The short-term impact of probiotic consumption on the oral cavity microbiome

The dysbiosis of the oral microbiome is associated with both localized and systemic diseases. Modulating the resident microbial communities by the dietary consumption of probiotics has become an appealing means to promote host health by either restoring host-microbe balance or preventing dysbiosis. Most probiotics strategies target the intestinal microbiome, but little is known about their impact on the oral microbiome. We analyzed here the saliva microbiome from 21 volunteers, longitudinally collected before, during, and after consumption of a commercial probiotic and a standard yoghurt using 16S amplicon sequencing. The alpha diversity of the saliva microbiome had a statistically significant increase (P-value = 0.0011) in one of the groups that consumed the probiotic. The overall structure of the microbiome was however not significantly impacted by the probiotic, although oligotyping analysis revealed that both Streptococci and Lactobacilli present in the probiotic product persisted in the saliva microbiome. In contrast, non-probiotic yoghurt consumption had a lesser impact on the overall diversity and Lactobacillus and Streptococcus persistence. Our results suggest that consumption of commercial probiotics in healthy subjects increase the overall diversity of the oral cavity microbiome in the short term, but such dietary interventions are not able to substantially modify the structure of the microbiome.

group of volunteers consumed the probiotic product (at breakfast and lunch respectively) consisting of a commercial fermented milk drink ("Coop Probiotico Bianco" [COOP Italia s.c. Bologna (IT)], 100 gr) containing S. thermophilus, L. delbrueckii subsp. Bulgaricus and L. paracasei strains, while the second group consumed the same amount of a non-probiotic yoghurt ("Mila Bianco" [Mila latte montagna, Alto Adige (IT)], 100 gr). On the last day of sampling (T3), the first group reverted back to the normal diet while the second and the third groups ate the probiotic product. The baseline is defined at T1, without any probiotic or nonprobiotic yoghurt ingestion. Group 1 was designed to evaluate the influence of a single probiotic intake, while Group 2 was designed to assess the differences between probiotic and non-probiotic yoghurt consumption. The last group was used to investigate a repeated probiotic intake (after lunch at both T2 and T3). All samples were collected after lunch, while yoghurt or probiotic ingestion occurred at breakfast (~5 hours before sample collection) or at lunch (~1 hour before sample collection) as reported in the figure. (B) Taxonomic profiling at genus level for all groups and timepoints. (C) Principal Coordinates Analysis (PCoA) plot of OTU-based beta diversity (weighted Unifrac) with samples colored by group. (D) Quantitative taxonomic composition of the probiotic fermented milk drink ("Probiotic") and the non-probiotic commercial yoghurt ("Yoghurt") as derived by OTU analysis and (E) by oligotyping analysis. The saliva microbiome was investigated by 16S rRNA gene sequencing producing a total of 4.6 million reads (39,3 k ± 14,2 k reads per sample, Supplementary Table 2). We also sequenced the probiotic and non-probiotic products to assess their composition (56,3 k and 12,0 k reads respectively). By applying the QIIME pipeline 22 to the dataset (see Methods), we generated a total of 4503 operational taxonomic units (OTUs) clustered at 97% identity with an open-reference OTU picking strategy.
The three most abundant phyla in the saliva samples were Proteobacteria, Firmicutes and Bacteroidetes, with average relative abundances of 36.4%, 28.7%, and 18%, respectively. Fusobacteria (10.5%) and Actinobacteria (5.5%) were less abundant, but still present in all samples (minimum abundance of 0.2% and 0.9% respectively) (Supplementary Figure 1A). At the genus level, Prevotella, Neisseria, Haemophilus, and Streptococcus were the dominant genera (Fig. 1B), which is consistent with previous studies of the healthy oral microbiome 6,23 . The three sets of biological replicates we performed showed a very low technical variability (Supplementary Figure 1B).
