Host-Microbial Interactions in Systemic Lupus Erythematosus and Periodontitis

Background: Systemic lupus erythematosus (SLE) is a potentially fatal complex autoimmune disease, that is characterized by widespread inflammation manifesting tissue damage and comorbidities across the human body including heart, blood vessels, joints, skin, liver, kidneys, and periodontal tissues. The etiology of SLE is partially attributed to a deregulated inflammatory response to microbial dysbiosis and environmental changes. In the mouth, periodontal environment provides an optimal niche for local and systemic inflammation. Our aim was to evaluate the reciprocal impact of periodontal subgingival microbiome on SLE systemic inflammation. Methods: Ninety-one female subjects were recruited, including healthy (n = 31), SLE-inactive (n = 29), and SLE-active (n = 31). Patients were screened for probing depth, bleeding on probing, clinical attachment level, and classified according to CDC/AAP criteria with or without periodontal dysbiosis. Serum inflammatory cytokines were measured by human cytokine panel and a targeted pathogenic subgingival biofilm panel was examined by DNA-DNA checkerboard from subgingival plaque samples. Results: The results showed significant upregulation of serum proinflammatory cytokines in individuals with SLE when compared to controls. Stratification of subject's into SLE-inactive (I) and SLE-active (A) phenotypes or periodontitis and non-periodontitis groups provided new insights into SLE pathophysiology. Ten proinflammatory cytokines were upregulated in serum of SLE-I only and one in SLE-A only. Four molecules overlapped in SLE-A and SLE-I. Anti-inflammatory cytokines included IL-4 IL-10, which were upregulated in SLE-I sera (but not SLE-A), controlling clinical phenotypes. Out of 24 significant differential oral microbial abundances found in SLE, 14 unique subgingival bacteria profiles were found to be elevated in SLE. The most severe oral pathogens (Treponema denticola and Tannerella forsythia) showed increase abundances on SLE-A periodontal sites when compared to SLE-I and healthy controls. Inflammation as measured by cytokine-microbial correlations showed that periodontal pathogens dominating the environment increased proinflammatory cytokines systemically. Conclusions: Altogether, low-grade systemic inflammation that influenced SLE disease activity and severity was correlated to dysbiotic changes of the oral microbiota present in periodontal diseases.


Introduction
The human microbiome is in constant interaction with the host, modulating health and disease phenotypes. We now appreciate that oral bacteria play indispensable roles in shaping the systemic host physiological landscape and in dysbiosis. As part of the upper digestive tract, the oral cavity presents specific niches, such as gingival sulcus and gingival crevicular fluid (GCF), which in turn harbor commensal and pathogenic bacteria with potential impact to oral and systemic health. Through local activation of inflammation, oral pathogens have shown to worsen the burden of chronic diseases through time including, type 2 diabetes, premature labor, rheumatoid arthritis, systemic lupus erythematosus (SLE), Alzheimer's, cardiovascular conditions, and cancer [1][2][3][4] .
Locally, the oral commensal flora and tissue inflammation evolved to develop a relationship of homeostasis. In dysbiosis, pathogens become dominant, including species previously clustered according to their pathogenesis into the "red complex"-Tannerella forsythia, Porphyromonas gingivalis, and Treponema denticola 5 . In periodontitis, for example, continuous and unresolved inflammatory response affects toothsupporting tissue structures (periodontal ligament, connective tissue, and bone) 3,6 . As one of the most common infectious diseases globally, the etiology of periodontitis is multifactorial, and the microbial-inflammation imbalance leads to bone resorption and consequent tissue tooth loss 7 . Inflammation precedes tissue loss and the systemic impact of periodontal dysbiosis goes beyond the oral compartment.
Systemic lupus erythematosus (SLE) is a multisystem autoimmune disease with increasing incidence worldwide 8 . The Lupus foundation of America estimates that 1.5 million Americans have SLE, and at least five million people worldwide have a form of lupus 9,10 . SLE affects mostly women and most people with the condition manifests the symptoms between ages of 15-44. The current incidence of SLE is 16,000 new cases per year and patients experience significant symptoms, such as pain, and fatigue, hair loss, cognitive issues, physical impairments, oral and vaginal mucosa manifestations, affecting every facet of their lives 11 . According 6 to the American College of Rheumatology (ACR), an array of clinical exams and positive serology are used to diagnose disease, including malar rash, discoid rash, arthritis, kidney disorder, anemia or leukopenia, abnormal antinuclear antibodies (ANA) and anti-DNA or antiphospholipid antibodies 12 . In addition to molecular diagnosis, systemic lupus erythematosus disease activity index (SLEDAI) is used in the assessment of disease severity and response to treatment; manifestations of comorbidities and molecular abnormalities affecting multiple organs of the body often challenges diagnostics at early stages 13 . Although complex diseases have shown to be influenced by genetic and environmental forces, the gut microbiome, and recently the oral microbiome, showed direct impact on SLE subjects 14 , the complexity of chronic diseases such as SLE is beyond isolated body compartments, and it requires the integration of host-microbial interactions. The heterogeneity of disease presentation and organ involvement contribute to clinical challenges for diagnosis and effective management 15 . While several studies reported associations among human oral microbiota compositions in SLE [16][17][18][19] , co-occurrences of specific periodontal pathogens and inflammatory cytokines important for low grade inflammation and chronic disease remains to be explored 4 .
In this study we have investigated SLE phenotypic differences, oral microbiota associations with systemic inflammation to unravel forces shaping oral and broader systemic health. As such, we compared the subgingival microbial signatures from SLE and healthy controls to investigate host-microbial relationships.
Specifically, we first sought to establish if specific microbial compositions correlated to periodontal clinical phenotypes with systemic levels of inflammatory cytokines. The presence of periodontal disease was assessed using categories of local inflammation, clinical attachment levels and bone loss, while SLEDAI > or < 2 was assessed to categorize the individuals into SLE-active, SEL-inactive. The results indicate that chronic periodontitis and SLE present low-grade inflammation modulated periodontal diseases and specific microbial signatures. The association of subgingival microbial profiles with SLE and its association with periodontal 7 clinical status and inflammatory markers established novel links developing a new framework for oral-systemic studies.  23 , which was used as an additional method of information on disease severity to better characterize the studied population. SLE patients were subdivided into two groups based on their disease activity: active (SLE-A; SLEDAI > 2; n= 29), and inactive (SLE-I; SLEDAI ≤ 2; n=31).

