Inter-site and interpersonal diversity of salivary and tongue microbiomes, and the effect of oral care tablets [version 1; peer review: 3 approved]

Background: Oral microbiota has been linked to both health and disease. Specifically, tongue-coating microbiota has been implicated in aspiration pneumonia and halitosis. Approaches altering one's oral microbiota have the potential to improve oral health and prevent diseases. Methods: Here, we designed a study that allows simultaneous monitoring of the salivary and tongue microbiomes during an intervention on the oral microbiota. We applied this study design to evaluate the effect of single-day use of oral care tablets on the oral microbiome of 10 healthy individuals. Tablets with or without actinidin, a protease that reduces biofilm formation in vitro, were tested. Results: Alpha diversity in the saliva was higher than that on the tongue without the intervention. The core operational taxonomic units (OTUs) common to both sites were identified. The salivary and tongue microbiomes of one individual tended to be more similar to one another than to those of other individuals. The tablets did not affect the alpha or beta diversity of the oral microbiome, nor the abundance of specific bacterial species. Conclusions: While the salivary and tongue microbiomes differ significantly in terms of bacterial composition, they show interrather than intra-individual diversity. A one-day usage of oral care tablets did not alter the salivary or tongue microbiomes of healthy adults. Whether the use of oral tablets for a longer period on healthy people or people with greater tongue coating accumulation shifts their oral microbiome needs to be investigated.


Introduction
Oral microbiota is a collection of microorganisms that reside in the oral cavity. It has been linked to the promotion of both health and disease 1,2 . Among the different tissues in the oral cavity, the tongue is considered a dominant source of oral microbial populations 3,4 . Further, tongue coating is proposed to cause oral malodor 5 or, upon sudden dissociation, aspiration pneumonia in elderly people with impaired defense mechanisms 6,7 . In addition, the tongue coating is a risk indicator of aspiration pneumonia in edentate individuals 8 .
A variety of methods to reduce tongue coating have been developed and tested to reduce oral malodor 9,10 . Mechanical removal of the tongue coating using tongue brushes or tongue cleaners is one such popular method 9,11 . Other methods include using antimicrobials, e.g., in gels or mouthwashes, or using oral tablets 10,12 .
The tongue microbiota in elderly individuals has been classified into several types with characteristic bacterial composition. These types correlate with the risk to aspiration pneumonia 4,13 . Therefore, methods that could alter the tongue microbiota to a healthy microbiota type could contribute to oral health. We have previously reported that tongue brushing does not alter the alpha or beta diversity of oral microbiota in healthy adults 14,15 . By contrast, according to a recent study, the use of oral care tablets decreases the amount of volatile sulfur compounds (VSCs) produced by bacteria 16 . Further, oral care tablets that contain actinidin, a cysteine protease found in kiwifruit, reduce oral biofilm formation in vitro 12 . However, it is not clear whether these interventions affect the oral microbiota as a whole or the abundance of specific bacteria.
In the current study, we examined the effect of oral care tablets with and without actinidin on the salivary and tongue microbiomes of healthy individuals. We also investigated the diversity of the salivary and tongue microbiomes, and interpersonal microbiome diversity. We show that alpha diversity of the salivary microbiome was greater than that of the tongue microbiome; that an individual's salivary and tongue microbiomes were more similar to one another than to those of another individual; and that the oral care tablets did not affect the oral microbiomes in the population tested. These findings add to the knowledge of the interpersonal diversity and dynamics of the oral microbiota in humans.

Methods
Ten healthy adults participated in the study, with three different treatments tested: two different types of oral tablets (with or without protease), and a negative control (no tablet). For the tablet treatments, the saliva and tongue coating were collected between October 2016 and November 2017 at participants' home (mainly in Osaka, Japan, and in some cases nearby prefectures). DNA extraction and data analysis were conducted at Department of Bacteriology, Osaka Dental University (Hirakata, Japan).

