Decoding the autophagy-immune axis in periodontitis pathogenesis: an integrated bulk and single-cell transcriptomic approach

Introduction

Periodontitis is an inflammatory and destructive disease that damages the periodontal supporting tissues, including the gingiva, periodontal ligament, cementum, and alveolar bone. The prevalence of periodontitis among adults with teeth is approximately 62%, with severe cases affecting 23.6% of the population between 2011 and 2020 (1). Periodontitis ultimately leads to tooth mobility and eventual tooth loss, thereby compromising chewing function (2). Moreover, severe periodontitis may cause substantial functional impairment and psychological distress (3). Current diagnosis relies on clinical and radiographic assessments, which merely reflect tissue destruction and fail to predict disease progression risk (4). Importantly, even with standardized treatment protocols, periodontitis can persist, leading to tooth loss (5). This dual challenge of diagnosis and treatment highlights the urgent need for novel biomarkers and therapeutic targets.

Periodontitis is a chronic multifactorial inflammatory disease primarily caused by Porphyromonas gingivalis (P. gingivalis), Treponema denticola, and Tannerella forsythia (6). A dynamic imbalance between local immunity and the microbial community is a key driver of disease (7,8). The periodontal immune microenvironment is composed of diverse cells, extracellular matrix components, and cytokines that establish complex interactions and signaling networks (9). In recent years, uncontrolled immune responses and oxidative stress mediated by polymorphonuclear neutrophils have been recognized as central contributors to periodontitis (6). One study identified a novel activated-to-guide (AG) fibroblast subpopulation that modulates oral inflammation, and proposed that the AG fibroblast-neutrophil-type 3 innate lymphoid cell axis is involved in the pathological inflammatory process of periodontitis (10). Furthermore, the programmed death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1) pathway has been identified as a double-edged sword in disease pathogenesis (11-14). High PD-1/PD-L1 expression can suppress inflammation and reduce periodontal tissue destruction; however, it may also enable P. gingivalis to evade immune surveillance, facilitating persistent infection (15). The precise cellular and molecular mechanisms underlying these effects remain unclear.

Autophagy is an evolutionarily conserved intracellular degradation pathway through which misfolded proteins and damaged organelles are delivered to lysosomes for degradation (16). It has been shown to play an important role in various chronic oral inflammatory diseases (17), including apical periodontitis (AP) (18,19) and oral lichen planus (OLP) (20). Recent studies indicate that autophagy is critically involved in the onset, progression, and resolution of periodontitis (21). Autophagy is typically upregulated in infected cells to eliminate intracellular pathogens, and it can promote cell death in response to bacterial invasion (22). However, some periodontal pathogens have developed mechanisms to exploit autophagy, enabling them to evade host immune defenses (16). In addition, autophagy regulates classical pattern recognition receptors, modulates innate immune responses, and inhibits cytokine secretion, thereby dampening the periodontal immune response (23). Autophagy may also protect against periodontal cell apoptosis (23-25). Despite these insights, the specific signaling molecules and cellular targets of autophagy in periodontitis remain poorly characterized. To address this gap, our study employed a combined bulk and single-cell transcriptomic approach to investigate the functional involvement of autophagy-related genes in periodontitis pathogenesis.

Through bioinformatic analysis of transcriptomic data, we identified three autophagy-related hub genes with diagnostic potential, validated by receiver operating characteristic (ROC) curve analysis. We hypothesized that these genes regulate the progression of periodontitis, which we further investigated using single-cell RNA sequencing (scRNA-seq) to profile expression patterns. Their expression was validated in clinical samples using quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR), immunohistochemistry (IHC), and multiplex IHC (mIHC) assays (Figure 1). We present this article in accordance with the STREGA reporting checklist (available at https://fomm.amegroups.com/article/view/10.21037/fomm-2025-1-36/rc).

Figure 1 The study design. DEG, differentially expressed gene; GEO, Gene Expression Omnibus; GO, Gene Ontology; GSEA, gene set enrichment analysis; IHC, immunohistochemistry; KEGG, Kyoto Encyclopedia of Genes and Genomes; mIHC, multiplex IHC; qRT-PCR, quantitative real-time reverse transcription polymerase chain reaction.

Methods

Data acquisition

The GSE10334 and GSE171213 datasets were downloaded from the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo). GSE10334 includes 247 samples, comprising 64 healthy and 183 diseased gingival tissues from 90 patients with chronic or aggressive periodontitis. A mixed-effects linear model was applied to address the non-independent sample issue of GSE10334. Gene expression data were obtained from the local file “GSE10334.csv”. Preprocessing included automatic log₂ transformation, mean collapse of duplicate genes, and integration of sample metadata (group and patient_id) aligned by GSM identifiers.

GSE171213 contains single-cell transcriptomic data from 51,248 cells, including cells from the periodontal tissues of healthy controls, patients with severe chronic periodontitis, and patients within 1 month of initial periodontal treatment. To further investigate the cell-specific expression of hub genes under periodontitis conditions, this study only included scRNA-seq data from periodontal tissues of patients with severe periodontitis in the dataset. Raw sequence reads in FASTQ format were processed and aligned to the GRCh38 human reference transcriptome using the CellRanger pipeline (version 7.1.0) with default parameters.

Variance analysis

Differentially expressed genes (DEGs) were identified using the Limma package with a design matrix constructed as ~0 + group (healthy/disease). To account for within-patient correlations (due to non-independent data from multiple samples per patient), the duplicateCorrelation function was applied with patient_id as the blocking variable, and the estimated correlation was incorporated into linear model fitting via lmFit (using block and correlation arguments). The contrast of interest was defined as disease vs. healthy. Multiple testing correction was performed using the Benjamini-Hochberg (BH) procedure, with significance thresholds set at false discovery rate (FDR) <0.05 and |log₂fold change (log₂FC)| ≥1. DEGs were visualized as volcano plots (ggplot2) and heatmaps (pheatmap).

Functional enrichment analysis

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed on significant DEGs (FDR <0.05 and |log₂FC| ≥1) using the hypergeometric test via the “clusterProfiler” R package. These DEGs were first mapped to ENTREZ IDs. GO enrichment (biological process, cellular component, molecular function) was conducted with clusterProfiler::enrichGO, and KEGG enrichment (species = hsa) with clusterProfiler::enrichKEGG. Terms with FDR <0.05 were considered significant. Results were visualized as dotplots (clusterProfiler::dotplot), with the X-axis showing the rich factor and FDR values included.

