**3. Results**

#### *3.1. Phenolic Profile of Oolong Tea Infusion*

The total polyphenols content and phenolic profile of the oolong tea infusion used in this study were determined, and the results indicated that the total polyphenol content of the tea infusion was 2.83 ± 0.02 g/L. Following untargeted UHPLC Q-TOF-MS approach, the phenolic constituents present in the tea infusion were further analyzed. Table 1 gives the MS characteristics and tentative identification of each chromatographic peak. These chromatographic peaks, along with their proposed chemical structure, are depicted in Figure S1. In summary, 33 constituents were tentatively identified from the

tea infusion, including 2 alkaloids, 7 flavan-3-ols, 7 organic acids and esters, 4 proanthocyanidins, 11 flavonoid glycosides, 1 theaflavin, and 1 amino acid. Of the total chromatographic peak areas, caffeine (peak 16), epigallocatechin (peak 13), epicatechin (peak 20), gallic acid (peak 5), and caffeoyl-hexoside (peak 1) were the most abundant constituents.


**Table 1.** The phenolic profiles of the oolong tea infusion.

a Peaks were assigned from the chromatograms in Figure S1; b [M+H]<sup>+</sup> mode.

#### *3.2. Overall Salivary Bacterial Structure*

The salivary bacterial components of the three subjects during the 12-week experimental period were investigated and evaluated using the Illumina HiSeq sequencing analysis. A total of 1,983,489 (average length = 425 bp) quality filtered sequencing reads corresponding to the V3–V4 region of bacterial 16S rRNA genes were obtained. Good's coverage estimation values were within the range of 99.8%–100%, which indicated adequate sequence coverage to reliably describe the full bacterial communities present in all the samples. All sequences were clustered into 189 to 458 OTUs with a 97% similarity level for each sample. The summary of the sequencing results is listed in Table S1.

After the taxonomic assignment, these sequences were then annotated into 25 phyla and 260 genera. At the phylum level, *Firmicutes* (41.04%), *Bacteroidetes* (24.23%), and *Proteobacteria* (23.31%) comprised the majority of OTUs (88.59%). While at the genus level, *Streptococcus* (28.24%), *Haemophilus* (15.97%), *Prevotella* (14.64%), *Alloprevotella* (5.27%), and *Neisseria* (4.21%) were the most prevalent bacterial taxa throughout the three subjects, which in totality accounted for 69.05% of all salivary bacteria. The relative abundance of these bacterial taxa at the phylum level and genus level are presented in Figure 1. These findings were generally in line with the findings of Belstrøm et al., which indicated the five most predominant genera identified were *Streptococcus*, *Haemophilus*, *Prevotella*, *Rothia*, and *Neisseria*, accounting for around 50% of the identified OTUs [29].

**Figure 1.** Relative abundances of the most abundant phyla and genera in each salivary sample in the (**A**) phylum level and (**B**) genus level.

#### *3.3. Comparisons of Salivary Bacterial Communities*

Based on the relative abundance of all the OTUs, salivary bacterial community diversity (expressed by the Shannon and Simpson indexes) was investigated first, and the results are shown in Table 2. Compared with baseline (week 0), after eight weeks of tea consumption, a remarkable reduction in the community diversity was noticed across the three subjects, with the exception of the Shannon index of subject 3.



Values are expressed as the mean ± SD (*n* = 3). Means with different superscript letters (a, b) within a row sugges<sup>t</sup> significant differences (*p* < 0.05); means with the same superscript letters (a, b) within a row sugges<sup>t</sup> the differences are not significant (*p* ≥ 0.05), as determined by Duncan's multiple range test.

In order to adequately compare the homogeneity of salivary bacterial communities among the three subjects, MRPP and Anosim tests were then performed. In the pairwise comparisons, positive delta values from MRPP tests and R values from Anosim tests were observed, which indicated a higher similarity within the groups (Table 3). Thus, diversities of salivary microbiota among individuals were much larger than the variation within individuals over the course of tea consumption.

**Table 3.** Summary of multiple response permutation procedure (MRPP) and analysis of similarity (Anosim) tests between each subject.


The general profiles of salivary microbiota of each individual subject at different sampling times were further compared with PCA (Figure 2). For subject 1, the salivary bacterial communities in the baseline period were separated from the tea intervention and follow-up period, while in the follow-up period bacterial communities gathered with those in the tea consumption period. For subject 2, clear distinctions in the bacterial communities were discovered between week 0 and the other experimental periods, while the bacterial communities in week 12 and week 4 were overlapped. In the case of subject 3, relatively higher similarities were found among the different treatment periods, which might sugges<sup>t</sup> a slighter or lower impact of tea consumption on the salivary microbiota.

