The circ_0042103/TAF15/NER axis regulates inflammation and DNA damage in pulpitis

Abstract

Aim

Circular RNAs (circRNAs) have been identified as key regulators in inflammatory diseases, yet their function in pulpitis is unclear. This study investigates their potential role in the progression of pulpitis.

Methodology

Microarray and single-cell RNA sequencing were applied to assess DNA damage responses (DDR) in inflammatory pulp and its derived stem cells, respectively. qRT-PCR and Western blot were employed to detect the DNA double-strand break (DSB) marker γ-H2AX and inflammatory cytokines in pulp tissue. Bioinformatics analysis was used to identify upregulated circRNAs in inflamed DPSCs. Functional assays were performed to assess the impact of circ_0042103 on LPS-driven cellular damage and inflammation in DPSCs. The interaction between circ_0042103 and TAF15 was investigated using RNA FISH, pulldown, and nuclear-cytoplasmic fractionation assays. Transfection with circ_0042103/TAF15-siRNA in DPSCs was carried out to evaluate activation of the nucleotide excision repair (NER) pathway and its regulatory effects on DNA damage and inflammation.

Results

DDR was activated in both pulpitis and inflamed DPSCs. DNA damage showed a positive correlation with inflammation in pulpitis. In vitro, circ_0042103 upregulation amplified LPS-stimulated DDR and inflammatory signaling, whereas its knockdown alleviated both effects. Mechanistically, circ_0042103 bound TAF15, leading to decreased levels of the NER-related proteins (ERCC1 and PCNA) and increased DNA damage and inflammation.

Conclusion

By interacting with TAF15, circ_0042103 reduces the levels of the NER-related proteins ERCC1 and PCNA, leading to increased DNA damage and inflammation in hDPSCs, thereby defining a circ_0042103/TAF15/NER axis in pulpitis progression.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13287-025-04817-1.

Introduction

Pulpitis is a common inflammatory disease in dentistry and is multifactorial [1]; among these factors, persistent polymicrobial infection of the root canal system is a principal driver. Microbial products, including endotoxin (LPS/LOS) and protein exotoxins, can induce excessive release of proinflammatory cytokines such as IL-6, IL-8, and TNF-α [2, 3]. These events lead to severe pain and progressive, sometimes irreversible, destruction of dental-pulp tissue [4–6]. Current clinical management primarily relies on removal of infected/necrotic pulp tissue, yet this approach does not reliably control inflammatory spread or promote true tissue regeneration [7–9]. Therefore, elucidating the molecular mechanisms underlying pulpitis is critical for developing more targeted and effective therapeutic strategies.

Emerging evidence suggests that chronic inflammation can induce DNA damage and tissue injury [10]. Upon sensing DNA lesions, cells initiate a sophisticated DNA damage response (DDR), which orchestrates damage recognition, signaling transduction, and recruitment of DNA repair machinery. When repair fails, cells may undergo apoptosis or senescence [11]. Among various forms of DNA lesions, double-strand breaks (DSBs) are considered the most deleterious [12]. Notably, elevated expression of γ-H2AX, a hallmark of DSBs [13], has been observed in inflamed human dental pulp tissues [14, 15]. Our previous study also confirmed that DDR is activated during pulpitis [16], revealing the contribution of DNA damage to the progression of pulpitis.

Circular RNAs (circRNAs), a class of covalently closed non-coding RNAs, have recently emerged as important epigenetic regulators involved in both inflammation and DDR [17–19]. For instance, circ_0138960 has been shown to modulate inflammatory signaling via the miR-545-5p axis [2], while circ_0089282 knockdown exacerbates DNA damage in lung tissues by inhibiting FUS-mediated repair [20]. Despite these advances, it remains unclear whether specific circRNAs regulate DDR to drive pulpitis progression.

In this study, we aimed to test whether a specific circRNA links inflammatory signaling to DDR in pulpitis and contributes to disease progression, by integrating bioinformatics screening, in vitro DPSC experiments, and analyses of clinical specimens to inform potential therapeutic strategies.

