Background
Recurrent aphthous stomatitis (RAS) is the most common form of oral ulcerative disease that affects as much as 5–20% of the population. It is characterized by shallow round ulcers that afflicts pain on the patients [
1]. These lesions are benign and self-limiting, but are usually chronic and frequently recur [
2], leading to difficulties in oral functions [
3].
The etiology of RAS is still unclear, although association between RAS and a number of factors have been reported. These factors include local trauma [
4], saliva composition changes [
5], a series of systematic diseases such as HIV infection [
6] and Crohn’s disease [
7], genetic factors [
8], food allergy [
9], immunological factors [
10], stress [
11], nutritional deficiency [
12], and microbial agents [
13]. The lack of clear understanding of the etiology of RAS hinders the efficient treatment of this disease.
The role of several bacterial species in RAS has been implicated in previous investigations by culture-dependent techniques.
Helicobacter pylori has been found on RAS lesions [
14], and association between
H. pylori and RAS has been suggested [
13], although this association has been controversial [
15]. Several
Streptococcus species have been suspected to be involved in the development of RAS [
16], and this involvement was suggested to be the result of autoimmune reaction of streptococcal heat-shock proteins [
17]. Despite these investigations, no definitive connection between microbial infection and RAS has been demonstrated.
The emergence of high-throughput sequencing and various other high-throughput microbial techniques allowed in-depth culture-independent analysis of microbial colonization, and has been proven successful in detecting key pathogens for various diseases [
18]. To date, several studies have been performed in attempt to understand the bacterial community composition in RAS-affected patients. Marchini et al. compared the microbiomes of 10 healthy and 10 RAS-affected subjects using 16S rDNA library-dependent cloning techniques, and found different microbiome structures [
19]. Bankvall et al. compared the microbiomes of 60 healthy and 60 RAS-affected patients using Terminal-Restriction Fragment Length Polymorphism (T-RFLP) of 16S rDNA amplicons, and found differences in T-RFLP patterns, but were unable to pinpoint the key pathogens involved in RAS [
20]. Seoudi et al. compared the saliva microbiomes of 26 healthy subjects and 8 RAS-affected patients using human oral microbe identification microarray analysis, and found decreased levels of
Rothia,
Neisseria, and
Veillonella in RAS-affected patients [
21]. Kim et al. compared the microbiomes of oral mucosa (
n = 18) and saliva (
n = 7) of RAS-affected patients with healthy subjects (n = 18) using 454 pyrosequencing of 16S rDNA, and found the association of the decrease of
Streptococcus salivarius and the increase of
Acinetobacter johnsonii with RAS risk [
22]. Hijazi et al. performed 454 pyrosequencing of 16S rDNA from 18 RAS-affected patients and 17 healthy subjects, and found higher levels of Bacteroidales, Porphyromonadaceae and Veillonellaceae, along with decreased Streptococcaceae in association with RAS [
23]. These investigations have a small sample size and cannot identify a significant difference (clustering) in the overall bacterial community between healthy and disease-affected samples, and have a lower sequencing depth (read numbers) and therefore could lead to missing information. Therefore, a more detailed comparison of microbiota between healthy and RAS-affected subjects is warranted.
In this work, we performed an in-depth analysis and comparison of the microbiomes between healthy mucosa and RAS lesions from 24 RAS-affected patients by high throughput Illumina sequencing of 16S rDNA, with an average sequence depth of 68,633 reads per sample. Suggestions on the association of specific bacteria with RAS are made that require further mechanistic investigations for the confirmation of etiology.
Discussion
Understanding the etiology of RAS is a big step forward in finding effective cures for this common disease, and it has been suspected that microbial infection contributes to RAS [
13,
16]. Recent progress in high throughput sequencing techniques enables metagenomic approaches in understanding the microbiomes of biological samples, therefore allows us to pinpoint the specific pathogen responsible for diseases by comparing the microbiomes of pathological and normal tissues. Therefore, we exploited high-throughput sequencing technologies in this work in attempt to find specific association of bacterial community compositions with aphthous ulcers, which further leads to proposals of the etiology of RAS.
