Pinocembrin Alleviates Gingival Fibroblast Senescence in a Mouse Model of Periodontitis Via CYP1B1 Downregulation

Abstract

Introduction and aims

The severity of periodontitis, a chronic inflammatory disease, correlates with gingival tissue senescence. This study aimed to characterise the heterogeneous functions of these senescent cells and to define the mechanisms driving gingival fibroblast (GF) senescence, thereby assessing cellular senescence as a potential therapeutic target.

Methods

We analysed cellular senescence changes in gingival tissues during periodontitis by integrating human gingival single-cell RNA sequencing (scRNA-seq) datasets. Biomarkers for a senescence-based diagnostic model for periodontitis were identified, and a potential drug targeting a key biomarker was screened. The effects of this drug on GF senescence were validated using an in vitro model induced by lipopolysaccharide (LPS) in primary human GFs (hGFs), alongside an in vivo experimental periodontitis mouse model induced by ligature.

Results

ScRNA-seq revealed an increase in senescence across various cell types in periodontitis-affected gingival tissues, with GFs being the predominant senescent cell population. Senescence levels in GFs were elevated in both human and mouse periodontitis tissues. A diagnostic model for periodontitis was developed based on a cellular senescence-associated 3-gene signature. Further mechanistic investigation showed that CYP1B1 drives LPS-induced hGF senescence by suppressing fatty acid metabolism. Ultimately, pinocembrin was found to attenuate senescence of GFs and ameliorate experimental periodontitis in mice, an effect mediated primarily through the downregulation of CYP1B1.

Conclusion

Cellular senescence is increased in gingival tissues during periodontitis. Targeting senescent GFs presents a promising therapeutic strategy for the treatment of periodontitis.

Clinical Relevance

Periodontitis-related tissue destruction involves cellular senescence, yet the responsible gingival cell populations and molecular mechanisms remain undefined. CYP1B1-driven suppression of fatty acid metabolism underlies gingival fibroblast senescence, the predominant senescent process in periodontitis, and is attenuated by pinocembrin. CYP1B1 represents a novel therapeutic target for periodontitis. Pinocembrin warrants further preclinical evaluation as a promising adjunct to periodontal therapy.

Introduction

Periodontitis, a chronic inflammatory disease causing the progressive destruction of tooth-supporting tissues and eventual tooth loss,¹ affects about 60% of adults globally, with nearly a quarter of cases being severe.² Beyond its direct oral impact, it is linked to systemic conditions and imposes a major socioeconomic burden. Hence, deciphering the mechanisms underlying its development and progression is vital for identifying new treatment targets.

Cellular senescence, a state of irreversible cell cycle arrest triggered by various endogenous and exogenous stressors, is a key driver of both aging and chronic inflammation.^3,4 Senescent cells are characterised by cell cycle arrest, distinct morphological and metabolic alterations, and most notably, the senescence-associated secretory phenotype (SASP), through which they secrete a plethora of bioactive factors.4, 5, 6 The SASP components, in turn, induce stress and senescence in neighbouring normal cells, thereby creating a positive feedback loop that perpetuates tissue inflammation and damage, and heightens the risk for the onset and progression of additional inflammatory and age-related diseases.⁷ In periodontitis, senescent cells accumulate in the gingiva in proportion to disease severity,8, 9, 10, 11, 12 and their removal alleviates pathology, highlighting therapeutic potential. Consequently, targeting the process of cellular senescence and SASP is regarded as a highly promising strategy to disrupt disease progression and its destructive cycle. Emerging evidence, however, indicates that senescent cells exhibit considerable heterogeneity across different biological contexts.⁶ Defining the specific senescent landscape within the periodontal inflammatory microenvironment is therefore essential for developing precise, effective treatments.

Gingival fibroblasts (GFs), the main functional and immune-sentinel cells in the gingiva, are vital for periodontal homeostasis.¹³ Accumulating evidence indicates that the pathogenic microenvironment in periodontitis, comprising virulence factors from periodontal pathogens, chronic inflammation, and oxidative stress, drives GFs into senescence via sustained exposure.^10,14 Senescent GFs lose their normal proliferative and reparative capacities and are converted into effector cells exhibiting the SASP, characterised by the persistent release of abundant pro-inflammatory cytokines, chemokines, and matrix-degrading enzymes.¹⁵ This SASP can directly damage the periodontal supporting apparatus and further contribute to tissue breakdown by recruiting and activating immune cells and promoting osteoclast differentiation. However, the regulatory mechanisms that trigger GF senescence within the periodontal inflammatory microenvironment remain poorly understood.

Therefore, this study aims to characterise the senescent landscape of gingival tissues in the context of periodontitis and to identify the predominant cell types undergoing senescence. By integrating single-cell transcriptomic analysis, in vitro cellular assays, and in vivo intervention studies in a mouse model of periodontitis, we seek to elucidate the molecular mechanisms underlying the induction of senescence in GFs, identify specific targets within this process, and screen for potential therapeutic agents.

Materials and methods

Specimen collection

The study protocol was approved by the Ethics Committee of Southern Medical University (Approval No.: NFEC-2026-088). The baseline characteristics of the study participants are presented in Supplementary Table 1.

