**1. Introduction**

An estimation of 700 diverse bacterial species have been identified in human oral cavities, which constitute complex microbial communities [1]. These bacteria generally inhabit at different oral niches, including saliva, supragingival plaque, subgingival plaque, and mucosa. Of these niches, saliva harbors as much as 10<sup>8</sup> bacteria/mL and constitutes a reservoir of microorganisms regularly derived from dental plaque biofilms adhering to gingival crevices, periodontal pockets, the dorsum of the tongue, and other oral mucosal surfaces [2]. As an integral part of oral microbiota, salivary microbiota has been found to be differentiated between patients with a healthy oral cavity and those with dental caries and periodontitis [3]. Additionally, several studies discovered marked clinical importance of salivary microbiota on the general health of the host, such as by either preventing or causing infections [4]. Thus, salivary microbiota may provide further insight into the integral microbiota structure within the human oral cavity, and even the oral and general health status of individuals.

Since the oral cavity is exposed to the external environment, the salivary microbiota may be influenced by various factors, including oral hygiene, smoking, nutrients, mechanical stress, and the overall health condition of the host [5]. The impact of nutritional factors in shaping the oral microbial ecosystem cannot be ignored. Food residuals in the mouth can be utilized as substrates for oral bacteria; moreover, some food components have a selective effect on microbial growth, by either stimulating or suppressing some specific bacteria. For example, a regular consumption of polyphenol-rich beverages and foods, such as tea, cranberry, co ffee, grape, almond, and alcohol-free red wine, have been reported to inhibit oral pathogenic bacteria [6–8]. The suppression of oral, especially periodontal pathogenic, bacteria may ameliorate the control of plaque biofilms, and thus reduce the inflammatory and immunological processes of oral and periodontal diseases [9]. Recently, the impact of nutraceutical dietary aliments, such as antioxidants, probiotics, natural agents, and vitamins, on oral health is gaining more and more attention [10].

Tea (*Camellia sinensis*), second only to water, is the most widely consumed beverage in the world. The major constituents of tea leaves are the flavonoids, including flavonols, flavones, and flavan-3-ols, of which over 60% are the flavan-3-ols, commonly referred to as catechins. Based on the United States Department of Agriculture (USDA) Flavonoid Database, it has been estimated that the daily total flavonoid intake is mainly from flavan-3-ols (83.5%); while, the major source of flavonoids is tea (157 mg), and citrus fruit juices come second (8 mg) [11]. There is a large population of heavy tea consumers all over the world, especially in the southern part of China, where people consume a substantial amount of tea infusions on a daily basis. A number of health-promoting e ffects have been associated with tea consumption; these e ffects are generally attributed to the phenolic compounds in tea. Tea polyphenols are well known for their antimicrobial properties, including on *Streptococcus mutans* and *lactobacilli* [12], and thus, they are believed to possess anti-cariogenic e ffects [13,14]. Moreover, regular consumption of tea has proved to exert gu<sup>t</sup> microbiota regulation e ffect [15,16]. However, with regard to the normal balanced oral microbiota, little is known about the influence of tea drinking. Considering the wide range of biological properties, including anti-microbial, anti-oxidant, anti-inflammatory, anti-cariogenic, and gu<sup>t</sup> microbiota regulation e ffects of tea polyphenols, it is reasonable to assume that sustained tea drinking will result in certain oral ecological shifts. A better understanding of the oral ecological shifts under sustained and significant tea consumption may contribute to oral health managemen<sup>t</sup> for tea consumers.

It is also worth noting that, due to the variability in genes, social habits, hormonal fluctuation, diet, quality and quantity of saliva, etc., the oral environment di ffers between subjects and represents huge inter-individual variations [17]. Moreover, the responses of oral microbiota of di fferent individuals to certain nutritional factors maybe also be diverse. To understand the influence of tea consumption on oral microbiota, tracking the temporal dynamic of salivary microbiota of subjects separately may provide useful information free from interference of inter-individual variations. In the current study, it is hypothesized that sustained tea consumption will alter the composition of salivary microbiota and exert oral health benefits to the host. To test this hypothesis, three orally healthy subjects were recruited and instructed to consume a substantial amount of tea infusions on a daily basis and their salivary bacterial communities pre-, peri-, and post-treatment were quantified by utilizing a high-throughput HiSeq sequencing technique. Then, via several multivariate statistical analyses, the temporal dynamics of salivary microbiota of each individual were analyzed. Based on these, the impact of sustained consumption of tea on the normal balanced oral microbiota was discussed.

