Five Candidate Hub Genes Show Elevated Expression in Periodontitis and Lupus

A comprehensive multi-stage transcriptomic analysis paired with clinical quantitative real-time PCR validation has identified five candidate hub genes shared between periodontitis and systemic lupus erythematosus, illuminating the molecular and immunological architecture linking these distinct inflammatory disorders. Periodontitis is a chronic inflammatory disorder triggered by a microbial biofilm that forms around the teeth, leading to destruction of supporting periodontal tissues and a decline in quality of life. Globally, the disease affects an estimated twenty to fifty percent of the adult population and has been linked to various systemic disorders and an elevated risk of mortality. Systemic lupus erythematosus is a severe autoimmune condition characterized by a fluctuating relapsing-remitting trajectory, widespread immune-complex deposition, and multi-organ damage spanning the kidneys, joints, skin, and central nervous system. Epidemiological observations indicate that approximately seventy percent of patients with lupus have periodontitis, with previous studies reporting an association between periodontitis and greater disease severity or systemic inflammation in these patients. Despite documented clinical interactions, the precise overlapping transcriptomic networks connecting localized gingival destruction and systemic autoimmune pathophysiology have remained unresolved across disparate datasets.

To systematically map shared transcriptomic signatures, investigators retrieved four independent human microarray expression datasets from the National Center for Biotechnology Information Gene Expression Omnibus repository. The discovery phase utilized two primary training datasets: GSE16134, comprising gingival tissue specimens from 241 periodontitis patients and 69 healthy controls profiled on the Affymetrix GPL570 platform, and GSE61635, comprising peripheral blood mononuclear cell profiles from 99 systemic lupus erythematosus patients and 30 healthy controls on the same platform. Using the empirical linear modeling framework implemented in the limma software package, differential expression analysis was performed applying strict statistical thresholds of an absolute log2 fold-change exceeding 0.5 alongside a false discovery rate-adjusted p-value below 0.05. In the periodontitis discovery dataset, this screening identified 1,151 differentially expressed genes, encompassing 689 upregulated and 462 downregulated transcripts. Concurrently, the systemic lupus erythematosus training cohort yielded 3,617 differentially expressed genes, comprising 1,999 upregulated and 1,618 downregulated transcripts. Cross-referencing these two disease signatures using Venn diagram overlapping analysis revealed 124 common differentially expressed genes, consisting of 73 co-upregulated and 51 co-downregulated molecular candidates across both disease states.

Functional enrichment profiling was conducted on the 124 shared transcripts using the clusterProfiler package to delineate biological themes common to periodontal tissue damage and systemic lupus pathology. Gene Ontology analysis revealed marked overrepresentation within biological processes governing the inflammatory response, positive regulation of responses to external stimuli, immune response activation, and directional cell motility, including leukocyte chemotaxis, leukocyte migration, and chemokine-mediated signaling cascades. Complementary Kyoto Encyclopedia of Genes and Genomes pathway enrichment analyses revealed significant representation across several defined immunological routes, most notably the Toll-like receptor signaling pathway, the tumor necrosis factor signaling cascade, the nuclear factor-kappa B pathway, and the interleukin-17 signaling pathway. Additional enriched pathways included lipid and atherosclerosis, rheumatoid arthritis, pertussis, Chagas disease, and pantothenate and coenzyme A biosynthesis. Collectively, these enrichment results suggest that the overlapping differentially expressed genes are predominantly associated with inflammatory and immune-related biological processes and signaling pathways.

