Periodontal inflammation: Integrating genes and dysbiosis

Abstract Biofilm bacteria co‐evolve and reach a symbiosis with the host on the gingival surface. The disruption of the homeostatic relationship between plaque bacteria and the host can initiate and promote periodontal disease progression. Recent advances in sequencing technologies allow researchers to profile disease‐associated microbial communities and quantify microbial metabolic activities and host transcriptional responses. In addition to confirming the findings from previous studies, new putative pathogens and novel genes that have not previously been associated with periodontitis, emerge. For example, multiple studies have reported that Synergistetes bacteria are associated with periodontitis. Genes involved in epithelial barrier defense were downregulated in periodontitis, while excessive expression of interleukin‐17 was associated with a hyperinflammatory response in periodontitis and with a unique microbial community. Bioinformatics‐enabled gene ontology pathway analyses provide a panoramic view of the bacterial and host activities as they shift from periodontal health to disease. Additionally, host innate factors, such as genetic variants identified by either a candidate‐gene approach or genome‐wide association analyses, have an impact on subgingival bacterial colonization. Transgenic mice carrying candidate genetic variants, or with the deletion of candidate genes mimicking the deleterious loss‐of‐function variant effect, provide experimental evidence validating the biologic relevance of the novel markers associated with the microbial phenotype identified through a statistical approach. Further refinement in bioinformatics, data management approaches, or statistical tools, are required to gain insight into host‐microbe interactions by harmonizing the multidimensional “big” data at the genomic, transcriptional, and proteomic levels.


| INTRODUC TI ON
The current model of pathogenesis of periodontal disease underlines the complex interactions among plaque bacteria, the host's genetic factors, and acquired environmental stressors. 1 Bacteria can elicit inflammatory responses in different host cell populations through interactions of pathogen-associated molecular patterns and pattern recognition receptors. For example, lipopolysaccharide, a prototype of pathogen-associated molecular patterns from Escherichia coli, is a potent inflammation inducer by activating toll-like receptor-4, a classic pattern recognition receptor, on the cell surfaces of macrophages or dendritic cells. 2,3 Other pathogen-associated molecular patterns, such as the bacterial cell wall component peptidoglycan or flagellin of motile bacteria, can be recognized by toll-like receptor-2 and toll-like receptor-5 on the host cell surfaces, respectively. 4,5 In addition, inflammation can also be readily stimulated by activating intracellular pattern recognition receptors such as nucleotide-binding oligomerization domain-like receptors and retinoid acid-inducible gene I-like receptors. 6,7 Although bacteria present in the plaque biofilm initiate inflammation, intrinsic host factors and environmental stressors modulate the magnitude, duration, and extent of an inflammatory response. For example, occasionally patients with poor oral hygiene and abundant calculus present inflammation that is mostly confined to the gingiva while the deeper tooth-supporting structures are rarely affected. This clinical observation is echoed by historical longitudinal studies monitoring the rate of periodontal destruction, which undoubtedly suggests a difference in disease progression in populations with a similar oral hygiene condition. 8,9 These distinct susceptibilities to periodontitis clearly indicate a genetic component. In addition, smoking, nutrition, medications, and systemic diseases such as diabetes, also modify periodontal disease activity. Therefore, the genetic makeup and the epigenetic program modulated by environmental stimuli determine the periodontal inflammatory response upon challenge by plaque pathogens at the oral mucosal or gingival surface.
A comprehensive mapping of molecular events, such as transcriptome analysis in tissue samples, offers an integrated global overview of biologic process changes in diseases. Several transcriptome studies have elucidated common pathways or molecular regulatory networks that are specific to different states of periodontal diseases. [10][11][12][13] Other than those signature molecular events, several novel pathways, such as those associated with neural regulation or epithelial barrier defense function, have emerged. 13,14 These altered biologic activities, which have not been previously characterized in periodontal disease, provide a fresh perspective on understanding periodontal inflammation as a response to plaque bacteria.
Subgingival regions at the root surface or pocket epithelium harbor a favorable environment for microbial colonization. Plaque bacteria employ various strategies to foster a metastable microbial community that is in harmony with local gingival tissues. However, disturbances that disrupt this symbiosis can lead to periodontal disease. A majority of investigators agree that periodontitis is as- ribosomal RNA sequencing have not only confirmed the major findings from those classical studies, but have expanded our understanding of the bacterial etiology of periodontal disease by discovering novel taxonomic bacteria that were not previously associated with periodontitis. 18,19 Information from this ribosomal homology-based high-throughput sequencing technique and whole genomic sequencing with an even more refined resolution at the species level provide a global snapshot of bacterial diversities and community dynamics associated with periodontal health and disease conditions.
Similarly to other common polygenic diseases, such as diabetes, coronary disease, or hypertension, the susceptibility to periodontal disease can be explained by the "common disease common variant" hypothesis. In this hypothetic model, disease-associating genetic variants or single nucleotide polymorphisms present in the coding, regulatory, or intergenic sequences of genes are relatively common in the general population. [20][21][22] The disease phenotype is determined by the totality of variants, each of which contributes to the clinical disease in varying degrees. However, the effect size of those disease-associating single nucleotide polymorphisms is usually moderate compared with rare alleles found in rare familial disorders with high penetrance and more severe clinical disease presentations. 22 Genome-wide association analysis represents an unbiased discovery tool to identify disease trait-associating genetic variants. Several genome-wide association studies have uncovered a number of candidate genetic loci or haplotypes that are significantly associated with different clinical and biologic periodontal inflammatory phenotypes at the genomic scale. [23][24][25][26][27] Many of these disease-associated genomic signals exist in or approximate to genes that have an unknown function or have not been previously implicated in periodontitis.
Genetic variants also have an impact on the colonization of bacteria in biofilm. Two models have been proposed to explain this genetics-associated dysbiosis. 28 In the first model, single nucleotide polymorphism variants may compromise genes that are associated with pathways of bacterial sensing and recognition. For example, variants in the promoter or coding sequence of pattern recognition receptor genes, or genes encoding binding partners of pattern recognition receptors in the downstream signaling events, would clearly affect the colonizing capacities of certain bacteria in plaque biofilm.
In the second model, the variant-affected genes lead to an excessive inflammatory environment in the subgingival niche, which favors the growth of specific biofilm bacteria. The increased production of gingival crevicular fluid caused by a hyperinflammatory response provides nutrients for plaque bacteria. Indeed, the inflammation-associated metabolic changes of bacteria can influence the mass and structure of subgingival biofilm. In addition, tissue breakdown caused by an excessive inflammatory response also feeds pathogenic bacteria and affects biofilm formation. 29 In this paper, we review the recent advances gained from periodontal "-omic" studies that profile host-bacteria interactions with a global panoramic overview. We summarize the molecular mechanisms of the inflammatory response in periodontal transcriptiome studies, the changes in diversity and dynamics of subgingival microbial flora, and the host genetic underpinning of bacterial colonization in plaque.