Materials and methods

Microarray analysis

We used gene expression microarray data (GSE77459), comprising six irreversible pulpitis and six healthy pulp samples [21], to assess tissue-level differential activation of inflammation and the DNA damage response (DDR) between inflamed and healthy pulps. Data preprocessing, including background correction and normalization, was performed using the RMA algorithm in the oligo package. Differentially expressed genes were identified using the limma package, with thresholds of |log₂FC| ≥ 1 and FDR < 0.05. Functional enrichment (GO and KEGG) and gene set enrichment analysis (GSEA) were conducted using clusterProfiler, with p.adjusted < 0.05 considered significant. All analyses were performed at the tissue level; no cell-type inferences were made from this dataset.

scRNA-Seq analysis

Single-cell RNA sequencing data were obtained from the GEO database (GSE185222) [22]. Quality control excluded cells with fewer than 200 or more than 6,000 detected genes, greater than 50,000 unique molecular identifiers (UMIs), or mitochondrial gene content exceeding 15%. Data preprocessing and clustering were conducted using the Seurat v5 package [23]. Gene expression was normalized with a scale factor of 10,000, and the top 2,000 highly variable genes were selected for further analysis. Principal component analysis (PCA) was conducted on the top 16 components identified through the elbow plot method. Cell clustering was performed using the Louvain algorithm via the FindClusters function [24]. Dimensionality reduction and visualization were carried out using the UMAP algorithm. Differentially expressed genes (DEGs) were identified using the Wilcoxon rank-sum test, with a mean fold change >0.25 and detection in >25% of cells within each cluster. Cell type annotations were assigned based on canonical marker genes and literature references [25]. Cell-cell communication was inferred by analyzing ligand–receptor interactions, defined as significant when P < 0.05 and mean log-normalized expression >0.1 [26]. Within the DPSC cluster, module scores for inflammation and DDR gene sets were computed using Seurat’s AddModuleScore to evaluate whether these programs are preferentially activated in DPSCs from inflamed pulp relative to healthy controls; higher scores indicate stronger gene-set activity.

Tissue extraction and cell culture

The Ethical Review Committee of Guangzhou Medical University Dental Hospital (JCYJ2023005) reviewed and approved all experimental protocols using human cells and tissues. Collection of pulp tissue that were clinically not retainable. The obtained pulp tissue was preserved in liquid nitrogen for subsequent RNA and protein extraction. Human dental pulp stem cells (hDPSCs) were isolated from healthy pulp tissue of patients aged from 12 to 20 years old. For hDPSCs isolation, freshly dissected pulp tissue was minced and digested with Type I collagenase (1 mg/mL, 37 °C, 40 min). The resulting pulp fragments were then cultured in 20% FBS medium at 37 °C with 5% CO₂ for 2 weeks to establish primary hDPSCs. The derived cells were subsequently stored at 37 °C and 5% CO₂, and cultured in alpha-MEM medium (Gibco, USA) supplemented with 10% fetal bovine serum (Gibco, USA) and 1% penicillin-streptomycin solution (Gibco, USA). The stemness of hDPSCs has been characterized in previous studies [27]. LPS powder (E. coli-L2880; Sigma, USA) was dissolved in α-MEM blank medium at a master mix concentration of 1 mg/mL. For application, Penicillin/Streptomycin Solution-free α-MEM medium (containing 10% FBS) was prepared to dilute the master mix for use.

RNA extraction and qRT‒PCR

Cellular or tissue RNA was purified with TRIzol reagent (Invitrogen, USA) following standard protocols. cDNA synthesis was carried out with the Evo M-MLV Reverse Transcription Premix (Accurate Biology, China). For gene expression quantification, SYBR Green Pro Taq HS qPCR Master Mix (Accurate Biology, China) was employed on a QuantStudio5 thermocycler (Applied Biosystems, USA). Target gene expression was calculated via the 2^−ΔΔCT method, with GAPDH serving as the reference control. All oligonucleotide primers were commercially obtained from IGE Biotechnology (Guangzhou, China). The primer sequences used for qRT-PCR and PCR are provided in Supplementary Table S1.

Enzyme-linked immunosorbent assay (ELISA)

For quantification of IL-6 and IL-8 secretion levels in LPS-stimulated hDPSCs, commercial ELISA kits (Beyotime Biotechnology, Shanghai, China) were used following protocol guidelines.

