Significance

Periodontal diseases are the most prevalent oral inflammatory diseases in humans. Despite growing evidence of oral inflammatory diseases influencing human systemic health, it remains unclear how healthy oral tissues are affected. The human experimental gingivitis model with split-mouth design allows for this study. Here, we highlight measurable changes in both the host response and microbiome in healthy oral tissues distant from tissue sites experiencing acute microbially induced inflammation despite maintained oral hygiene. This host response in healthy tissues precedes a microbiome shift mirroring compositional changes observed in induced inflammatory sites. Furthermore, this nonlocalized response varies in magnitude and timing depending on the designated inflammatory response types of these individuals.

Abstract

Variation in human immune response to the same bacterial or viral pathogen is well established in the literature. Variation in immune response to microbial challenge has also been observed within the human oral cavity. Our recent study focused on characterizing observed variations in microbially induced gingival inflammation—resulting in three distinct clinical Inflammatory Responder Types (IRTs): High-IRT, Low-IRT, and Slow-IRT. Here, we applied a high-resolution temporal multiomic analysis during microbially induced inflammation in order to characterize the effects of localized oral inflammation on distant healthy tissues in young healthy adults. Our results highlight a nonlocalized subclinical effect with alterations in proinflammatory host mediators and an ecological shift toward dysbiosis within the subgingival microbiome in an IRT-dependent manner—despite maintained oral hygiene. Our results provide mechanistic insight into how healthy tissues within humans are influenced by distant localized inflammation and may ultimately become susceptible to disease.
Model systems for studying dynamic host–microbe interactions during inflammatory events directly in humans in a well-controlled and reversible way are limited. A unique advantage of studies in the human oral cavity is the availability of one such model, experimental gingivitis (EG). The EG model permits the high-resolution study of the initiation, development, and maturation of normal dental plaque, a diverse biofilm, along the gingival margin and within the periodontal pocket surrounding human teeth. This results in a relatively rapid host response as inflammation progresses to a clinically diseased state (15). Furthermore, when incorporating a split-mouth design this model also permits the investigation of the effects of localized inflammation in distant otherwise healthy tissues within the human oral cavity.
Periodontitis, a chronic and irreversible form of periodontal disease, is an age associated gingival inflammatory disease that affects more than 795 million adults globally and is even more prevalent than cardiovascular disease (6, 7). Periodontitis is associated with a dysbiotic subgingival dental plaque that is enriched with periodontal disease associated (PDA) bacterial taxa. Many PDA bacteria are gram-negative anaerobes within the Bacteroidetes and Spirochetes phyla, including Porphyromonas, Prevotella, Tannerella, and Treponema genera (812). If left untreated, periodontitis results in a dysregulated and chronic host immune response leading to irreversible structural tissue damage and even alveolar bone loss—both of which are critical for normal tooth health and function. Periodontitis has even been associated with other systemic inflammatory diseases in humans, including rheumatoid arthritis, endocarditis, bacterial pneumonia, type-2 diabetes, and Alzheimer’s (1316). Gingivitis, a reversible and milder form of periodontal disease, is often considered a precursor in the development of proportion of periodontitis cases (17). However, even though a direct connection between the etiopathogenesis leading from gingivitis to periodontitis remains unresolved, it remains a major focus in current periodontal research (18). Additionally, gingivitis and periodontitis generally occur within localized tooth sites, though susceptible individuals may develop multiple diseased sites over different time periods and across their lifespan.
As part of a recent EG study that incorporated a split-mouth design, effectively providing intraoral controls for each subject over the duration of the study, we provided a high-resolution analysis of host mediator, microbiome, and clinical data for 21 young healthy individuals [aged 20 to 27 ( x=22  )] over a period of 7 wk [8 study visits generating a total of 672 subgingival plaque and GCF (Gingival crevicular fluid) samples] with 21 d of induced inflammation. Results from this study not only verified previously observed clinical variations in host response to microbially induced inflammation in the human oral cavity (High and Low responders), but also identified a previously unrecognized delayed clinical response type, Slow responders (1). These variations in gingival inflammation observed in humans are now represented by our three proposed clinical Inflammatory Responder Types (IRTs): High-IRT, Low-IRT, and Slow-IRT. Each of these phenotypes revealed distinctly defining clinical, microbial, and host mediator dynamics within unbrushed test sites. In brief, High-IRT has a rapid plaque growth rate, compositional maturation with rapid increase in PDA bacteria, and high levels of inflammation. Low-IRT shares the characteristics of a rapid plaque growth rate and compositional maturation with rapid increases in PDA bacteria found in High-IRT. However, Low-IRT exhibits lower levels of clinical inflammation. The Slow-IRT has both a delayed plaque growth rate and compositional maturation. Notably, these recently recognized Slow-IRT individuals have higher and sustained levels of gram-positive Streptococcus, delayed enrichment of PDA bacteria, and delayed onset of clinical inflammation. However, these individuals still get high levels of inflammation, similar to the High-IRT (1).
One of the most successful options for treating periodontal diseases, including periodontitis and gingivitis, is the physical debridement or removal of dental plaque (bacterial biofilm) using more invasive oral hygiene methods within affected site(s), such as scaling and root planning (19). Notably, one study has even shown that when these more invasive hygiene methods are applied to diseased tooth sites in patients with periodontitis, it resulted in a reduction of clinical inflammatory measures (increasing health) within distant sites in the mouth (20). These observations indicate there are inducible host immunoregulatory mechanism(s) that link diseased and distant tooth sites within an individual. However, to our knowledge, no study has comprehensively characterized the effects of microbially induced inflammation in distant healthy tissues during the initiation and development of these inflammatory processes at the molecular level.
In the present study, we sought to characterize temporal changes observed in healthy control tissues and tooth sites in relation to natural plaque accumulation and maturation occurring within distant test sites over a period of 21 d. Here, we report significant alterations in healthy control sites that directly result from microbially induced inflammation occurring within distant test sites located contralaterally in the human oral cavity. Our results highlight distinct temporal shifts in multiple host mediators associated with gingival inflammation that precedes a shift in the subgingival microbiome composition, which becomes more similar to distant test sites during EG—although delayed and in lower magnitude. Notably, these temporal changes occur despite maintained oral hygiene, limited visible plaque accumulation, and in the absence of significant clinical inflammation of tissue within these healthy control sites. Furthermore, this phenomenon occurs in an IRT-dependent manner being most evident in the High- and Slow-IRTs while less evident in the Low-IRT.
Results from this study bolster our understanding of the variation in observed host response to localized microbially induced inflammation within the human oral cavity. Here, we highlight significant nonlocalized effects on both host mediator profiles and microbial community composition within healthy tooth sites in response to varying severity of inflammation originating from distal tooth sites located elsewhere in the mouth. These observed changes provide multiple layers of evidence that illustrates how microbially induced gingival inflammation may influence nonlocalized healthy tissues—contributing to the understanding of how healthy sites become enriched in PDA bacteria and may ultimately become susceptible to periodontal diseases, including gingivitis and periodontitis, despite maintained oral hygiene of those sites.

Results

Experimental Design and Clinical Inflammation Measures for Test and Control Sites during EG.