Following routine local anaesthesia, a gingival tissue specimen measuring approximately 5 mm × 3 mm was excised during crown lengthening or periodontal flap surgery. The samples were immediately stored at −80°C; alternatively, samples were preserved using a 4% paraformaldehyde solution.

Animal studies

All animal experiments were conducted using 6-week-old male C57BL/6 mice (specific pathogen-free grade). The study protocol was approved by the Animal Ethics Committee of Nanfang Hospital (Approval No.: IACUC-LAC-20231220-004). Periodontitis was induced by placing 5-0 silk ligatures around the bilateral maxillary second molars for 14 days. Mice in the pinocembrin group received local injections of 50 μM pinocembrin into the gingiva adjacent to the ligatures on alternating days, starting immediately post-ligation.

Micro-computed tomography analysis

The left maxillae were subjected to scanning using a micro-computed tomography (micro-CT) system. The vertical distance from the cemento-enamel junction (CEJ) to the alveolar bone crest (ABC) (CEJ-ABC distance) and the bone volume fraction (BV/TV) were quantified for comparison.

Tissue embedding

Human gingival tissues and mouse maxillae were fixed in 4% paraformaldehyde at 4°C for 48 h. Subsequently, mouse maxillae were decalcified in EDTA decalcification solution for 2 weeks. All tissues were then routinely dehydrated, cleared, and embedded in paraffin. Sections were cut into 4 μm slices for subsequent analysis.

Haematoxylin and eosin staining

Haematoxylin and eosin staining (H&E staining) was performed using a commercial kit according to the manufacturer’s protocol.

Immunofluorescence staining

Immunofluorescence staining was performed using an immunofluorescence staining kit according to the manufacturer’s instructions.

Cell isolation, culture, and treatment

Primary human GFs (hGFs) were isolated from healthy donor gingiva. Cells (passages 3-7) were seeded and allowed to adhere for 12 hours. The medium was then replaced with control medium, medium containing 10 μg/mL Porphyromonas gingivalis lipopolysaccharide (LPS), or LPS medium supplemented with 50 μM pinocembrin. After 24 hours of treatment, cells were harvested for analysis.

Quantitative real-time polymerase chain reaction

Total RNA was extracted from cell and tissue samples using the EZ-press RNA Purification Kit. Reverse transcription was performed with the PrimeScript RT Reagent Kit. Quantitative real-time polymerase chain reaction (qRT-PCR) was carried out on a LightCycler 480 system using SYBR Green Master Mix. The primer sequences are listed in Supplementary Table 2.

Western blot

Proteins were extracted using RIPA Buffer, separated by SDS-PAGE, and transferred to PVDF membranes. After blocking, the membranes were sequentially incubated with primary and HRP-conjugated secondary antibodies. Protein signals were detected and subsequently quantified.

BODIPY staining

BODIPY staining was performed using a commercial kit according to the manufacturer’s protocol.

FAOBlue staining

FAOBlue staining was performed using a commercial kit according to the manufacturer’s protocol.

Single-cell RNA sequencing data analysis and diagnostic modelling

In this study, a comprehensive bioinformatics pipeline was systematically established to dissect the senescent microenvironment of periodontitis. Briefly, we first integrated 3 independent single-cell RNA sequencing (scRNA-seq) datasets, which were subjected to stringent quality control and ambient RNA decontamination. Subsequently, Harmony was employed for robust multi-dataset batch correction, facilitating precise dimensionality reduction, unsupervised clustering, and the annotation of 14 major cell lineages. To specifically delineate the senescent landscape, cellular senescence was quantified utilising multiple reference gene sets, enabling the stratification of gingival fibroblasts for downstream differential expression and functional enrichment analyses. Intercellular ligand-receptor crosstalk within the dynamic periodontal microenvironment was further inferred using CellChat. For clinical translation, the bulk transcriptomic profiles were intersected with senescence signatures to identify a core 3-gene biomarker panel and construct a Random Forest-based diagnostic model. Finally, the specific mechanistic and therapeutic potential of CYP1B1 was explored by computationally simulating its transcriptomic perturbation via in silico virtual knockout, followed by a systematic drug-gene interaction screening that ultimately prioritised pinocembrin as a promising senotherapeutic candidate.

Statistical analysis

Data analysis was performed using GraphPad Prism 7 software (San Diego, California, USA). The results presented in the figures were expressed as the mean ± standard deviation (SD). Differences between 2 groups were assessed by unpaired 2-tailed Student’s t-tests, while comparisons among more than 2 groups were analysed by one-way ANOVA. Statistical significance is defined as P < .05 and was indicated in the Figures as follows: *P < .05, **P < .01, ***P < .001 and ****P < .0001.

A comprehensive description of detailed methods can be found in the Supplementary Materials and Methods section.

Results

Senescence atlas of cells in periodontitis

To investigate the presence of cellular senescence in gingival tissues, we collected gingival tissue samples from both healthy individuals and patients with periodontitis. Periodontitis tissues exhibited disrupted collagen structure and significant inflammatory infiltration (Figure S1A), accompanied by elevated mRNA levels of interleukin-6 (IL6) and interleukin-8 (IL8) (Figure S1B). We further examined the expression of senescence markers in the gingival samples. Western blot analysis showed that the expression levels of the senescence markers p16 and γ-H2A.X were significantly increased in periodontitis gingival tissues compared with those in healthy controls (Figure 1A). These results indicate that the level of cellular senescence is markedly upregulated in gingival tissues under periodontitis conditions.