Gene set enrichment analysis (GSEA) was performed using the “clusterProfiler” R package. A pre-ranked gene list was generated by sorting all genes based on their t-statistics from the limma differential expression analysis. The analysis prioritized MSigDB-KEGG gene sets from the “msigdbr” package for implementation with GSEA(); if unavailable, it fell back to gseKEGG(). Normalized enrichment score (NES) >0 indicated enrichment in the Disease group relative to the Healthy group. The top 5 upregulated and downregulated pathways in the Disease group were visualized as overlapping enrichment plots, respectively, and a pathway summary table (including NES and FDR) was exported.

Identification of hub genes and transcription factor enrichment analysis

Hub genes were identified by intersecting DEGs with autophagy-related genes obtained from the Human Autophagy Database (HAD). The intersection was visualized with a Venn diagram.

Transcription factor enrichment analysis was performed using the iRegulon plugin to identify key transcriptional regulators. iRegulon predicted enriched transcription factor motifs and tracks of potential hub genes, and the results were visualized as transcription factor-hub gene regulatory networks.

Single-gene analysis and GSEA

Single-gene analyses were performed for each identified hub gene. Using the GSE10334 dataset, samples were stratified into high- and low-expression groups based on the median expression level of each hub gene. The Limma package was employed to identify DEGs, with statistical significance determined by BH FDR <0.05 and |log₂FC| ≥1. DEGs were visualized using volcano plots.

The t-statistics from the Limma analysis were used as pre-ranked scores for GSEA. Gene symbols were converted to ENTREZ IDs using org.Hs.eg.db, with the maximum absolute t-value selected for multiple mappings. GSEA was performed using clusterProfiler with MSigDB C2:CP:KEGG gene sets (obtained via msigdbr), with parameters set as nPerm =1,000, minGSSize =5, and maxGSSize =1,000. For each hub gene, the top 5 enriched pathways (NES >0, indicating high-expression group enrichment) were visualized using enrichplot::gseaplot2. Statistical significance was defined as FDR <0.05.

Immune cell infiltration analysis

Immune cell composition was determined by deconvoluting the bulk expression matrix using the CIBERSORT algorithm with the LM22 signature matrix (representing 22 immune cell types). The analysis was performed with 1,000 permutations (with automatic fallback if unavailable), and the output values were normalized to a sum of 1 per sample, generating a sample × cell proportion matrix. For hub genes, samples were divided into high- and low-expression groups based on the median expression. Immune cell type scores between the two groups were compared using the Wilcoxon rank-sum test, with significance denoted by asterisks, and statistical tables were exported.

Hub gene expression and diagnostic performance analysis

Data from the GSE10334 dataset were used to assess hub gene expression. For each hub gene, box plots were generated using “ggplot2” to display expression differences between Healthy and Disease samples, with significance determined by the Wilcoxon test and denoted by asterisks. ROC curves were constructed using the “pROC” package, with the area under the curve (AUC) calculated and plotted. The model was built with all samples from the GSE10334 dataset.

Single-cell data analysis

The Seurat R package was used to import and process original gene expression matrices. Three quality control metrics were calculated: total number of genes, total number of transcript molecules (UMI counts), and percentage of mitochondrial gene transcripts. Cells were excluded if they contained fewer than 200 or more than 6,000 genes, fewer than 200 or more than 30,000 UMIs, or >30% mitochondrial gene expression.

Data were normalized and scaled to a mean of zero and a standard deviation (SD) of one. Principal component analysis (PCA) was then applied for dimensionality reduction based on the 3,000 most variable genes. The Harmony package (with default parameters) was subsequently used on the top principal components to correct for batch effects. Subsequently, nearest neighbors were identified from these corrected components, and cell clusters were identified at a resolution of 0.8.

Differential expression analysis was performed between clusters to identify marker genes for each cluster. Cell types were then annotated by comparing these marker genes with established cell-type-specific signatures reported in the literature. To investigate the activity of specific gene programs, signature scores for individual cells based on the expression of three hub genes were computed using the AddModuleScore function in Seurat and visualized using a DotPlot.

Autophagy score calculation and cell communication analysis

Autophagy scores for individual cells were computed using the PercentageFeatureSet function in Seurat and visualized with DotPlot. Cells were stratified into high- and low-autophagy groups according to the median autophagy score to enable subsequent intercellular communication analysis.

Cell-cell communication was analyzed using the CellChat package with the human-specific interaction database “CellChatDB.human”. The filterCommunication function was applied with min.cells =10, and all other parameters were left at default settings. Communication networks were visualized using the netVisual_circle and netVisual_bubble functions.

Patient samples collection

Inflamed gingival tissues were collected from ten patients with stages III–IV who underwent modified Widman flap surgery in the Department of Periodontology between February and June 2024. Control samples consisted of healthy pericoronal gingival tissues collected from periodontally healthy subjects undergoing mandibular third molar extraction during the same period. Inclusion criteria for these controls included: probing depth (PD) not exceeding 3 mm, bleeding on probing (BOP) rate below 10%, absence of clinical attachment loss (CAL), no history of periodontitis, and no history of pericoronitis. Exclusion criteria were as follows: (I) age <18 or >60 years; (II) specific physiological states (e.g., pregnancy, lactation); (III) smoking history; (IV) diagnosis of diabetes, autoimmune diseases, infectious diseases, or malignant tumors; and (V) use of antibiotics within 3 months before enrollment. All subjects were consecutively enrolled, and the two groups were matched in terms of key demographic characteristics (Figure S1). Detailed baseline data are presented in Table S1. This study did not interfere with patient treatment plans and adhered to all relevant ethical standards. Four tissues from both the experimental and control groups were immediately immersed in ice-cold 0.9% saline (4 ℃), transported to the laboratory, and processed for homogenization and RNA extraction. Six tissues from each group were placed in centrifuge tubes containing 4% paraformaldehyde and fixed for 24 h. Samples were then dehydrated using an automated dehydrator, embedded in paraffin, and sectioned for subsequent experiments.

qRT-PCR

Gingival tissues were homogenized using a cryogenic tissue homogenizer (Servicebio, SWE-FP). Total RNA was extracted with an Animal RNA Extraction Kit (Beyotime, R0026) and reverse-transcribed into complementary DNA (cDNA) using a cDNA Synthesis Kit (Yeasen, 11141ES10). qRT-PCR was performed using PCR-grade reagents (Takara, RR420A) on a LightCycler 480 System (Roche Diagnostics). Cycling parameters were as follows: initial denaturation at 95 ℃ for 30 s (1 cycle), followed by 40 cycles of 95 ℃ for 5 s and 60 ℃ for 30 s. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as the reference gene for normalization after CT values were obtained. Primer sequences are listed in Table S2.