An overview of the experimental design, timeline, sample collection, and analysis is highlighted in Fig. 1A. In brief, 21 young (2027) generally healthy adults were enrolled into a 7-wk EG study with 21 d of naturally induced inflammation (Days 0 to 21) that included five sampling intervals at Days 0 (baseline), 4, 7, 14, and 21 (total of 420 subgingival plaque and GCF samples). All study participants received a professional cleaning at the time of inclusion (Day −14) in order to normalize the oral health status across individuals. At baseline (Day 0), study subjects refrained from brushing randomly assigned test sites, which were protected by a personalized intraoral stent (Materials and Methods), while regular oral hygiene with fluoride containing toothpaste was maintained for the rest of the mouth, including distant control sites located contralaterally to test sites, for the duration of the study. Data from this study were stratified by clinical IRT and no significant differences between sex and age between IRT groups were observed (SI Appendix, Table S1). Additionally, all study test sites responded as expected with abstained oral hygiene—resulting in significant accumulation and maturation of dental plaque, represented by plaque index (PI) (21), and Bacterial Load (16S rRNA copy numbers). These clinical parameters also correspond to the onset of significant clinical inflammation within those same test sites, represented by the gingival index (GI) (22) and bleeding on probing (BOP) (23). Stratifying the data using previously defined responder groups revealed each IRT had significant increases in plaque accumulation (PI and Bacterial Load) (Fig. 1 C and K) and gingival inflammation (GI and BOP) (Fig. 1 E and G) within test sites during this inflammatory Induction phase. Analysis of clinical parameters including PI, GI, BOP, GCF volume, and Bacterial Load between IRT test and control sites over the induction phase are highlighted in SI Appendix, Figs. S1 and S2.
Fig. 1.
EG incorporating a split mouth design reveals control sites are not static and vary by IRT. (A) Study overview highlighting the clinical EG model. (B) Proportion of clinical IRTs (Materials and Methods). (C) PI stratified by IRT test and control sites during the Induction phase (Days 0 to 21). (D) Mean PI by IRT test and control sites during the Induction phase. (E) GI stratified by IRT test and control sites during the Induction phase. Red dashed line represents a GI value of 1.5 and represents the difference between high and low levels of clinically observed inflammation. (F) Mean GI by IRT test and control sites during the Induction phase. (G) BOP stratified by IRT test and control sites during the Induction phase. (H) Mean BOP by IRT test and control sites during the Induction phase. (I) GCF volume stratified by IRT test and control sites during the Induction phase. (J) Mean GCF volume by IRT test and control sites during the Induction phase. (K) Bacterial Load (16S rRNA gene copies) stratified by IRT test and control sites during the Induction phase. (L) Mean Bacterial Load by IRT test and control sites during the Induction phase. Significance level indicated by asterisks: *P < 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001. All significance test results between test and control sites provided in SI Appendix,Tables S6–S11 and Figs. S1 and S2).
PI, which represents visible plaque accumulation along the gingival margin (gum line separating the exposed tooth crown and the gingival sulcus) significantly increased after just 4 d of abstained oral hygiene across IRT test sites. Control sites, which had maintained oral hygiene, remained rather stable during the Induction phase (Days 0 to 21). Only the Slow-IRT showed significant changes in PI during the later stages of the Induction phase (Linear Regression, P < 0.05, Days 0 to 14; P < 0.01, Days 0 to 21) yet lower in magnitude in comparison to respective test sites (Fig. 1 C and D). In contrast to all IRT test sites, no significant clinical inflammation, represented by GI, was observed in any IRT control sites within these generally healthy individuals over the Induction phase (Fig. 1 E and F). Although we observed increasing trends in percent BOP in control sites (%BOP: average of 12 sites for 3 teeth over 20 s) (Fig. 1 G and H), this was only statistically significant within High-IRT (Linear Regression, P < 0.001, Days 0 to 14 and Days 0 to 21) and effected less than 25% of sites measured. Interestingly, bacterial load, which represents the total number of 16S rRNA gene copies within a sample, also trended upward in control sites during the Induction phase within High-IRT (Linear Regression, P > 0.05, Days 0 to 14) and was statistically significant in the Slow-IRT (Linear Regression, P < 0.05, Days 0 to 21).

Changes in Subgingival Plaque Diversity are Observed in Healthy Control Sites and Vary by IRT.

In order to investigate low-magnitude microbial fluctuations within distant healthy tooth sites during the later time points of the Induction phase (Days 0 to 21) in more detail, microbial diversity was assessed for each test and control site stratified by clinical IRT. Alpha diversity was assessed using observed amplicon sequence variants (ASVs) and Shannon indices (Fig. 2A). Although there was no statistically significant increase in alpha diversity within control sites, both High- and Slow-IRT tended to have an increase in richness within those sites over the Induction phase while Low-IRT maintained a rather stable community across individuals.
Fig. 2.
Changes in subgingival plaque diversity are observed in healthy control sites and varies by IRT. (A) Alpha diversity measured by Observed ASVs and Shannon Indices. (B) Bray–Curtis Dissimilarity Index for each IRT test site compared to baseline (Day 0) over the Induction phase (Days 0 to 21). (C) Bray–Curtis Dissimilarity Index for each IRT control site compared to baseline over the Induction phase. (D) Beta Diversity calculated using Weighted Unifrac Distances and visualized by PCoA for each IRT test and control sites. Marginal Boxplots indicate PC1 and PC2 over the Induction phase. Boxes represent the distribution of data and medians ± interquartile ranges; whiskers and outliers >1.5 IQR below (above) the 25th (75th) percentile. Trend lines represent loess regression mean values across all time points. Dashed lines represent test sites, and solid lines represent control sites. Statistical analysis was performed using the nonparametric Wilcoxon-Rank Sum Test adjusted by FDR. Weighted Unifrac Beta Diversity was assessed for significance using the PERMANOVA, 999 permutations. Significance level indicated by asterisks: *P < 0.05, **P ≤ 0.01.
Beta diversity was visualized using a principal coordinate analysis (PCoA) of Weighted Unifrac distances and stratified by IRT test and control sites over the Induction phase (Fig. 2D). While test sites showed clear shifts in beta diversity between Days 0 to 21 for all IRT test sites (PERMANOVA, perm = 999; High-IRT, P < 0.001; Slow-IRT, P < 0.01; Low-IRT, P < 0.05), only High- and Slow-IRTs displayed a similar, although more gradual, shift over the same period within distant control sites (PERMANOVA, perm = 999; High-IRT, P < 0.05, Slow-IRT, P > 0.05). Again, Low-IRT controls showed a relatively stable community over the Induction phase (PERMANOVA, perm = 999; P > 0.05) (SI Appendix, Table S2).
In order to further understand how beta diversity changed over the Induction phase, we assessed Bray–Curtis dissimilarity within IRT test and control sites compared to their respective baseline (Day 0) composition (Fig. 2 B and C and SI Appendix, Table S3). Here, we observed a significant increase in dissimilarity among all IRT test sites (Wilcoxon; High-IRT, P < 0.01; Low-IRT, P < 0.01, Days 0 to 21; Slow-IRT, P < 0.01, Days 4 to 21). Notably, we observed a significant increase in dissimilarity within High- and Slow-IRT control sites over the same period (Wilcoxon; High-IRT, P < 0.01, Days 0 to 21; Slow-IRT, P < 0.05, Days 4 to 21). In addition, no significant changes in dissimilarity were observed in the Low-IRT control sites—further supporting this groups rather stable community composition within healthy control sites throughout the Induction phase.
Regardless of diversity metric, our analysis indicated that healthy control tooth sites, despite having maintained oral hygiene for the study duration and being distal to the test sites, were not static. Rather, these healthy control sites seemed to mimic changes in microbial diversity that were observed during the natural subgingival plaque biofilm development and maturation in the respective IRT test sites and was most evident in the High- and Slow-IRTs.

Healthy Control Sites Reflect Taxonomic Changes Observed in Distant Test Sites in a Responder-Dependent Manner.