To pinpoint the central molecular coordinators embedded within the 124 overlapping genes, investigators assembled a functional protein-protein interaction network using the STRING database applying an interaction confidence threshold of 0.400. The mapped interaction network assembled 119 nodes and 135 connecting edges, which were exported into Cytoscape software version 3.9.1 for topological evaluation. Four distinct topological algorithms within the cytoHubba plugin—specifically Degree, Density of Maximum Neighborhood Component, Maximal Clique Centrality, and Maximum Neighborhood Component—were systematically executed to rank top-tier nodes. Intersecting the highest-ranked candidates generated by all four algorithmic scoring metrics revealed five consensus hub genes: C-C motif chemokine ligand 3 (CCL3), lymphocyte antigen 96 (LY96), the proto-oncogene FOS, matrix metalloproteinase 9 (MMP9), and prostaglandin-endoperoxide synthase 2 (PTGS2). Each of these five hub genes exhibited significant differential expression between diseased cohorts and matched healthy controls within both initial training datasets, establishing them as prominent candidate genes bridging chronic oral osteolysis with systemic lupus autoimmunity.

The expression patterns of the five candidate hub genes were subsequently interrogated in two independent validation datasets retrieved from the Gene Expression Omnibus to assess biological reproducibility across independent clinical cohorts. The periodontitis validation cohort GSE10334, containing gingival tissue samples from 183 patients and 64 controls analyzed on the GPL570 platform, successfully verified all five hub genes, demonstrating statistically significant differential expression identical in direction to the discovery cohort. However, evaluation in the independent systemic lupus validation cohort GSE121239, consisting of 292 patients and 20 controls profiled on the GPL13158 platform, revealed notable divergence among candidates. Specifically, FOS, LY96, and PTGS2 retained significant differential expression between lupus patients and controls, whereas CCL3 and MMP9 did not demonstrate statistically significant alterations within this dataset. These findings indicate that three of the five hub genes were consistently validated across the systemic lupus datasets, whereas CCL3 and MMP9 exhibited dataset-dependent variation that may reflect heterogeneity in study populations, disease severity, treatment status, or experimental platforms.

Diagnostic and discriminatory utility was quantitatively evaluated using receiver operating characteristic curve analysis, calculating the area under the curve across both training and independent validation cohorts with the pROC software package. In the periodontitis discovery dataset GSE16134, area under the curve values for CCL3, FOS, LY96, MMP9, and PTGS2 were 0.734, 0.724, 0.844, 0.796, and 0.694, respectively. In the systemic lupus training cohort GSE61635, discriminatory performance was particularly pronounced, with corresponding values reaching 0.696 for CCL3, 0.993 for FOS, 0.969 for LY96, 0.721 for MMP9, and 0.840 for PTGS2. Within the periodontitis validation series GSE10334, discrimination was preserved across the panel, yielding areas of 0.732 for CCL3, 0.711 for FOS, 0.812 for LY96, 0.778 for MMP9, and 0.689 for PTGS2. In the independent lupus validation cohort GSE121239, the area under the curve values were 0.573 for CCL3, 0.753 for FOS, 0.804 for LY96, 0.499 for MMP9, and 0.754 for PTGS2, with FOS, LY96, and PTGS2 showing relatively higher discriminatory performance while CCL3 and MMP9 exhibited limited discriminatory ability.

To provide empirical biological confirmation beyond computational microarray reanalyses, investigators executed quantitative real-time PCR on prospective clinical specimens collected at the First Affiliated Hospital of Nanchang University between November 2024 and August 2026. The clinical cohort comprised thirty systemic lupus erythematosus study participants—fifteen treatment-naive lupus patients fulfilling 1997 American College of Rheumatology diagnostic criteria alongside fifteen matched healthy volunteers—from whom peripheral blood mononuclear cells were isolated. Concurrently, thirty periodontal participants contributed gingival biopsies of approximately two cubic millimeters, comprising fifteen patients presenting with clinical attachment loss of at least 3 millimeters and probing depths of at least 5 millimeters, alongside fifteen control subjects undergoing third molar extractions. Real-time amplification using specific oligonucleotide primers and SYBR Green chemistry normalized against glyceraldehyde-3-phosphate dehydrogenase demonstrated that relative messenger RNA levels of CCL3, LY96, FOS, MMP9, and PTGS2 were all significantly elevated in periodontitis gingival tissues compared to healthy controls, and similarly showed statistically significant upregulation across lupus mononuclear cells.