| FIND ING S OF TR AN SCRIP TOME S TUD IE S IN G ING IVITIS AND PERI ODONTITIS
With the improved resolution of sequencing technology that has become more readily available, transcriptome data provide a genomewide atlas of altered gene expression associated with periodontal disease processes. Several studies aiming to explore the distinct gene expression profiles and pathways unique to periodontitis have indeed confirmed the anticipatory upregulation or downregulation of genes involved in inflammation or bone resorption. 14,30 Examining aggressive and chronic periodontitis gingival biopsy tissues through gene ontology, Demmer et al 14 identified that p38, c-Jun N-terminal kinase, and mitogen-activated protein kinase pathways were significantly upregulated in diseased tissues, while several members of the transforming growth factor-beta family were downregulated in diseased tissues compared with disease-free control samples. In addition, they also reported upregulation for a majority of chemokines, platelet-derived growth factor and tumor necrosis factor signaling, and cell adhesion molecules, all of which were previously found to be positively associated with disease. Using a reverse engineering approach, the same group reconstructed the regulatory network based on the co-expression data in gingival tissues and subsequently interrogated the transcription factor or master regulator-target gene interactions to identify periodontitis-specific pathways. 10 Of all the differentially regulated pathways, those related to the immune response and immune system development were the most frequently enriched. For example, B-cell development, leukocyte extravasation signaling, phosphoinositide 3-kinase signaling in B lymphocytes, B-cell receptor signaling, and granulocyte adhesion and diapedesis were all among the top 10 most commonly enriched master regulator-associated pathways. Importantly, some of the major differentially regulated pathways reported in those 2 studies 10,14 have also been identified in other independent studies 30 . For example, pathways involved in cytokine and chemokine activities, B-cell receptor signaling, and defense and immunity proteins in both innate and adaptive immune responses, were reported to be among those most upregulated in periodontitis gingival tissues when assessed using RNA sequencing. 30 However, novel pathways, or genes that have not been previously associated with periodontitis, emerged in those "hypoth- infection. 37 Interleukin-17 receptor A null mice are more susceptible to oropharyngeal candidiasis caused by the commensal fungus Candida albicans than T helper 1-deficient mice. 38,39 However, exaggerated production of interleukin-17 was also found in gingival tissues affected by periodontitis. 29 Therefore, the homeostasis of this key cytokine, which balances interleukin-17-mediated mucosal defense and inflammatory response, is indispensable for maintaining periodontal health.
Transcriptome analysis at the genomic scale was also performed in gingivitis, a reversible inflammatory state that can be induced experimentally and safely in human participants.
Employing a well-established stent-induced biofilm overgrowth model in human subjects, 40