-Subjects
The control group was composed of 31 systemically healthy women recruited at the University Hospital in Brasília during the same time period. All patients answered a health questionnaire investigating medical 8 and dental history. A visual investigation of the oral cavity was performed at enrollment to investigate the number of remaining teeth, other oral diseases and periodontal health.
Periodontal Status. After recruitment of subjects, participants were referred to the Dental School of Brasília (Brasilia, Brazil) for clinical periodontal examination and sample collection. The clinical periodontal examination was performed by a single trained examiner before the collection of GCF, subgingival plaque and blood samples prior to oral and non-surgical periodontal treatment. The following parameters were investigated by using a UNC-15 millimeter periodontal probe at 6 sites/tooth: pocket probing depth (PD), clinical attachment level (CAL) and gingival bleeding index 24 . Plaque index was assayed by staining the dental plaque with disclosing solution at mesial, buccal, distal and lingual sites 25 .
After periodontal examination, periodontal disease was classified according to CDC and AAP criteria [25][26][27] including: A) Mild periodontitis: 2 or more interproximal sites with ≥ 3 mm of CALs and ≥ 4 mm PD (not on the same tooth) or 1 site with PD ≥ 5mm; B) Moderate periodontitis: 2 or more interproximal sites with ≥ 4 mm of CAL (not on the same tooth) and 2 or more interproximal sites ≥ 5 mm PD (not on the same tooth); C) Severe periodontitis: 2 or more interproximal sites with ≥ 6 mm of CALs (not on the same tooth) and 1 or more interproximal site(s) with ≥ 5 mm PD. Total prevalence of periodontitis was determined by the sum of mild, moderate and severe periodontitis. The extent of disease was reported by the severity of disease at 5, 10 and 30% of sites and teeth. After radiographic and clinical examination followed by collection of blood, GCF and subgingival plaque samples, patients were submitted to nonsurgical periodontal treatment consisted of supra and subgingival scaling and root planing, when necessary, dental polishing and oral hygiene instruction. 9

-Sample Collection
Subgingival plaque and GCF sample collection were collected prior to any treatment for all subjects, including non-surgical periodontal treatment. GCF and subgingival plaque samples were collected from four sites showing the deepest probing depth in each patient. The area was isolated with cotton rolls to prevent contact with saliva and air dried. Supragingival plaque was removed with Gracey curettes previous to subgingival plaque collection. The subgingival plaque was retrieved by the introduction of PerioPaper strips into gingival sulcus for 30 seconds and stored in sterilized tubes at -80ºC till processing. Blood samples were collected after routine examination for SLE monitoring. Control subjects were asked to perform similar examinations at the Hospital. The samples were stored at -80ºC up to processing.
PerioPaper strips were centrifuged at 3000 x g for 15 min at 4ºC in PBS elution for the dosage of the tissue destruction and bone resorption markers by Multiplex (Human MAP, Millipore, USA).