Fig. 1.

Increased senescence of GFs in periodontitis.

(A) Western blot analysis of senescence-related proteins in human gingival tissues (n = 3 per group).

(B) UMAP visualisation of 144,195 integrated single cells, coloured by annotated major cell types.

(C) Violin plots showing the distribution of senescence-associated gene expression scores across cell types. Scores were calculated per cell using the AddModuleScore function in Seurat with the classical Fridman senescence gene set as a reference.

(D) Density plots of senescence scores across cell types, ordered by median score. The red dashed line indicates the threshold for the top 25% of scores across all cells.

(E) Pie chart showing the composition of the top 25% high-senescence cell population.

(F) Representative immunofluorescence images of human gingival tissues co-stained for the GF marker S100A4 and the senescence markers p16 and γ-H2A.X (n = 3 per group). Yellow arrows indicate senescent hGFs.

(G) Quantitative analysis of the percentage of senescent hGFs (n = 3 per group).

(H) Representative immunofluorescence images of mouse maxillae co-stained for the GF marker S100A4 and the senescence markers p16 and γ-H2A.X (n = 5 per group). Yellow arrows indicate senescent mGFs.

(I) Quantitative analysis of the percentage of senescent mGFs (n = 5 per group).

To characterise cellular senescence in gingival tissues, we integrated 3 independent scRNA-seq datasets derived from human gingival tissues (Figure S2A-D and Figure 1B). We scored each of the 14 annotated cell types for senescence-associated gene expression using the AddModuleScore function with the classical Fridman senescence gene set as a reference. The results revealed heterogeneity in senescence scores across different cell types. GFs exhibited a relatively high senescence signature score and represented the most abundant population among cells with high senescence-related scores (Figure 1C-E and Figure S3A).

We further classified the senescence patterns of distinct cell types using the senescent cell identification (SenCID) program¹⁶ (Figure S3B). Epithelial cells exhibited the highest score for senescence identity (SID) 2, whose core features are epidermal development and lipid metabolism. Mural cells were predominantly assigned to SID3, characterised by mitochondrial dysfunction and antioxidant stress responses. Endothelial cells and lymphatic endothelial cells displayed SID4 as their dominant senescence identity, a subtype defined by the unfolded protein response and endoplasmic reticulum stress. GFs showed the highest score for SID5, marked by the activation of anti-apoptotic pathways and permanent cell cycle arrest. In contrast, immune cells were mainly classified as SID6, whose core characteristics include loss of TP53 function and telomere dysfunction. These results reveal cell-type-specific senescence programs with distinct molecular signatures in gingival tissue. Finally, we focused on genes positively correlated with the SID5 senescence score in GFs, which included core SASP components, cell cycle arrest regulators, and DNA damage response genes (Figure S3C). These findings suggest that the SASP may be the primary mechanism through which senescent GFs aggravate periodontitis, in line with previous studies.

Given that GFs represent the primary cell type undergoing senescence in periodontitis, we assessed their senescence status in patient gingival tissues. Cellular senescence was indicated by the expression of p16 and γ-H2A.X, while hGFs were identified using the marker S100A4. Immunofluorescence analysis revealed increased expression of the senescence markers in hGFs from periodontitis tissues compared with healthy controls (Figure 1F and G). These findings suggest that hGFs exhibit an elevated level of senescence under periodontitis conditions.

To further delineate the senescence-associated alterations in GFs under periodontitis conditions, we validated our findings in a mouse experimental periodontitis model. A ligature-induced periodontitis model was successfully established in mice (Figure S4A-D). Compared with the control group, the ligature-induced model group exhibited significantly stronger fluorescence signals for both p16 and γ-H2A.X within S100A4-positive mouse GFs (mGFs) in gingival tissues (Figure 1H and I). These results indicate that mGFs also undergo elevated senescence in the experimental periodontitis model.

Functional and communicative attributes of senescent gingival fibroblasts in periodontitis

We further performed senescence scoring on GFs. Cells in the top 25% of scores were classified as high-senescence (Fridman-hi, herein 'senescent GFs') subpopulations, whereas the remaining cells were classified as low-senescence (Fridman-low, herein 'normal GFs') subpopulations (Figure S5A-C). AddModuleScore analysis using 3 additional, independent senescence-related gene sets consistently demonstrated significantly higher senescence scores in the Fridman-hi group compared to the Fridman-low group (Figure S5D), confirming the validity of this classification. Further analysis revealed a significantly higher proportion of Fridman-hi cells in periodontitis tissues than in healthy tissues (Figure 2A). Cell-cycle analysis demonstrated a markedly increased percentage of G1-phase cells in the Fridman-hi group relative to the Fridman-low group (Figure 2B and C), indicating that senescent GFs are arrested in the G1 phase, a hallmark of cellular senescence.

Fig. 2.

Functional profiling and microenvironmental crosstalk of senescent GFs.

(A) Grouped bar plot comparing the proportional distribution of senescent GFs and normal GFs in the Healthy and Periodontitis groups.

(B) Alluvial plot illustrating the distribution of cell cycle phases (G1, S, G2/M) within senescent GF and normal GF subpopulations.