The complexity of the saliva microbial communities in subjects who ate the probiotic drink tended to be higher than for those who ate the standard yoghurt (Supplementary Figure 2A), confirming our previous finding 21 although the result was not statistically significant (P-value 0.081). The samples associated with repeated probiotic intake were characterized by a higher alpha diversity than standard diet and single probiotic intake ones (Supplementary Figure 2C) but, again, this was not supported by statistical significance.
Samples did not show a strong grouping by probiotic intake when displaying their inter-sample beta diversity with ordination plots (Fig. 1C), which was expected based on our previous observations 21 . Instead samples tended to cluster by subject, according to both standard OTU analysis and subspecies level oligotyping analysis 24 (Supplementary Figure 3) of the four most abundant genera (Streptococcus, Haemophilus, Prevotella, and Neisseria), confirming the presence of a personal microbial signature in the saliva 25,26 (Supplementary Figures 4 and 5). We also profiled both the probiotic fermented milk drink and the non-probiotic yoghurt (Fig. 1D) finding that, in both communities, Streptococcus and Lactobacillus were the most dominant genera as these encompass the bacteria responsible for fermentation of the products. In the probiotic product, Streptococcus was enriched (76%) compared to the non-probiotic (53%), while Lactobacillus was characterized by an opposite trend (20% in probiotic, 41% in non-probiotic yoghurt). This is compatible with the known composition of the probiotic, enriched in S. thermophilus, L. delbrueckii subsp. bulgaricus and L. paracasei strains. Both the probiotic and non-probiotic product shared the most abundant Streptococcus oligotype and the four most abundant Lactobacillus oligotypes (Fig. 1E).
Evidence of increased complexity of the oral microbiome after short-term probiotic consumption. The longitudinal analysis of the alpha diversity in the three groups ( Fig. 2 and Supplementary Figure 6) revealed that probiotic intake is significantly associated with an increase in the microbial community diversity for Group 1 (P-value of 0.0011 between T1 and T2). The non-probiotic yoghurt consumption group also showed an increased alpha diversity but to a lesser extent than its probiotic counterpart (P-value of 0.0204 between T1 and T2 in Group 2). Group 3 also consumed the probiotic, but the observed increase in average alpha diversity was not significant. Probiotic consumption is thus associated with a higher, albeit limited, enrichment in the microbial complexity of the saliva, compared to the non-probiotic yoghurt and standard diet.
We further investigated the differences in microbiome composition across time-points and study groups. Ordination analysis showed no apparent clustering by group, suggesting that the overall structure of the saliva microbiome is not affected by short-term probiotic intake (Fig. 1C). Indeed, no statistically relevant difference was found in beta diversity analysis across timepoints and groups (Supplementary Figure 7). Moreover, no significant variations were observed when comparing intra-and inter-samples distances at different timepoints (Supplementary Figure 8). Although the probiotic consumption does not change the microbiome composition to a large extent, the large overall variability could potentially hide the small-scale changes driven by probiotic Overall, despite an increase in microbiome complexity after probiotic intake, no strong variation in the taxonomic structure of the saliva microbiome was observed.
