**4. Discussion**

In the present study, an oolong tea infusion containing a total of 2.83 ± 0.02 g/<sup>L</sup> polyphenols, including catechin, epicatechin, epigallocatechin gallate, and at least 30 other components, was used to evaluate its salivary microbiota modification e ffect. Three subjects were required to consume 1.0 L of tea infusions daily, which equaled approximately 52 mg/kg body weight of tea polyphenols. The preparation method and intake amount of tea infusions followed the general tea drinking habits of individuals in the southern part of China, which would provide a more pragmatic and appropriate insight. Under this consumption amount, diverse responds of salivary microbiota were observed among the three subjects.

Following the Illumina high-throughput sequencing, a highly diverse salivary bacterial community was observed. A total of 8801 OTUs with a 97% similarity level was identified from the 36 saliva samples and annotated into 25 phyla and 260 genera. In addition to the complexity, a high inter-individual variation in salivary microbiota was also discovered. It was revealed that the salivary microbial communities within the three subjects were significantly distinct from each other, exhibiting host-specific microbiota profiles; their overall collective responses to tea consumption also varied among each participant. The positive delta values from MRPP tests and R values from Anosim tests indicated the di fferences of salivary microbiota profiles among subjects were far more significant than that among di fferent time points within one subject (Table 3). When all 36 salivary microbiota data sets from the three subjects were depicted into one PCA plot, no clear cluster was observed (Figure S2). The host-specific salivary microbiota were also confirmed by the distinct correlation networks of each participant. These results were consistent with the findings of Belstrøm et al., since the authors confirmed that the five individuals in their study had a personalized salivary bacterial fingerprint [29]. Hall et al. also stated that the oral bacterial community fingerprint varied from person to person in their study [17]. Thus, in order to minimize the inter-individual variations, the data sets from di fferent subjects were analyzed separately, otherwise the e ffect of tea may be obscured by the inter-individual variations, as shown in Figure S2.

In general, it was revealed that oolong tea consumption led to a profound reduction in diversity of the salivary bacterial communities of subject 1 and subject 2. Takeshita et al. stated in their population-based study that good oral health was associated with a lower phylogenetic diversity of the salivary microbiome [5]. Moreover, Vestman et al. reported that the diversity of the tooth biofilm samples was reduced after probiotics supplementation [30]. The increase of diversity of gu<sup>t</sup> microbiota is normally associated with better gu<sup>t</sup> health conditions, such as through the extension of the functional genes for facilitation of the absorption of nutrients and energy, or for appropriate development of immunity. In contrast to the commensal microbiota residents in the intestinal tract, which typically live in harmony with the host, the oral microbiota is responsible for the two most common diseases, including dental caries and periodontal diseases [31]. The increase of diversity in salivary microbiota may be associated with the flourish of dental plaque which resulted from the accumulation of attached bacteria, and thus increase the risk of dental caries and periodontal diseases; while, the decrease in taxonomic diversity in saliva may indicate the shrinking of bacterial communities in dental plaque biofilms, and thus lead to healthier oral ecological conditions. However, for subject 3, the decrease of the salivary microbial community diversity was not significant, except for Simpson index of week 8, which might indicate a lower modulation e ffect of tea on subject 3. Furthermore, according to PCA, significant overall shifts of salivary microbiota composition were noted in subjects 1 and 2. However, in the case of subject 3, a higher variation was discovered amongs<sup>t</sup> di fferent sampling time points, which may also sugges<sup>t</sup> a lower e ffectiveness of tea consumption upon the salivary microbiota of subject 3.

The oral cavity, as the portal of entry to the gastrointestinal tract, is one of the most complex microbial colony sites within the human body [32]. In order to better understand the complex ecologic system, a correlation network was employed in this study to simplify and visualize the co-occurrence patterns of salivary bacteria. The bacteria taxa with robust connections with other salivary bacteria were defined as "hub salivary bacteria". Subsequently, via a heatmap plot, the temporal dynamic of each individual hub salivary bacteria was clearly presented. Afterwards, through a Venn diagram, the shared hub microbiota across the three subjects were further identified. Separated correlation network analysis revealed the detailed influence of tea consumption on the salivary microbiota composition within the same contactable environment. Thus, it minimized the inter-individual variations between subjects. While, an additional Venn diagram further helped in seeking the common influences of tea consumption.