Hematoxylin and eosin (HE) staining

Sections were first heated at 65 ℃ for 1 h, dewaxed, and rehydrated by soaking in xylene three times for 10 min each, followed by graded ethanol solutions (100%, 90%, 80%, and 70%) for 5 min each. Samples were stained with hematoxylin (Servicebio; G1005-1) for 6 min and rinsed with tap water. They were then treated with hematoxylin differentiation solution (Servicebio; G1039) for 3 s, rinsed, and restored to blue coloration with tap water. Eosin solution (Servicebio; G1005-2) was applied for 1.5 min. Sections were dehydrated with anhydrous ethanol twice for 5 min each, further dehydrated with fresh anhydrous ethanol for 5 min, and cleared twice with xylene for 5 min each. Finally, the sections were mounted with neutral gum.

IHC staining

After heating, dewaxing, and rehydration as described above, sections were boiled in hot repair solution (Servicebio; G1202) for 30 min and cooled to room temperature in the same solution. A 3% hydrogen peroxide solution was applied for 20 min, followed by three phosphate-buffered saline (PBS) washes (5 min each). Sections were blocked with 10% goat serum (Beyotime; C0265) for 1 h and incubated with primary antibody (FOS; 1:150; Absin, abs115049) at 4 ℃ overnight. The following day, sections were incubated at room temperature for 30 min and washed with PBS three times. They were then incubated with a poly horseradish peroxidase (HRP)-conjugated secondary antibody (Servicebio; G1302) at room temperature for 20 min and washed three times with PBS. Target protein staining was performed using the DAB HRP Color Development Kit (Beyotime; P0202). The reaction was stopped with tap water. Sections were counterstained with Mayer’s hematoxylin solution (Solarbio; G1080) for 1 min 20 s, rinsed, differentiated, and counterstained. Samples were dehydrated with graded ethanol concentrations (75%, 85%, 95%, and 100%) for 5 min each, cleared twice with xylene for 10 min each, and sealed with neutral gum.

Immunofluorescence

Following deparaffinization, rehydration, and antigen retrieval, sections were blocked for 1 h with QuickBlock™ Blocking Buffer for Immunol Staining (Beyotime; P0260). They were then incubated overnight at 4 ℃ with primary antibodies, including DNAJB9 (SAB; #49066; 1:300) and CD138 (Proteintech, 67155-1-lg; 1:300). Secondary antibodies, YSFluor™ 488 Donkey Anti-Rabbit IgG (H+L) (Yeasen; 34206ES60; 1:200), and YSFluor™ 594 Donkey Anti-Mouse IgG (H+L) (Yeasen; 34112ES60; 1:200), were applied at 37 ℃ for 1 h. Nuclei were stained with 4',6-diamidino-2-phenylindole (DAPI) (Beyotime; C1005) for 15 min, and sections were mounted with an anti-fade fluorescence medium (Biosharp; BL739B). Images were acquired using a ZEISS fluorescence microscope.

mIHC staining

Multiplex immunofluorescence staining was performed using a commercial kit (WASci; WAS1504). After dewaxing and rehydration, sections underwent thermal repair and blocking, followed by overnight incubation at 4 ℃ with primary antibodies. HRP-conjugated secondary antibodies (WASci; WAS1201) were then applied at room temperature for 30 min and labeled with fluorophore-conjugated tyramide signal amplification (TSA) reagents (TSA520, WAS1002; TSA570, WAS1003; TSA690, WAS1006). These steps were repeated for sequential incubation with CXCR4 (Proteintech; 84904-1-RR), CD3 (Proteintech; 17617-1-AP), and CD20 (CST; 48750T) antibodies. Nuclei were stained with DAPI (WASci; WAS1301). Results were visualized using a confocal microscope.

Statistical analysis

Images were processed using ImageJ software, and statistical analyses were performed with GraphPad Prism v9.5.1. Differences between groups were assessed using unpaired t-tests. A P value <0.05 was considered statistically significant. Data are expressed as mean ± SD.

Ethical considerations

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the institutional ethics committee of Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine (No. SH9H-2021-T112-1). Written informed consent was obtained from all participants before enrollment.

Results

DEGs in periodontitis and functional enrichment analysis

Differential expression analysis of the GSE10334 dataset identified 133 DEGs, including 102 upregulated and 31 downregulated genes. These DEGs were visualized using volcano plots and heat maps (Figure 2A,2B).

Figure 2 DEGs in periodontitis and functional enrichment analysis. (A) Heatmap of DEGs in GSE10334. (B) Volcano plot of DEGs in GSE10334. (C) KEGG pathway enrichment results in GSE10334. (D) The enriched GO terms of DEGs in GSE10334; (E) DEG-related upregulated pathways in GSEA. (F) DEG-related downregulated pathways in GSEA. BP, biological process; CC, cellular component; DEG, differentially expressed gene; FC, fold change; FDR, false discovery rate; GO, Gene Ontology; GSEA, gene set enrichment analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; MF, molecular function; NS, not significant.

To investigate their biological significance, KEGG pathway and GO term enrichment analyses were performed. The most significantly enriched KEGG pathways included viral protein interaction with cytokine and cytokine receptor, interleukin-17 (IL-17) signaling pathway, cornified envelope formation, and cytokine-cytokine receptor interaction, etc. (Figure 2C). GO analysis showed that enriched biological processes were mainly related to the production of molecular mediators of immune response, neutrophil chemotaxis, immunoglobulin production, and the B cell receptor signaling pathway, etc. Enriched cellular components were primarily located on the immunoglobulin complex, within the blood microparticle, and in the cornified envelope. The major molecular functions included CXCR chemokine receptor binding, structural constituent of skin epidermis, chemokine activity, and antigen binding (Figure 2D).

GSEA identified distinct pathway activation patterns associated with DEGs. Upregulated pathways were enriched in immune-inflammatory signaling cascades, such as cell adhesion molecules, cytokine-cytokine receptor interaction, hematopoietic cell lineage, and natural killer cell-mediated cytotoxicity, collectively forming a core regulatory module for inflammatory responses (Figure 2E). In contrast, downregulated pathways showed suppression of cell cycle, oxidative phosphorylation, proteasome, ribosome, and spliceosome, suggesting reduced proliferative capacity and impaired bioenergetic homeostasis in the disease state (Figure 2F).

Transcription factor enrichment analysis and single-gene analyses of the three hub genes

Cross-referencing the 133 DEGs with the HAD identified three hub genes: CXCR4, FOS, and DNAJB9 (Figure 3A). Transcription factor enrichment analysis revealed five potential upstream regulators: MEF2D, IKZF2, MEF2A, SRF, and MEF2C (Figure 3B).