| MI CROB IAL DYS B I OS IS IN PERIODONTAL DIS E A S E
The current periodontal microbiology theory proposes that infectious oral diseases are the result of a "dysbiotic" biofilm rather than the direct effect of specific pathogenic bacteria in the host. 43 This theory suggests that microbial synergy among biofilm colonizers shapes and stabilizes a disease-provoking microbial profile that disrupts equilibrium with the host, leading to a diseased state. Several lines of experimental evidence, mostly centered on the classic periodontal pathogen, Porphyromonas gingivalis, support this notion.
Porpyromonas gingivalis is not a potent stand-alone inducer of inflammation, and often contradictory host responses to P. gingivalis are observed in vitro and in vivo. For example, P. gingivalis lipopolysaccharide can antagonize toll-like receptor 4, unlike other highly pro-inflammatory lipopolysaccharides from most gram-negative bacteria. 44 Similarly, in the absence of commensal bacteria, P. gingivalis fails to induce periodontitis when used as a mono-infection in germfree mice. 44 The dysbiosis theory then hypothesizes that P. gingivalis acts like a "keystone" pathogen, which even in the presence of a small fraction of the microbial community can elevate the pathogenicity of biofilm bacteria by disrupting host-bacteria homeostasis. 43 In fact, results from major human microbiome studies align well with the microbial dysbiosis theory in periodontitis pathogenesis.
The advent of 16S ribosomal RNA gene-sequencing technologies has allowed the evaluation of phylogenetic relatedness among bacterial species. Diaz et al 15 summarized that: (a) health-and periodontitis-associated microbial communities differ; (b) there is more bacterial diversity in subjects with periodontitis (eg, more species phylotypes are enriched in periodontitis) than in subjects with periodontal health; (c) health-associated species are not lost or replaced but rather are suppressed; and (d) periodontitis is associated with shifts in the species that numerically dominate subgingival communities rather than with de novo colonization by new species.

| Dysbiosis in gingivitis
The oral microbial transitions from periodontal health to disease involve microbial successions and adaptations to changing environments. Such transitions are also influenced by the host response in the gingival sulcus. 45 Transition of the microbial ecology from periodontal health to gingivitis is probably best characterized by the reversible nature of experimental gingivitis in response to the lack of plaque control. 9 Recent 16S ribosomal RNA high-throughput sequencing analyses have confirmed early findings regarding the shift from gram-positive cocci to gram-negative morphotypes (rods, filaments, and spirochetes). Kistler et al 46 evaluated the changes of microbial community diversity and shifts in periodontally healthy subjects, 1 and 2 weeks after oral hygiene abstention. Significant shifts in the microbiome were recorded during this transition, with particular attention focused on the correlation between the composition of the microbiome and bleeding on probing scores. They found that the bacteria which were negatively correlated with bleeding on probing were predominantly aerobic and facultatively anaerobic gram-positive cocci and rods, including members of the genera Actinomyces, Rothia, and Streptococcus (classic early colonizers of the tooth surface). 47 The species that increased in relative abundance as gingivitis developed and showed a positive correlation with bleeding on probing were mostly gram-negative taxa of the genera Campylobacter, Fusobacterium, Lautropia, Leptotrichia, Porphyromonas, Selenomonas, and Tannerella (mostly obligate anaerobes).