-DNA extraction, Amplification and Checkerboard measurements
DNA-DNA checkerboard hybridization was performed using the method developed by Socransky et al. (1998). Briefly, to prepare standard probes, forty selected bacterial species were grown on agar plates and were suspended in 1 mL of TE buffer (10 mM Tris-HCl, 0.1 mM EDTA, pH 7.6). Cells were pelleted at 1300 x g for 10 min and washed with TE buffer. Gram-negative strains were lysed with 10% SDS and Proteinase K (20 mg/mL) while Gram-positive strains were lysed with 15 mg/mL lysozyme (Sigma) and 5 mg/mL achromopeptidase (Sigma). Following, cells were sonicated and incubated at 37°C for 1 hour.
DNA was isolated and purified as described previously by Smith et al. (1989). DNA probes were generated for each strain by labeling with digoxigenin (Roche, Basel, Switzerland) 28 . Next, the DNA content of the subgingival plaque samples was amplified using multiple displacement amplification 29 . Briefly, subgingival DNA templates were mixed with random hexamer primers in 50 mM Tris-HCl (pH 8.2) and 0.5 mM EDTA. The solution was heat denatured at 95°C for 3 min in a thermocycler. DNA polymerase with dNTPs was then added and incubated at 30°C for 2 hours. The subgingival plaque samples along with 1 ng and 10 ng of known DNA standards were analyzed by using checkerboard DNA-DNA hybridization as previously described by Socransky et al., (2004). In brief, samples were boiled for 10 min in 1 mL of TE buffer and were loaded on a nylon membrane (Roche, Basel, Switzerland) using a Minislot 30 apparatus (Immunetics, Cambridge, MA, USA). Then cross-linked by ultraviolet light using UV-crosslinker (Stratalinker 1800, La Jolla, CA, USA). Probes were subsequently bound perpendicular to the samples on the membrane and were detected using anti-digoxigenin antibody conjugated with alkaline phosphatase and a chemifluorescent substrate. Signal intensities of the plaque samples were measured using a Typhoon TRIO+ Scanner (GE Healthcare). Finally, absolute counts were generated by comparing with signal intensities of standards using Phoretix Array Software (TotalLab) 30,31 .

-Principal Coordinate analysis
The differences between control and diseased groups were investigated by applying Principal Coordinates Analysis (PCoA) on the binary Jaccard distance metric using ClusterApp tool (available at http://dorresteinappshub.ucsd.edu:3838/clusterMetaboApp0.9.1/) and visualized via EMPeror software 32 .
Pairwise permutation MANOVA (ADONIS within RVAideMemoire package in R) with 999 permutations were used to test significant differences between sample groups.

-Phylogenetic analysis
The 16S rRNA gene sequences were downloaded from the Human Oral Microbiome Database (HOMD; version 14.51, January 2017), and were aligned using ClustalW software within Molecular Evolutionary Genetics Analysis (MEGA) software version 7 33 . The phylogenetic tree was plotted by applying the Neighbor-Joining method using p-distance model and by implementing 1,000 bootstrap replications. Species abundance data and the phylogenetic tree were jointly visualized further using the interactive tree of life (iTOL) version 4 (available at http://itol.embl.de/) 34 . Cytokine abundance levels were reported in pg/mL by the assay system. Quality control of the machine-generated raw data was performed using xPONENT 4.2 software as per manufacturer's guidelines (Affymetrix eBioscience, San Diego, USA).