Figure 3 Screening of hub genes and single-gene analysis. (A) Venn diagram showing the intersection of DEGs and the HAD. (B) Enrichment analysis of upstream transcription factors of hub genes. (C) Gene volcano plot of differential expression between CXCR4 high- and low-expression groups in GSE10334. (D) GSEA-enriched functional pathways from CXCR4 high- and low-expression groups. (E) Gene volcano plot of differential expression between DNAJB9 high- and low-expression groups in GSE10334. (F) GSEA-enriched functional pathways from DNAJB9 high- and low-expression groups. (G) gene volcano plot of differential expression between FOS high- and low-expression groups in GSE10334. (H) GSEA-enriched functional pathways from FOS high- and low-expression groups. DEG, differentially expressed gene; FC, fold change; FDR, false discovery rate; GSEA, gene set enrichment analysis; HAD, Human Autophagy Database; KEGG, Kyoto Encyclopedia of Genes and Genomes; NES, normalized enrichment score; NS, not significant; TNF, tumor necrosis factor.

Single-gene analysis was then performed for each hub gene. Volcano plots illustrated DEGs between high- and low-expression groups, and GSEA was performed using a pre-ranked gene list based on the t-statistics derived from this comparison.

In the CXCR4 high-expression group, RAC2, THEMIS2, and IRAG2 were highly expressed, whereas ELOVL4, BPIFC, and WAKMAR2 were expressed at significantly lower levels in the low-expression group (Figure 3C). GSEA indicated enrichment in the B cell receptor signaling pathway (Figure 3D).

For DNAJB9, the high-expression group showed elevated ANKRD36BP2, SEL1L, and PLPP5 expression, while CAPNS2, FAM83C, and SLC27A6 were significantly reduced in the low-expression group (Figure 3E). GSEA results demonstrated enrichment in protein processing in endoplasmic reticulum (ER) (Figure 3F).

For FOS, FOSB, CCN1, and ATF3, they were upregulated in the high-expression group, while no significantly downregulated genes were observed in the low-expression group (Figure 3G). GSEA revealed enrichment in tumor necrosis factor (TNF) signaling pathways (Figure 3H).

Immune cell infiltration of hub genes

Immune cell infiltration analysis was performed by stratifying patients according to the median expression levels of CXCR4, DNAJB9, and FOS, followed by CIBERSORT deconvolution. Patients with high CXCR4 expression demonstrated significantly increased infiltration of neutrophils, plasma cells, CD4 memory activated T cells, and gamma delta T cells compared with the low-expression group (Figure 4A). DNAJB9-high patients exhibited similar enrichment of plasma cells, neutrophils and gamma delta T cells (Figure 4B). By contrast, high FOS expression was associated with increased infiltration of M0 macrophages and CD4 memory resting T cells, accompanied by reduced infiltration of CD4 naïve T cells (Figure 4C).

Figure 4 CIBERSORT-based deconvolution analysis revealed distinct immune microenvironment profiles associated with differential expression patterns of hub genes. (A) The distinct immune cell proportions between CXCR4-low and CXCR4-high expression groups. (B) The distinct immune cell proportions between DNAJB9-low and DNAJB9-high expression groups. (C) The distinct immune cell proportions between FOS-low and FOS-high expression groups. *, P<0.05; **, P<0.01; ***, P<0.001; n.s., not significant. NK, natural killer.

Expression and diagnostic significance of hub genes in periodontitis

Expression profiles of the three hub genes were analyzed in the GSE10334 dataset. CXCR4 (Figure 5A), DNAJB9 (Figure 5B), and FOS (Figure 5C) were significantly upregulated in periodontitis tissues compared with healthy tissues. ROC curve analysis showed that CXCR4 (AUC =0.891) (Figure 5D), DNAJB9 (AUC =0.822) (Figure 5E), and FOS (AUC =0.711) (Figure 5F) exhibited strong diagnostic potential for periodontitis.

Figure 5 Diagnostic potential of hub genes in periodontitis (GSE10334 dataset). (A-C) Box plots showing differential expression of (A) CXCR4, (B) DNAJB9, and (C) FOS in healthy vs. periodontitis-affected gingival tissues. Center lines represent medians, boxes indicate IQRs, and whiskers extend to 1.5× IQR. (D-F) ROC curves evaluating the diagnostic performance of (D) CXCR4, (E) DNAJB9, and (F) FOS as biomarkers for periodontitis. AUC values denote predictive accuracy. ****, P<0.0001. AUC, area under the curve; IQR, interquartile range; ROC, receiver operating characteristic.

Single-cell RNA-seq analysis and qRT-PCR results of hub genes in periodontitis

Quality control of the GSE171213 scRNA-seq dataset was performed, with key metrics including total number of genes, transcript counts, and percentage of mitochondrial gene transcripts assessed before and after filtering (Figure 6A). UMAP analysis identified 14 initial clusters, which were consolidated into 10 distinct cell types: TRAC T cells, MS4A1 B cells, IGHG1 plasma cells, PECAM1 endothelial cells, FCGR3B neutrophils, MS4A6A monocytic cells, COL1A1 fibroblasts, TPSAB1 mast cells, KRT6A epithelial cells, and LTF MDSCs (Figure 6B-6D).

Figure 6 Single-cell transcriptomic landscape characterization and qRT-PCR results. (A) Quality control metrics showing gene/cell distributions pre- and post-filtering. (B) UMAP visualization of 14 initial clusters. (C) Consensus cell type annotation using lineage-specific markers. (D) UMAP showing 10 major cell populations. (E) Dot plot displaying the expression of three hub genes in each cell cluster. (F) qRT-PCR results of three hub genes in gingival tissues from periodontitis and healthy groups. MDSC, myeloid-derived suppressor cell; qRT-PCR, quantitative real-time reverse transcription polymerase chain reaction; UMAP, Uniform Manifold Approximation and Projection.

Differential expression dot plots revealed distinct cell type-specific expression patterns. CXCR4 was expressed predominantly in lymphoid cell clusters (T/B cells), DNAJB9 was enriched in plasma cells, and FOS showed broad expression across all annotated populations (Figure 6E). qRT-PCR confirmed that CXCR4, DNAJB9, and FOS were upregulated in gingival tissues from periodontitis patients compared with healthy controls (Figure 6F).