To investigate taxonomic changes in microbial community composition between test and control sites during the Induction phase (Days 0 to 21), subgingival plaque samples were initially examined at phylum level for each clinical IRT (SI Appendix, Fig. S4). The top six most abundant phyla and their mean relative abundance within test and control sites over the Induction phase are highlighted in Fig. 3A with statistical significance reported in SI Appendix, Table S4.
Fig. 3.
Microbiome compositions shift within healthy control sites in a responder-dependent manner. (A) Mean relative abundance of phylum level agglomerated data are shown for each clinical IRT Test and Control sites over the Induction phase (Days 0 to 21). Dashed lines represent Test sites, and solid lines represent Control sites. Statistical significance for changes in relative abundance during the Induction phase are reported in table 1. (BG) Fold Change using Genus level agglomerated data over the Induction Phase compared to baseline (Day 0) stratified by IRT Test and Control sites. Genera positively associated with gingival inflammation, gingivitis, and/or periodontitis are highlighted in red. (H - J) Summary of the number of Genera with a Fold Change greater than 1.0 by IRT Test and Control site over the Induction phase. Inset figures of Bacterial Load (Log10 transformed 16S copy numbers) are shown for each IRT Test and Control site over the Induction phase. A dashed red line is shown across all IRTs for reference. Dashed lines represent Test sites, and solid lines represent Control sites. (KM) Percent relative abundance of agglomerated data at the phylum level grouped by IRT Test and Control sites over the Induction period where dashed lines represent Test sites and solid lines represent Control sites (e.g., HTB = High-IRT Test Bacteroidetes; HTF = High-IRT Test Firmicutes; HCB = High-IRT Control Bacteroidetes; HCF = High-IRT Control Firmicutes). Graphical interpretation of the temporal shifts in community composition are highlighted by IRT, respectively (e.g., HT Shift = High-IRT Test Shift; HC Shift = High-IRT Control Shift). Trend lines represent mean values by IRT test or control over the Induction phase. Whiskers represent SE.
From these results, there are several trends that were apparent, including distinct temporal shifts across IRT test and control sites. Specifically, we observed an increase in gram-negative-associated Bacteroidetes (Wilcoxon-FDR; Slow-IRT, padj < 0.01, Days 0 to 21) and a decrease in gram-positive-associated Firmicutes (Wilcoxon-FDR; High-IRT, padj < 0.05, Days 0 to 21; Slow-IRT, padj < 0.05, Days 0 to 14 and 0 to 21) within healthy control sites (Fig. 3A and SI Appendix, Table S4). These results mirrored similar taxonomic changes observed within distant test sites over the same period, although delayed and in lower magnitude despite maintained oral hygiene of control sites. Interestingly, not all phyla shared similar trajectories between test and control sites over the Induction phase. Rather, Actinobacteria were observed to generally increase across all IRT control sites in relation to respective IRT test sites, and this increase was statistically significant within High-IRT (Wilcoxon-FDR; padj < 0.01, Days 0 to 21) (Fig. 3A and SI Appendix, Table S4). Together these observed changes suggest they are not just the result of normal maturation after a professional cleaning (Materials and Methods).
As the Bacteroidetes and Firmicutes were the two most abundant phyla represented in this study, we sought to investigate their temporal dynamics over the Induction phase within IRT test and control sites (Fig. 3 KM). This revealed distinct timepoints during the Induction phase in which we observed compositional shifts within these sites from a commensal/mutualist dominant community (Firmicutes enriched) to a community enriched in PDA gram-negative bacteria (Bacteroidetes). The exact timing was dependent on IRT and differed for test and control sites. Notably, this compositional shift across IRT test sites positively correlates with significant plaque accumulation (Fig. 1 C and D) and onset of clinically observed inflammation represented by GI (Fig. 1 E and F) in an IRT-dependent manner. Specifically, we observe a rapid shift by Day 4 in High- and Low-IRT test sites with a delayed shift until Day 7 in Slow-IRT test sites. Interestingly, we observed similar changes with respect to these two major phyla, though delayed and in lower magnitude, within respective IRT control sites. This was most evident in the High- and Slow-IRTs.
In order to identify specific genera of interest within healthy control sites, particularly those within the Firmicutes and Bacteroidetes phyla, taxonomic changes were examined at the genus level (SI Appendix, Table S5). Additionally, we investigated the fold change in relative abundance using genus-level data over the Induction phase compared to baseline (Day 0) for both IRT test and control sites (Fig. 3 BG). Looking at PDA genera showed an enrichment of several gram-negative anaerobes across healthy sites, but this was highly variable with no clear temporal enrichment between test and control sites (Fig. 3 BG genera in red). Additionally, healthy sites showed low cell number increases (represented by 16S rRNA copy numbers) (Fig. 1 K and L) and low fluctuations in microbial diversity (Fig. 2A) and relative abundance (Fig. 3A) throughout the study.
In the absence of strong signals, we sought to identify timepoints where the largest number of changes occurred within healthy control sites during the Induction phase for each IRT. For this, we quantified the number of genera increasing between each time point of the Induction phase with a fold change in relative abundance greater than 1. These results are summarized in Fig. 3 HJ. For the High-IRT we observed the highest number of enriched genera by Day 14 (n = 21) and by Day 21 (n = 22) for the Slow-IRT. Interestingly, the Low-IRT had a max number of 5 genera increasing by Day 21 across the entire Induction phase using this cutoff.
We observed a direct relationship between the genera enrichment and cell number increases represented by 16S rRNA (Fig. 1 K and L; Inset Fig. 3 HJ). The maximum number of enriched genera for High-IRT by Day 14 correlated to a Days 0 to 14 0.6 log fold increase in cell numbers. For Slow-IRT Days 0 to 21 corresponded to a 0.7 log fold change (LFC). Consistent with the low levels of inflammation and the maximum enrichment of only 5 genera in the Low-IRT, very little change was seen in healthy site cell numbers for the Low-IRT group.
Combining these results, we were able to validate that the subgingival community within each IRT site undergoes a compositional shift, which we define as a decrease in commensal/health-associated Firmicutes with a simultaneous increase in PDA Bacteroidetes (Fig. 3 KM). The High- and Low-IRT test sites undergo this shift by Day 4, whereas the shift in Slow-IRT test sites was delayed until Day 7 of the Induction phase. Using the relative abundance results for the Firmicutes and Bacteroidetes phyla (Fig. 3 A and KM) and the quantified genus level changes (Fig. 3 HJ), we observed that the High- and Slow-IRTs control sites experience a similar Firmicutes/Bacteroidetes phyla shift. This compositional shift followed the test site shifts by 7 to 10 d (Day 14 for High- and Day 21 for Slow-IRT). However, this characteristic shift was not observed to any significant degree in Low-IRT control sites using our defined criteria.

PDA Bacterial Strains are Detected Contralaterally between Test and Control Sites within Individuals across Responder Type.

Because the temporal enrichment of PDA genera (gram-negative anaerobes) seen in test sites was also observed in control sites, albeit delayed and at a lower magnitude, we sought to determine whether the same bacterial strains could be detected within both an individuals’ test and control sites. Furthermore, we sought to identify the temporal relationship of shared strains between test and control sites over the Induction phase (Days 0 to 21). The focus was on key genera within the Bacteroidetes phyla (Porphyromonas, Tannerella, Alloprevotella, Prevotella) as well as other PDA genera reported in the literature (Selenomonas, Aggregatibacter, Saccharibacteria, Treponema) (Fig. 4D) (24).
Fig. 4.
ASVs are detected contralaterally between test and control sites. ASV level data were converted to a presence–absence matrix and plotted as a heatmap in order to identify contralateral detection between test and control sites over the induction period (Days 0 to 21) by clinical IRT. Gram-negative and CPR genera that were enriched in test sites were selected for this analysis resulting in a total of 3,509 ASVs. Red text represents ASVs that were detected simultaneously between test and control sites by subject. Blue text represents ASVs that were detected in test sites prior to any detection in control sites by subject. (A) High-IRT by subject resulted in 29 (0.83%) ASVs that were contralaterally detected. 6/6 subjects had a contralaterally detected ASV. (B) Low-IRT by subject resulted in 15 (0.43%) ASVs that were contralaterally detected. 3/6 subjects had a contralaterally detected ASV. (C) Slow-IRT by subject resulted in 89 (2.5%) ASVs that were contralaterally detected. 8/9 subjects had a contralaterally detected ASV. (D) A Table of the gram-negative and CPR genera used in this analysis, the number of species detected within those genera as well as the number of corresponding ASVs for each genera.
Individual strains were represented by ASVs. ASVs represent exact sequences and can be used to resolve strains with 16S rDNA gene sequencing data (25). During the Induction phase, we detected a total of 11,712 ASVs with 3,509 classified to our genera of interest, including: Aggregatibacter (7 Species, 155 ASVs), Alloprevotella (7 Species, 285 ASVs), Porphyromonas (10 Species, 414 ASVs), Prevotella (38 Species, 1,602 ASVs), Selenomonas (19 Species, 433 ASVs), Saccharibacteria (8 Species, 269 ASVs), Tannerella (3 Species, 118 ASVs), and Treponema (23 Species, 233 ASVs) (Fig. 4D). Using a presence/absence heatmap of the 3,509 ASVs within these PDA-enriched genera, 29 ASVs (0.83%) were detected contralaterally within 6/6 High-IRT subjects (Fig. 5A). 15 ASVs (0.43%) were detected contralaterally among 3/6 Low-IRT subjects (Fig. 5B). 89 ASVs (2.5%) were detected contralaterally in 8/9 Slow-IRT subjects (Fig. 5C). While all IRTs show contralateral ASV detection of these PDA genera, High- and Slow-IRTs had a higher number of ASVs and higher prevalence in comparison to the Low-IRT, consistent with the control site results (Figs. 2 and 3).
Fig. 5.
Induced inflammation in test sites changes chemokine profiles in control sites. (A) CCL2 by IRT test and control during the Induction phase (Days 0 to 21). (B) IL-10 by IRT test and control during the Induction phase. (C) CCL1 by IRT test and control during the Induction phase. (D) Macrophage MIF by IRT test and control during the Induction phase. (E) Row-wise z-scored heatmap of host inflammatory mediators (28) clustered using the k-means algorithm for IRT test and control sites over the Induction phase. Select chemokines associated with periodontal inflammation and health are highlighted in red. (F) Trajectories of clustered chemokines over the Induction phase represented by the SD from the mean chemokine value. Gray dashed line represents major host mediator shift within test sites. Boxes represent data and medians ± interquartile ranges; whiskers and outliers >1.5 IQR below (above) the 25th (75th) percentile. Trend lines represent loess regression mean values across all time points. Statistical analysis was performed using logistic regression adjusted by FDR. Significance level indicated by asterisks. Significance levels: *P < 0.05, **P < 0.01, and ***P < 0.001. Clusters containing statistically significant changing host mediators and those associated with periodontal inflammation are highlighted in red.