Participants
Participants were recruited from faculty members and graduate students working at the Osaka Dental University hospital, as well as from dentists who were acquainted with an author of this study. Ten healthy volunteers (6 males and 4 females; age: 27-60 years [39.8 ± 3.1 (mean ± SD)]) were enrolled in the study and were anonymized randomly as A-G, O, Q, and R ( Table 1). The inclusion criteria were as follows: healthy men and women over 20 years of age. The exclusion criteria were as follows: daily smoking; treatment with local or systemic antibiotics within 1 month prior to the study; and allergy to kiwifruit. The method and objective of this study were explained to the participants, who provided written informed consent before participating. The Osaka Dental University Medical Ethics Committee approved this study (approved on 3/31/2015; approval number 110864) and the investigations were conducted following the rules of the Declaration of Helsinki. The committee did not consider the study to be interventional in nature and therefore is not a clinical trial.

Oral tablets
Two types of oral care tablets for tongue cleaning were tested in the current study. One type contained actinidin, a cysteine protease extracted from kiwifruit ("protease tablet") and the other did not ("plain (placebo) tablet"). Both tablets were provided by Ezaki Glico Co., Ltd (Osaka, Japan). The protease tablets are identical to those marketed as BREO EX (Ezaki Glico Co.). Tablet composition was described previously 12 . To use the tablets, the participants placed one tablet on the dorsum of the tongue and waited until it dissolved naturally. One tablet takes approximately 5-7 min to completely dissolve.

Study design
The study design is illustrated in Figure 1. The tongue tablet experiment was a placebo-controlled double-blind crossover study. The 10 participants were randomly divided into 2 groups of 5 participants each, by using computer-generated random numbers. All participants performed an initial tongue cleaning (by brushing) at the beginning of the study. The participants were asked not to eat, drink, or perform oral cleaning before each sampling. After a washout period of 10 days during which the participants did not perform any tongue cleaning, they collected their saliva and tongue coating into separate containers in the morning immediately after waking up (sample D1). Then, the participants in each group took tablets, with or without the protease. The participants and the researchers who analyzed the data were not informed about the tablet types given to the participants. The participants were asked to use the tablet three times on the day of the experiment-in the morning (between 9-12 am), in the afternoon (1-4 pm), and in the evening (7-10 pm)-taking one tablet each time. The following morning, the participants collected their saliva and tongue coating separately immediately after waking up (sample D2). After a washout period of 10 days, the participants took the other type of tablet that they had not previously received, and collected the saliva and tongue samples as before. Control experiments (no tablet usage) were conducted with the same participants, after they conducted treatment using the tablets. The duration between the tablet treatments and the control experiment ranged from 10-60 days, depending on the participant. In these experiments, after an initial tongue cleaning and a 10-day washout period, all participants collected samples on two consecutive days (D1 and D2) without taking any tablet in between.

Sample collection
The saliva and tongue-coating samples were collected immediately after the participants woke up, in the morning of the day of tongue cleaning by tablet (D1) and the next morning (D2). The participants first collected 3 mL of saliva in a 25-mL sterile plastic tube. The tongue coating was collected by scrubbing the tongue with a swab and then soaking the tip of the swab in 0.6 mL of phosphate-buffered saline (PBS(-); Wako Pure Chemical Industries, Ltd., catalogue number 166-23555) to suspend the coating. Because this collection method involves scrubbing the tongue with a swab, the tongue coating was collected from the left half of the tongue for D1 and from the right half of the tongue for D2, to minimize the carryover effects of scrubbing. The collected samples were maintained at 4°C for up to 1 day and transported to the laboratory. The saliva samples (3 mL) were homogenized by repetitive pipetting. Then, 0.5-mL aliquots were transferred to sterile tubes. The saliva (0.5 mL) and tongue-coating (0.6 mL) samples were then centrifuged at 10,000 × g for 4 min. The supernatant was discarded and the pellet was stored at −20 °C until DNA extraction. All samples were frozen no later than on the day of D2 sampling.