Association between autophagy activity and immune communication networks in periodontitis

Comprehensive autophagy scoring across all cell populations (Figure 7A) allowed categorization of cells into high- and low-autophagy groups based on median scores for subsequent communication analysis. Our results revealed significantly enhanced multidirectional communication between T cells and various stromal and immune cell subsets in the high-autophagy group. Specifically, T cells exhibited stronger interactions with stromal cells, including endothelial cells, epithelial cells, and fibroblasts, via multiple ligand-receptor pairs such as TIGIT-PVR/NECTIN2, CD96-PVR/NECTIN1, and SELL-PODXL/CD34 (Figure 7B,7C). These differential communication patterns suggest that autophagy may modulate immune responses within the microenvironment by altering both the diversity and intensity of T cell-mediated signaling. Additional intercellular communication networks are presented in Figure S2.

Figure 7 Autophagy scoring and intercellular communication analysis in scRNA-seq data. (A) Dot plot displaying differential autophagy scores in different cell clusters. (B) Activated ligand-receptor pairs mediating T-cell communication with other cell types in the high-autophagy-score group. (C) Activated ligand-receptor pairs mediating T-cell communication with other cell types in the low-autophagy-score group. MDSC, myeloid-derived suppressor cell; scRNA-seq, single-cell RNA sequencing.

Validation of FOS expression level in periodontitis

Gingival samples from six patients with periodontitis and six healthy individuals were analyzed. HE staining revealed extensive infiltration of inflammatory cells in the connective tissue of the gingiva in patients with periodontitis (Figure 8A). The expression level of FOS was significantly lower in healthy gingival tissues than in periodontitis-affected gingival tissues (11.26%±1.571%, P<0.001) (Figure 8B). In the periodontitis group, FOS expression was elevated in inflammatory cells within the connective tissue, vascular endothelial cells, and certain epithelial cells (Figure 8A).

Figure 8 Validation of three hub genes in clinical samples. (A) Representative images of HE and FOS IHC staining in gingival tissues from periodontitis and healthy groups (arrows indicate inflammatory cell infiltration). (B) Quantification of FOS expression in periodontitis compared with healthy groups (n=6). ImageJ was used to set thresholds and calculate the proportion of FOS brown-stained areas in the field of view. All data are presented as points in the graph. P<0.001. (C) Representative images of immunofluorescence staining of CD138 and DNAJB9. Blue (DAPI), green (DNAJB9), red (CD138). (D) Quantification of CD138+DNAJB9+ cell density in periodontitis and healthy groups (n=6). ImageJ was used to count the total number of cells in the field of view and the number of CD138⁺DNAJB9⁺ double-positive cells. The proportion of double-positive cells was calculated. All data are presented as points in the graph. P<0.001. (E) Multiplex immunofluorescence staining of CXCR4, CD3, and CD20. Blue (DAPI), green (CXCR4), yellow (CD3), red (CD20). (F) Quantification of CD3⁺ CXCR4⁺ cell density in periodontitis and healthy groups (n=6). (G) Quantification of CD20⁺CXCR4⁺ cell density in periodontitis and healthy groups (n=6). ImageJ was used to count the total number of cells in the field of view, as well as CD3⁺CXCR4⁺ cells and CD20⁺CXCR4⁺ cells. The proportions of the two types of double-positive cells were calculated separately. All data are presented as points in the graph. P<0.001. DAPI, 4',6-diamidino-2-phenylindole; HE, hematoxylin and eosin; IHC, immunohistochemistry.

Experimental verification of DNAJB9 and CXCR4 in periodontitis

DNAJB9 has been reported to be elevated in the plasma cells of immunoglobulin G4 (IgG4)-related disease, a multisystem fibroinflammatory disease (26,27). To further investigate the relationship between DNAJB9 and plasma cells in periodontitis, CD138 was used as a plasma cell marker, and immunofluorescence staining was performed to assess the expression patterns of CD138 and DNAJB9 in both periodontitis and healthy gingival tissues. Our results revealed that the expression levels of both CD138 and DNAJB9 were significantly elevated in periodontitis tissues compared with those in healthy controls (Figure 8C). Moreover, notable co-expression of these markers was observed, with the number of co-expressing cells being significantly higher in the periodontitis group than in the control group (13.49%±1.594%, P<0.001) (Figure 8D). These findings suggest a potential synergistic role of CD138 and DNAJB9 in the pathogenesis of periodontitis, warranting further exploration of their mechanistic interplay.

CD3, a T-cell marker, and CD20, a B-cell marker, are associated with CXCR4, a chemokine receptor critical for immune cell migration and activation (28,29). The co-expression of CXCR4 with CD3, as well as CXCR4 with CD20, suggests that CXCR4 may be involved in the inflammatory proteolytic process (30). Consequently, mIHC was used to assess the expression of CD3, CD20, and CXCR4 in periodontitis and healthy gingival tissues. The results revealed significantly elevated expression of CD3, CD20, and CXCR4 in periodontitis tissues compared with controls (Figure 8E). CXCR4 was co-expressed with CD3 and CD20, and the densities of CXCR4⁺CD3⁺ and CXCR4⁺CD20⁺ cells were significantly higher in periodontitis samples (increased by 18.95%±3.155% and 12.02%±2.452%, respectively; P<0.001) (Figure 8F,8G). These findings indicate that CXCR4-mediated interactions with T and B cells contribute to immune dysregulation in periodontitis.

Discussion

Periodontitis is characterized by inflammation involving both innate and adaptive immune responses. Resident cells, including epithelial cells and fibroblasts, together with innate immune cells, play pivotal roles in mediating inflammatory responses to bacterial invasion. Bulk and single-cell analyses have provided novel insights into the mechanisms of immune cell infiltration in periodontal disease (31). Liu et al. demonstrated that B/plasma cell infiltration was significantly elevated in periodontitis patients, driven by the activation of the macrophage migration inhibitory factor pathway (32). Qian et al. revealed that endothelial cells expressing HLA-DR and CXCL13⁺ fibroblasts were highly correlated with immune regulation in periodontitis (33). Caetano et al. reported that memory B and T cells are abundant in mild periodontitis but decline significantly in severe cases (34). Agrafioti et al. found that both pro-inflammatory and anti-inflammatory markers are expressed in macrophages in periodontitis, with their polarization being non-mutually exclusive (35).

Previous studies on the relationship between autophagy and the immune microenvironment in periodontitis have predominantly focused on individual molecules or pathways. Bioinformatics has introduced novel approaches to uncover this relationship. Bian et al. (36) identified differentially expressed autophagy-related genes and constructed a competing endogenous RNA network. By integrating qRT-PCR results, they found that the epidermal growth factor receptor may play a crucial role in the development of periodontitis. Zhang et al. (37) systematically evaluated the impact of autophagy on periodontal immune characteristics and reported that the CXCR4-BCR signaling pathway was the most positively correlated immune pathway, whereas the PEX3-TNF family member receptors exhibited the strongest negative correlation.