-Statistical analyses
Statistical analyses were applied to the microbial and cytokine abundance data to detect any features that are significantly different between diseased and control groups. Rank-based tests were employed as a robust nonparametric group comparison tool for those non-normally distributed microbiome and cytokine data. Using Wilcoxon rank-sum test, we further analyzed differences between groups i.e., control versus SLE-I, control versus SLE-A and SLE-I versus SLE-A. For significance testing, FDR-corrected (Benjamini-Hochberg) p-values were set at 5% threshold. For binary abundances we dichotomized the relative abundances data into presence (i.e. abundance > 0) and absence (i.e. abundance = 0) binary values. We used a logistic regression model to analyze the relationship between the binary abundance data and two clinically important categorical covariates: SLE group and periodontal condition 36 . For the disease group, we have three levels (control, SLE-A, and SLE-I); for the periodontal condition, we have two levels (control versus periodontitis).
Potential associations between microbial species and cytokine abundances were evaluated by Spearman's rank-based correlations. Spearman's correlation coefficients were calculated using R package psych, and significant correlations were plotted with R package corrplot. Pairwise correlations were computed separately for control and SLE samples. To avoid false positives, we selected correlations with adjusted p-values <0.05. For visual simplicity, we showed only the significant correlations. Topological parameters were analyzed by importing the pairwise correlations using network analyzer algorithm within Cytoscape, 3.7.1. Hierarchical clustering was applied on the relative abundance data of species in each sample based on Spearman rank correlation and the heatmap was generated using ggplots package in R.

-Clinical Findings of the Cohort
To investigate the role of the periodontal microbiome-induced inflammation on SLE subjects, whole blood, crevicular fluids, and plaque were collected from subjects enrolled in the study. General characteristics of patients are shown in Table 1. A total of 91 subgingival samples were analyzed from women aged 18-65. SLE activity was determined by SLEDAI > 2 37 . Of 91 subjects, 31 individuals were 13 controls, 29 were SLE-I and 31 were SLE-A. The average SLE disease activity index (SLEDAI) for SLE-A subjects significantly increased (7.29 ± 4.31), while SLE-I showed a lower SLEDAI (1.07 ± 0.99).
Patients with SLE-I had a longer duration of the disease and aged significantly older, whereas the SLE-A patients were relatively younger than controls and SLE-I subjects (32.58 ± 8.9 years, p = 0.045). SLE-I individuals showed increased clinical attachment loss (CAL) and worsen periodontal clinical outcomes (8.31 ± 9.438). The majority of oral health indices including teeth number, bleeding points, BOP (Bleeding on probing), stained teeth, plaque index, and probing depth were similar between SLE groups. Ethnicity was equally distributed among the three groups of the studied population.

-Inflammatory Cytokines Upregulation in SLE
Various inflammatory cytokines have been involved in regulating disease activity and in organ pathologies of patients diagnosed with SLE 38,39 . In order to identify specific markers responsible for systemic inflammation in SLE patients, we investigated whole blood serum concentrations of pro and anti-inflammatory cytokines from SLE groups compared to healthy control. We measured IL1β, IL-1ɑ), and E-selectin was unique to SLE-A only ( Figure 1B). SLE-I sera also demonstrated antiinflammatory increase by upregulation of IL-10, this was not evident on SLE-A. Among the significant differences from SLE, inactive and active groups, there were four molecules that overlapped (MCP-1, IL- 14 8, IP-10, IL-6). These results suggest that increased low-grade systemic inflammation was observed in SLE patients when compared to healthy controls.

-Subgingival Microbial Compositions in SLE
To visualize variations in the composition of subgingival biofilm-associated bacterial species across samples within SLE with or without periodontitis from controls (n = 31), SLE-inactive (n = 28), and SLE-active (n = 31) groups; we performed a collection of samples by probing the gingival site with PerioPaper strips, following elution. After hybridization with specific periodontal pathogen probes, DNAcheckerboard revealed pathogen signatures and abundances. A principal coordinate analysis (PCoA) on the Jaccard distances was generated taking into consideration of periodontal status stratification ( Figure   2A). These visual patterns were confirmed by beta-diversity analysis using pairwise permutation Multivariate analysis of variance (MANOVA), revealing significant differences in bacterial species in the SLE-active group without periodontitis and SLE-inactive group with periodontitis compared to control group with periodontitis (p = 0.045; Table 2). Lastly, no differences were found between SLE-inactive groups and healthy controls (Supplementary Table 2). Taken together, we found significant variations in the microbiome composition in the SLE when compared to controls, with increased differences when SLE groups were stratified into SLE-A and SLE-I.