DNA extraction and library construction
Bacterial DNA was extracted from the pellets via chemical and mechanical lysis using a QIAamp UCP Pathogen Mini kit (QIAGEN, catalogue number 50214), as previously described 17,18 . Briefly, thawed pellets were immediately suspended in  Sequence processing and data analysis Demultiplexed paired-end sequences were processed using QIIME 2 (v.2020.2) and its associated plugins 19 in a Docker container. Sequences obtained from independent Miseq runs were denoised separately using DADA2 (via q2-dada2) 20 applying previously-optimized parameters 14 (trim-left-f = 20; trim-left-r = 20; trunc-len-f and trunc-len-r were set between 241 and 248 depending on the sequence quality; other parameters followed the default settings, including chimera-method = "consensus"). The resulting exact amplicon sequence variants (ASVs) were merged (via q2-feature-table). For taxonomy assignment to each ASV, a naïve Bayes taxonomy classifier trained (via q2-feature-classifier) 21 on the V3-V4 region of the 16S rRNA sequences in the expanded human oral microbiome database (eHOMD; v.15.2) 22 was used. All ASVs were aligned using MAFFT 23 and used to construct a phylogeny with FastTree 2 (via q2-phylogeny) 24 . Sample metadata format was validated using the cloud-based tool Keemei 25 .
Alpha diversity was assessed by calculating the number of observed features (ASVs) and the Shannon index (via q2-diversity), after samples were subsampled without replacement (rarefied) to 40,000 sequences per sample. The non-parametric Kruskal-Wallis test was used to test for significant differences in alpha diversity between sample groups (P < 0.05).
Beta diversity was computed based on the unweighted UniFrac distance 26 (via q2-diversity) and visualized as threedimensional principal coordinate analysis (PCoA) plots using EMPeror (via q2-emperor) 27 . Permutational analysis of variance (PERMANOVA) 28 was used to test the significant difference in bacterial composition among samples (P < 0.05). Differential abundance of bacterial taxonomic groups was tested using the analysis of composition of microbiomes (ANCOM) (via q2-composition) 29 .
Clustered operational taxonomic unit (OTU) table was also used to calculate the beta diversity and differential abundance of specific taxonomic groups. Open-reference clustering was chosen for a high-quality taxonomic assignment to a curated database 30 Table 2.

Study overview
The study design is illustrated in Figure 1. Ten healthy adults participated in the study, with three different treatments tested: two different types of oral tablets (with or without protease), and a negative control (no tablet). For the tablet treatments, the saliva and tongue coating were collected before (D1) and after Inter-individual diversity of the salivary and tongue microbiomes We first analyzed the microbiome of the saliva and tongue coating, to determine the baseline for the study. In total, 30 D1 samples (10 participants, three independent treatments) each of the saliva and tongue coating were analyzed. Alpha diversity of the tongue microbiome was significantly lower than that of the salivary microbiome, using both the number of observed ASVs (P = 8.9e-7, Kruskal-Wallis test) (Figure 2a) and Shannon index (P = 2.0e-7, Kruskal-Wallis test) as measures (Figure 2b). This was consistent with a previous report 3 . The difference in diversity is probably associated with the tongue acting as a specialized niche for specific microorganisms, and the saliva containing a mixture of microbiota from different sites in the oral cavity. Interestingly, the number of observed ASVs varied among the individuals, ranging from approximately 170 to 325 in the saliva (P = 0.027, Kruskal-Wallis test), and from approximately 125 to 225 in the tongue coating (P = 0.022) (Figure 2c), suggesting a difference in oral microbiome among individual.
We next assessed the differences in bacterial composition among samples (beta diversity) (Figure 3). A significant difference between the salivary and tongue microbiomes was detected both in PCoA based on unweighted UniFrac distances using the full ASV table (P = 0.001, PERMANOVA) ( Figure 3a) and MDS based on generalized UniFrac distances 32 using the clustered OTU table (P = 0.003, PERMANOVA) ( Figure 3c).
The data also revealed a significant difference between individual microbiomes (Figure 3b and 3d; P = 0.001, PERMANOVA using the clustered OTU table). As indicated by the plots in Figure 3, the similarity of the salivary and tongue microbiomes within an individual was greater than the similarity of the salivary or tongue microbiomes between individuals. This suggests there is stability in an individual's oral microbiome, at least within the relatively short time period of the study (several weeks). This observation is consistent with earlier studies that highlight the stability of an individual's oral microbiome 37 .