(C) Grouped bar plot quantifying the percentage of cells in each cell cycle phase (G1, S, G2/M) for senescent GFs and normal GFs.

(D) Dot plot showing KEGG pathway enrichment analysis results for the senescent versus normal GF comparison.

(E) Network diagram illustrating the pathway interaction landscape among significantly enriched KEGG terms in senescent GFs.

(F) Dot plot displaying Gene Ontology (GO) biological process enrichment results for ligands specifically upregulated in senescent GFs compared to normal GFs.

(G) Grouped bar plots comparing the communication strength from senescent GFs and normal GFs toward each of the 8 major immune cell types.

(H) Horizontal bar plot quantifying the difference in communication strength between senescent and normal GFs directed toward each immune cell type.

(I) Horizontal bar plot displaying the top differentially active ligand-receptor interaction pairs from GFs toward immune cell populations. Interactions are ranked by the difference in activity between senescent (red) and normal (blue) GFs.

To characterise biological differences between senescent and normal GFs, Gene Set Enrichment Analysis (GSEA) was performed. Senescent GFs showed significant enrichment of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways associated with inflammatory response, immune activation, and tissue remodelling, as well as the p53 signalling pathway (Figure 2D). The p53 signalling pathway indicated the activation of this core senescence- and stress-related pathway. In contrast, metabolic and biosynthetic pathways were downregulated, indicating metabolic reprogramming with reduced anabolic activity. KEGG analysis further revealed that differentially expressed genes formed an interaction network centred on a “p53–cell cycle–cellular senescence” axis linked to inflammatory stress and matrix degradation pathways (Figure 2E).

Cell-cell communication networks were significantly remodelled in periodontitis (Figure S5E and F). Further comparison showed that the communication strength between senescent GFs and other cell types in the microenvironment was markedly higher than that of normal GFs (Figure S5G and H). Ligands specifically upregulated in senescent GFs were predominantly enriched in pathways related to immune cell recruitment, directed migration, and inflammatory activation (Figure 2F). In interactions with immune cells, senescent GFs exhibited significantly enhanced communication with nearly all immune cell subsets (Figure 2G and H). Notably, the interaction probabilities between the classic SASP components, midkine (MDK) and fibronectin 1 (FN1), and their corresponding receptors were significantly elevated in senescent GFs, suggesting their potential broad involvement in modulating the immune microenvironment (Figure 2I).

Prediction models for periodontitis based on senescence signature

To establish a senescence-based diagnostic model for periodontitis, we integrated periodontitis transcriptomic data with the classical Fridman senescence gene set, identifying 3 overlapping genes: murine double minute 2 (MDM2), cytochrome P450 family 1 subfamily B member 1 (CYP1B1), and endothelial cell-specific molecule 1 (ESM1) (Figure 3A). Random forest feature importance analysis, together with LASSO regression and SVM-RFE-based feature selection, confirmed that each of these 3 genes contributed to distinguishing periodontitis from healthy samples (Figure 3B). Evaluation of diagnostic efficacy revealed that each individual gene possessed a certain level of diagnostic value (Figure 3C). Moreover, a combined diagnostic model constructed with all 3 genes demonstrated a significantly increased area under the curve, representing the optimal diagnostic performance (Figure 3D).

Fig. 3.

Identification and validation of a core senescence gene signature for periodontitis diagnosis.

(A) Venn diagram illustrating the overlap between differentially expressed genes identified in the bulk RNA-seq dataset GSE173078 and the Fridman senescence gene set, yielding 3 shared candidate genes at the intersection.

(B) Horizontal bar plot displaying the feature importance of the 3 genes, ranked by Mean Decrease Accuracy.

(C) Individual Receiver operating characteristic (ROC) curves for each of the 3 signature genes evaluated independently in the GSE173078 dataset.

(D) ROC curves comparing the diagnostic performance of 2 classification models in the GSE173078 dataset.

(E) UMAP visualisations of the GF compartment comparing senescence classification by the original Fridman method and the new 3-gene signature model.

(F) Confusion matrix quantifying the concordance between the Fridman-based classification and the new 3-gene signature classification across all GFs.

(G) Dot plot displaying KEGG pathway enrichment results for the 3-gene senescence signature.

(H) Network diagram illustrating the pathway interaction landscape among significantly enriched KEGG terms for the 3-gene signature.

(I) qRT-PCR analysis of MDM2, CYP1B1, and ESM1 mRNA levels in human gingival tissues (n = 3 per group).

(J) qRT-PCR analysis of Mdm2, Cyp1b1, and Esm1 mRNA levels in mouse gingival tissues (n = 5 per group).

(K) Violin plots displaying the single-cell expression distribution of MDM2, CYP1B1, and ESM1 across all annotated cell types in the integrated single-cell atlas.

To validate the association of the 3-gene signature with classical cellular senescence states, we compared their expression profiles with the senescence classification based on the classical Fridman senescence gene set. The 3-gene signature effectively distinguished the senescent and normal GF subpopulations (Figure 3E). Confusion matrix analysis demonstrated a high level of agreement between the 2 classification methods at the single-cell level (Figure 3F), indicating that this gene signature reliably reflects the cellular senescence state of GFs. Further functional analysis revealed that biological processes associated with these genes were significantly enriched in pathways related to p53-mediated cellular senescence, stress response, and immune regulation, functioning through a coordinated network (Figure 3G-H). These findings provide functional support for their role as senescence-associated biomarkers in periodontitis.