Western blot analysis

Cells or tissues were lysed on ice with RIPA lysis buffer (Beyotime Biotechnology, China). Protein quantification was conducted with a BCA assay kit (Beyotime Biotechnology). Appropriate total protein samples were electrophoresed on 10% denaturing polyacrylamide gels (Epizyme, China) and subsequently transferred onto 0.22 μm PVDF membranes (Millipore, USA). The membranes were blocked with blocking solution (Beyotime Biotechnology, China) for 1 h at room temperature, followed by overnight incubation at 4 °C with primary antibodies, including anti-γ H2A.X (phospho S139; ab26350, Abcam, UK), anti-GAPDH (ab9485, Abcam, UK), anti-TAF15 (ab134916, Abcam, UK), anti-DGCR8 (ab191875, Abcam, UK), anti-FUS (ab243880, Abcam, UK), anti-IGF2BP2 (ab124930, Abcam, UK), anti-PCNA (13110, Cell Signaling Technology, USA) and anti-ERCC1 (ab129267, Abcam, UK). After TBST elutioning, membranes were incubated with IRDye 800CW goat anti-rabbit secondary antibody (926-32211, LI-COR, USA) and IRDye 800CW goat anti-mouse secondary antibody (925-32210, LI-COR, USA). Protein signals were visualized and quantified using an Odyssey LI-COR imaging system.

CircRNA sequencing

circRNA sequencing analysis was conducted on LPS-treated (1 µg/ml, 5 h) and untreated hDPSCs through next-generation sequencing, with technical support from BGI (Shenzhen, China). Sample preparation was performed according to the standard procedure of Arraystar [28]. To isolate circular RNA species, total RNA underwent enzymatic digestion with RNase R to remove linear transcripts, followed by purification with RNeasy MinElute Cleanup Kit (Qiagen, China). For library generation, we utilized Illumina’s VAHTS Total RNA-seq (H/M/R) preparation kit according to manufacturer specifications. The workflow involved (1) fragmentation of enriched circRNA molecules; (2) reverse transcription to generate cDNA; (3) fragment size selection using VAHTS magnetic beads; (4) end repair and polyadenylation; and (5) adapter ligation for Illumina sequencing. To ensure strand specificity, uracil-N-glycosylase (UNG) was applied to digest secondary cDNA strands before final PCR amplification. Sequencing was performed on Illumina HiSeqTM 2500 platform. Following raw data quality control, sequence alignment and circRNA identification were executed using Tophat2 software (version 2.0.3.12). Differentially expressed circRNAs (DE-circRNAs) were identified using the edgeR package under a stringent threshold: |log₂ FC| ≥1 (equivalent to twofold difference) and a false discovery rate (FDR) ≤ 0.05. Upregulated circRNAs were defined as having higher expression in LPS-treated hDPSCs compared to controls, while downregulated circRNAs showed the inverse pattern.

Separation of nuclear and cytoplasmic RNA

Nuclear and cytoplasmic RNAs were isolated using the PARIS subcellular fractionation system (Life Technologies, USA) according to the manufacturer’s protocol with the following parameters. Approximately 3 × 10⁶ hDPSCs were resuspended in ice-cold Cell Fractionation Buffer, incubated on ice for 5 min, and centrifuged at 5,000 × g for 5 min at 4 °C to separate the cytoplasmic supernatant and nuclear pellet. The supernatant was transferred to a fresh tube as the cytoplasmic fraction. The nuclear pellet was resuspended in pre-chilled Disruption Buffer, incubated on ice for 10 min, and centrifuged at 2,000 × g for 4 min at 4 °C to remove chromatin. Each fraction was mixed with Lysis Buffer at a fixed ratio, precipitated with absolute ethanol, and purified with two washes using the kit Wash Buffer to obtain nuclear and cytoplasmic RNAs. Purified RNAs were treated with DNase I and reverse-transcribed with equal RNA input per fraction. Furthermore, U6 snRNA (nuclear-enriched) and GAPDH mRNA (cytoplasmic-enriched) were assayed in parallel as fractionation controls to verify purity.