Microbially Induced Inflammation Results in Changes in Chemokine Profiles within Distant Clinically Healthy Tissues.

Having observed similar, although delayed, IRT-dependent shifts in the microbial communities (Figs. 1 K and L, 2, and 3) between test and control sites during the inflammatory Induction phase (Days 0 to 21), we sought to investigate whether host mediator (chemokines and cytokines) profiles also showed similar shifts between sites. We examined GCF, a rich serum exudate, collected from test and control sites stratified by IRT using a panel of 41 host mediators associated with gingival tissue homeostasis, bone homeostasis, and periodontal inflammation. Although our initial analysis revealed several trends, only a handful of host mediators showed statistically significant changes within the distal healthy control sites. These included C-C motif ligand 2 (CCL2/MCP-1) (Linear Regression; P < 0.05) in the High-IRT, interleukin-10 (IL-10) (Linear Regression; P > 0.05) in the Low-IRT, and C-C motif ligand 1 (CCL1/I-309) (Linear Regression; P < 0.05), and macrophage migration inhibitory factor (MIF) (Linear Regression; P < 0.05) in the Slow-IRT (Fig. 5 AD and SI Appendix, Tables S6–S11).
In order to investigate the temporal dynamics of these mediators in more detail, mediator data were then visualized using a row-wise normalized Z score and (Fig. 5E) clustered using the k-means algorithm which resulted in 12 unique mediator clusters. This analysis revealed shared characteristics between IRT test sites previously reported (1) as well as highlighted how the distant healthy control sites varied during this inflammatory Induction phase. As shown previously, High-IRTs had high levels of host mediators while Slow-IRTs had high levels of mediators with a delayed response compared to High-IRT (1). Low-IRTs displayed several standard deviations below the mean mediator value in most chemokines assessed in this study in comparison to other IRT groups.
To investigate the temporal dynamics of each cluster over the Induction phase, cluster trajectories were calculated as the standard deviation from the mean mediator value stratified by IRT test and control sites (Fig. 5F). Of particular interest were clusters 5 and 12 which contain several proinflammatory mediators associated with periodontal diseases, including: interleukin-6 (IL-6), interleukin-8 (IL-8), tumor necrotizing factor alpha (TNF-a), myeloperoxidase (MPO), and interleukin-1-beta (IL-1b). All of these appear to share similar trajectories to their respective IRT test sites over the Induction phase, despite maintained oral hygiene and no significant clinical inflammation within healthy control sites. In summary, we again observed healthy control sites were neither static nor randomly changing as there were multiple host mediators showing coregulatory patterns that varied by IRT and time compared to respective test sites.

Microbially Induced Inflammation in Test Sites Influences Host Mediator Changes in Distant Control Sites that Precede a Shift in Control Site Community Composition.

Having independently established that both the subgingival community and host mediators shift within distant healthy control sites, with similar characteristics to what were observed within respective IRT test sites, we sought to characterize the temporal relationship between the host and subgingival community by integrating our different data types (clinical, microbiome, mediator). To better resolve the timing of host mediator changes in healthy control sites during the Induction phase (Days 0 to 21), we calculated the fold change at each time point compared to Baseline (Day 0) (Fig. 6 AC). This revealed a coordinated shift among many of the host mediators within each IRT, including several well-establish progingival inflammation signals (IL-6, IL-8, IL-1B, and TNF-a) (Fig. 6 GI). Interestingly, the mediator changes within the healthy controls corresponded to the timing of microbial and clinical changes in the test sites. The largest changes in mediators occurred alongside the significant accumulation and maturation of dental plaque as well as the onset of clinically observed inflammation (gingival index: GI) within distant IRT test sites (Figs. 1 D and E and 6 DF, Insets). Furthermore, these changes in the test sites and mediator shifts in the healthy controls correlated with the shift from Firmicutes to Bacteroidetes in the test sites (Fig. 6 JL). In contrast, the Firmicutes to Bacteroidetes shift occurred much later in the control sites and did not quite occur in the Low-IRT. A graphical representation of the relationship between these observed temporal shifts within the subgingival community as well as host mediators across the different IRTs is highlighted in Fig. 6 MO.
Fig. 6.
Maturing plaque in test sites induces host mediator changes in control sites which precede a shift in the control site microbiome. (AC) A row-wise z-scored heatmap of LFC of chemokines compared to baseline (Day 0) among different clinical IRT control sites. Key inflammatory mediators are highlighted in red text. (DF) Quantification of the number of mediators with a LFC greater than 1 over the Induction phase by IRT control site. Shaded boxes highlight the shift in host mediators within respective control sites. Inset clinical Gingival Inflammation (GI) plots for the Induction phase by IRT test and control sites. Red dashed line represents GI = 1.5. (GI) IL-8 (Left y axis), IL-6, and TNF-a (Right y axis) among IRT control sites over the Induction phase. Shaded boxes highlight the shift in host mediators within respective control sites. (JL) Percent relative abundance of Firmicutes (Left y axis) and Bacteroidetes (Right y axis) using agglomerated data at the phylum level. Trendlines represent the mean value. Solid lines represent IRT control sites and dashed lines represent IRT test sites. Labels represent IRT test and control sites by Phylum (i.e., HTF—High Test Firmicutes, HCF—High Control Firmicutes, HTB—High Test Bacteroidetes, HCB—High Control Bacteroidetes). Shaded boxes highlight the shift in host mediators within respective control sites. (MO) Graphical interpretation of the temporal relationships of the microbiome and host mediators between test and control sites over the Induction phase.