The expression of MDM2, CYP1B1, and ESM1 was examined in gingival tissues. The results showed that the expression levels of all 3 genes were significantly higher in human periodontitis tissues than in healthy controls (Figure 3I). Similarly, in a ligature-induced experimental periodontitis mouse model, the expression of these genes was significantly upregulated in the gingival tissues of the ligature group compared to the control group (Figure 3J). Further analysis of the single-cell transcriptomic data revealed that CYP1B1 expression was markedly elevated in senescent GFs (Figure 3K).

Upregulation of CYP1B1 is associated with LPS-induced senescence and abnormal fatty acid metabolism in gingival fibroblasts

To clarify the potential mechanisms underlying hGF senescence, we established an in vitro model of gingival fibroblast senescence by stimulating hGFs with LPS. Senescence-associated β-galactosidase (SA-β-gal) staining showed that the proportion of senescence-positive cells was significantly increased following LPS stimulation (Figure 4A). Western blot analysis revealed that the expression levels of the senescence marker proteins p16 and γ-H2A.X were markedly upregulated (Figure 4B). Concurrently, the mRNA expression of SASP components was also significantly elevated (Figure 4C). These results indicate that LPS can successfully induce cellular senescence in hGFs.

Fig. 4.

Fatty acid metabolism in LPS-induced senescent gingival fibroblasts.

(A) Representative images of senescence-associated β-galactosidase (SA-β-gal) staining in LPS-induced senescent hGFs (n = 3 per group).

(B) Western blot analysis of senescence-related proteins in LPS-induced senescent hGFs (n = 3 per group).

(C) qRT-PCR analysis of inflammatory cytokine mRNA levels in LPS-induced senescent hGFs (n = 3 per group).

(D) qRT-PCR analysis of MDM2, CYP1B1, and ESM1 mRNA levels in LPS-induced senescent hGFs (n = 3 per group).

(E) Protein-protein interaction network of CYP1B1.

(F) GO enrichment analysis of CYP1B1-interacting proteins via the STRING database.

(G) Representative images of FAOBlue staining indicating fatty acid oxidation activity in LPS-induced senescent hGFs (n = 3 per group).

(H) Relative fluorescence intensity of FAOBlue staining in LPS-induced senescent hGFs (n = 3 per group).

(I) Representative images of BODIPY 493/503 staining visualising neutral lipid droplets in LPS-induced senescent hGFs (n = 3 per group).

(J) Relative fluorescence intensity of BODIPY 493/503 staining in LPS-induced senescent hGFs (n = 3 per group).

In the LPS-induced senescence model, the expression levels of MDM2, CYP1B1, and ESM1 were all significantly upregulated (Figure 4D). Given that single-cell data indicated a more pronounced upregulation of CYP1B1 in senescent GFs, we hypothesised that CYP1B1 might be involved in regulating the cellular senescence process. Functional enrichment analysis of CYP1B1 was performed based on the protein-protein interaction (PPI) network constructed using the STRING database, revealing a significant enrichment in the ‘negative regulation of fatty acid metabolic process’ (Figure 4E and F). Previous studies have suggested that CYP1B1 can inhibit fatty acid oxidation (FAO) and promote lipid accumulation,¹⁷ processes closely linked to cellular senescence and inflammation.¹⁸ Prompted by these findings, we further examined changes in FAO and lipid accumulation in the LPS-induced senescence model. FAOBlue staining showed that FAO was significantly reduced after LPS stimulation (Figure 4G and H), while BODIPY fluorescence staining revealed a marked increase in intracellular lipid droplet accumulation (Figure 4I and J). These results indicate that the upregulation of CYP1B1 is associated with disrupted fatty acid metabolism in LPS-induced senescent hGFs.

Identification of CYP1B1-targeted compounds for attenuating senescence

To elucidate the role of CYP1B1 in regulating cellular senescence, we simulated CYP1B1 knockout using a bioinformatics approach and performed pathway enrichment analysis on the differentially expressed genes. The results showed that following CYP1B1 knockout, the differentially expressed genes were significantly enriched in signalling pathways closely associated with cellular senescence, such as the longevity regulation pathway (Figure 5A and Figure S6A and B).

Fig. 5.

Targeting CYP1B1 with pinocembrin ameliorates LPS-induced senescence in gingival fibroblasts.

(A) Gene regulatory network diagram centred on CYP1B1 knockout (KO), illustrating the downstream effector genes of CYP1B1 derived from transcriptomic perturbation analysis. Lines connect CYP1B1 to differentially expressed downstream effectors (red, upregulated upon KO; blue, downregulated upon KO).

(B) Dot plot displaying curated CYP1B1 drug-gene interactions retrieved from the Drug-Gene Interaction Database (DGIdb). Each point represents a drug-gene interaction pair, colour-coded by drug category.

(C) Scatter plot evaluating candidate drugs across 2 composite therapeutic dimensions: Anti-aging Score derived from molecular docking and inhibition data, and Anti-inflammatory Score derived from published evidence of anti-inflammatory activity. Dot size reflects the Combined Score integrating both dimensions; dot colour encodes drug category.