Core OTUs in the Salivary and Tongue Microbiomes
To identify the core members of the oral microbiome, we focused on OTUs that were present in ≥95% of the saliva or tongue D1 samples. Seven OTUs were present in ≥95% of both, the saliva and tongue samples. These were F. periodonticum Effect of tablet taking on the salivary and tongue microbiomes Using the above data as the base line, we finally assessed the effect of taking oral tablets (with or without protease) on the salivary and tongue microbiomes. Alpha diversity in D1 and D2 samples was not significantly different between any treatments (Figure 6a). In the control (no tablet) treatment, whereas  the observed number of ASVs seemed to slightly increase in the saliva and to slightly decrease in the tongue, they were not statistically significant (Figure 6a). This indicates that some fluctuation of the oral microbiome may occur naturally. Further, MDS analysis indicated that the beta diversity between D1 and D2 samples was not significantly different in any treatment, for either the salivary or tongue microbiome (P > 0.7, Kruskal-Wallis test) ( Figure 6B).
We next examined whether any bacterial species were specifically impacted by oral tablet usage. Both ANCOM using the full ASV table or Kruskal-Wallis test using the clustered OTU table revealed that no OTU was differentially abundant before (D1) or after (D2) tablet use, in any of the treatments (no tablet, protease tablet, and plain tablet). The OTU abundance in each treatment group is summarized in Figure 7a-c. Although according to a recent study oral tablet use decreases the abundance of Fusobacterium nucleatum on the tongue of healthy young adults 16 , we did not detect any significant decrease of OTUs that correspond to F. nucleatum.
Further, in the current study, whereas 7.6% of all OTUs from the tongue microbiome were assigned to the genus Fusobacterium (Figure 4a), the majority of them were classified as F. periodonticum (7.5% of total) and only <0.1% of all OTUs was assigned to F. nucleatum at species level.