| Dysbiosis in periodontitis
Most studies analyzing subgingival microbial profiles agree that patients with periodontitis harbor a different bacterial community than people with periodontal health, usually with a greater abundance and a more diverse range of taxa in disease. [48][49][50][51] Such a different microbial profile is usually a result of differences in the relative abundance of taxa shared by both periodontal disease and health; that is to say, it is not entirely caused by novel colonizers. Therefore, the microbial shift from periodontal health to the diseased state is more than likely a microbial succession process in which the proportion of existing species changes as new colonizers emerge, and in which the taxa associated with periodontal health are not replaced. 48 While gram-negative anaerobic species are significantly enriched with increased gingival inflammation, the species that are enriched in gingivitis are not exactly the same as those associated with periodontitis. For instance, Diaz et al 15 reported that only about 20% of gingivitis-associated taxa are simultaneously associated with periodontitis. Also, within periodontitis-associated taxa, different "clusters" can be present, depending on the severity of the disease. with no known genetic defects, was usually increased compared with patients with periodontal health. 48,54,55 In addition, analysis of plaque samples from patients with leukocyte adhesion deficiency-I periodontitis demonstrated that Aggregatibacter actinomycetemcomitans, a species that is usually present in patients with aggressive periodontitis who do not have a known genetic or immune defect, was usually undetectable or only detected at a very low level. Interestingly, several species that are not usually associated with either chronic or aggressive periodontitis were uniquely present in patients with leukocyte adhesion deficiency-I periodontitis. For example, Pseudomonas aeruginosa, which is not associated with periodontal disease but is an infectious agent in immunocompromised patients, was readily detected in patients with leukocyte adhesion deficiency-I. 55,56 This antibiotic-resistant bacterium, which can survive in hostile environments, may influence the formation of a microbial community that is unique to patients with leukocyte adhesion deficiency-I periodontitis.
Other colonizers specific to leukocyte adhesion deficiency-I include Leptotrichia spp. that are not usually associated with common periodontitis. The genetics-determined inflammatory state of patients with leukocyte adhesion deficiency-I may have an impact on the leukocyte adhesion deficiency-I-specific microbial colonization. One unique inflammation signature of leukocyte adhesion deficiency-I periodontitis is the uninhibited production of interleukin-17 in local gingiva by increased numbers of T helper 17 cells. 29 It is likely that the excessive secretion of interleukin-17 in patients with leukocyte adhesion deficiency-I periodontitis, who also harbor a genetic mutation in integrin beta chain-2, promotes a dysbiosis that is distinct from the microbial communities associated with common periodontitis.

| Dysbiosis and metabolic signatures associated with periodontitis
A DNA sequencing-based profiling approach that explores the microbial community cannot readily provide information about whether the identified bacteria are metabolically active or even alive or dead, as the sequencing data do not measure the biologic activi- and Treponema phyla were significantly associated with periodontitis parameters. [57][58][59] Synergistetes spp. were also associated with 2 novel dipeptides: cyclo(-Phe-Pro) and cyclo(-Leu-Pro). In addition to quorum-sensing molecules, these metabolites possess bacteriolytic activity and therefore inhibit the growth of certain bacteria in biofilm. [60][61][62] It is hypothesized that high levels of those cyclodipeptides promote dysbiosis by thwarting commensal bacteria and favoring the overgrowth of Synergistetes bacteria, which are clinically associated with severe periodontal disease. Another study reported using ribosome RNA-sequencing to access aggressive periodontitis-associated subgingival microbial composition and metatranscriptomic RNA-sequence data to analyze microbial gene-expression profiles from metabolically active microorganisms in the same plaque samples. The results showed that the disease-associated microbial communities had overall significantly fewer species (alpha-diversity) and less dispersed distribution (beta-diversity) in diseased pockets than in periodontal health-associated sites. 63 Genes involved in conserved metabolic pathways, such as lysine fermentation to butyrate, histidine catabolism, nucleotide biosynthesis, and pyruvate formation, were consistently upregulated in diseased sites, and probably contribute to the disease process. The sequencing data also suggest that although subgingival plaque-colonizing bacteria usually present