Agarose gel electrophoresis

A 3.0% (w/v) agarose gel was prepared by dissolving 4.5 g of agarose powder in 150 ml of 1xTBE buffer. The mixture was heated in a microwave oven until fully dissolved. After the solution cooled slightly, 5 µl of a nucleic acid gel stain (EB substitute) was added and mixed thoroughly. The solution was poured into a gel tray (mold), a comb was inserted, and the gel was left to solidify. Once solidified, the comb was removed, and the gel was placed into an electrophoresis chamber. The chamber was filled with 1xTBE buffer until the gel was submerged. The qRT-PCR products from the gDNA and cDNA samples were collected. For sample loading, 10 µl of the PCR product was mixed with 2 µl of 6x loading buffer. The total 12 µl mixture was then loaded into the wells. 5 µl of a DNA Marker (Ladder) was loaded as a reference. Electrophoresis was run at 110 V for 30 min. After electrophoresis, the gel was visualized using a gel imaging system. To validate the circular structure of circ_0042103, divergent primers spanning the back-splice junction and convergent primers were designed. PCR was performed using both cDNA and genomic DNA (gDNA) templates. The convergent primers for the housekeeping gene GAPDH were used as a positive control to confirm the integrity and quality of the gDNA and cDNA templates.

RNase R treatment

Linear RNA was digested using RNase R (K3061, Apexbio, USA). Total RNA (1000 ng) extracted from hDPSCs was divided into two groups: an RNase R treatment group and a control group. The treatment samples were incubated with RNase R at 37 °C for 10 min. Immediately following incubation, the reaction was stopped and the enzyme was removed by purifying the RNA according to the manufacturer’s protocol. Following treatment, quantitative real-time PCR (qRT-PCR) was performed to detect the expression levels of the linear parent gene MYOCD, the general linear control GAPDH, and hsa_circ_0042103. The expression of each gene in the RNase R-treated group was compared to the control group, assuming equal starting amounts of total RNA. All experiments were independently repeated three times, with three parallel samples per group.

Fluorescence in situ hybridization (FISH)

Sequence-specific 6-FAM–labeled probes targeting the circ_0042103 back-splice junction were synthesized by Sangon Biotech (Shanghai, China). For cellular localization, hDPSCs grown on glass coverslips in 24-well plates were fixed in 4% paraformaldehyde, permeabilized with proteinase K, and dehydrated through an ethanol series. The coverslips were then hybridized with diluted circ_0042103 probes for 16 h at 37 °C. Following hybridization and subsequent washing steps, nuclei were counterstained with DAPI. Fluorescence images were acquired on a Leica SP8 confocal microscope (Leica, Germany) and captured under identical exposure settings across all conditions. Images were analyzed using ImageJ software, where RGB images were separated into individual channels and processed in grayscale. After optical density calibration post-inversion, cells were selected with the Freehand ROI tool to measure the total integrated fluorescence intensity of the 6-FAM channel (A). The nuclear area was defined as a region of interest (ROI) using the DAPI channel signal, and the nuclear fluorescence (B) was measured as the integrated intensity of the 6-FAM channel within this DAPI-defined nuclear ROI. Cytoplasmic fluorescence was calculated as C = A - B, and the nuclear-to-cytoplasmic (N/C) ratio was determined as N/C = B / (A - B). For each biological replicate, 8–10 randomly selected non-overlapping fields were analyzed, with at least 50 cells quantified per replicate. Data are presented as mean ± SD from n = 3 independent experiments, with P < 0.05 considered significant.

Immunofluorescence staining (IF)

hDPSCs were fixed in 4% paraformaldehyde for 15 min, washed with PBS, permeabilized with 0.5% Triton X-100 for 10 min at room temperature, and blocked with 5% BSA for 1 h. Cells were incubated overnight at 4 °C with primary antibody against TAF15 (1:100, Proteintech, China), washed, and then incubated for 1 h at room temperature with Coralite 555 conjugated goat anti-rabbit IgG (1:100, Proteintech, China). Nuclei were counterstained with DAPI (Beyotime, China). Images were acquired on a Leica confocal microscope. Biological replicates, n = 3.

RNA interference and overexpression

Small interfering RNAs (siRNAs) targeting circ_0042103 and TAF15, as well as overexpression plasmids (pcDNA3.1-circ_0042103 and pcDNA3.1-TAF15) and corresponding negative controls (NC), were designed and synthesized by GenePharma (Suzhou, China). Transfections were performed in hDPSCs using Lipofectamine 3000 (Invitrogen, USA) according to the manufacturer’s instructions; the P3000 reagent was used for plasmid transfection. For stimulation experiments, hDPSCs were treated with LPS (1 µg/mL) 5 h after vector transfection. siRNA sequences are provided in Supplementary Table S2.