Persistence of the probiotic strains in the saliva microbiome. We then investigated whether bacteria present in the probiotic product (S. thermophilus, L. delbrueckii subsp. bulgaricus and L. paracasei) were detectable in the saliva samples to evaluate the presence and persistence of the probiotic strains in the saliva. As a genus, Streptococcus was present in all subjects, and even OTU analysis cannot reach the resolution needed to track the probiotic strain. Nevertheless, finer resolution oligotyping showed a higher abundance of the probiotic Streptococcus oligotypes in subjects who ate the probiotic with respect to non-probiotic oligotypes of the same genus (37.1% average increase after probiotic consumption in Group 1 and 28.7% average increase in Group 3, P-values: 0.37 and 0.51, respectively) (Fig. 3). When the subjects reverted to the standard diet also the probiotic-specific Streptococcus oligotypes returned to a condition similar to the baseline, confirming the short time persistence of oral probiotic organisms (P-value: 0.17, 51.6% average decrease in the only group with standard diet following the probiotic consumption timepoint, Group 1). The same oligotyping analysis performed for probiotic Lactobacillus highlighted a stronger effect compared to Streptococcus, probably because the former is more likely to be confounded with other Streptococci commonly present in the human saliva. The probiotic Lactobacillus OTUs and oligotypes were indeed found enriched (P-values OTUs for the three study groups: 0.21, 0.12 and 0.22; P-values oligotypes: 0.27, 0.008 and 0.004) only in groups which consumed the fermented milk product (Lactobacillus oligotypes relative abundances increased on average 66.7% and 83.3% after probiotic consumption at breakfast in Group 1 and at lunch in Group 3, respectively), with the only exception of the first time-point of the second group (associated with standard diet), in which the Lactobacillus oligotypes were present at relatively high abundance (28.6%). In contrast, no significant enrichment was found associated with the non-probiotic yoghurt consumed at breakfast (Fig. 3). In general, the probiotic-associated Lactobacillus OTUs and oligotypes showed a longer term persistence in the oral microbiome than the Lactobacillus OTUs and oligotypes associated with the non-probiotic yoghurt intake.

Discussion
In this work we extended our previous analysis 21 to verify the hypothesis that the intake of commercially available probiotic products has a directly effect on the diversity and composition of the saliva microbiome, at least at short timescales. Our longitudinal study design showed that the diversity of the saliva microbiome is increased after consumption of the probiotic product. The overall taxonomic and abundance distribution of bacterial genera is however minimally influenced by probiotic intake even though we verified that the bacteria present in the probiotic product persist to a certain extent in the saliva microbiome. In particular, the impact of probiotic Streptococci is limited in that it only slightly modifies the overall abundance of this already resident genus. Stronger support for short-term persistence was found for Lactobacillus, probably because this organism is not common in the oral microbiome. Altogether, our study confirms that the oral microbiome can potentially be modulated by commercial probiotic products, but that this modulation is limited in both the breadth of the change and in its temporal persistence. Additional studies are required to verify whether continued (e.g. daily) intake of probiotics for longer periods can have a larger and more long-lasting impact on the structure of the oral microbial communities.

Methods
Cohort and Study design. Saliva samples were collected from 21 age-matched healthy volunteers (10 males, 11 females, avg. age 23 s.d. 1,23) that were sampled at three different time-points, following the protocol in Fig. 1A and as described below. We collected 21 samples for each time-point and, for the first time-point, three samples were collected in duplicate as controls for a total of 66 samples. All participants gave their informed consent, and this study was approved by the ethical committee of the University of Trento (Reference number 2015-014); all experiments were performed in accordance with the relevant guidelines and regulations for human subjects research. Each volunteer completed a questionnaire about oral care habits (including frequency of teeth brushing, use of dental floss or mouthwash). Current smokers and individuals using antibiotic treatments in the last six months were excluded from the study. Volunteers were instructed to self-collect the saliva samples and were asked not to brush their teeth, use dental floss or mouthwash before sample collection. They were also requested to avoid intimate kissing. Self-collection was performed under supervision and directly in the laboratory. Volunteers were randomly assigned into three groups of seven subjects (Fig. 1A). For all the three groups, no administration of probiotic or non-probiotic yoghurt was perfomed for the first time-point in order to define the baseline. The second collection after the probiotic administration for the first and the third group and the non-probiotic yoghurt for the second group. At the third sampling, the first group returned to the standard diet, the second switched to the probiotic and the third group repeated the probiotic intake. The probiotic was a commercial fermented milk drink ("Coop Probiotico Bianco" [COOP Italia s.c. Bologna (IT)], 100 g) enriched in S. thermophilus, L. delbrueckii subsp. bulgaricus, L. paracasei strains and supplemented with vitamin B6 and vitamin D, conserved at +4 °C until consumption. The same amount of non-probiotic commercial yoghurt ("Mila Bianco" [Mila latte montagna, Alto Adige (IT)], 100 g) was administered as a control.