Discussion

The oral cavity serves a critical role in human innate immunity in which a complex network of specialized tissues and immune cells provide a constant state of immune surveillance (15, 26, 27). This barrier defense, similar to the human gut, relies on an array of commensal microbes that have coevolved with humans over tens of thousands of years in order to maintain normal function through healthy homeostasis (29, 30). As bacterial community composition and increase in bacterial load are both associated with an increase in host inflammatory response in vivo, it remains unclear if one or the other or a combination of both are required to elicit such a response (811, 15, 31, 32). This shift usually involves the niche expansion of gram-negative anaerobes, often Bacteroidetes, with specialized mechanisms to evade the human immune system, such as Porphyromonas and Tannerella spp. Periodontal diseases, including gingivitis and periodontitis, are localized oral inflammatory diseases that result in the progressive destruction of the supporting tissues surrounding human teeth. However, the etiopathogenesis of how healthy teeth become susceptible, how these periodontal diseases spread beyond localized tooth sites in the mouth, or how these localized oral inflammatory diseases (gingivitis, periodontitis) directly impact systemic health in humans remains less clear (33). Due to difficulties capturing the spontaneous transition from health to naturally occuring gingivitis or from gingivitis to periodontitis within periodontal tissues, which would require constant longitudinal sampling of large populations, mechanistic insights into these processes are limited (34). However through EG models, the shift from health to gingivitis in humans can be induced in a controlled manner by allowing unabated plaque growth with sampling performed directly at the host–microbe interface where inflammation occurs. A limitation of the EG model is that it may not reflect the microbial and host factors obeserved in patients with chronic gingivtiis and or severe periodontitis. Despite this limitation, a major finding that has come out of EG studies focusing on the development and resoultion of this acute microbial-induced inflammation is that patients show differential rates and severity of gingival inflammation allowing them to be grouped into clinical IRTs (1, 3, 35).
In the present study, we incorporated the EG model with split-mouth design, which allowed us to uniquely characterize the effects of natural plaque accumulation and onset of clinically localized microbially induced inflammation within test sites on distant healthy tooth sites with maintained oral hygiene over a period of 21 d. Our results highlight a nonlocalized subclinical inflammatory effect with distinct temporal shifts in host mediator profiles including well-established proinflammatory mediators associated with periodontal diseases as well as shifts in microbial diversity and community composition within subgingival plaque. As with test sites, these changes differed according to our defined IRTs: High-IRT, Low-IRT, and Slow-IRT (1), and were most evident in the groups that demonstrated susceptibility to high inflammation (High- and Slow-IRTs).
It is generally reported that once oral hygiene is stopped, microbial growth and shifts in the community from gram-positive dominated to gram-negative enrichment occurs (36, 37). Clinically visible inflammation is then observed as measured by increases in GI and BOP (1, 3). In subjects that are initially healthy, these processes can occur with variations in rate and severity, now defined by IRTs. We have shown that in the High-IRT, test sites have a characteristic rapid plaque growth and maturation rate that results in the significant accumulation of dental plaque (Plaque Index: PI) along the gingival margin after just 4 d of abstained oral hygiene (Fig. 1 B and D) (1). During this increase in subgingival plaque, distinct temporal shifts in microbial diversity (Fig. 2) and subgingival community composition (Fig. 3) occur within High-IRT test sites over the same period. Specifically, we observe a decrease in commensal/health-associated Firmicutes (−11%, Days 0 to 21; Wilcoxon, P = 0.004) and enrichment of PDA Bacteroidetes (generally gram-negative anaerobes) within High-IRT test sites (14%, Days 0 to 21; Wilcoxon, P = 0.004) (Fig. 3K). Additionally we observed increases in PDA Spirochaetes (~3%, Days 4 to 14; Wilcoxon, P = 0.041) (Fig. 3A) and Saccharibacteria (~0.5%, Days 0 to 21; Wilcoxon, P = 0.026) (SI Appendix, Fig. S3) phyla over the same period. It is this rapid increase in load (Fig. 1 C and K), decrease in commensal Firmicutes, and shift toward a PDA-enriched composition (Fig. 3 A and K) that results in the rapid shift in most host mediators we assessed (Fig. 5E). This microbially induced shift in host responses results in the onset of inflammation. Clinical inflammation was detected with GI and BOP increases as early as Day 4 in the High-IRT (Fig. 1 E and G).
Notably, when investigating distant High-IRT healthy control sites located contralaterally in the mouth (Fig. 1A), we observed a coordinated shift in most host mediators by Day 4 (~66%, 27/41 mediators) (Fig. 6A), including well-established proinflammatory mediators associated with periodontal inflammation: IL-6, IL-8, and tumor necrosis factor-alpha (TNF-a) (Fig. 6G). The shift in key proinflammatory signals within healthy High-IRT control tissues temporally aligns with the onset of significant microbially induced inflammation occurring within distant High-IRT test sites (Figs. 1E and 6 D, Inset). This suggests that the microbially induced inflammation occurring in distant High-IRT test sites is somehow triggering a nonlocalized response that is detectable in distant healthy tooth sites located contralaterally in the mouth. Even more interesting is that this observed shift in host mediators within healthy tooth sites precedes a shift in both microbial diversity (Fig. 2D) and community composition (Fig. 3K) within healthy tooth sites, including among the two most abundant phyla detected in this study (Fig. 6J). Notably, as observed in distant High-IRT test sites, we see the decrease in commensal/health Firmicutes (−14%, Days 0 to 21; Wilcoxon, P = 0.026) and simultaneous increase in PDA Bacteroidetes starting by Day 7 of the Induction phase (8%, Days 7 to 21; Wilcoxon, P = 0.13) in High-IRT controls sites. This observation is further supported by significant correlation between these two phyla over the Induction phase (Spearman, Rho −0.56, P < 0.001) (SI Appendix, Fig. S4). These host and microbial changes within healthy control sites are evident despite maintained oral hygiene of these distant tooth sites and absence of significant clinically observed inflammation.
We then investigated if there are any similar temporal relationships occurring in the Slow-IRT group. Importantly, the defining phenotype of these individuals is that they display a delayed plaque growth rate and delayed onset of inflammation compared to High-IRT yet achieve the same high levels of inflammation by Day 21 within test sites (Fig. 1E) (1). Interestingly, Slow-IRT control sites showed a similar coordinated shift in most mediators to that seen in High-IRT controls. However, consistent with Slow-IRT test sites showing delayed shifts compared to High-IRT test sites, this shift was delayed in comparison to the mediator changes in High-IRT control sites (Fig. 6A) and occurred between Day 4 and Day 7 (~68%, 28/41 mediators) (Fig. 6C). This included established proinflammatory mediators associated with periodontal diseases (Fig. 6I). Once again, the mediator shift in healthy sites temporally aligned with the onset of significant inflammation observed in distant test sites (Fig. 1E). Similar to High-IRT control sites, this observed shift in host mediators within distant healthy Slow-IRT control sites preceded an observed shift in both microbial diversity (Fig. 2D) and subgingival community composition (Fig. 3M), including a decrease in Firmicutes and increase in PDA Bacteroidetes (Fig. 6L). Specifically, we observed a decrease in Firmicutes (−13%, Days 0 to 21;Wilcoxon, P = 0.014) and enrichment of PDA Bacteroidetes (~10%, Days 0 to 21;Wilcoxon, P = 0.008) in distant healthy Slow-IRT control sites with an increase of ~8% between Days 14 and 21 (Fig. 6L). This observation was also supported by significant correlation between these two phyla over the Induction phase (Spearman, Rho −0.56, P < 0.001)(SI Appendix, Fig. S4). Additionally, this observed enrichment of PDA Bacteroidetes (Fig. 6L) in Slow-IRT control sites occurs a week after the observed shift in host mediators (Fig. 6 C and F)—similar to what was observed within the High-IRT group (Fig. 6J).
Overall, the temporal shifts in Slow-IRT control sites are consistent with this clinical inflammatory responder phenotype, which is delayed plaque growth and delayed onset of microbially induced gingival inflammation. Although there are minor differences between High- and Slow-IRT control sites, both share similar characteristics of a detectable yet subclinical proinflammatory response that temporally aligns with the onset of significant clinical inflammation occurring in distant test sites. Furthermore, these observed shifts in host mediators in distant healthy control sites within the High- and Slow-IRTs precedes a significant reduction in commensal Firmicutes and enrichment of PDA Bacteroidetes in control sites, similar to what was observed in respective test sites, despite maintained oral hygiene and absence of significant inflammation.
The defining characteristic of the Low-IRT phenotype is that they are uniquely able to modulate their immune response by Day 7—ultimately resulting in lower levels of observed inflammation in comparison to High- and Slow-IRT groups (Fig. 1 E and F). This occurs despite the early similarities to High-IRT (Fig. 1E) resulting in inflammation by Day 4, rapid plaque growth (Fig. 1C), similar shifts in microbial diversity (Fig. 2), decrease in Firmicutes (−12%, Days 0 to 21;Wilcoxon, P = 0.041) (Fig. 3L), and enrichment of PDA Bacteroidetes (12%, Days 0 to 21; Wilcoxon, P = 0.004). However, Low-IRT showed no significant correlation between the phyla (Spearman, Rho −0.29, P = 0.12) (SI Appendix, Fig. S4). In keeping with lower inflammation levels within test sites, Low-IRT healthy sites consistently showed no significant changes. There was no clear observed shift in host mediators (Fig. 6 B and E), nor shift within the subgingival community composition as observed in the other IRT groups (Figs. 3I and 6K). The nonlocalized effect observed in High- and Slow-IRT control sites seems to be modulated or muted in Low-IRT control sites, potentially due to this groups unique ability to modulate their response to microbially induced inflammation within test sites.
Importantly, these Low-IRT individuals serve as a natural negative control group within the current study and support that this observed effect in humans is highly likely to be a nonlocalized response to microbially induced inflammation occurring within distant test sites located contralaterally in the mouth. Such a nonlocalized response within the human oral cavity is further supported by prior observations regarding improved health parameters in distant tooth sites when subgingival plaque is removed from diseased sites within periodontitis patients (20). Furthermore, it has been well established that the systemic changes associated with pregnancy results in a constant influence of hormonal, metabolic, and immunological factors, which are suspected drivers influencing the subgingival microbiome composition (pregnancy-induced gingivitis) (3840).
Two key components of this longitudinal study were the availability of early time points during the Induction phase (Days 0 to 21) and analyzing the results within respective responder types (IRTs). Using these, we were able to resolve the temporal relationships between the localized effect induced in test sites and the nonlocalized response observed in distant healthy control sites, highlighting key shifts in both host mediators and the subgingival microbiome. A graphical representation of these temporal relationships is presented in Fig. 6 MO. Together, these data indicate the presence of a nonlocalized effect on distant tooth sites within the human oral cavity where: 1) test sites allow normal accumulation and maturation of dental plaque (bacterial biofilm), which 2) initiates a proinflammatory response in the host test sites (localized), that 3) is simultaneously detected within the GCF in distant healthy control sites (nonlocalized) that, 4) potentially allowing for the niche expansion of PDA gram-negative bacteria within those distant sites (nonlocalized effect). Importantly, this nonlocalized effect in healthy control sites varies in magnitude and time in relation to the severity and timing of inflammation observed within test sites and is dictated by an individual’s clinical IRT.
Remarkably, these coordinated shifts in host mediator and microbiome composition occur within High- and Slow-IRT control sites despite maintained normal oral hygiene and absence of significant gingival inflammation (GI). Specifically, we can observe these compositional changes within control sites without major changes in the total subgingival cell numbers (as estimated by total 16S rRNA gene copies). Given clear compositional shifts in the High-IRT and Slow-IRT, it is unclear why this remains subclinical and does not trigger a full host inflammatory response as observed clinically within respective test sites. This may suggest that the relative percent increase of periodontal pathobionts alone may not be sufficient to trigger clinically observable changes (GI). This is relevant to human oral health as it changes the perception that a community which is shifted toward a dysbiotic state, normally observed in later stages of gingivitis and periodontitis, must result in a host response and ultimately inflammation. Rather, it is likely both pathobiont enrichment (such as Bacteroidetes >20% relative abundance in subgingival plaque) and exceeding a particular threshold in total cell numbers may induce a clinically observable inflammatory response—highlighting that the level of host tolerance or capacity even in susceptible individuals (i.e., High- and Slow-IRT groups) is an important factor. These observed nonlocalized changes are a result of inflammation and therefore may represent precursor events leading to increased risk of gingival inflammation and ultimately more severe periodontal disease in susceptible responder types.