RNA pulldown

A biotin-labeled probe for circ_0042103 was designed and produced by GenePharma (Suzhou, China). The binding of circ_0042103 to TAF15 was verified using an RNA pull-down kit (Bersin Bio, China) according to the manufacturer’s instructions.

Bioinformatics analysis

The CircAtlas, Starbase, and RBPmap databases were used to predict the possible RNA-binding proteins that may be involved in the molecular mechanism of circ_0042103. Predicted RNA-binding proteins were first subjected to GO/KEGG enrichment (FDR < 0.05). Candidate TAF15-interacting proteins were retrieved from starBase (accessed Jan 2024), yielding 5,439 putative TAF15-binding proteins. KEGG enrichment of these candidates identified 78 proteins mapping to DNA damage/repair pathways; this 78 genes set was used as the input for PPI analysis (full list in Supplementary Material 3). Networks were built in STRING using the default evidence channels with a medium confidence threshold (combined score ≥ 0.400). Edges were restricted to interactions among the 78 input proteins; disconnected nodes were removed. Figures were exported as vector graphics with node size = degree and edge width = STRING combined score.

Comet assay

DNA damage in human dental pulp stem cells was assessed by comet assay [29]. Slides were first precoated with 100 µL of 0.07% normal-melting-point agarose (NMA), covered with a coverslip, and allowed to solidify for 30 min at room temperature (RT). A PBS-resuspended cell suspension was mixed with 0.07% low-melting-point agarose (LMPA), and 70 µL of the mixture was spread onto the NMA-coated slide. After 15 min at RT, the coverslip was removed. Slides were then immersed in pre-cooled lysis buffer and incubated in the dark at 4 °C for 1.5 h. Next, slides were equilibrated in electrophoresis buffer at low temperature for 30 min, followed by electrophoresis at 25 V for 30 min in the dark. DNA was stained with 20 µL Nucleic Acid Red (Beyotime, Shanghai, China) and visualized using a fluorescence microscope (AMG EVOS). For each sample, ≥ 50 nuclei were analyzed with CaspLab Comet Assay software, and Olive tail moment (OTM) was used as the readout of DNA damage [30]. OTM reflects the extent of DNA migration (tail length × fraction of DNA in tail) rather than overall nuclear fluorescence intensity. Therefore, nuclei that appear brighter in the head region may still indicate low DNA damage if DNA migration into the tail is minimal. Tail DNA% yielded concordant results with OTM.

RNA stability assay

hDPSCs were transfected with si-TAF15(1), si-TAF15(2) or si-NC and cultured to 70–80% confluence. To block transcription, cells were treated with actinomycin D (ActD, 5 µg/mL; KKL, USA) and harvested at 0, 4, 8, and 12 h. Total RNA was isolated, reverse transcribed, and ERCC1 and PCNA mRNA was quantified by qRT-PCR. Values were normalized to the reference gene and expressed relative to 0 h. Biological replicates (n = 3).

Statistical analysis

Data were analyzed with SPSS version 21.0 (IBM, USA) and GraphPad Prism 9.0 (GraphPad Software, USA). For comet assays, at least 50 nuclei per biological replicate were scored and averaged; all hypothesis tests were performed at the replicate level to avoid pseudoreplication (Lovell and Omori, 2008). Olive tail moment (OTM) values were log transformed prior to analysis. Normality of replicate-level data was assessed with the Shapiro–Wilk test. Differences between two groups were evaluated using two-tailed Student’s t-tests. For experiments with more than two groups, one-way ANOVA followed by Dunnett’s post hoc test was used to compare each treatment with the control. Correlations were assessed by linear regression. Data are presented as mean ± SD from n = 3 independent experiments, and P < 0.05 was considered statistically significant.