Our study conducted a differential expression analysis of transcriptomic data (GSE10334) and identified 133 DEGs. To further investigate the relationship between autophagy and periodontitis, we cross-referenced the DEGs with HAD and identified three autophagy-related hub genes: CXCR4, FOS, and DNAJB9. Further single-gene analysis showed that these hub genes were associated with specific pathways: CXCR4 with the B-cell receptor signaling pathway, DNAJB9 with protein processing pathways, and FOS with TNF signaling pathway. ROC curve analysis demonstrated their diagnostic value for periodontitis, with all AUC values exceeding 0.7, supporting their reliability as potential diagnostic markers. Single-cell transcriptomic analysis revealed distinct cellular distribution patterns: CXCR4 was predominantly localized in T and B cells, DNAJB9 was primarily expressed in plasma cells, and FOS exhibited pan-cellular distribution. These expression patterns were validated at multiple levels by qRT-PCR, IHC, and mIHC. Notably, autophagy scoring analysis and cell communication networks showed that T cells in the high-autophagy group exhibited enhanced multidirectional communication, suggesting that autophagy may regulate immune cell interactions.

Our transcription factor enrichment analysis identified five potential upstream regulators (MEF2D, IKZF2, MEF2A, SRF, MEF2C), among which MEF2D and SRF have established links to autophagy. MEF2D undergoes direct degradation via chaperone-mediated autophagy (38), while SRF is subject to autophagy-dependent destruction (39), suggesting that autophagy may modulate these transcription factors to regulate hub gene expression. This finding connects upstream transcriptional regulation to the autophagy-immune axis in periodontitis pathogenesis, while the specific roles of IKZF2, MEF2A, and MEF2C in this context require further investigation.

FOS, also known as c-FOS, forms the AP-1 complex in conjunction with members of the Jun family (40). It is rapidly activated in response to diverse cellular stimuli (41). Chan et al. (42) demonstrated that FOS expression can be stimulated by protease-activated receptor activation during gingival injury or inflammation. Zhang et al. (43) observed through immunohistochemical staining that FOS expression was significantly upregulated in the gingival tissues of rats with periodontitis compared to controls. This observation aligns with our findings in human gingival tissues. Our study provides the first immunohistochemical evidence of pan-cellular overexpression of FOS in human gingival tissues with periodontitis, suggesting its potential involvement in inflammatory signaling and cellular stress responses during periodontal pathogenesis.

CXCR4, a member of the G protein-coupled receptor superfamily, plays a pivotal role in B-cell development, regulation of bone marrow homeostasis, and leukocyte transport and distribution in peripheral tissues (44,45). Abe et al. (46) observed that CXCR4 is upregulated in gingival tissues affected by periodontitis, whereas DSC-1 is downregulated. In contrast, our study revealed that in the low CXCR4 expression group, DSC-1 cells also exhibited low expression. Because DSC-1 plays a crucial role in intercellular junctions, this finding suggests that CXCR4 influences intercellular communication in periodontitis. Xu et al. (47) identified CXCR4 as a key immune-related pathogenic gene in periodontitis using bioinformatics analysis and confirmed that CXCR4 influences disease progression by mediating neutrophil dynamics. Our findings on CXCR4 coexpression in both B and T cells suggest that CXCR4 may exacerbate periodontal tissue destruction by modulating lymphocyte recruitment and functionality, although further mechanistic validation is warranted.

DNAJB9, also known as MDG1, MDJ7, or ERDJ4, plays a critical role in mediating the ER stress response. Tsaryk et al. (48) reported that DNAJB9 is expressed in the perinuclear region of endothelial cells, in certain lymphocyte nuclei, and in both the cytoplasm and nucleus of granulocytes within the granulation tissue of chronic wounds. The involvement of DNAJB9 in periodontitis has been minimally explored, with only two bioinformatic studies by Zhang et al. (37) and Bian et al. (36) reporting its dysregulation among autophagy-related genes, consistent with our findings. Importantly, our work advances this observation by identifying plasma cells as a specific cellular niche for DNAJB9 expression and by providing experimental validation using mIHC. Mahanonda et al. (49) observed a substantial presence of CD138⁺ plasma cells distributed throughout the gingival connective tissue in periodontitis, particularly at the advancing front of lesion progression. Another study (50) highlighted that an increased number and density of plasma cells are hallmarks of advanced periodontitis. Elevated expression of DNAJB9 in plasma cells therefore suggests its potential as a novel biomarker for the clinical diagnosis of advanced periodontitis.

Our study further revealed that T cells in the high-autophagy group exhibited active multidirectional intercellular communication. Current evidence shows that activated T cells in periodontitis can produce receptor activator of nuclear factor-κB ligand, which induces osteoclast-mediated alveolar bone resorption. Additionally, activated Th1, Th2, and Th17 cells secrete pro-inflammatory cytokines, including IL-1β, IL-17E (IL-25), and IL-17, which subsequently activate neutrophils, dendritic cells, and B cells (51). The precise mechanisms underlying the autophagy-dependent enhancement of T-cell communication remain to be fully elucidated.

Through the integration and validation of multi-dimensional technologies, this study has achieved three core advances: first, the combined analysis of bulk and single-cell transcriptomics has improved the localization resolution of cell-specific immune niches, accurately clarifying the cellular expression characteristics of CXCR4, DNAJB9, and FOS; second, the co-expression validation of hub genes and cell markers via mIHC has provided direct experimental evidence for clinical translation; third, the integration of autophagy scoring and cell-cell communication network analysis has revealed the regulatory role of autophagy in T-cell crosstalk, offering a new perspective for understanding the mechanism of immune dysregulation in periodontitis. We propose that these three hub genes function as core nodes within a periodontitis autophagy–immune axis. This pathogenic axis is defined by a cascade from autophagy activation within the inflammatory periodontal niche, to differential hub gene expression, to the subsequent regulation of immune cell functions, culminating in local immune dysregulation—thereby providing a testable mechanistic framework for periodontitis.

From a translational perspective, these findings hold clear potential diagnostic and therapeutic value. For diagnosis, unlike traditional clinical examinations that only assess periodontal structural damage, this study enables more targeted and reliable evaluation of “severe periodontitis patients undergoing flap surgery” through the combined expression and distinct cell-specific characteristics of three hub genes. Based on the precise correlation of “gene localization-cellular function-pathological features”, CXCR4’s specific expression in T/B cells may reflect inflammatory activity, DNAJB9’s enrichment in plasma cells correlates with core pathological features of advanced periodontitis, and FOS’s pan-cellular distribution indicates inflammatory spread, providing functional references for refined clinical assessment. For therapy, these cell-specific expression patterns offer clear directions for targeted interventions: optimizing local anti-inflammatory strategies by focusing on CXCR4-mediated immune cell recruitment, DNAJB9-related plasma cell functions, or FOS-associated inflammatory pathways, thus overcoming the lack of precise therapeutic direction in traditional treatments.

