taxa such as
Veillonella,
Actinomyces and
Fusobacterium have been found to be increased in the oral cavity of smokers compared to non-smokers, although contrasting results exist [
33‐
36] possibly due to differences in biogeography of samples and technical limitations of the 16S gene in delineating microbial species (as mentioned above). In addition, we identified association of a
Bergeyella and a member of candidate phylum SR1 with non-smokers, which was also found in non-smokers from a New York City cohort [
37]. One notion of how smoking influences the oral microbial community composition is that cigarette smoke creates anaerobic conditions favouring anaerobic microorganisms such as
Veillonella,
Fusobacterium and
Actinomyces, and suppresses aerobes such as
Neisseria [
32,
35,
37]. This inference is consistent with our data showing associations of other anaerobic or microaerophilic taxa such as
Fusobacterium,
Leptotrichia,
Porphyromonas,
Lactobacillus and
Treponema in smokers from the tonsillitis and healthy cohorts, although whether they are linked to a predisposition to disease in smokers remains an open question. Most notably,
Fusobacterium,
Leptotrichia and
Rothia in the oral cavity have been associated with squamous cell carcinoma [
38] and oral leukoplakia [
39], and thus their association with smokers suggests that smoking habits may exacerbate oral cancers through the enrichment and/or activity of these microorganisms. However, as mentioned earlier, 16S amplicon sequencing does not provide sufficient resolution in identifying species level associations. Additional studies are required to validate whether the microbial taxa implicated here indeed have oncogenic potential, and play a role in the link between smoking, microorganisms and oral cancer [
40].
Although it is widely accepted that antibiotics exert a strong influence on the composition of human microbiomes, we detected no statistical difference in the oral rinse communities of patients provided with systemic antibiotics compared to those without. We wish to clarify that we likely saw no measurable effects of antibiotics consumption on the oral microbiome in this cohort because the use of antibiotics among subjects was highly variable (15 of 43 subjects received antibiotics; several classes of antibiotics) and not controlled as these were prescribed by general practitioners prior to subjects’ admission into the hospital. Therefore, findings relating to the consumption of antibiotics should be interpreted with caution.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.