Results

DNA damage accompanies inflammation in inflamed pulp tissues

We analyzed microarray data from six inflamed and six healthy dental pulps to determine whether DNA damage response (DDR) is activated in pulpitis. Differential expression analysis identified 752 upregulated and 227 downregulated genes in inflamed tissues (Fig. 1A). Functional enrichment of these upregulated genes revealed significant involvement in inflammatory signaling and DDR-related pathways (Fig. 1B). Furthermore, to examine the relationship within diseased tissues, we assessed γ-H2AX by Western blot across inflamed pulps (Fig. 1C) and correlated it with IL-6 measured from the same samples, observing a significant positive association (Fig. 1D), supporting a link between DNA damage and inflammation in pulpitis. Given the key immunomodulatory role of dental pulp stem cells (DPSCs) in tissue repair under inflammatory conditions, we next examined this population (Fig. 1E; clustering markers, Fig. S1). Single-cell RNA sequencing revealed a reduced proportion of DPSCs in inflamed tissues (Fig. 1F); however, these inflammatory DPSCs displayed higher inflammation and DNA damage scores compared to those from healthy samples (Fig. 1G), indicating that DDR is also activated in the inflammatory DPSC subset.

Fig. 1.

DNA damage accompanies inflammation in inflamed pulp tissues and inflamed DPSCs. A Volcano plot of differentially expressed genes (DEGs) in pulpitis versus healthy tissues (|log₂ FC| ≥ 1; p value < 0.01). B GO and KEGG pathway enrichment analysis of upregulated DEGs in pulpitis samples. C Expression levels of γ-H2AX protein in clinical dental pulp tissues. D Expression of γ-H2AX in clinical dental pulp tissues and its correlation with IL-6. (n = 10, r = 0.8155, p < 0.001). E UMAP visualization of 10 major cell subpopulations identified in healthy and Inflamed pulp tissues. F Proportional changes of identified cell populations between healthy and inflamed pulp tissues. Panels E-F share a single legend; the color scheme is identical across both panels. G Inflammation and DNA damage scores in DPSC subpopulations derived from healthy and inflamed pulp tissues. Data are shown as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001

DDR activation accompanies inflammation in LPS-Stimulated hDPSCs

In order to establish an in vitro inflammatory model, DPSCs were stimulated with LPS at concentrations of 0.5, 1, 5 or 10 µg/mL for different periods of time (Fig. S2). qRT-PCR and ELISA analyses showed that stimulation with 1 µg/mL LPS led to a time-dependent increase in IL6 and IL8 expression at both mRNA and protein levels, with peak expression observed at 5 h (Fig. 2A-D). Dose-dependent analysis confirmed that 1 µg/mL was the most effective concentration for cytokine induction (Fig. 2E-H), identifying 1 µg/mL for 5 h as the optimal condition for inflammation modeling in hDPSCs. Under these same conditions, we assessed the activation of DDR. Western blot analysis revealed elevated γ-H2AX expression, peaking at 5 h post-treatment with 1 µg/mL LPS (Fig. 2I-J). Comet assays showed consistent results, with maximal DNA fragmentation occurring at the same time point and concentration (Fig. 2K-L). These findings confirm that DDR is concomitantly activated in the LPS-induced inflammatory cell model. Accordingly, hDPSCs stimulated with 1 µg/mL LPS for 5 h were used for further mechanistic investigations.

Fig. 2.

DNA damage accompanies inflammation in inflamed DPSCs model. A, B IL6 and IL8 mRNA expression in DPSCs treated with LPS at different time points (1, 3, 5 h, 1 µg/mL). C, D Secreted protein levels of IL-6 and IL-8 in DPSCs treated with LPS at different time points (1, 3, 5 h, 1 µg/mL). E, F IL6 and IL8 mRNA expression in DPSCs treated with LPS at different concentrations (0.5, 1, 5, 10 µg/mL, 5 h). G, H Secreted protein levels of IL-6 and IL-8 in DPSCs treated with LPS at different concentrations (0.5, 1, 5, 10 µg/mL, 5 h). I Western blot analysis and quantification of γ-H2AX protein in DPSCs treated with LPS at different time points (1, 3, 5 h, 1 µg/mL). J Western blot analysis and quantification of γ-H2AX protein in DPSCs treated with LPS at different concentrations (0.5, 1, 5, 10 µg/mL, 5 h). K Comet assay images and quantification of tail moment in DPSCs treated with LPS at different time points (1, 3, 5 h, 1 µg/mL). Scale bars = 200 μm. L Comet assay images and quantification of tail moment in DPSCs treated with LPS at different concentrations (0.5, 1, 5, 10 µg/mL, 5 h). Scale bars = 200 μm. Data are shown as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001

