Introduction

Periodontal disease is a global health issue affecting a substantial proportion of the population1. Periodontitis is a chronic inflammatory disease that primarily damages the supporting tissues surrounding the teeth, including the gingiva, periodontal ligament, and alveolar bone, and can cause tooth mobility and eventual tooth loss2. Furthermore, periodontitis is associated with systemic diseases, such as cardiovascular diseases, aspiration pneumonia, and cancer, emphasizing its potential impact on overall health3.

Subgingival microbiota, a community of microorganisms residing in the gingival sulcus or periodontal pockets, has been implicated in the onset and progression of periodontitis by inducing harmful inflammatory responses in the host4,5. Therefore, numerous previous studies have attempted to profile the subgingival microbiota to elucidate the etiology of periodontal disease6,7,8,9,10. The advent of the PacBio sequencing system and full-length 16S ribosomal RNA (rRNA) sequencing offer new opportunities for studying microbial communities with high accuracy and resolution11,12. Although full-length PacBio sequencing has been applied to periodontal and peri-implant samples in a few studies, its potential in oral microbiome research, particularly for analyzing subgingival plaque microbiota, remains underexplored13,14,15,16,17. Accurate identification of subgingival bacterial species could provide key information about the pathogens associated with periodontitis and deepen our understanding of the etiology of periodontal disease.

The difficulty in profiling the subgingival microbiota is partly due to the proximal presence of supragingival plaque and the difficulty in discriminating bacterial species from the subgingival specimen. We collected supragingival plaque (SUPP), subgingival plaque (SUBP), and tongue coating (TC) samples from 34 Japanese adults diagnosed with severe periodontitis who visited dental clinics. A comparison of the subgingival microbiota with other niche microbiota, particularly those of supragingival microbiota, enabled the identification of bacterial species specific to subgingival plaque. The relationship between microbiota in different oral niches has not yet been determined using full-length PacBio sequencing. We accurately determined the bacterial composition of SUPP, SUBP, and TC samples using full-length 16S rRNA gene sequencing and an amplicon sequence variant (ASV) approach, and characterized the subgingival plaque microbiota by comparing them.

Results

Participants’ characteristics and full-length 16S rRNA gene sequences

We examined 34 participants (14 men and 20 women, aged 29–77 years) with ≥ 20% of probing sites with probing depth (PD) ≥ 4 mm, who visited eight dental clinics in Japan. Their detailed characteristics are summarized in Table 1. The mean number of present teeth was 25.6 ± 3.8 teeth (mean ± SD). Notably, all participants exhibited a high percentage of periodontal sites with PD ≥ 4 mm (53.9 ± 19.4%), and the mean percentage of bleeding on probing (BOP) sites was 60.4 ± 25.1%. A total of 102 samples were analyzed using PacBio long-read sequencing with full-length 16S rRNA gene amplicon analysis to determine the bacterial composition. Denoising with DADA218 resulted in 792,354 denoised reads (4,893 ± 2,803 reads per SUBP sample, 5,172 ± 3,918 reads per SUPP sample, and 13,239 ± 5,490 reads per TC sample) and 7,835 ASVs. Of all ASVs, 7,412 ASVs (97.7% of all reads) with ≥ 98.5% identity with the reference sequences in eHOMD were clustered into 446 species-level taxa.