#### **2. Materials and Methods**

#### *2.1. Oolong Tea Infusion Preparation and Phenolic Profile Analysis*

The tea used in this study was an oolong tea variety, purchased from a local market in Fujian Province, China. The oolong tea was prepared in accordance with the tea consumption method of local residents. A certain amount of dry oolong tea (whole leaves) was immersed in 20 times the volume of distilled boiling water (temperature around 90–95 ◦C) for 1 min, then the tea leaves were filtered, and the liquor was retained as an oolong tea infusion.

The phenolic profile of the tea infusion was then analyzed by utilizing ultra-high performance liquid chromatography (UHPLC) coupled to quadrupole time-of-flight mass spectrometer (Q-TOF MS/MS) approach, as previously described [15]. Briefly, chromatography separation was performed on an Acquity UHPLC system (Waters, Milford, MA, USA) with HSS T3 column (100 mm × 2.1 mm, 1.7 μm). A sample of 1 μL was injected and eluted with the mobile phase at 0.3 mL/min at 40 ◦C; detection was at 280 nm. The mobile phase consisted of (A) 0.1% formic acid solution (v/v) and (B) acetonitrile with 0.1% formic acid (v/v), while the gradient program was as follows: 99%–93% (A) in 0–2 min; 93%–60% (A) in 2–13 min; 60%–1% (A) in 13–14 min. The eluent was then introduced to a SYNAPT G2-Si high-definition mass spectrometer (Waters, Milford, MA, USA) equipped with an electrospray ionization (ESI) source. The analyses were performed in negative-ion mode and positive-ion mode, with a sampling cone voltage of 40.0 V, and a capillary voltage of 2500 V. The source temperature was 120 ◦C, with a desolvation gas flow of 800 L/h at a temperature of 450 ◦C. The time-of-flight (TOF) acquisition rate was 0.2 s/scan with 0.01 s inter-scan delay. Data were collected in centroid mode from 100 to 1200 Da in full scan during 0–14 min. The mass data were corrected during acquisition using a lock-mass calibrant of leucine enkephalin (200 ng/mL), via a lock spray interface at a flow-rate of 50 μL/min, generating a reference ion for positive ion mode ([M+H]<sup>+</sup> = 556.2771) and negative ion mode ([M–H]− = 554.2615) to ensure accuracy during the MS analysis. All data analyses were conducted using the MarkerLynx application manager software (version 4.1, Waters, Milford, MA, USA). The total polyphenols content in the tea infusions was then measured by utilizing the Folin–Ciocalteu method [18]. Briefly, 1 mL sample, 5 mL Folin–Ciocalteu's reagen<sup>t</sup> (diluted 10 times), and 4 mL sodium carbonate (7.5%, w/v) were mixed. After 60 min, the absorbance at 765 nm was measured. Total phenolic content was expressed as a mass percentage on dry matter basis. Gallic acid was used as an external standard.

#### *2.2. Subject Enrollment, Study Design, and Salivary Sample Collection*

The inclusion criteria for this study included: healthy adult individuals sharing a relatively similar living environment; no tea and antibiotics taken in the previous 3 months; and no smoking. After the screening process, three healthy adult Chinese individuals (2 females and 1 male), 23 years of age, were enrolled from the campus of Fuzhou University, Fuzhou, China. The plaque and gingival status was examined before and after tea intervention. No obvious change was observed either before or after tea usage. In addition, no adverse reaction was reported throughout the experimental period by participants. Written informed consent was obtained from each participant. This study was approved by the ethical committee of the Institute of Food Science and Technology of Fuzhou University (approval number: IFSTFZU20180301).