In particular, seven shared hub OTUs across the three subjects were identified from the highly complex and personalized oral ecosystem. Notably, OTU\_1 (*Streptococcus* sp.), as the most predominant taxon, also acted as a shared hub microbiota and favorably interacted with other oral bacteria. Due to the biofilm formation and acid production ability of *Streptococcus*, multiple members of this genus, including *Streptococcus mutans, Streptococcus sobrinus, Streptococcus salivarius, Streptococcus constellatus*, and *Streptococcus parasanguinis*, were considered as opportunistic pathogens [33]. With regard to the shifts of *Streptococcus* sp., a significant decrease (−16.94%, *p* = 0.035) was found after eight weeks of tea consumption. Therefore, a *Streptococcus* inhibitory effect of tea was observed in this study, and the effect may assist in the prevention of dental caries. There is a preponderance of evidence to support the beneficial role of tea in protecting against this oral pathogen. Narotzki et al. reviewed the clinical and biological studies regarding the correlation between green tea and oral health and concluded that green tea may reduce dental caries through bacterial growth repression and enzyme activity inhibition [14]. With the exception of green tea, it has also been reported that black tea extracts could inhibit *S. mutans* adhesion in vitro [34]. Kawarai et al. compared the *S. mutans* biofilm formation inhibitory effect of Assam tea (a black tea variety) and green tea and ascertained that Assam tea exhibited a stronger biofilm inhibition activity than green tea [35]. The inhibitory activity of specific teas against oral pathogens are commonly attributed to the phenolic components within the tea [14].

Similar inhibitory effects were also observed on OTU\_33 (*Ruminococcaceae* sp.) and OTU\_696 (*Haemophilus* sp.), both of which were also hub microbiota across the three subjects. *Haemophilus* are a common bacteria which inhabit the mouth, vagina, and intestinal tract. The genus includes commensal organisms, along with some pathogenic species such as *H. influenzae* and *H. ducreyi*. The inhibitory effect of tea on *Haemophilus* may also reduce the risk of infection. *Ruminococcaceae*, one of the most typical gu<sup>t</sup> microbiotas, can be found in significant numbers in the intestines of humans. However, the biological meaning regarding the depletion of this bacterium induced by tea drinking was not clear.

Along with oolong tea consumption, a significant elevation of OTU\_133 (*Veillonella* sp.), OTU\_23 (*Veillonella* sp.), OTU\_42 (*Actinomyces odontolyticus*), and OTU\_6 (*Gemella haemolysans*), which were all demonstrated as robust network nodes across the three subjects, was observed in this study. Lim et al. illustrated a significant negative association between *Haemophilus* and *Veillonella* [36], which was consistent with our findings. Furthermore, it was reported that the establishment of some certain oral commensals was linked to oral health, such as the bacterial species belonging to *Neisseria*, *Veillonella*, and *Actinomyces* [37], although details regarding the exact mechanisms are not ye<sup>t</sup> available. Moreover, the elevated effect on these four hub bacteria continued throughout the follow-up period, which demonstrated the sustained effect of tea drinking.

With regard to the mechanisms behind the modification effect of tea on salivary microbiota, several hypotheses have been invoked to account for this particular effect: (i) tea polyphenols possess antimicrobial properties, which are believed to aid in the inhibition of certain bacteria, including some pathogens [13,14]; (ii) tea polyphenols as antioxidants may alleviate oral oxidative stress and inflammation, which may further impact the oral immune system and induce a drift of the bacterial community [14]; (iii) tea polyphenols can precipitate salivary proteins and inhibit the activity of salivary alpha-amylase, and thus, induce the decrease of fermentation of carbohydrates involved in caries formation [38]. However, the precise mechanism is still ambiguous, resulting in the necessity for further studies. Numerous epidemiologic studies and clinical trials have validated that regular tea consumption could reduce the risk of cardiovascular disease, including coronary heart disease, stroke, and peripheral arterial disease [39]. Recent studies show a correlation between periodontal disease and

cardiovascular disease [40,41]. Thus, from the perspective of alleviating systematic inflammatory and immunological processes, explicating the underlying mechanisms (e.g., to link the levels of endogenous mediators, such as endothelin [42] and vitamins [9] of tea consumption may open an innovative avenue toward the development of new antibiotics with good safety and tolerability margin.

It was also acknowledged that the inadequate number of subjects in this study might limit the statistical analysis. As explained previously, using limited subjects and following the time course of each individual may help to minimize the inter-individual variations. However, further studies with larger sample sizes are warranted to validate these findings.