Table 1 Characteristics of the study participants.
Full size table

Microbial differences and the link between tongue and plaque microbiota

We compared the overall bacterial composition of SUBP samples with those of SUPP and TC samples. Both SUBP and TC samples exhibited significantly higher species diversity than SUPP samples (P < 0.05, Fig. 1). Principal coordinate analysis (PCoA) based on the Bray–Curtis distance at the species level demonstrated a high degree of similarity in bacterial composition between the SUBP and SUPP samples, which was significantly different from that of the TC samples (P < 0.01, Fig. 2, P < 0.001, Supplementary Fig. S1). Eight classes and 14 genera were identified as the predominant classes and genera with ≥ 3% mean relative abundance (Fig. 3 and Supplementary Fig. S2). Porphyromonas and Peptostreptococcus were significantly more abundant in the SUBP microbiota than in the SUPP microbiota (P < 0.05). In contrast, Capnocytophaga, Leptotrichia, and Actinomyces were the most abundant genera in SUPP microbiota. Veillonella, Prevotella, Rothia, and Schaalia were the predominant genera in TC microbiota. According to the species-level heatmap, Streptococcus salivarius, Prevotella melaninogenica, Neisseria perflava, and Rothia mucilaginosa were abundant and specific to the TC microbiota (Supplementary Fig. S3). Although the SUPP and SUBP microbiota share several bacterial species, a clustering approach identified a SUBP-specific group, including Porphyromonas gingivalis, Streptococcus intermedius, and Parvimonas micra.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Comparison of alpha diversity across subgingival plaque (SUBP), supragingival plaque (SUPP), and tongue coating (TC) microbiota. Alpha diversity was assessed using the number of species in samples collected from 34 patients with severe periodontitis. Statistical differences between niches were analyzed using the Wilcoxon signed-rank tests with Benjamini–Hochberg false discovery rate (FDR) adjustment. *P < 0.05.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Principal coordinate analysis (PCoA) plot based on the log-transformed Bray–Curtis distance at the species level. Samples from subgingival plaque (SUBP), supragingival plaque (SUPP), and tongue coating (TC) are depicted in different colors (n = 34). Ellipses represent 67% confidence intervals for each niche. **P < 0.01.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Compositional differences across subgingival plaque (SUBP), supragingival plaque (SUPP), and tongue coating (TC) microbiota. Fourteen genera with ≥ 3% mean relative abundance in any of the three microbiota are displayed. Statistical differences were identified using the Wilcoxon signed-rank test with Benjamini–Hochberg false discovery rate (FDR) adjustment. *P < 0.05, **P < 0.01, ***P < 0.001.

In addition to compositional differences, the extent of the microbial link between the tongue and plaque microbiota was further examined. Among 206.8 ± 124.0 (mean ± SD), 94.2 ± 53.5, and 104.9 ± 59.7 ASVs observed in TC, SUPP, and SUBP, respectively, 20.1 ± 16.4, 22.4 ± 17.8, and 41.4 ± 30.3 ASVs were shared between TC and SUPP, TC and SUBP, and SUPP and SUBP, respectively, in each participant (Supplementary Fig. S4). In the TC microbiota, ASVs shared with SUPP and SUBP accounted for 13.2 ± 16.8% and 12.6 ± 14.0% (mean ± SD, Supplementary Fig. S4). Meanwhile, ASVs shared with TC occupied 37.2 ± 25.6% of SUPP microbiota and 39.6 ± 26.0% of SUBP microbiota. Between dental plaques, 62.8 ± 24.0% of SUPP and 59.2 ± 23.6% of SUBP were shared.

Subgingival plaque-specific bacteria identification

Among the 446 species included in the species-level analysis, we focused on identifying bacteria that were specifically enriched in SUBP. Species were considered SUBP-specific if their mean relative abundance in the SUBP samples was higher than that in the SUPP and TC samples, with an FDR-adjusted P-value < 0.05. Thus, 17 species, including Porphyromonas gingivalis and Tannerella forsythia, met the criteria and they collectively accounted for 18.9 ± 20.2% (mean ± SD) of the SUBP microbiota (Figs. 4 and 5). Although these species were also detected in SUPP and TC samples, their relative abundance was considerably lower, at 2.1 ± 3.9% and 2.1 ± 5.6%, respectively (Fig. 5). Among the SUBP-specific species, Mogibacterium timidum, T. forsythia, Parvimonas micra, and Desulfobulbus sp. HMT-041 exhibited high SUBP specificity (Q-value < 0.001 for both SUBP versus SUPP and SUBP versus TC comparisons). In contrast, although P. gingivalis demonstrated high specificity in the comparison between SUBP and SUPP (Q-value < 0.001), its specificity was lower than when comparing SUBP to TC (Q-value < 0.05; Fig. 4).

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

Oral niche specificity of species enriched in subgingival plaque (SUBP) microbiota. The scatter plot shows the results of Wilcoxon signed-rank tests with Benjamini–Hochberg false discovery rate (FDR) adjustment, comparing the relative abundance of bacterial species between SUBP and supragingival plaque (SUPP), as well as between SUBP and tongue coating (TC). Human microbial taxon (HMT) numbers in the eHOMD are shown in parentheses after bacterial names. The x-axis and y-axis represent the negative log10-transformed Q-values (adjusted P-values) for comparing SUBP and SUPP (Q1) and SUBP and TC (Q2), respectively. Dashed lines at − log10(0.05) indicate the significance thresholds (Q-value = 0.05). Colored points represent species that are significantly enriched in SUBP (Q-value < 0.05 for both comparisons) and have higher relative abundance in SUBP than in both SUPP and TC. Gray points represent species that have higher relative abundance in SUBP than in both SUPP and TC but do not reach statistical significance (Q-value ≥ 0.05 in either or both comparisons). The size of each point represents the mean relative abundance of the species in the SUBP microbiota.