-Microbial Signatures Associated with SLE and Periodontal Phenotypes
We next sought to understand the underlying microbial associations among SLE subjects and their clinical phenotypes. To investigate the subgingival plaque bacterial composition, we thus applied Wilcoxon rank-sum test on two datasets related to SLE phenotypes. The first dataset was based on relative abundance and the second dataset was based on logistic regression model binary abundances 15 (presence/absence) to evaluate prevalence. Based on the relative abundance profiles, and the hierarchical clustering using Spearman's rank correlation coefficients, we identified two distinct groups of bacterial species in SLE individuals with and without periodontitis when compared to their respective controls ( Figure 2B, Figure 3A). Intriguingly, these comprised of species representatives of pathogenic groups previously classified in clusters including the red complex 30 Table 2). A total of twenty-four bacterial species had significant differential abundances among SLE and control subjects, with nine in the non-periodontitis and fifteen in the periodontitis groups ( Figure 3A  We further characterized the periodontal pathogens in sites from SLE-I versus SLE-A individuals. Proportions of C. gingivalis, S. gordonii, P. nigrescens, C. ochracea, F. nucleatum and S. sanguinis were significantly reduced on SLE subjects ( Figure 2B, Figure 3A, Supplementary Table 2). Based on our analysis the red complex pathogen, T. forsythia, was enriched in periodontitis subjects from SLE-active, but not in their healthy counterpart. In addition to these changes, we found S. noxia, S. oralis and A.
gerencseriae at higher abundance in the SLE-I group compared with control individuals with periodontitis ( Figure 2B, Figure 3A). A. gerencseriae, S. oralis, C. ochracea, P. nigrescens and T. forsythia were also found to be significantly different between SLE-I and SLE-A groups. These results suggest unique pathogen signatures associated with SLE-A phenotype and that commensals are associated with better systemic and periodontal health outcomes. A more stringent evaluation of bacterial abundance profiles via binary abundances revealed significant differences in the presence/absence of C. ochracea, C. gingivalis, P. nigrescens, T. forsythia, F. nucleatum, S. gordonii and S. sanguinis in periodontal sites of the SLE-active group compared with the control group (Supplementary Table 3).
Unique and overlapping microbial species were identified among periodontitis versus nonperiodontitis groups ( Figure 3B). Among fourteen unique microbial species, four bacterial species overlapped among non-periodontitis and periodontitis individuals (F. nucleatum, A. gerencseriae, S. sanguinis and C. ochracea). While four unique bacterial species (T. denticola, S. intermedia, F. polymorphum and A. odontolyticus) were significantly abundant only in non-periodontitis subjects, six species were unique in periodontitis subjects (T. forsythia, S. oralis, P. nigrescence, S. gordonii, S. noxia and C. gingivalis; Figure 3B). Together these results suggest periodontal bacterial diversity were lower in SLE patients with enrichment of specific periodontal pathogens dominating the microbial environment.

-Bacterial Species and Cytokine Co-Occurrences
Periodontitis is caused by the complex interplay between subgingival microbiota composition and In SLE-inactive individuals, we found co-occurrence of F. nucleatum with E-selectin levels ( Figure 4B noxia was higher in the control group while C. gingivalis had no connectivity in control individuals ( Figure   5). These findings suggest that subgingival bacterial species associated with SLE govern systemic host cytokine patterns impacting overall health.