| G ENE TI C DE TERMINANTS OF SUBG ING IVAL BAC TERIAL COLONIZ ATION
The homeostasis at the mucosal surface is co-determined by the colonizing bacteria and host immune response, underpinned by the host's genetics. A large amount of evidence has shown that resident microbes foster normal immune-system development and mediate both innate and adaptive immune responses to achieve a symbiosis at different mucocutaneous niches of the host, such as skin, gut, and oral cavity. Convincing experimental data from germ-free mice have provided clues as to how commensal bacteria "educate" immune systems by avoiding excessive immune reactions to achieve such a symbiosis. For example, polysaccharide A from Bacteroides fragilis in gut mucosa can exploit the host toll-like receptor-2 pathway to specifically suppress T helper 17 cell development. Such a B. fragilisassociated toll-like receptor-2-mediated inhibition of the T helper 17-response promotes colonization of this commensal bacteria. 68 By contrast, host genetic information shapes the composition of the microbiota residing in those niches by modifying immune responses.
A good example that illustrates the host genetic influence on gut microbe colonization is from mice genetically deficient in the toll-like receptor-5 gene, 69 which exhibit hallmark features of metabolic syndrome. Recipient wild-type, germ-free mice that were transferred with gut microbiota collected from those toll-like receptor-5-deficient mice also developed many features of metabolic syndrome. Therefore, the colonization of gut biofilm by pathogenic bacteria is selective and takes place under the pressure of genetics-mediated host immune responses.
It has long been hypothesized that the host genetic variants affect the colonization of bacteria in the subgingival ecological niche.
In addition to a hyper-or hypo-inflammatory response, the variant-mediated host response may also determine the composition and microbial load of subgingival bacteria aggregating in plaque biofilm. This genetic effect on subgingival bacterial colonization is well illustrated in mice with a specific ablation of genes mimicking the extreme loss-of-function genetic variants or humans with genetic defects. For example, mice deficient in the developmental endothelial locus 1 gene deficiency harbored a significantly higher oral bacterial count than age-matched control mice, which had a qualitatively different oral microbial community. 70 Leukocyte adhesion deficiency-I patients with molecular defects induced by mutations in the integrin beta chain-2 (integrin subunit beta-2) gene exhibited a significantly greater bacterial load in subgingival plaque and presented a unique disease-associating composition of microbes, such as enrichment of Treponema spp., Saccharibacteria phylum, Porphyromonas endodontalis, P. aeruginosa, Leptotrichia spp., and Scardovia wiggsiae. 55 These results indicate that microbial colonization and pathogenicity are profoundly affected by genetic determinants.

| Genetic determinants of bacterial colonization by a candidate-gene association approach
The "inside-out" relationship, in which the host's intrinsic genetic factors affect microbial colonization at the gingival surface, has been explored using a candidate gene approach that links several

| Genetic determinants of bacterial colonization using a genome-wide association approach
The limited success of the candidate gene approach is usually confined by the lack of power inherent to the small sample size frequently seen in those studies, increased risk of bias because of the case-control design, which is based on care-seeking rather than a population-based method, and high population heterogeneity with an absence of stratification. 78 . This variant site is in the promoter region of potassium channel subfamily K member 1, a potassium channel gene whose overexpression is linked to heart failure. 86 The same single nucleotide polymorphism variant is also close to mitogen-activated protein kinase kinase kinase 21, a gene-encoding mixed lineage kinase 4 that negatively regulates lipopolysaccharide-mediated toll-like receptor-4 signaling. 87 Another variant at rs1932040 (A/G) that is associated with a high level of colonization with bacteria of the orange complex is located in an intergenic area encompassed by the chloride intracellular channel protein 5 gene and the runt-related transcription factor 2 gene, which encodes an osteogenic marker that is involved in periodontitis. 88 Other suggestive lead single nucleotide polymorphism variants are located in proximity to genes encoding proteins involved in inflammation and the immune response (eg, interleukin-33, vesicle-