However, this study has limitations. First, we relied only on a microarray dataset and scRNA-seq data from periodontitis and control groups obtained from different populations in public databases. Second, the sample sizes of periodontitis and normal gingival tissues collected for validation were relatively small. Third, despite strong observed correlations, the direct causal relationships between hub gene expression, autophagic regulation, and immune cell function require further verification in future functional studies. Therefore, we will subsequently modulate these three hub genes and their pathways in periodontitis animal models to explore the regulatory mechanisms of autophagy on their cell-type-specific expression, and expand the clinical sample size to analyze associations between gene expression profiles and clinical parameters, providing more robust mechanistic support for the translational application of these target molecules.

Conclusions

Our study integrated periodontitis pathogenesis with autophagy processes, identified disease-specific autophagy-related genes, and elucidated their potential connections with dysregulation of the immune microenvironment. The cell-type-specific expression patterns of these hub genes provide novel diagnostic biomarkers and therapeutic targets for periodontitis management.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STREGA reporting checklist. Available at https://fomm.amegroups.com/article/view/10.21037/fomm-2025-1-36/rc

Data Sharing Statement: Available at https://fomm.amegroups.com/article/view/10.21037/fomm-2025-1-36/dss

Peer Review File: Available at https://fomm.amegroups.com/article/view/10.21037/fomm-2025-1-36/prf