Characterization of the local and systemic cellular landscapes using the CIBERSORT computational deconvolution algorithm revealed shared immunological perturbations across periodontitis gingival lesions and systemic lupus peripheral blood. Deconvoluting the twenty-two standard leukocyte subsets indicated that both disease groups exhibited statistically significant shifts relative to healthy controls in five common cell compartments: naive B cells, antibody-secreting plasma cells, activated memory CD4-positive T lymphocytes, activated dendritic cells, and mature neutrophils. Pearson correlation analyses between estimated leukocyte abundances and candidate gene levels revealed extensive, heterogeneous association networks. Furthermore, computational modeling of upstream transcriptional and post-transcriptional regulators via the NetworkAnalyst platform identified 90 transcription factors in the TRRUST database and 102 microRNAs within the miRTarBase repository interacting with the five hub genes, forming a complex regulatory network of 197 nodes and 228 edges. In the predicted network, STAT1 exhibited a high degree among transcription factors, while miR-146a showed a high degree among microRNAs, having been linked to aggressive periodontitis in patients and reported to alleviate kidney injury in lupus mice.

Exploration of potential pharmacological associations was performed by querying the Drug Signatures Database with the consensus hub genes, generating ten top-ranked candidate therapeutic compounds based on combined score metrics: dinoprostone, pyrrolidine dithiocarbamate, capsaicin, phencyclidine, flurbiprofen, propranolol, simvastatin, bisindolylmaleimide, DL-mevalonic acid, and beclomethasone. Mechanistically, each of the five hub genes contributes defined pathophysiological functions to chronic inflammation. LY96, encoding MD-2, associates directly with Toll-like receptor 4 to facilitate bacterial lipopolysaccharide recognition and downstream nuclear factor-kappa B signaling in response to periodontal pathogens such as Porphyromonas gingivalis. The protein product of FOS can dimerize with JUN family members to form activator protein-1 complexes, which have been implicated in the dysregulation of interleukin-2 expression in lupus and are markedly upregulated in antibody-secreting cells from patients with the disease. PTGS2 plays an important role in inflammatory responses through the production of pro-inflammatory prostaglandins, while MMP9 participates in degrading collagens in periodontal tissues and is thought to contribute to vascular wall injury in systemic lupus erythematosus.

While these integrative findings offer a coherent molecular framework linking periodontitis with systemic lupus erythematosus, several important methodological constraints warrant careful consideration. Primarily, the underlying transcriptomic datasets were derived from discordant anatomical sources, contrasting resident and infiltrating structural cells within gingival biopsies against circulating mononuclear cells from peripheral venous blood. Additionally, although prospective real-time PCR verified significant upregulation across all five targets, the clinical validation was restricted to single-center cohorts of fifteen individuals per arm. The researchers emphasize that the identified hub genes, regulatory transcription factor circuits, and predicted therapeutic compounds represent exploratory targets rather than established clinical biomarkers. Definite clinical translation, risk stratification, and therapeutic validation will necessitate larger multi-center longitudinal investigations and targeted functional animal models to confirm causal mechanisms.

Subject of Research: Identification and clinical validation of shared hub genes connecting periodontitis and systemic lupus erythematosus

Article Title: Identification and Validation of Hub Genes Associated With Periodontitis and Systemic Lupus Erythematosus

Article References: Li, Y., Xu, X., Wu, Z., Li, F., & Zhang, J. (2026). Identification and Validation of Hub Genes Associated With Periodontitis and Systemic Lupus Erythematosus. Immunity, Inflammation and Disease, 14(11), Article e70588. https://doi.org/10.1002/iid3.70588

Image Credits: AI Generated

DOI: 10.1002/iid3.70588

Keywords: periodontitis, systemic lupus erythematosus, differentially expressed genes, hub genes, transcriptomics, qPCR validation, CCL3, LY96, FOS, MMP9, PTGS2, CIBERSORT, immune infiltration, receiver operating characteristic, Gene Expression Omnibus, protein-protein interaction, STRING database, cytoHubba, Toll-like receptor signaling, microRNA regulation