| Validation of candidate genetic markers
The identified variant loci or candidate gene targets that are associated with either specific bacterial pathogens or the subgingival plaque microbial community from the aforementioned coronary cohort by genome-wide association study need to be validated in other populations. Independent genome-wide association investigations have also identified periodontitis-associated loci close to genes that were reported in the genome-wide association study in the atherosclerosis risk in communities population. For example, the variants encompassing the neuropeptide Y gene, one of the four target genes which met the gene-centric statistical significance threshold in participants of the atherosclerosis risk in communities study, have also been marked by the genome-wide genotyping data from a separate F I G U R E 2 Plakophilin 2 (PKP2) lossof-function inhibits cell proliferation and cell-to-cell contact. PKP2 knockdown in primary gingival epithelial cells significantly hindered cell proliferation, as assayed by the MTS assay (A) and impaired cell spreading with increased gaps among cells 1-h post-seeding (B) community-dwelling adult population in a US and a German population. 24 Significantly more P. gingivalis was present in TRAF3 interacting protein-2-null mice than in control wild-type mice, in the plaque samples collected after oral infection. Such a delay in clearance of P. gingivalis was associated with lower neutrophil infiltration of gingival tissues in knockout mice ( Figure 3B). Therefore, a weakened epithelial defense because of inefficient neutrophil recruitment in those interleukin-17 F I G U R E 3 TRAF3 interacting protein-2 (Traf3ip2) −/− mice harbored more Porphyromonas gingivalis A7436 and had less neutrophil infiltration than age-matched wild-type (WT) control animals. Mice were orally inoculated with P. gingivalis A7436 for 14 d. Plaque samples were collected 2 d after the last inoculation and P. gingivalis was quantified by real-time quantitative PCR against a standard. (A) Significantly more P. gingivalis was present in oral plaque from Traf3ip2 −/− mice than in oral plaque from wild-type controls (*P = .008 Mann-Whitney test). (B) Fewer neutrophils (leukocyte antigen-6-gene-positive cells stained by immunohistochemistry, as indicated by arrows) were present in the gingival tissue of Traf3ip2 −/− mice (upper panel) than in the WT controls (lower panel); E, enamel

| FUTURE D IREC TI ON S
The sophisticated sequencing tools now available provide ample opportunities to predict host-pathogen interactions at a global scale on the gingival surface. The metabolomic and proteomic work further adds functional dynamics to the interplay between the host and the subgingival microbiota. However, we are still at the infancy of comprehending those global changes of and interactions between the host response and plaque bacteria activities that determine an individual's susceptibility to periodontal disease and shape the disease's course. One of the challenges we are confronting is how to harmonize and integrate these "-omics" F I G U R E 4 Global profiling or "-omics" data sets obtained from the host and plaque biofilm are integrated through refined bioinformatics, computational biology, and statistics tools to evaluate the host's determinates on bacterial colonization in plaque biofilm and host responses to a dysbiotic microflora in the subgingival environment. GWA, genome-wide association analysis; rRNA, ribosomal RNA

Bioinformatics, computational biology and statistics
Responses modulated by dysbiotic bacteria

Host determinants
profiling data to comprehensively understand periodontal disease processes. More refined data-management tools, statistical and mathematical models, and bioinformatics are necessary for improved data integration. Such a holistic "multidimensional" approach allows investigators to assess the host's genetic impact on bacterial colonization in plaque biofilm and host responses to dysbiotic microflora in the subgingival niche ( Figure 4). Future efforts are also required to replicate and biologically validate the candidate targets screened from "-omic" data. For example, novel genomic markers suggested by the genome-wide association studies from the atherosclerosis risk in communities population need to be replicated by other studies using microbial pathogens as a phenotypic component. The variant-associated subgingival microbial community shifts and metabolic changes are largely unknown and need to be explored. The role of genes that have not been previously implicated in periodontal disease can be evaluated in vivo using transgenic mice in various periodontal bone loss models. In addition, candidate single nucleotide polymorphisms can be genomically engineered through the clustered regularly interspaced short palindromic repeats/CRISPR-associated protein 9 (CRISPR/CAS9) system in cells and mice where the variant effects can be assessed. 99

| CON CLUS IONS
Global profiling data enable a comprehensive assessment of biofilm bacteria-host interactions in periodontal disease. Indeed, the multidimensional information derived from these "-omics" data sets confirmed certain findings from previous studies that have shaped our current understanding of disease processes. Nevertheless, novel single nucleotide polymorphisms, genes, pathways, metabolites, and bacterial species that have not been previously associated with periodontal disease are emerging. Those new candidate targets clearly require replication in other studies with a similar design but performed in different populations. More importantly, these novel disease-associating markers need to be validated biologically in vitro and in vivo using animal models. In addition, more advanced bioinformatic tools are called for to harmonize these "big" data of different dimensions to advance precision medicine to measure an individual's risk for periodontitis and to provide the rationale for personalized periodontal therapies.

ACK N OWLED G M ENT
The authors would like to recognize Dr. Steven Offenbacher DDS, PhD, MMSc for his never-ending support and continued mentorship of their careers in dental academics.