(D) qRT-PCR analysis of CYP1B1 mRNA levels in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

(E) Western blot analysis of CYP1B1 in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

(F) Representative images of FAOBlue staining indicating fatty acid oxidation activity in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

(G) Relative fluorescence intensity of FAOBlue staining in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

(H) Representative images of BODIPY 493/503 staining visualising neutral lipid droplets in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

(I) Relative fluorescence intensity of BODIPY 493/503 staining in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

(J) Representative images of senescence-associated β-galactosidase (SA-β-gal) staining in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

(K) Western blot analysis of senescence-related proteins in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

(L) qRT-PCR analysis of inflammatory cytokine mRNA levels in LPS-induced senescent hGFs after pinocembrin treatment (n = 3 per group).

Based on the Drug–Gene Interaction database (DGIdb), we screened for potential compounds targeting CYP1B1 (Figure 5B). Pinocembrin demonstrated excellent performance in both anti-senescence and anti-inflammatory scoring, positioning it within the ideal zone (Figure 5C). Further evaluation of the candidate drug’s effects on senescence- and periodontitis-related biological processes revealed that pinocembrin most potently suppressed the SASP, while its inhibitory effect on collagen synthesis was relatively mild (Figure S6C). Although the dose–response curve suggested that its direct inhibitory potency against CYP1B1 was not the highest among the candidates (Figure S6D), pinocembrin emerged as a priority candidate for targeting CYP1B1 to intervene in periodontitis and associated cellular senescence, based on its overall favourable profile in delaying senescence, suppressing inflammation, and maintaining extracellular matrix homeostasis.

Pinocembrin ameliorates LPS-induced senescence in gingival fibroblasts through restoration of fatty acid metabolic homeostasis

To evaluate the therapeutic potential of pinocembrin, we treated LPS-induced senescent gingival fibroblasts with this compound. The results showed that pinocembrin treatment significantly downregulated both the mRNA and protein expression levels of CYP1B1 (Figure 5D and E), and significantly reversed the LPS-induced reduction in FAO and the LPS-induced increase in intracellular lipid accumulation (Figure 5F-I). Moreover, pinocembrin intervention reduced the proportion of SA-β-gal-positive cells and decreased the expression levels of the senescence markers p16 and γ-H2A.X, as well as SASP components IL6 and IL8 (Figure 5J-L), indicating that it can markedly alleviate LPS-induced cellular senescence and inflammatory secretion in hGFs.

To evaluate the therapeutic effect in vivo, pinocembrin was administered via local gingival injection in a mouse experimental periodontitis model. The results showed that pinocembrin treatment significantly downregulated the mRNA of CYP1B1 and SASP components, and attenuated inflammatory cell infiltration in periodontal tissues and reduced alveolar bone loss (Figure 6A-E). Concurrently, the senescence level of mGFs in the gingival tissues was markedly reduced (Figure 6F and G).

Fig. 6.

Pinocembrin alleviates gingival fibroblast senescence in mouse experimental periodontitis.

(A) qRT-PCR analysis of Cyp1b1 mRNA levels in gingival tissues from ligature-induced periodontitis mice treated with pinocembrin (n = 5 per group).

(B) qRT-PCR analysis of inflammatory cytokine mRNA levels in gingival tissues from ligature-induced periodontitis mice treated with pinocembrin (n = 5 per group).

(C) Representative H&E staining of maxillary periodontal tissue from ligature-induced periodontitis mice treated with pinocembrin (n = 5 per group).

(D) Representative micro-CT images of the maxillae from ligature-induced periodontitis mice treated with pinocembrin (n = 5 per group).

(E) Quantitative analysis of micro-CT, including cemento-enamel junction to alveolar bone crest (CEJ-ABC) length and bone volume fraction (BV/TV, bone volume/tissue volume) (n = 5 per group).

(F) Representative immunofluorescence images of mouse maxillae co-stained for the GF marker S100A4 and the senescence markers p16 and γ-H2A.X (n = 5 per group). Yellow arrows indicate senescent mGFs.

(G) Quantitative analysis of the percentage of senescent mGFs (n = 5 per group).

Discussion

We report that cellular senescence is a hallmark of periodontitis-affected gingiva. Single-cell transcriptomics identified GFs as the major senescent population, and functional analysis revealed their role in pathogenic microenvironment remodelling. A senescence gene signature derived from this analysis proved effective for periodontitis diagnosis and prediction. We mechanistically linked one signature gene, CYP1B1, to the promotion of GF senescence through impaired fatty acid metabolism. Subsequently, we identified pinocembrin as a CYP1B1-targeting compound that alleviated cellular senescence and tissue destruction in experimental periodontitis (Figure 7).

Fig. 7.

Schematic diagram of the underlying mechanism by which pinocembrin alleviates periodontal tissue destruction.

Under periodontitis conditions, LPS induces the senescence of GFs via the CYP1B1-driven dysregulation of fatty acid metabolism. These senescent GFs subsequently secrete SASP components to exacerbate periodontal tissue destruction. Treatment with pinocembrin effectively inhibits CYP1B1 expression, thereby attenuating cellular senescence and alleviating tissue damage in the experimental periodontitis model.