Discussion
The oral microbiota has been associated with specific diseases in susceptible populations. In the current study, we examined the effect of oral care tablet use, with or without actinidin, on the salivary and tongue microbiomes. We showed that whereas there are some differences between the tongue and salivary microbiomes, the microbiomes were not affected by the oral tablet use, regardless of the tablet type. This does not preclude the possibility that a persistent oral tablet use would alter the oral microbiome. Controlled alteration of the oral microbiome has potential for disease prevention.
We here identify the core OTUs that are common between saliva and tongue. Among these OTUs, S. oralis and Campylobacter sp. have been determined to be the core OTUs common in the saliva and tongue previously 3 . F. periodonticum and Granulicatella adiacens have been found in tongue microbiome of adult, P. melaninogenica and H. parainfluenzae have been found in infant and adult tongue in common 38 . TM7 species have been identified in oral microbiomes including tongue 39,40 and supragingival plaque 41 . It should be noted that although the high prevalence of Saccharibacteria (TM7) [G-1] sp. HMT-352 (≥95% in both saliva and tongue) as shown in our present study has not been reported previously, it could also be a result of clustering similar sequences into a single OTU.
Although there are some differences in the classification of the core or predominant oral OTUs between the current and other studies 1,3 , the majority of the species were identified as the core oral OTUs across the studies. Since low-abundance rather than highly abundant OTUs may contribute more to the difference in oral bacterial communities 42,43 , detailed analysis of low-abundance OTUs would be important in future research. O. sinus and S. moorei were previously classified as core OTUs common to the salivary and tongue microbiomes 3 , but in our present study were identified as tongue-specific. This seems reasonable considering that S. moorei plays an important role in halitosis [44][45][46] . The effective separation of saliva-and tongue-specific OTUs suggest the usefulness of our study design in analyzing the salivary and tongue microbiomes simultaneously. Our present study shows a variety among individuals in the number of observed ASVs in the salivary and tongue microbiomes. On the contrary, a significant interpersonal diversity in the supragingival plaque and salivary microbiomes, but not in the tongue plaque microbiomes was previously reported, using Faith's phylogenetic diversity as a measure 3 . Since the same V3-V4 region was targeted for amplicon sequencing in both studies, the discrepancy concerning the interpersonal differences in tongue microbiome might be associated with the differences in the alpha-diversity measure used, and/or in the participants' age [25.3 ± 3.1 years in Hall et al. study 3 and 39.8 ± 10.9 years (mean ± SD) in the current study (see Methods)].
The tongue microbe is associated with various diseases, including halitosis 47,48 and aspiration pneumonia 4 ; alterations in salivary microbiome has also linked to increasing numbers of oral and non-oral diseases 49 . With the advancement of microbiome studies, methods to predict host traits that predispose to various diseases or conditions based on microbiome analysis have been developed 50,51 . Lu et al. have shown that tongue coating microbiome data can be used to distinguish individuals with pancreatic head carcinoma (PHC, one of pancreatic adenocarcinoma which occurs in the head of the pancreas) from healthy subjects 39 . Although the evaluation of an individual's disease status based on the tongue microbiota data is possible, the collection methods of the tongue coating samples may not be reliable when performed by a non-specialist, especially because of the anterior to posterior gradient of the bacterial communities in the tongue surface 43 . We here showed that the microbiomes of the saliva and tongue of an individual tend to be more similar to one another than to the salivary or tongue microbiomes from other individuals. Considering this fact together with the stability of oral microbiome over a prolonged period of time 37 , salivary collection could perhaps be used in the future as a standard method to predict diseases associated with the tongue-coating microbiota, as well as those linked to that of the saliva.
Oral care tablets have been shown previously to reduce tongue coating load and VSCs 12,16 . We here analyzed the effect of oral care tablets on the salivary and tongue microbiome.
To avoid individual varieties in the amount of tongue coating or saliva flow affecting the analysis, we recruited only healthy adults to participate in this study. We did not detect any significant differences in the alpha diversity, beta diversity, or abundance of specific OTUs at species level after oral tablet use. There are several possible explanations for these observations. First, the participants of the current study were healthy adults with no apparent tongue coating accumulation. Although accumulated tongue coating can be reduced with oral tablets 12 , the amount of tongue coating analyzed herein may have been insufficient for detecting the differences in the microbiota. Second, the tablet intervention period in the current study was only 1 day and the samples were collected 1 day after the tablet use. In contrast, twice daily tongue scraping for three days, together with sampling within 15 mins after intervention, have shown to reduce the gram-negative anaerobes on the tongue 9 . Although we chose to collect samples 1 day after the intervention, the 1-day period could have been long enough for the resilience of the oral microbiota to revert any shift in the oral microbiomes invoked by the tablet use. Considering these factors, analyzing oral care tablet intervention in individuals with a higher tongue coating index and/or over a longer period of time together with immediate sampling after intervention may provide more information on whether and how oral care tablets alter the oral microbiota, contributing to the maintenance of oral health.
Oral care tablets containing actinidin reduces tongue coating, and actinidin prevents biofilm formation by degrading cell-surface proteins in vitro 12 . We here attempted to elucidate the effect of the protease, supplied in oral tables, on the oral microbiome. Unfortunately, we were unable to assess the effect of the protease, because oral tablet treatments failed to alter the oral microbiome or specific bacterial taxa, regardless of the presence or absence of actinidin. As above, including participants with a higher tongue coating and a longer intervention period with immediate sampling may have allowed detection of the effect of actinidin in the tablets. Alternatively, an in vitro culturing system could be used to analyze the effect of actinidin on the oral microbiome, with the effects of the compound tested in a controlled manner. For example, nitric oxide 17 or statins 52 have been shown to alter the abundance of specific bacterial species. Using such system would allow the analysis of the effect of actinidin on oral microbiota separately from the effect of mechanical removing of the tongue coating.
Various lines of evidence suggest a link between oral microbiota and health or disease 1,2,53 . The current and other 3 studies have highlighted interpersonal differences in the oral microbiota. Several types of tongue microbiota have been shown to exist in individuals with different susceptibility to pneumonia 4 . Hence, personalized treatment based on an individual's oral microbiota is required, as has been already pointed out in the context of periodontal disease 54 . Analysis of how different types of oral microbiota are affected by certain interventions (e.g., oral care tablet or antibiotic treatment) would enable a more precise control over the oral microbiome in the future. In vitro culturing systems mentioned above are powerful tools for elucidating responses of bacterial communities taken from different individual to various interventions, and the contributing factors.
In conclusion, we have shown that while the salivary and tongue microbiomes differ significantly in terms of bacterial composition, they show inter-rather than intra-individual diversity. We have also identified bacterial species that are common to the salivary and tongue microbiome, as well as those that are specific to either of these. In addition, we showed that oral care tablets may not alter the bacterial composition of the saliva or the tongue, at least over short periods of time in healthy individuals. Considering the link between oral microbiota and health or disease, analyzing the differences in how individual oral microbiota responds to external factors will pave the way to more effective therapeutic and diagnostic approaches and, ultimately, contribute to the development of personalized dental medicine.  Discussion Figure 5 should be cited in the second paragraph of the discussion.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact research@f1000.com