Materials and Methods

Induced EG in Humans.

Study methodology was based on the established protocol by Loe, Theilade, and Jensen (41) for induction of reversible bacterial-induced inflammation via cessation of oral hygiene in humans as described in Bamashmous et al. (1). Study protocols were approved by the University of Washington’s Institutional Review Board (HSD# 50151). In brief, twenty-one generally healthy adults aged 18 to 35 y were consented and enrolled based on inclusion and exclusion criteria previously reported (1). The average age of study participants was 22 y (range: 20 to 27), and no significant differences between sex of age were observed (sex, P = 0.7; age, P 0.3) (SI Appendix, Table S1). The study consisted of three phases: a hygiene phase (Day −14 to 0); 2) an induction phase (Days 0 to 21); and 3) a resolution phase (Days 21 to 35). As previously described, all study subjects were given customized intraoral stents for use during regular brushing with detailed instructions in plain health language in order to prevent accidental brushing of test sites (1). The intraoral stent covers the entire tooth buccal and palatal surfaces with extensions past the mesial and distal sites of the test site teeth (test side randomized to either teeth #3,4,5 or #12,13,14) and was only used during brushing as instructed (approximately 4-min per day) (SI Appendix, Fig. S5). All remaining teeth, including contralateral control sites, maintained normal oral hygiene for the duration of the study. Fidelity monitoring of the intervention was conducted at each time point throughout the experiment by clinical assessment of the PI, which was overseen by a single calibrated examiner. Proposed clinical IRTs were resolved retrospectively and reported previously (1). The effects of microbially induced inflammation beyond localized sites in the human mouth remain unresolved. The basis of the present study was to temporally characterize IRT healthy control sites in coordination with the onset of microbially induced inflammation within distant test sites elsewhere in the mouth during a recent EG study (1).

Characterization of Microbial Changes in Response to Plaque Accumulation.

The full methods are described in Bamashmous et al. (1). In brief, subgingival plaque samples were collected using sterile paper points inserted into the gingival sulcus of test and control teeth for 30 s and then pooled to avoid contamination arising from low-biomass. DNA was extracted and 16S rRNA libraries were generated as previously described (28, 42) (SI Appendix, Supplemental Materials and Methods).

Identifying Clinical IRTs during EG.

Proposed clinical IRTs were determined by a retrospective clustering analysis using the k-means algorithm on clinical measures (PI, GI, BOP) from test sites over the induction phase (Days 0 to 21) as previously reported (1).

Investigating Clinical Parameters Stratified by IRTs.

Clinical data, including PI (21), GI (2), BOP, GCF volume, and bacterial load (total 16S copy numbers), were investigated for each clinical IRT stratified by test and control sites. PI and GI measures were collected and averaged by individual tooth sites, which were then summarized as averages by test or control side (6 sites per tooth × 3 teeth per study side). BOP was assessed and averaged by test or control side (4 sites per tooth × 3 teeth per study side). Mean values across test and control sites for respective IRTs were calculated and plotted using loess linear regression (SI Appendix, Supplemental Materials and Methods).

Microbial Diversity in Test and Control Sites Stratified by IRTs.

Filtered sequence data including singletons, was assessed using established alpha diversity indices (Shannon, Observed, Simpson, Chao1) for each IRT stratified by test and control sites. Beta diversity was calculated using filtered sequence data, with singletons removed, using Weighted Unifrac distances. Beta Dissimilarity was calculated using filtered count data without singletons using Day 0 as a reference (SI Appendix, Supplemental Materials and Methods).

Microbial Composition in Test and Control Sites by IRT.

Exact amplicon sequence variant (ASV) sequences generated by the DADA2 (43) workflow were classified using the expanded Human Oral Microbiome Database (v. 15.1) (eHOMD) (44) using Qiime2 (45). Taxonomic count data were then processed and visualized at the Phylum, Genus, Species, and ASV levels (SI Appendix, Supplemental Materials and Methods).

Characterization of In-Vivo Chemokine Responses during EG.

Local chemokine responsiveness was measured using GCF, a serum exudate, as previously reported (1). Samples were collected at each time point by inserting sterile paper strips (Periopaper; Oraflow Inc.) into the gingival crevice for 30 s. Samples were then immediately quantified (Periotron 8010, OralFlow, inc), pooled by test and contralateral control sites, respectively, and transported to the lab on ice for further processing (SI Appendix, Supplemental Materials and Methods).
Individual changes in host mediators were investigated by IRT with overlayed test and control sites over the induction phase (Days 0 to 21). Additionally, host mediators identified to change significantly within respective IRT control sites were then investigated for trends across individual subjects within the respective IRT groups using linear regression (SI Appendix, Supplemental Materials and Methods).
In our secondary analysis, host mediator data were then stratified by IRT test and control sites, respectively, for the induction phase – resulting in six separate datasets (High-IRT Test, High-IRT Control, Low-IRT Test, Low-IRT Control, Slow-IRT Test, Slow-IRT Control). Each individual dataset was then assessed to determine the optimal number of clusters over the induction phase (SI Appendix, Supplemental Materials and Methods). This resulted in a range of 6-15 optimal clusters for each IRT test and control datasets. A comprehensive dataset including all IRT test and control sites over the induction phase was then assessed using a row-wise z-scored normalized heatmap. It was determined that increasing the number of optimal clusters beyond 12 did not result in any additional clusters; rather, it resulted in isolated host mediators. Host mediators in each cluster were then isolated and plotted in order to visualize host mediator dynamics within clusters for each IRT test and control site over the induction phase and is represented by the standard deviation from the mean value at each time point (SI Appendix, Supplemental Materials and Methods).
In our tertiary analysis, we aimed to investigate the log-fold-change (LogFC) of host mediators within control sites for each IRT over the induction phase compared to baseline (Day 0) (SI Appendix, Supplemental Materials and Methods).