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

Differentially abundant species in subgingival plaque (SUBP) compared to supragingival plaque (SUPP) and tongue coating (TC). The box plots show the relative abundance (%) of 17 bacterial species that were significantly enriched in the SUBP microbiota compared to both SUPP and TC microbiota. Human microbial taxon (HMT) numbers in the eHOMD are shown in parentheses following bacterial names. Species were considered significantly enriched if they had an adjusted P-value < 0.05, as determined by the Wilcoxon signed-rank tests with Benjamini–Hochberg false discovery rate (FDR) adjustment.

Discussion

We employed PacBio full-length 16S rRNA gene amplicon sequencing to accurately characterize and compare the microbial communities inhabiting the SUBP, SUPP, and TC niches in 34 participants with severe periodontitis. Our primary objective was to use the improved species-level resolution provided by this method to identify microbial species with strong specificity for the subgingival environment, which is a critical site in the pathogenesis of periodontal disease. Species-level analysis revealed a panel of 17 species that were significantly more abundant in the SUBP microbiota than in both the SUPP and TC communities (Figs. 4 and 5). These largely consisted of previously reported discriminating bacterial species in periodontitis SUBP samples9,19. Similar results were obtained using another comparative approach, a discriminant analysis (Supplementary Fig. S5). This group of SUBP-specific bacteria consists of obligate anaerobic bacteria, including well-known periodontopathogens such as P. gingivalis and T. forsythia20, and together accounted for 18.9% of the SUBP microbiota. In contrast, the relative abundance of these species was significantly lower in the SUPP (2.1%) and TC (2.1%) microbiota, underscoring their strong ecological preferences for the subgingival niche. This high-resolution profiling of SUBP microbiota provides precise information on important targets for diagnostic or therapeutic strategies for periodontal disease.

Detailed analysis of specificity profiles (Fig. 4) revealed clear differences among the 17 SUBP-specific bacteria. Four species–Mogibacterium timidum, T. forsythia, Parvimonas micra, and Desulfobulbus sp. HMT-041 exhibited particularly high SUBP specificity with low Q-values (< 0.001) compared to both SUPP and TC. These species are promising candidates for subgingival plaque biomarkers or may play a central role in maintaining the distinct ecology of this niche21,22,23,24,25. Interestingly, among the 17 SUBP-specific bacteria, P. gingivalis exhibited stronger specificity for the SUBP niche than for SUPP (Q < 0.001), whereas its distinction from TC was less pronounced (Q < 0.05), indicating a moderate degree of SUBP preference for tongue coating. In contrast, T. forsythia demonstrated significantly higher specificity for SUBP than for both SUPP and TC, with low Q-values in both comparisons. This specificity was not observed in other predominant Tannerella species, such as Tannerella species HMT-286, suggesting a more exclusive adaptation of T. forsythia to the subgingival environment (Supplementary Fig. S3). The ASV-level ternary plot revealed that two ASVs assigned to T. forsythia were specific to SUBP, whereas some of the eight ASVs assigned to P. gingivalis were broadly distributed toward TC (Supplementary Fig. S6). This difference in SUBP specificity suggests that, although both are key subgingival plaque species, certain P. gingivalis may possess a slightly broader capacity for survival under microaerophilic conditions or transient colonization in other oral niches, such as the tongue coating, compared to the more fastidious, strictly anaerobic T. forsythia26,27,28,29. These differences in niche specificity may influence how these species spread, persist, and contribute to the onset, progression, and recurrence30.