Sample collection and DNA extraction. Samples, including technical replicates, were collected through a passive drooling procedure in sterile 50 ml screw top tubes (approximately 3 ml of saliva per sample). About 30 minutes before collection, all volunteers performed a mouth rinse with drinking water. Samples were vortexed before 2 ml of each was centrifuged at 10,000 × g for 10 minutes to collected the bacterial pellets. Total DNA was extracted using the PowerSoil DNA Isolation Kit (MoBio), according to manufacturer's instructions. The same extraction procedure was also performed on 2 ml of the probiotic drink, the non-probiotic yoghurt and pure water as a negative control. DNA was eluted in 100 μl molecular-grade water and the DNA purity, yield, and concentration was assessed with the NanoDrop ND-2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). DNA was stored at −20 °C until used.
16S rRNA hypervariable region V4 amplification library preparation and sequencing. The V4 hypervariability region of 16 rRNA gene was amplified using Illumina adapted primer 515F (GTGYCA GCMGCCGCGGTAA), and Illumina adapted barcoded 806R primers (GGACTACNVGGGTWTCTAAT) as described previously 28,29 . The complete list of sample-specific barcodes is included as supplementary material (Supplementary Table 1 The duplicate amplicon libraries were pooled, visualized and extracted from a 1.5% agarose gel using the Wizard SV Gen and PCR Clean-Up System (Promega Italia s.r.l., Milano, Italy), as described by the manufacturer. The libraries were further cleaned using Agencourt AMPure Xp purification (Beckman Coulter s.r.l, Milan, Italy), following manufacturer's instructions, and quantified using the Qubit Fluorometric Quantitation system (ThermoFisher Scientific-Invitrogen; Life Technolo-gies Europe BV, Monza, Italy). Finally, equal molarities of amplicons were pooled and sequencing was performed with Illumina MiSeq Reagent Kit v2 (300 cycles) as described in 29 .
Data analysis. The Illumina MiSeq sequencing run produced 4.6 million paired-end reads after merging with fastq-join 30 using default parameters. Joined reads had an average length of 250 bp. We obtained on average 39.3 thousands reads for each sample. Average phred quality score per read was 37. QIIME 22 was used to demultiplex and trim low quality reads, reducing the initial reads to 3.38 million (26.5% reduction on average). Trimming was performed with 25 as minimum acceptable quality and 75% as minimum acceptable read length after trimming. FastQC 31 was used to analyze overall quality of the raw data. The median sequence length of joined reads after pre-processing was 253 bp. Operational Taxonomic Units (OTU) were built using 5 as minimum OTU cardinality using UCLUST 32 . The few OTUs present also in the negative controls were discarded. RDP 33 was then used to perform the taxonomic profiling, using Greengenes 13.8 dataset as reference sequences 34 . QIIME was also used to measure alpha diversity and beta diversity. All ordination plots are based on PD-whole tree metric, if not otherwise specified. Biomarker detection was performed using LEfSe 27 . Oligotyping 24 analysis was performed by extracting reads assigned by QIIME to each genus (Streptococcus, Lactobacillus, Haemophilus, Neisseria, Prevotella) and using 10 entropy components for Streptococcus and Lactobacillus, 5 for Neisseria, 9 for Haemophilus and 15 for Prevotella; minimum substantive abundance filter threshold was set to 50 reads in all cases. Ethics approval and consent to participate. This study was approved by the ethical committee of the University of Trento (Reference number: 2015-014) and all participants gave their signed informed consent prior to enrollment. Consent for publication. Not applicable because no individuals person's data are reported in any form.
Data availability. Raw sequences were deposited in the NCBI Short Read Archive (http://www.ncbi.nlm.nih. gov/sra) and are available with accession numbers SRR2093962 to SRR2094047.