In this work, we found that the increase of
E. coli and
Alloprevotella, as well as the decrease of
Streptococcus in bacterial communities is significantly associated with aphthous ulcers. The decrease of
Streptococcus in aphthous ulcers is in agreement with previous findings [
22,
23]. However, the increase of
E. coli in aphthous ulcers is a new and particularly intriguing finding.
E. coli is one of the most common bacteria in the human microbiome, particularly intestinal microbiome [
24]. Inoculation of
E. coli to oral mucosa is easy and common via the fecal-oral pathway. Considering 40% of the human population suffers from RAS, the cause of this disease has to be a common factor. This common occurrence is in coincidence with
E. coli colonization which is also a very common phenomenon. Therefore, the significant association of aphthous ulcers with
E. coli abundance leads to the proposal that
E. coli colonization could be the cause of RAS. Previous investigations suggested that
Helicobacter pylori could be the cause of RAS, but results from this work do not suggest a significant correlation between aphthous ulcers and
H. pylori (
p = 0.185). Therefore, we doubt that
H. pylori has a direct role in the formation of aphthous ulcers, in agreement with the previous suggestion that
H. pylori does not play a role in RAS [
15].
A number of previous studies investigated the microbiota of RAS [
18‐
23]. These investigations are mostly qualitative rather than quantitative, and cannot lead to the identification of significantly enriched groups in the bacterial community of aphthous ulcers. Three previous studies quantified the microbial abundance of bacteria in aphthous ulcers using microarray or pyrosequencing approaches [
21‐
23]. These investigations either compared the oral bacterial community composition of healthy subject and patients and therefore suffered from background noise caused by differences between individuals [
22], involved saliva microbiome which could naturally have different bacterial community composition with the mucosa as saliva is a natural disinfectant [
21,
22], or has a relatively small sample size (
n = 8 or 12) [
21,
23]. In particular, the two investigations with pyrosequencing only had respectively 3000 and 9500 tags/sample [
22,
23], which could lead to loss of information due to lower sequencing depth and smaller sample volume. The methods used in this work ruled out differences between individuals by comparing the normal oral mucosa and aphthous ulcers of the same individual, had a larger sample volume (
n = 24), and had a better sequencing depth (> 68, 000 tags/sample). Therefore, we are able to more effectively detect bacterial groups specific to aphthous ulcers in this work. It needs to be noted that samples were taken from only RAS patients intentionally without collecting samples from healthy individuals as controls, because it was decided that individual diversity may contribute significantly towards differences in bacterial community leading to difficulties in finding bacteria that are associated with RAS. Including healthy individuals will only complicate the study rather than help it. Also, not having healthy controls does not weaken the findings of this work as this work aims to find RAS-associated, localized, ulcerative mucosa-bearing microbes, and a proper control is the healthy mucosa of the same individual. A large number of taxa were found differently represented in oral mucosa and aphthous ulcers (Additional file
2). With more stringent statistical analysis like LEfSe, we are capable of identifying
E. coli and
Alloprevotella as the bacterial groups specific to aphthous ulcers, which was never observed before. We also confirmed previous findings that the reduction of
Streptococcus (Streptococcaceae) and
Rothia is associated with aphthous ulcers [
22,
23], while the increase of Porphyromonadaceae is associated with aphthous ulcers [
23]. Previous reports on the positive association of
Acinetobacter and Bacteroidales with aphthous ulcers, as well as the negative association of
Neisseria with aphthous ulcers were not confirmed by our results [
21‐
23]. The role of Veillonellaceace on aphthous ulcers was controversial [
21,
23], and our results couldn’t suggest a significant correlation between this group of bacteria with RAS.
The work we performed suggested that the colonization of E. coli or Alloprevotella, more likely the former, may be the cause of RAS. However, it needs to be pointed out that this suggestion is not conclusive, as finding an association is not equivalent to finding the causality. We cannot rule out the possibility that RAS leads to increased abundance of E. coli and Alloprevotella, in contrary to our hypothesis that increased abundance of E. coli and/or Alloprevotella leads to RAS. Furthermore, consideration on other possible complications influencing oral environment and bacterial community structures, such as other underlying conditions and drug use, was not included in this investigation, which was due to the assumption that they are not major drivers of the oral microbiomes and these effects may be minimized by stringent statistics. A much larger surveillance is still needed to identify detailed factors influencing oral microbiomes. Nevertheless, the findings of this work, in particularly the coincidence that E. coli colonization and RAS occurence are both common, points to a high possibility to the etiology of RAS. Further in-depth pathological work is needed to confirm this possibility. These findings have the potentials to guide the discovery of new cures for RAS, which may include targeting oral E. coli colonization and removing it using antibiotics.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.