Our findings also suggest a microbial link across sites, in addition to site specificity. Among the three sites, ASVs composing SUPP and SUBP microbiota were highly shared (62.8 ± 24.0% of SUPP and 59.2 ± 23.6% of SUBP) (Supplementary Fig. S4). Considering that these bacteria are in contact with the tooth surface, bacterial sharing is reasonable. Meanwhile, only 13.2 ± 16.8% and 12.6 ± 14.0% of TC microbiota were shared with SUPP and SUBP microbiota, suggesting that most TC microbiota is composed of TC-specific bacteria (Supplementary Fig. S3). Interestingly, 37.2 ± 25.6% of SUPP microbiota and 39.6 ± 26.0% of SUBP microbiota were shared with TC microbiota. These results suggest a microbial link between the tongue and plaque microbiota, although they are not anatomically adjacent. We have previously demonstrated that the bacterial composition of the tongue microbiota was associated with dental plaque accumulation or dental caries experience31,32,33, suggesting that the tongue microbiota could influence the bacterial composition of the dental plaque microbiota and its pathogenicity. Considering the site-specialist hypothesis, bacteria shared with TC in the SUPP and SUBP may play an accessory role rather than being structurally integrated members of the dental plaque community34. The co-occurrence network analysis revealed that the major cohabiting group in SUBP samples was composed of only SUBP-specific bacteria (Supplementary Fig. S7). The functional and structural associations between tongue bacteria and dental plaque communities should be further examined in future studies. Furthermore, considering that the role of bacteria is different in each niche, it also highlights the importance of considering and modeling the distinct clinical environments of oral niches in mechanistic and antimicrobial studies.

The present study had several limitations. First, this comparative analysis was confined to plaque and tongue samples. A more comprehensive assessment and understanding of the specificity of the SUBP microbiota can be achieved by including other distinct oral niches, such as the buccal mucosa, keratinized gingiva, and hard palate, as characterized previously35,36. Second, dental plaque samples were pooled from multiple sites per participant. Although this approach provides an overview of the plaque microbiota, it may obscure potential site-specific differences in bacterial composition between deep and shallow periodontal or peri-implant pockets37,38,39,40. Future studies employing site- and depth-specific sampling strategies may reveal finer spatial heterogeneity within the subgingival microbial communities. Third, because most participants were recruited at the first visit, no standardized instructions were available before sampling, such as refraining from eating or drinking. Fourth, we could not clinically characterize the participants according to the current classification system of periodontitis41 owing to the lack of information on clinical attachment level, radiographic data, smoking habits, and diabetic status. Fifth, the recruitment of 34 patients with severe periodontitis from dental clinics may have introduced selection bias, potentially limiting the generalizability of our findings to broader populations or individuals with milder forms of periodontal disease. Future studies should, therefore, involve larger and more diverse cohorts to validate these observations. Sixth, because the present findings were based on the 16S rRNA gene amplicon analysis, the functional features of SUBP-specific species were not determined. Metagenomic and metatranscriptomic approaches can reveal the details of their involvement in periodontal diseases.

In conclusion, we conducted a full-length 16S rRNA gene amplicon analysis to compare bacterial communities in SUBP, SUPP, and TC at a high resolution. We identified 17 species that were significantly enriched in the subgingival niche, underscoring the selective ecological pressures within periodontal pockets. Furthermore, quantification of SUBP specificity revealed that four species, including T. forsythia and P. micra, exhibited exceptionally high SUBP specificity. Moreover, we uncovered subtle but potentially significant differences in SUBP specificity among established periodontopathogens, such as P. gingivalis and T. forsythia. These findings enhance our understanding of the subgingival microbiota in periodontitis and highlight highly specific subgingival microbial targets, which may inform the development of effective diagnostic tools or targeted therapeutic strategies for the management of periodontal disease.

Methods

Study participants

This study included 42 patients who visited eight dental clinics in Japan. We only included patients who had ≥ 20% of probing sites with PD ≥ 4 mm to focus on the subgingival plaque microbiota and not used antibiotics within one month before sample collection42,43. Among these, 13 were included in a previous study44. Patients who used antibiotics within 1 month before sample collection were not recruited. After excluding eight participants with missing data (n = 5), insufficient samples for analysis (n = 2), and failure to obtain appropriate PCR amplicons for sequencing (n = 1), a total of 34 participants (14 men and 20 women) were included in the final analysis.

Sample collection and clinical information

Sample collection and dental examinations were performed according to a previously described protocol45. Plaque samples were collected from the upper half of the jaw to simplify the sampling steps. SUPP samples were collected from all tooth surfaces on the upper half of the jaw, specifically from the side with the highest number of teeth, using sterile curettes and pooled into sterile tubes. SUBP samples were subsequently obtained and pooled from the gingival crevices in the same region using sterile curettes. TC samples were collected by scraping the dorsal surface of the tongue using a sterile plastic spatula. The collected TC samples were immediately transferred into sterile tubes containing 500 µL of lysis buffer (9 mM Tris-HCl, 0.9 mM EDTA, and 1% sodium dodecyl sulfate). All samples were transported to the laboratory on ice and stored at − 80 °C until further analysis. Following sample collection, the PD and BOP were measured at six sites per tooth (mesiobuccal, midbuccal, distobuccal, mesiolingual, midlingual, and distolingual) using a periodontal probe.