Discussion
At the center of autoimmune pathogenesis, this study provides evidence that inflammatory response to microbiome controls the severity, and magnitude of the SLE disease 40 . The data also indicates that periodontal levels are also associated with low-grade systemic inflammation. Oral microbes, including subgingival bacteria, are involved in homeostasis and maintenance of health, and also in the initiation and progression of chronic periodontitis (CP) which leads to systemic inflammation such as SLE [41][42][43][44][45][46][47] . Dual relationships among systemic and oral diseases play a role in the pathogenesis, and low-grade inflammation modulating the biological compartments 19 , and here we assayed how the microbial species influence specific systemic cytokines. Evidence of the exact repertoire of subgingival pathogens influencing specific lupus phenotypes has been scarcer in the context of oral and systemic disease. We 19 explored two distinct populations, healthy and SLE positive patients to determine these correlations. Our clinical cross-sectional study is unique because we have stratified chronic SLE patients, which are usually under treatment, (SLE-inactive) and a more acute group, and recently diagnosed SLE subjects (SLEactive). We have also carefully stratified the subjects in periodontally compromised versus healthy controls. Our results suggest that chronic and SLE-inactive subjects are positively associated with severe periodontitis states. Among healthy and SLE groups, age was not significantly different, but when stratifying SLE active and inactive groups, younger subjects were found significantly present on SLE-A when compared to SLE-I. This was expected due to time of disease since diagnosis 48 . Patients with SLE in general showed clear molecular patterns related to chronic inflammation and immune response to microbiome when compared to healthy controls.
A more distinct host cytokine dysregulation was associated with SLE subjects when compared to healthy controls. The data presented herein show no evidence that SLE-active subjects had higher serum concentrations of pro-and anti-inflammatory cytokine. But, both SLE groups showed higher expressions than healthy control. Interestingly IL-10 which is a anti-inflammatory cytokine response was found increased in SLE-I inactive when compared to active and controls (p=0.01, Figure 1A). Our results showed significantly increased levels of IFN-ɑ, IL-17A, MIP-1-ɑ, IL-4, IFN-ɣ, and IL-10 in SLE-inactive individuals. sICAM-1, GM-CSF, IL-12p70, IL-1β, IL-1ɑ in serum were unique of SLE-inactive subjects, while four cytokines were significantly increased, with one unique profile including E-selectin of SLE-A subjects ( Figure 1B). A high number of cytokine perturbations in SLE-I individuals can be associated with diverse clinical manifestations, activity and the severity of the disease. As mediators of inflammation, cytokine production feeds forward cell-cell and cell-tissue communications guiding health and disease phenotypes and organ disruptions. Here, we show that proinflammatory cytokines MCP-1, IL-8, IP-10, and IL-6 were significantly elevated in SLE-I and SLE-A patients in comparison to non-SLE subjects 20 ( Figure 1A). Of this proinflammatory cytokines, IL-6, IL-17, TNF-α, and IFN-α suppression have been reported as therapeutic targets for clinical management of lupus and other chronic inflammatory diseases 4,[49][50][51] .
Systemic and local periodontal inflammation can impact subgingival bacterial compositions, which can in turn further enhance systemic inflammation, leading to tissue loss, mainly in subjects with SLE condition 49 . The bacterial abundances of fourty-selected subgingival species were evaluated through checkerboard and we found that two distinct groups of subgingival bacteria were present in SLE individuals, especially when stratifying for the presence of absence of periodontitis ( Figure 3A). S. sanguinis and C. ochracea are known to be prevalent in healthy subgingival sites [52][53][54] . In line with previous studies, proportions of T. denticola, S. sanguinis and C. ochracea were significantly higher in periodontally healthy individuals than in individuals with SLE. A. gerencseriae, and F. nucleatum were more evidently abundant in SLE individuals without periodontitis. We found a higher abundance of S. noxia, S. oralis and A. gerencseriae in SLE subjects with periodontitis. On the other hand, we found significant depletion of C. gingivalis, S. gordonii, P. nigrescens, C. ochracea, F. nucleatum and S. sanguinis, which were previously found to be associated with SLE pathogenesis (Figure 3A,   Supplementary Table 2, Supplementary Table 3). It is noteworthy that SLE-I microbial compositions was not different from healthy controls in the non-periodontitis group, indicating that they are similar in periodontally healthy subjects (Supplementary Table 2). On the contrary, we show that SLE-A microbial positions maintains differences in both periodontitis and non-periodontitis groups, and that one form of inflammation (local or systemic) control these differences associated with SLE systemic inflammatory profiles.
Co-occurrence of subgingival bacteria and cytokines also differed among controls, SLE-I and SLE-A subjects. Co-occurrence among cytokines was more evident (77 positive connecting edges) in the 21 SLE states, suggesting positive association among cytokines are an important driver of SLE pathogenesis ( Figure 4A-C). We did not find any correlations between bacteria and increased levels of IL-6 and IL-17.
Despite IFN-α and TNF-α co-occurred with certain subgingival bacteria, these bacteria were not differentially abundant. Beyond cytokine to microbe interactions, SLE-A showed higher cytokine- inflammatory profiles on subjects with inactive disease when compared to active, offer a mechanistic hypothesis to the lack of disease comorbidities of this suppressed condition, but the host is still able to produce low-grade inflammation systemically. In future studies using next generation sequencing strategies of highly diverse SLE subjects, it will be possible to map the microbial ecology of multiple sites of the human body, including oral sites. As the science of host-microbes is advancing, highly 22 individualized, rather than simple binary distinctions of healthy and disease or single-microbe etiology across populations, might be found 56 .
The results presented here show that oral-systemic pathogenic burden is evident in lupus subjects.
These data are also consistent with complex ecological interactions among multiple human chronic diseases 57 . Among the relationships found, co-occurrence within the cytokines was evident on SLE-A group (a total number of 77 connecting edges, Figure 4C) when compared to controls and SLE-inactive states ( Figure 4A and Figure Table 2). Indicator species with significant differences in abundances between periodontitis and nonperiodontitis subjects are indicated with asterisks.