Funding: This study was supported by the Shanghai Science and Technology Program Project (No. 20ZR1448700) and the Fundamental Research Program Funding of Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine (No. JYZZ245).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://fomm.amegroups.com/article/view/10.21037/fomm-2025-1-36/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the institutional ethics committee of Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine (No. SH9H-2021-T112-1). Written informed consent was obtained from all participants before enrollment.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Trindade D, Carvalho R, Machado V, et al. Prevalence of periodontitis in dentate people between 2011 and 2020: A systematic review and meta-analysis of epidemiological studies. J Clin Periodontol 2023;50:604-26. [Crossref] [PubMed]
2. Kwon T, Lamster IB, Levin L. Current Concepts in the Management of Periodontitis. Int Dent J 2021;71:462-76. [Crossref] [PubMed]
3. Slots J. Periodontitis: facts, fallacies and the future. Periodontol 2000 2017;75:7-23. [Crossref] [PubMed]
4. Heitz-Mayfield LJA. Conventional diagnostic criteria for periodontal diseases (plaque-induced gingivitis and periodontitis). Periodontol 2000 2024;95:10-9. [Crossref] [PubMed]
5. Yang B, Pang X, Li Z, et al. Immunomodulation in the Treatment of Periodontitis: Progress and Perspectives. Front Immunol 2021;12:781378. [Crossref] [PubMed]
6. Wang J, Zhou Y, Ren B, et al. The Role of Neutrophil Extracellular Traps in Periodontitis. Front Cell Infect Microbiol 2021;11:639144. [Crossref] [PubMed]
7. He L, Liu L, Li T, et al. Exploring the Imbalance of Periodontitis Immune System From the Cellular to Molecular Level. Front Genet 2021;12:653209. [Crossref] [PubMed]
8. Loos BG, Van Dyke TE. The role of inflammation and genetics in periodontal disease. Periodontol 2000 2020;83:26-39. [Crossref] [PubMed]
9. Han N, Liu Y, Du J, et al. Regulation of the Host Immune Microenvironment in Periodontitis and Periodontal Bone Remodeling. Int J Mol Sci 2023;24:3158. [Crossref] [PubMed]
10. Kondo T, Gleason A, Okawa H, et al. Mouse gingival single-cell transcriptomic atlas identified a novel fibroblast subpopulation activated to guide oral barrier immunity in periodontitis. Elife 2023;12:RP88183. [Crossref] [PubMed]
11. Zhu D, Liu F, Dai F, et al. Expression of programmed death-1 and programmed death ligand-1 in the peripheral T-lymphocytes from patients with chronic periodontitis. Zhonghua Kou Qiang Yi Xue Za Zhi 2014;49:216-9.
12. Shelby A, Pendleton C, Thayer E, et al. PD-L1 correlates with chemokines and cytokines in gingival crevicular fluid from healthy and diseased sites in subjects with periodontitis. BMC Res Notes 2020;13:532. [Crossref] [PubMed]
13. Figueira EA, de Rezende ML, Torres SA, et al. Inhibitory signals mediated by programmed death-1 are involved with T-cell function in chronic periodontitis. J Periodontol 2009;80:1833-44. [Crossref] [PubMed]
14. Esberg A, Isehed C, Holmlund A, et al. Serum proteins associated with periodontitis relapse post-surgery: A pilot study. J Periodontol 2021;92:1805-14. [Crossref] [PubMed]
15. Liu X, Yang L, Tan X. PD-1/PD-L1 pathway: A double-edged sword in periodontitis. Biomed Pharmacother 2023;159:114215. [Crossref] [PubMed]
16. Yang Y, Huang Y, Li W. Autophagy and its significance in periodontal disease. J Periodontal Res 2021;56:18-26. [Crossref] [PubMed]
17. Tan YQ, Zhang J, Zhou G. Autophagy and its implication in human oral diseases. Autophagy 2017;13:225-36. [Crossref] [PubMed]
18. Wu Z, Duan S, Li M, et al. Autophagy regulates bone loss via the RANKL/RANK/OPG axis in an experimental rat apical periodontitis model. Int Endod J 2024;57:1315-25. [Crossref] [PubMed]
19. Li J, Yue Y, Chan W, et al. RGS10 negatively regulates apical periodontitis via TFEB-mediated autophagy in BABL/c mice model and in vitro. Int Endod J 2023;56:854-68. [Crossref] [PubMed]
20. Zhang N, Zhang J, Tan YQ, et al. Activated Akt/mTOR-autophagy in local T cells of oral lichen planus. Int Immunopharmacol 2017;48:84-90. [Crossref] [PubMed]
21. Kang S, Liu S, Dong X, et al. USP4 depletion-driven RAB7A ubiquitylation impairs autophagosome-lysosome fusion and aggravates periodontitis. Autophagy 2025;21:771-88. [Crossref] [PubMed]
22. Tsuda H, Ochiai K, Suzuki N, et al. Butyrate, a bacterial metabolite, induces apoptosis and autophagic cell death in gingival epithelial cells. J Periodontal Res 2010;45:626-34. [Crossref] [PubMed]
23. Liu C, Mo L, Niu Y, et al. The Role of Reactive Oxygen Species and Autophagy in Periodontitis and Their Potential Linkage. Front Physiol 2017;8:439. [Crossref] [PubMed]
24. Greabu M, Giampieri F, Imre MM, et al. Autophagy, One of the Main Steps in Periodontitis Pathogenesis and Evolution. Molecules 2020;25:4338. [Crossref] [PubMed]
25. İnan S, Barış E. The role of autophagy in odontogenesis, dental implant surgery, periapical and periodontal diseases. J Cell Mol Med 2024;28:e18297. [Crossref] [PubMed]
26. Lin W, Zhang P, Chen H, et al. Circulating plasmablasts/plasma cells: a potential biomarker for IgG4-related disease. Arthritis Res Ther 2017;19:25. [Crossref] [PubMed]
27. Stuchfield-Denby E, Mattutzu V, Grobost V, et al. Immunohistochemical expression of DnaJ homolog subfamily B member 9 in immunoglobulin G4-related disease: a pilot study. Clin Exp Rheumatol 2024;42:718-25. [Crossref] [PubMed]
28. Jung SH, Won KJ, Lee KP, et al. DJ-1 protein regulates CD3+ T cell migration via overexpression of CXCR4 receptor. Atherosclerosis 2014;235:503-9. [Crossref] [PubMed]
29. Pavlasova G, Mraz M. The regulation and function of CD20: an "enigma" of B-cell biology and targeted therapy. Haematologica 2020;105:1494-506. [Crossref] [PubMed]
30. Tanios F, Pelisek J, Lutz B, et al. CXCR4: A Potential Marker for Inflammatory Activity in Abdominal Aortic Aneurysm Wall. Eur J Vasc Endovasc Surg 2015;50:745-53. [Crossref] [PubMed]
31. Razzouk S. Single-cell sequencing, spatial transcriptome ad periodontitis: Rethink pathogenesis and classification. Oral Dis 2024;30:2771-83. [Crossref] [PubMed]
32. Liu L, Chen Y, Wang L, et al. Dissecting B/Plasma Cells in Periodontitis at Single-Cell/Bulk Resolution. J Dent Res 2022;101:1388-97. [Crossref] [PubMed]
33. Qian SJ, Huang QR, Chen RY, et al. Single-Cell RNA Sequencing Identifies New Inflammation-Promoting Cell Subsets in Asian Patients With Chronic Periodontitis. Front Immunol 2021;12:711337. [Crossref] [PubMed]
34. Caetano AJ, Yianni V, Volponi A, et al. Defining human mesenchymal and epithelial heterogeneity in response to oral inflammatory disease. Elife 2021;10:e62810. [Crossref] [PubMed]
35. Agrafioti P, Morin-Baxter J, Tanagala KKK, et al. Decoding the role of macrophages in periodontitis and type 2 diabetes using single-cell RNA-sequencing. FASEB J 2022;36:e22136. [Crossref] [PubMed]
36. Bian M, Wang W, Song C, et al. Autophagy-Related Genes Predict the Progression of Periodontitis Through the ceRNA Network. J Inflamm Res 2022;15:1811-24. [Crossref] [PubMed]
37. Zhang X, Jin Y, Wang Q, et al. Autophagy-mediated regulation patterns contribute to the alterations of the immune microenvironment in periodontitis. Aging (Albany NY) 2020;13:555-77. [Crossref] [PubMed]
38. Yang Q, Mao Z. Dysregulation of autophagy and Parkinson's disease: the MEF2D link. Apoptosis 2010;15:1410-4. [Crossref] [PubMed]
39. Luo J, Jin FQ, Yin M, et al. Regulation of SRF protein stability by an autophagy-dependent pathway. Biochem Biophys Res Commun 2020;521:279-84. [Crossref] [PubMed]
40. Ray N, Kuwahara M, Takada Y, et al. c-Fos suppresses systemic inflammatory response to endotoxin. Int Immunol 2006;18:671-7. [Crossref] [PubMed]
41. Sun Y, Lin Z, Liu CH, et al. Inflammatory signals from photoreceptor modulate pathological retinal angiogenesis via c-Fos. J Exp Med 2017;214:1753-67. [Crossref] [PubMed]
42. Chan CP, Chang MC, Wang YJ, et al. Thrombin activates Ras-CREB/ATF-1 signaling and stimulates c-fos, c-jun, and c-myc expression in human gingival fibroblasts. J Periodontol 2008;79:1248-54. [Crossref] [PubMed]
43. Zhang L, Chen W, Li Y, et al. Effect of 650-nm low-level laser irradiation on c-Jun, c-Fos, ICAM-1, and CCL2 expression in experimental periodontitis. Lasers Med Sci 2020;35:31-40. [Crossref] [PubMed]
44. Giorgiutti S, Rottura J, Korganow AS, et al. CXCR4: from B-cell development to B cell-mediated diseases. Life Sci Alliance 2024;7:e202302465. [Crossref] [PubMed]
45. Pozzobon T, Goldoni G, Viola A, et al. CXCR4 signaling in health and disease. Immunol Lett 2016;177:6-15. [Crossref] [PubMed]
46. Abe D, Kubota T, Morozumi T, et al. Altered gene expression in leukocyte transendothelial migration and cell communication pathways in periodontitis-affected gingival tissues. J Periodontal Res 2011;46:345-53. [Crossref] [PubMed]
47. Xu X, Li T, Tang J, et al. CXCR4-mediated neutrophil dynamics in periodontitis. Cell Signal 2024;120:111212. [Crossref] [PubMed]
48. Tsaryk R, Bartholomä NM, Simiantonaki N, et al. Endoplasmic reticulum-resident chaperones modulate the inflammatory and angiogenic responses of endothelial cells. Br J Dermatol 2015;173:416-27. [Crossref] [PubMed]
49. Mahanonda R, Champaiboon C, Subbalekha K, et al. Human Memory B Cells in Healthy Gingiva, Gingivitis, and Periodontitis. J Immunol 2016;197:715-25. [Crossref] [PubMed]
50. Thorbert-Mros S, Larsson L, Berglundh T. Cellular composition of long-standing gingivitis and periodontitis lesions. J Periodontal Res 2015;50:535-43. [Crossref] [PubMed]
51. Figueredo CM, Lira-Junior R, Love RM T. Int J Mol Sci 2019;20:3949. [Crossref] [PubMed]