#### *3.4. Correlation Networks of Salivary Microbiota*

Based on the Illumina sequencing results, 67 OTUs were defined as the predominant salivary microbiota of the three subjects, with relative abundance over 0.1%. Pearson's correlations were calculated among the predominant salivary microbiota of each subject, and the strong connections (|r| > 0.9 and *p* < 0.05) were further visualized as networks (Figure 3A,C,E). When comparing the networks of the three subjects, subject 1 had the most complicated co-occurrence patterns of salivary bacteria, with a total strong connection number of 128. For subjects 2 and 3, the strong connection numbers were 49 and 41, respectively.

**Figure 2.** Principal component analysis (PCA) score plots based on the relative abundance of all operational taxonomic units (OTUs) of each subject.

#### *3.5. Hub Salivary Microbiota Identification*

The size of each node in the network represents the number of strong connections with other nodes. Thus, the OTUs with a larger node size were identified as the hub salivary microbiota of each subject, which had more connections with other bacteria. In this study, approximately 20 hub OTUs from each subject were intentionally selected. In particular, for subject 1, 21 OTUs were defined as the hub microbiota (Figure 3A). The relative abundance changes of these bacteria were further visualized as a heatmap plot (Figure 3B). Of these, 8 OTUs (OTU 133, 23, 42, 5, 6, 7, 8, and 9) increased after tea intervention, while the remaining 13 OTUs decreased; moreover, OTU 1, 42, and 5 increased during the follow-up period (week 12). For subjects 2 and 3, 20 and 25 OTUs were identified as hub microbiota (Figure 3C,E). The successions of these hub salivary microbiota during the 12-week experimental period are illustrated in Figure 3D,F.

Through a Venn diagram, seven OTUs, including OTU\_1 (*Streptococcus* sp.), OTU\_133 (*Veillonella* sp.), OTU\_23 (*Veillonella* sp.), OTU\_33 (*Ruminococcaceae* sp.), OTU\_42 (*Actinomyces odontolyticus*), OTU\_6 (*Gemella haemolysans*), and OTU\_696 (*Haemophilus* sp.), were identified as the shared hub microbiota of the three subjects (Figure 4A). The unique hub salivary microbiome is also shown in Figure 4A. Based on the relative abundance of these shared hub bacteria during the entire experimental period for the three subjects, a PCA plot was further depicted (Figure 4B). A clear separation of the baseline period (week 0) from other score points was observed, which revealed a significant change with regard to these seven OTUs which occurred after tea infusion drinking. The PCA score plots of week 4 and week 8 were gathered into two discrete clusters, which indicated a time-dependent response of these bacteria to tea drinking. For week 12, this cluster was in-between those of week 4 and week 8, indicating a relatively similar bacterial profile pattern in the follow-up period with tea treatment. The temporal shifts of these seven shared hub salivary microbiota during the 12-week experimental period are reflected in Figure 5. In general, compared with the baseline period, in week 4, *Ruminococcaceae* sp. (OTU\_33) and *Haemophilus* sp. (OTU\_696) were suppressed significantly (*p* < 0.05), while *Veillonella* sp. (OTU\_133), *Actinomyces odontolyticus* (OTU\_42), and *Gemella haemolysans* (OTU\_6) were promoted significantly (*p* < 0.05). After eight weeks of tea consumption, *Streptococcus* sp. (OTU\_1), *Ruminococcaceae* sp. (OTU\_33), and *Haemophilus* sp. (OTU\_696) were suppressed significantly (*p* < 0.05), while *Veillonella* spp. (OTU\_133 and OTU\_23), *Actinomyces odontolyticus* (OTU\_42), and *Gemella haemolysans* (OTU\_6) were promoted significantly (*p* < 0.05). In the follow-up period, only *Streptococcus* sp. (OTU\_1) was return to its initial level (*p* > 0.05).

**Figure 3.** Correlation networks of the predominant salivary microbiota (**A**,**C**,**E**) and heatmaps of the hub salivary microbiota (**B**,**D**,**F**) in each subject. In correlation networks, each node represents an OTU; the color of nodes indicates the phylum information; the size of nodes represents the number of linkages; lines between nodes represent a strong correlation between these two OTUs (|r| > 0.9 and *p* < 0.05, Pearson's correlation); red line represents a positive correlation and blue line represents a negative correlation. The nodes with high strong connection numbers were selected as the "hub microbiota" and their dynamic shifts of relative abundance were further depicted on heatmaps. The color of the data matrix in heatmaps corresponds to the normalized relative abundance of the OTUs; the color bar on the top right indicates the scale.

**Figure 4.** (**A**) The Venn diagram of the hub salivary microbiota in each subject. (**B**) PCA score plots based on the relative abundance of the shared hub microbiota across the three subjects.

**Figure 5.** The temporal shifts of the shared hub salivary microbiota during the 12-week experimental period.