This study consisted of a 3-day baseline period, an 8-week oolong tea infusion intervention period, and a 4-week follow-up period. During the intervention period, the three subjects (subject 1, subject 2, and subject 3) were required to consume 1.0 L of oolong tea infusion per day (0.5 L in the morning and 0.5 L in the afternoon). Moreover, they were also instructed to circulate or swish the infusion around in their mouths prior to swallowing the tea infusion. During the follow-up period, the subjects were asked not to consume any tea drinks. In addition to this, the subjects were asked to maintain their regular diet and oral hygiene habits, with the exception of the sampling occasions. Salivary samples were collected at 4 different stages, each stage included 3 sequential days: (A) 3 sequential days of the baseline period, which was prior to the intervention period; (B) 3 sequential days after 4 weeks of the tea intervention; (C) 3 sequential days after 8 weeks of the tea intervention; and (D) 3 sequential days at the end of the follow-up period, which accounted for 4 weeks post-intervention. All salivary sample collections were conducted in the morning. Each subject was asked not to eat, drink, or brush their teeth before the sample collection. Then, 2 mL of unstimulated saliva were collected from the subjects by expectoration into a tube. In total, 36 salivary samples from the 3 subjects were sampled.

#### *2.3. Salivary Bacterial DNA Extraction*

Salivary bacterial DNA was extracted from the 36 salivary samples by utilizing a rapid DNA extraction kit (BioTeke Corporation, Beijing, China), following the manufacturer's instructions. The extracted bacterial DNA was then checked by agarose gel electrophoresis.

#### *2.4. Illumina Sequencing of Salivary Bacteria*

Bacterial primers 341-F (5-CCT AYG GGR BGC ASC AG-3) and 806-R (5-GGA CTA CNN GGG TAT CTA AT-3) with specific barcodes were used to amplify the V3–V4 region of bacterial 16S rRNA genes. The sequencing library of bacterial 16S rRNA genes was generated for high-throughput sequencing, employing the TruSeq® DNA PCR-Free Sample Preparation Kit (Illumina, San Diego, CA, USA). Next, the library was sequenced on an Illumina HiSeq2500 platform by Novogene Bioinformatics Technology Co., Ltd. (Beijing, China).

## *2.5. Bioinformatic Analysis*

Raw sequencing reads, obtained from the Illumina platform, were then merged by using FLASH software (Version 1.2.7) [19] and filtered using QIIME software (Version 1.7), with the default parameter setting of 'split\_libraries\_fastq.py' script [20,21]. All quality filtered sequencing reads were then clustered into operational taxonomic units (OTUs) with a threshold of 97% sequence similarity, by utilizing UPARSE software (Version 7.0) [22]. The representative sequence (most abundant) for each bacterial OTU was then annotated by utilizing the GreenGene Database [23] and Human Oral Microbiome Database (HOMD) [24]. The least total sequences number was 30,070 in this study. The total reads of each sample was normalized to 30,070 sequences/sample, and the OTUs abundance information was normalized correspondingly for further analysis.

Based on these annotated and normalized output data, different statistical methods were used to interpret the similarities of diverse data sets, or to plot the correlation network among the salivary microbiota. First, community diversity estimators including Shannon and Simpson indexes were calculated by R software (Version 3.2.5) with vegan package. Second, the multiple response permutation procedure (MRPP) and analysis of similarity (Anosim) were employed to compare the statistical differences within and between subjects in salivary microbiota profiles, by using R software with vegan package [23]. Third, principal component analysis (PCA) was applied to evaluate and visualize the differences of samples in OTU-level complexity, by using R software with mixOmics package. Next, the correlations among the OTUs with relative abundance over 0.1% of each subject were calculated, based upon Pearson's correlation coefficients, by using R software with Hmisc package. The strong connections (|r| > 0.9, *p* < 0.05) were further imported into Gephi software (Version 0.8.2), so as to generate correlation networks of these predominant microbiota [25]. The nodes (OTUs) with high strong connection numbers were defined as the "hub microbiota", which were likely to be more connected to other nodes when compared to non-hub nodes [26,27]. Moreover, the relative abundance of the hub microbiota was further visualized into heatmaps, by utilizing R software with pheatmap package. Hierarchical clustering of the columns (samples) was further calculated based on Euclidean distance and ward.D method, and indicated on the heatmaps. Lastly, in order to identify the shared and unique hub salivary microbiome of these three subjects, a Venn diagram was built according to the method as descripted by Heberle et al. [28].

Other data are expressed as mean ± SD. Furthermore, the statistical significance among different data sets was analyzed by Student's t-test or Duncan's multiple range test using SPSS software (Version 19.0.0), while the significance threshold was established at 0.05.