Periodontitis and tissue aging are increasingly recognised as interrelated processes. Clinical studies indicate that the prevalence and severity of periodontitis increase with age.19, 20, 21 Kim et al. reported a significant correlation between the expression of the aging marker p16 in gingival tissues and clinical parameters such as bleeding on probing (BOP) and clinical attachment loss (CAL).²² Furthermore, periodontitis itself can accelerate cellular senescence within gingival tissues independently of chronological aging. Increased senescence has been observed in GFs, periodontal ligament stem cells, and certain immune cells under periodontitis conditions.^15,23,24 Emerging concepts such as "ferro-aging" further suggest that cellular senescence comprises heterogeneous subtypes with distinct regulatory programs, analogous to the diversity in programmed cell death.²⁵ However, although the role of senescence in periodontitis pathogenesis is established, the functional heterogeneity of senescent cells within the periodontal microenvironment remains poorly understood. To address this, we applied a machine learning-based SenCID program to systematically profile senescence across gingival cell types.¹⁶ We found that the senescence state of each cell type is closely associated with its specific functional modules. These insights advance our understanding of senescent cell complexity in periodontitis and support the development of cell-type-specific senotherapies, which could enhance treatment precision while minimising off-target effects.

The present study confirms that GFs are the primary cell type undergoing senescence in periodontitis-affected gingival tissues. The role of these senescent GFs in disease progression is increasingly recognised. For instance, Guo et al. reported that senescent GFs highly express SASP components such as interleukin-1 beta (IL-1β), IL-6, transforming growth factor-beta (TGF-β), and IL-8, which promote neutrophil extracellular trap formation and drive macrophages toward an M1-polarised phenotype.¹⁰ Similarly, Yin et al. observed elevated expression of fibroblast activation protein (FAP) in senescent GFs, which contributes to periodontal tissue destruction.¹⁵ Extending these findings, our work further demonstrates that the functional profile of senescent GFs is significantly enriched in pathways related to inflammatory stress and extracellular matrix degradation, and their communication with various immune cells in the gingiva, including natural killer (NK) cells, macrophages, and neutrophils, is substantially enhanced. Key among the ligands mediating this cross-talk are MDK and FN1. Previous research has shown that MDK-mediated inflammatory signalling contributes to the dysregulation of the inflammatory microenvironment in psoriasis.²⁶ Separately, Ruze et al. reported that in both postmenopausal osteoporosis patients and ovariectomised mice, MDK inhibits bone formation and promotes the expression of inflammatory factors, whereas targeted inhibition of MDK alleviates bone loss in ovariectomised mice.²⁷ Furthermore, the FN1-CD44 interaction has been shown to regulate cell adhesion and migration.^28,29 Collectively, these observations suggest that senescent gingival fibroblasts may remodel the local immune microenvironment via molecules such as MDK and FN1, thereby exacerbating the pathological progression of periodontitis. In addition to modulating immune cells to accelerate periodontitis progression, senescent gingival fibroblasts may also profoundly affect stromal and structural cells within the local microenvironment. Previous studies have demonstrated that in the periodontitis microenvironment, gingival fibroblasts induce epithelial-mesenchymal transition (EMT) in adjacent gingival epithelial cells, by secreting abundant cytokines and matrix metalloproteinases (MMPs) and weakening the overlying epithelial barrier, thereby facilitating the invasion of periodontal pathogens into deeper tissues.^13,30,31 However, while the crosstalk between pro-inflammatory fibroblasts and epithelial cells is well-documented, no studies to date have deeply explored the specific mechanisms by which senescent gingival fibroblasts, particularly through their unique senescence-associated secretory phenotypes, affect gingival epithelial cells. This specific mechanism of cellular crosstalk warrants further investigation.

In the periodontitis prediction model established in this study, CYP1B1 was identified as a key senescence-associated gene, showing significantly upregulated expression in senescent GFs. The role of CYP1B1 in GF senescence has rarely been reported. Previous studies have suggested that CYP1B1 can promote the senescence of various cell types, such as exacerbating mesenchymal stem cell senescence by inducing mitochondrial dysfunction³² or participating in PM2.5-induced cardiomyocyte senescence by elevating reactive oxygen species (ROS) levels.³³ In the present study, we found that CYP1B1 inhibits FAO and promotes lipid droplet accumulation in GFs. CYP1B1 has recently been demonstrated to mediate the pathological remodelling of cellular lipid homeostasis.³⁴ Previous studies have demonstrated that elevated CYP1B1 expression in muscle stem cells impairs CD36-mediated fatty acid oxidation, thereby contributing to the pathogenesis of sarcopenia.³⁵ Cyp1b1-knockout mice exhibit suppressed expression of stearoyl-CoA desaturase 1 (Scd1), pyruvate dehydrogenase kinase 4 (Pdk4), and malic enzyme 1 (Me1), driving hepatic metabolic reprogramming from triglyceride deposition toward mitochondrial FAO and attenuating pathological lipid accumulation.³⁶ Moreover, the upregulation of CYP1B1 can activate mTOR signalling to inhibit the expression and translocation of TFE3, a central regulator of lipophagy, resulting in impaired phagolysosome biogenesis and subsequent ectopic lipid deposition.³⁷ Thus, we propose that CYP1B1-mediated suppression of FAO and promotion of lipid droplet accumulation may play an important role in driving the senescence of GFs under periodontitis conditions.