Statistical Analysis.

All statistical analysis performed on clinical, host mediator, and microbiome data was carried out using various R packages within the R suite (v. 4.0.4) (47) using RStudio (v. 1.4.1106) (46). Each data type was independently assessed for normality. Results from normality testing guided statistical analysis for each respective dataset in relationship to what was being compared (SI Appendix, Supplemental Materials and Methods).
Clinical data were summarized into single average scores for test and control sites for all study subjects at each time point. Results were then visualized using boxplots showing medians and interquartile ranges for each time point as well as the mean value over the Induction phase (Days 0 to 21) (SI Appendix, Figs. S1 and S2 and Supplemental Materials and Methods).
Microbiome diversity and composition analysis were investigated at various taxonomic levels. Alpha and Bray Curtis dissimilarity indices were then visualized using boxplots showing medians and interquartile ranges for each time point as well as the mean value over the Induction phase (Days 0 to 21) (SI Appendix, Supplemental Materials and Methods).
Host mediator data were investigated within IRT and between IRT groups over the Induction phase using linear-mixed models (LMMs) adjusted for individual random effects for each subject and fixed effects for time and site (Day, Test/Control) with an adjusted post hoc FDR (SI Appendix, Supplemental Materials and Methods). Additionally, statistical tests based on a LMMs were used to compare clinical and chemokine data between different IRT groups during the induction phase (Days 0, 4, 7, 14, and 21), and within each responder group between baseline (Day 0) and different time points during the induction phase (Days 4, 7, 14, and 21). Similar results for clinical comparisons between Wilcoxon test and LMMs were observed.

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix. Raw read and sample meta data have been deposited in the National Center for Biotechnology Information Sequence Read Archive, https://www.ncbi.nlm.nih.gov/sra (BioProject PRJNA615201) (48). The data that support the findings of this paper and the R code used for generating the analysis has been made publicly available published on a GitHub repository (https://github.com/mcleanlab/HVGI_Contralateral_Effect) (49). This repository contains all necessary raw and processed files, (i.e., .biom, metadata.txt, .tree, and raw data files) to recreate these analyses as well as additional supplemental material derived from this study.

Acknowledgments

The research was funded through a Colgate-Palmolive clinical research grant (R.P.D., PI) and in part by the NIH TL1 TR002318, KL2 TR002317, T90 DE021984, (to K.A.K.), DE029872 (to G.A.K.), and DE023810 (to J.S.M.). This funding source had no role in the design of this study and did not have any role during its implementation, analyses, interpretation of the data, or decision to submit results.

Author contributions

K.A.K., S.B., D.C., H.M.T., R.P.D., and J.S.M. designed research; K.A.K., S.B., G.A.K., R.P.D., and J.S.M. performed research; K.A.K., S.B., E.L.H., G.A.K., B.G.L., D.D.D., F.A.R., R.P.D., and J.S.M. analyzed data; and K.A.K., S.B., E.L.H., G.A.K., B.G.L., D.D.D., F.A.R., D.C., H.M.T., R.P.D., and J.S.M. wrote the paper.

Competing interests

The authors declare no competing interest.

Supporting Information

Appendix 01 (PDF)

References

1
S. Bamashmous et al., Human variation in gingival inflammation. Proc. Natl. Acad. Sci. U.S.A. 118, e2012578118 (2021).
2
H. Loe, J. Silness, Periodontal disease in pregnancy. I. Prevalence and severity. Acta Odontol. Scand. 21, 533–551 (1963).
3
L. Trombelli et al., Experimental gingivitis: Reproducibility of plaque accumulation and gingival inflammation parameters in selected populations during a repeat trial. J. Clin. Periodontol. 35, 955–960 (2008).
4
L. Trombelli et al., Modulation of clinical expression of plaque-induced gingivitis. III. Response of "high responders" and "low responders" to therapy. J. Clin. Periodontol. 31, 253–259 (2004).
5
L. Trombelli et al., Modulation of clinical expression of plaque-induced gingivitis. II. Identification of "high-responder" and "low-responder" subjects. J. Clin. Periodontol. 31, 239–252 (2004).
6
C. Janakiram, A. Mehta, R. Venkitachalam, Prevalence of periodontal disease among adults in India: A systematic review and meta-analysis. J. Oral. Biol. Craniofac. Res. 10, 800–806 (2020).
7
R. Nocini, E. J. Favaloro, F. Sanchis-Gomar, G. Lippi, Periodontitis, coronary heart disease and myocardial infarction: Treat one, benefit all. Blood Coagul. Fibrinolysis. 31, 339–345 (2020).
8
R. P. Darveau, Periodontitis: A polymicrobial disruption of host homeostasis. Nat. Rev. Microbiol. 8, 481–490 (2010).
9
G. Hajishengallis, Periodontitis: From microbial immune subversion to systemic inflammation. Nat. Rev. Immunol. 15, 30–44 (2015).
10
S. P. Hajishengallis et al., A low-abundance biofilm species orchestrates inflammatory periodontal disease through the commensal microbiota and the complement pathway. Cell Host Microbe 10, 497–506 (2011).
11
G. Hajishengallis, R. P. Darveau, M. A. Curtis, The keystone-pathogen hypothesis. Nat. Rev. Microbiol. 10, 717–725 (2012).
12
B. J. Paster et al., Bacterial diversity in human subgingival plaque. J. Bacteriol. 183, 3770–3783 (2001).
13
K. Abbayya, N. Y. Puthanakar, S. Naduwinmani, Y. S. Chidambar, Association between periodontitis and Alzheimer’s disease. N. Am. J. Med. Sci. 7, 241–246 (2015).
14
F. Q. Bui et al., Association between periodontal pathogens and systemic disease. Biomed. J. 42, 27–35 (2019).
15
G. Hajishengallis, T. Chavakis, Local and systemic mechanisms linking periodontal disease and inflammatory comorbidities. Nat. Rev. Immunol. 21, 426–440 (2021).
16
L. Winning, G. J. Linden, Periodontitis and systemic disease: Association or causality? Curr. Oral. Health Rep. 4, 1–7 (2017).
17
N. P. Lang, M. A. Schatzle, H. Loe, Gingivitis as a risk factor in periodontal disease. J. Clin. Periodontol. 36 (suppl. 10), 3–8 (2009).
18
S. Murakami, B. L. Mealey, A. Mariotti, I. L. C. Chapple, Dental plaque-induced gingival conditions. J. Periodontol. 89 (suppl. 1), S17–S27 (2018).
19
D. Herrera, Scaling and root planning is recommended in the nonsurgical treatment of chronic periodontitis. J. Evid. Based Dent. Pract. 16, 56–58 (2016).
20
A. P. Pawlowski et al., Clinical effects of scaling and root planing on untreated teeth. J. Clin. Periodontol. 32, 21–28 (2005).
21
J. Silness, H. Loe, Periodontal disease in pregnancy. Ii. Correlation between oral hygiene and periodontal condtion. Acta Odontol. Scand. 22, 121–135 (1964).
22
H. Loe, J. Silness, Periodontal disease in pregnancy. I. Prevalence and severity. Acta Odontol. Scand. 21, 533–551 (1963).
23
N. P. Lang, R. Adler, A. Joss, S. Nyman, Absence of bleeding on probing. An indicator of periodontal stability. J. Clin. Periodontol. 17, 714–721 (1990).
24
L. M. Sedghi, M. Bacino, Y. L. Kapila, Periodontal disease: The good, the bad, and the unknown. Front. Cell. Infect. Microbiol. 11, 766944 (2021).
25
B. J. Callahan, P. J. McMurdie, S. P. Holmes, Exact sequence variants should replace operational taxonomic units in marker-gene data analysis. ISME J. 11, 2639–2643 (2017).
26
S. Metcalfe, N. Anselmi, A. Escobar, M. B. Visser, J. G. Kay, Innate phagocyte polarization in the oral cavity. Front. Immunol. 12, 768479 (2021).
27
N. M. Moutsopoulos, J. E. Konkel, Tissue-specific immunity at the oral mucosal barrier. Trends Immunol. 39, 276–287 (2018).
28
J. Y. An et al., Rapamycin rejuvenates oral health in aging mice. Elife 9, e54318 (2020).
29
P. N. Deo, R. Deshmukh, Oral microbiome: Unveiling the fundamentals. J. Oral Maxillofac. Pathol. 23, 122–128 (2019).
30
R. Khan, F. C. Petersen, S. Shekhar, Commensal bacteria: An emerging player in defense against respiratory pathogens. Front. Immunol. 10, 1203 (2019).
31
R. P. Darveau, C. M. Belton, R. A. Reife, R. J. Lamont, Local chemokine paralysis, a novel pathogenic mechanism for Porphyromonas gingivalis. Infect. Immun. 66, 1660–1665 (1998).
32
R. P. Darveau, A. Tanner, R. C. Page, The microbial challenge in periodontitis. Periodontol. 2000 14, 12–32 (1997).
33
I. L. C. Chapple et al., Periodontal health and gingival diseases and conditions on an intact and a reduced periodontium: Consensus report of workgroup 1 of the 2017 World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions. J. Periodontol. 89 (suppl. 1), S74–S84 (2018).
34
S. Joseph, M. A. Curtis, Microbial transitions from health to disease. Periodontol. 2000 86, 201–209 (2021).
35
L. Trombelli et al., Interleukin-1 beta levels in gingival crevicular fluid and serum under naturally occurring and experimentally induced gingivitis. J. Clin. Periodontol. 37, 697–704 (2010).
36
R. J. Lamont, H. Koo, G. Hajishengallis, The oral microbiota: Dynamic communities and host interactions. Nat. Rev. Microbiol. 16, 745–759 (2018).
37
L. Sedghi, V. DiMassa, A. Harrington, S. V. Lynch, Y. L. Kapila, The oral microbiome: Role of key organisms and complex networks in oral health and disease. Periodontol. 2000 87, 107–131 (2021).
38
M. Abouzaid, N. Howidi, Z. Badran, G. Mohammed, N. A. Mousa, The potential role of the gingival crevicular fluid biomarkers in the prediction of pregnancy complications. Front. Med. (Lausanne) 10, 1168625 (2023).
39
P. Balan et al., Keystone species in pregnancy gingivitis: A snapshot of oral microbiome during pregnancy and postpartum period. Front. Microbiol. 9, 2360 (2018).
40
C. Ye, Y. Kapila, Oral microbiome shifts during pregnancy and adverse pregnancy outcomes: Hormonal and Immunologic changes at play. Periodontol. 2000 87, 276–281 (2021).
41
H. Loe, E. Theilade, S. B. Jensen, Experimental gingivitis in man. J. Periodontol. 36, 177–187 (1965).
42
E. I. Lamont et al., Modified SHI medium supports growth of a disease-state subgingival polymicrobial community in vitro. Mol. Oral Microbiol. 36, 37–49 (2021).
43
B. J. Callahan et al., DADA2: High-resolution sample inference from Illumina amplicon data. Nat. Methods 13, 581–583 (2016).
44
T. Chen et al., The human oral microbiome database: A web accessible resource for investigating oral microbe taxonomic and genomic information. Database (Oxford) 2010, baq013 (2010).
45
E. Bolyen et al., Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nat. Biotechnol. 37, 852–857 (2019).
46
R. Team, RStudio: Integrated Development Environment for R (RStudio, PBC, Boston, MA, 2020).
47
R. C. Team, R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, Austria, 2021).
48
S. Bamashmous, K. A. Kerns, R. P. Darveau, J. S. McLean, Human variation in gingival inflammation (HVGI) study. NCBI Sequence Read Archive (SRA). https://www.ncbi.nlm.nih.gov/bioproject/PRJNA615201. Accessed 1 April 2020.
49
K. A. Kerns, J. S. McLean, Human variation in gingival inflammation (HVGI) contralateral effect. Github. https://github.com/mcleanlab/HVGI_Contralateral_Effect. Deposited 7 September 2023.