Full-length 16S rRNA gene amplicon sequencing

Different pretreatment procedures were performed for each type of sample before DNA extraction. The SUPP and SUBP samples were resuspended in 200 µL of lysis buffer. For the TC samples, 200 µL of the resuspended supernatant was transferred into new sterile tubes after the spatulas were removed following centrifugation (15,000 rpm, 15 min, 4℃). After the pretreatment procedures, the DNA was extracted from each sample using the bead-beating method described previously46. The full-length 16S rRNA gene, containing all variable regions, was amplified using the primers 8 F (5’-AGA GTT TGA TYM TGG CTC AG-3’) with the sample-specific 8-base barcode sequence and modified 1492R (5’-GGH TAC CTT GTT ACG ACT T-3’)47. PCR amplification and purification were performed as previously described46. The purified amplicons were sequenced on a PacBio Sequel IIe system (Pacific Biosciences; Menlo Park, CA, USA) using the Sequel II Sequencing Kit 2.0 at Kazusa Genome Technologies, and circular consensus sequencing (CCS) reads were generated. Only high-fidelity (HiFi) CCS reads with ≥ 3 full-pass subreads and ≥ 20 quality values were finally used for analysis.

Data analysis and taxonomy assignment

The HiFi reads were primarily quality-checked using the R software (version 3.6.2; R Foundation for Statistical Computing, Vienna, Austria) and were excluded from the analysis when they exhibited < 1,000 bases or did not include the correct forward and reverse primer sequences. The remaining HiFi reads were demultiplexed by examining 8-base tag sequences at the forward end, and the forward and reverse primer sequences were trimmed. Quality filtering, denoising, and chimera filtering procedures were performed using the DADA2 pipeline (version 1.26.0) with default settings for PacBio reads, and an ASV table was subsequently created18. The class- and genus-level taxonomies of each sequence variant were determined using the RDP classifier with a minimum support threshold of 80% and RDP taxonomic nomenclature48. Sequence variants were taxonomically assigned at the species level using BLAST49 against 1,015 oral bacterial 16S rRNA gene sequences in eHOMD (eHOMD 16S rRNA RefSeq version 15.22)50. The nearest-neighbor species with ≥ 98.5% identity was selected as candidates for each sequence, and a species-level taxonomic table was generated by aggregating ASVs that were assigned to identical reference sequences in the eHOMD. ASVs without hits were excluded from the species-level analysis.

Ethics statement

The Ethics Committee of Kyushu University approved the present study and the procedure for obtaining informed consent (approval number: 2019 -105). All experiments in this study were performed in accordance with the relevant guidelines and regulations, including the principles of the Declaration of Helsinki. Written informed consent was obtained from all the participants prior to inclusion in the study.

Statistical analysis

All statistical analyses were performed using the R software. For all multiple comparisons, P-values were adjusted using the Benjamini–Hochberg method to control the false discovery rate (FDR). An FDR-adjusted P-value (Q-value) < 0.05 was considered statistically significant. Alpha diversity was assessed using the number of observed species calculated following rarefaction of 1,000 reads per sample using the diversity function in the vegan package. Differences in the number of species in the three oral niches (SUBP, SUPP, and TC) were compared using the Wilcoxon signed-rank test. Beta diversity was evaluated using Bray–Curtis distances based on a log-transformed species-level table. Significant differences in the overall microbiota structure among the niches were assessed by performing a permutational multivariate analysis of variance using the adonis2 function in the vegan package with 999 permutations. Community similarities were compared by comparing the inter-niche log-transformed Bray–Curtis distances using the Wilcoxon signed-rank test. In this study, ASVs detected in the TC and SUPP microbiota were defined as TC-SUPP-shared ASVs for each participant, and ASVs shared between the TC and SUBP microbiota were regarded as TC-SUBP-shared ASVs. The number and total abundance of the shared ASVs were calculated for each participant. Differential abundance analysis was performed at both the genus and species levels. Pairwise comparisons were performed using the Wilcoxon signed-rank test to identify genera with significantly different abundances among the three oral niches. SUBP-specific species were identified by comparing the relative abundance of each species in SUBP samples with its abundance in SUPP and TC samples using Wilcoxon signed-rank tests.