Previous studies have underscored a bifunctional, highly context-dependent role of FAO in cellular senescence, fundamentally dictated by stress etiology, temporal progression, and tissue specificity.38, 39, 40 During acute stress or oncogene-induced senescence (OIS), a compensatory upregulation of FAO generates acetyl-CoA to promote histone acetylation, thereby driving p16 expression and fuelling the SASP.^41,42 Conversely, chronic aging or replicative senescence features a pathological downregulation of PPARα and CPT1A, which leads to ectopic lipid accumulation, triggering lipotoxicity, mitochondrial ROS surges, and a sustained DNA damage response (DDR).^43,44 Consistent with this latter mechanism, our study demonstrates that diminished FAO and concurrent lipid accumulation serve as prominent hallmarks of LPS-induced senescence in hGFs. In summary, acute FAO upregulation supports senescence establishment as a metabolic adaptation, whereas its chronic downregulation drives mitochondrial dysfunction, lipotoxicity, and a self-reinforcing senescent state. Understanding this bidirectional regulation may guide stage-specific senolytic and metabolic strategies in age-related diseases.

Prior studies have demonstrated that targeted clearance or inhibition of senescent cells can delay the progression of periodontitis. For example, the combination of dasatinib and quercetin has been shown to suppress senescence in gingival tissues, attenuate inflammation, and reduce alveolar bone loss in mice.^11,45 Rapamycin has also been confirmed to alleviate periodontal bone destruction by reducing the number of senescent GFs and restoring the FAP/osteolectin (OLN) balance.¹⁵ Against this background, and to intervene in CYP1B1-mediated GF senescence under periodontitis conditions, the present study identified pinocembrin as a potential CYP1B1 inhibitor and validated its efficacy in mitigating cellular senescence and periodontal tissue destruction through both in vitro and in vivo experiments. Previous studies have demonstrated that pinocembrin exhibits a dual mode of action against CYP1B1, involving both high-affinity direct protein binding to reduce stability and the suppression of ROS/mitogen-activated protein kinase (MAPK)-mediated transcriptional induction.⁴⁶ Supported by our in silico docking predictions and the observed downregulation of CYP1B1 mRNA and protein, we hypothesise that pinocembrin exerts a similar comprehensive inhibitory effect in the periodontitis microenvironment. In other disease models, pinocembrin has also exhibited favourable anti-inflammatory, anti-cell-death activity, and a good safety profile.46, 47, 48, 49 Our study further extends the potential application of pinocembrin in the context of periodontitis and cellular senescence regulation, providing experimental evidence to support its development as a candidate therapeutic agent for periodontitis intervention.

This study has several limitations. Regarding the animal model, although the ligature-induced experimental periodontitis mouse model is widely used in research on gingival tissue senescence, its pathological progression is relatively rapid and may not fully recapitulate the long-term course of human periodontitis.^15,50 Therefore, our findings warrant further validation in chronic periodontitis models induced by periodontal pathogens such as P. gingivalis. In terms of the in vitro model, besides LPS, other stimuli such as ROS¹⁰ and bleomycin¹⁵ are also commonly used to induce cellular senescence. LPS is a widely utilised stimulus in research on bacteria-associated oral inflammatory diseases, including periodontitis⁵¹ and pulpitis,⁵² and is well-documented to induce profound phenotypic and functional cellular alterations. However, it must be acknowledged that a direct, single-factor LPS challenge may not fully recapitulate the complex and multifaceted senescent phenotypes observed within the multifactorial microenvironment of human periodontitis tissues. Furthermore, senescence phenotypes induced by different stimuli may exhibit heterogeneity. Thus, alterations in fatty acid metabolism and their functional implications in GF senescence models under other stimulus conditions require further investigation.

Conclusions

Taken together, our findings demonstrate that senescent GFs exacerbate periodontal tissue destruction through the secretion of SASP components. Mechanistically, CYP1B1-mediated dysregulation of fatty acid metabolism is a key event driving GF senescence. Notably, pinocembrin effectively attenuates senescence and alleviates tissue damage in an experimental periodontitis model by inhibiting CYP1B1 expression.

Data availability

All data are available from the corresponding author upon reasonable request.

Ethics approval

The research was approved by the Ethics Committee of Southern Medical University (Approval No.: NFEC-2026-088), and written informed consent was obtained from all participants prior to sample collection. The animal study was approved by the Animal Ethics Committee of Nanfang Hospital (Approval No.: IACUC-LAC-20231220-004). All animal procedures were performed in accordance with the ARRIVE guidelines.

Author contributions

Zehao Chen, Qianwen Tang, Ruoshu Tang, Yumeng Yang, and Weilun Cai contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript; Jun Shao and Ruiming Guo contributed to data acquisition, analysis, and interpretation; Fuchun Fang contributed to conception, design, and critically revised the manuscript. All authors gave their final approval and agreed to be accountable for all aspects of the work.

Funding

This study was supported by the National Natural Science Foundation of China (82270982, 82470975), Natural Science Foundation of Guangdong Province (2024A1515010840), Guangzhou Key Research and Development Program (2024B03J0667), Basic and Applied Basic Projects of Huadu District in Guangzhou (24HDQYLH16), and Shenzhen Medical Research Fund (A2502016).

Conflict of interest

The authors declare no conflicts of interest.