Information & Authors

Information

Published in

Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 120 | No. 41
October 10, 2023
PubMed: 37782795

Classifications

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix. Raw read and sample meta data have been deposited in the National Center for Biotechnology Information Sequence Read Archive, https://www.ncbi.nlm.nih.gov/sra (BioProject PRJNA615201) (48). The data that support the findings of this paper and the R code used for generating the analysis has been made publicly available published on a GitHub repository (https://github.com/mcleanlab/HVGI_Contralateral_Effect) (49). This repository contains all necessary raw and processed files, (i.e., .biom, metadata.txt, .tree, and raw data files) to recreate these analyses as well as additional supplemental material derived from this study.

Submission history

Received: April 13, 2023
Accepted: August 23, 2023
Published online: October 2, 2023
Published in issue: October 10, 2023

Keywords

  1. mucosal inflammation
  2. host response
  3. subgingival microbiome
  4. experimental gingivitis
  5. periodontal disease

Acknowledgments

The research was funded through a Colgate-Palmolive clinical research grant (R.P.D., PI) and in part by the NIH TL1 TR002318, KL2 TR002317, T90 DE021984, (to K.A.K.), DE029872 (to G.A.K.), and DE023810 (to J.S.M.). This funding source had no role in the design of this study and did not have any role during its implementation, analyses, interpretation of the data, or decision to submit results.
Author Contributions
K.A.K., S.B., D.C., H.M.T., R.P.D., and J.S.M. designed research; K.A.K., S.B., G.A.K., R.P.D., and J.S.M. performed research; K.A.K., S.B., E.L.H., G.A.K., B.G.L., D.D.D., F.A.R., R.P.D., and J.S.M. analyzed data; and K.A.K., S.B., E.L.H., G.A.K., B.G.L., D.D.D., F.A.R., D.C., H.M.T., R.P.D., and J.S.M. wrote the paper.
Competing Interests
The authors declare no competing interest.

Notes

This article is a PNAS Direct Submission.

Authors

Affiliations

Department of Periodontics, University of Washington, Seattle, WA 98195
Department of Oral Health Sciences, University of Washington, Seattle, WA 98195
Department of Periodontology, Faculty of Dentistry, King Abdulaziz University, Jeddah 21589, Saudi Arabia
Erik L. Hendrickson
Department of Periodontics, University of Washington, Seattle, WA 98195
Department of Oral Biology, Rutgers, Newark, NJ 07103
Department of Periodontics, University of Washington, Seattle, WA 98195
Department of Biostatistics, University of Washington, Seattle, WA 98195
Diane D. Daubert
Department of Periodontics, University of Washington, Seattle, WA 98195
Department of Periodontics, University of Washington, Seattle, WA 98195
Department of Oral Health Research, Colgate Palmolive Company, Piscataway, NJ 08854
Harsh M. Trivedi
Department of Oral Health Research, Colgate Palmolive Company, Piscataway, NJ 08854
Department of Periodontics, University of Washington, Seattle, WA 98195
Department of Microbiology, University of Washington, Seattle, WA 98195
Department of Periodontics, University of Washington, Seattle, WA 98195
Department of Oral Health Sciences, University of Washington, Seattle, WA 98195
Department of Microbiology, University of Washington, Seattle, WA 98195

Notes

1
To whom correspondence may be addressed. Email: [email protected] or [email protected].

Metrics & Citations

Metrics

Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.


Citation statements




Altmetrics

Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

View Options

View options

PDF format

Download this article as a PDF file

DOWNLOAD PDF

Get Access

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Personal login Institutional Login

Recommend to a librarian

Recommend PNAS to a Librarian

Purchase options

Purchase this article to access the full text.

Single Article Purchase

Localized microbially induced inflammation influences distant healthy tissues in the human oral cavity
Proceedings of the National Academy of Sciences
  • Vol. 120
  • No. 41

Media

Figures

Tables

Other

Share

Share

Share article link

Share on social media