circ_0042103 was upregulated in hDPSCs inflammatory model

circRNA-seq analysis was performed on DPSCs before and after stimulation (Fig. 3A). KEGG enrichment analysis of differentially expressed circRNAs revealed significant enrichment in pathways such as TNF, AGE-RAGE, and IL-17 signaling, which are associated with oxidative stress-induced DNA damage and repair regulation (Fig. 3B). qRT-PCR confirmed circ_0042103 was the most significantly upregulated circRNA among the top 10 candidates predicted by bioinformatics. (Fig. 3C). circ_0042103 is derived from exons 2–5 of the MYOCD gene via back-splicing, with its junction site validated by Sanger sequencing (Fig. 3D). Agarose gel electrophoresis using divergent and convergent primers showed that circ_0042103 was amplified only by divergent primers in cDNA, confirming its circular structure (Fig. 3E). RNase R digestion further demonstrated its high stability (Fig. 3F). Nucleocytoplasmic fractionation assays and FISH revealed that circ_0042103 is predominantly localized in the cytoplasm of inflamed DPSCs (Fig. 3G-H). Together, these results demonstrate that circ_0042103 is a highly stable, cytoplasm-localized, and significantly upregulated circRNA in LPS-induced inflammatory DPSCs.

Fig. 3.

circ_0042103 was upregulated in inflamed DPSCs model. A Volcano plot of differentially expressed circRNAs (DEcircs) in DPSCs treated with 1 µg/mL LPS for 5 h versus control (|log₂FC| ≥ 1; p value < 0.01). B KEGG pathway enrichment analysis of DEcircs. C The expression levels of the top 10 upregulated circRNAs by qRT-PCR. D Genomic location, back-splicing pattern, and Sanger sequencing validation of the head-to-tail junction of circ_0042103. E Agarose gel electrophoresis showed that the divergent primers amplified circ_0042103 in cDNA but not genomic DNA (gDNA). GAPDH as a common housekeeping linear transcript was used as a control. F Relative RNA expression of circ_0042103, its linear host gene MYOCD, and GAPDH in hDPSCs after RNase R treatment. G Detection of the localization of circ_0042103 in cells by qRT-PCR. U6 and GAPDH were used as nuclear and cytoplasmic controls, respectively. H FISH assay and quantification for the cytoplasmic localization of circ_0042103 (green) in hDPSCs. Nuclei were counterstained with DAPI (blue). Scale bars = 50 μm. Data are shown as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001

circ_0042103 promotes DNA damage and inflammation

To investigate the role of circ_0042103 in LPS-induced DNA damage and inflammation, two small interfering RNAs (si-1 and si-2) and an overexpression plasmid targeting circ_0042103 were constructed (Fig. 4A, B). Transfection efficiently modulated circ_0042103 levels without affecting its host gene, MYOCD (Fig. 4C). After knockdown, western blot analysis showed that γ-H2AX protein levels were significantly reduced in circ_0042103-silenced cells compared to controls (Fig. 4D, E). Comet assay results were consistent, revealing a marked decrease in tail moments following circ_0042103 knockdown (Fig. 4F). Meanwhile, ELISA results demonstrated a significant reduction in IL-6 and IL-8 protein secretion, although IL-6 mRNA levels, as assessed by qRT-PCR, showed no significant change (Fig. 4G-J). These results indicate that circ_0042103 knockdown alleviates LPS-induced DNA damage and IL-8-centered inflammatory response in DPSCs. Conversely, overexpression of circ_0042103 aggravates these effects in LPS-stimulated DPSCs (Fig. S3). Furthermore, in clinical pulp tissues, circRNA_0042103 expression showed a significant positive correlation with increasing DNA damage and exacerbated inflammatory responses (Fig. 4K, L).

Fig. 4.

circ_0042103 promotes DNA damage and inflammatory expression. A, B qRT-PCR analysis of circ_0042103 knockdown and overexpression efficiency. C MYOCD mRNA expression after circ_0042103 knockdown or overexpression. D, E Western blot and quantification of γ-H2AX in DPSCs after LPS stimulation and circ_0042103 knockdown. F Comet assay and tail moment quantification in DPSCs after LPS stimulation and circ_0042103 knockdown. Scale bars = 200 μm. G, H IL6 and IL8 mRNA levels in DPSCs after LPS treatment and circ_0042103 knockdown. I, J IL-6 and IL-8 secretion in DPSCs after LPS treatment and circ_0042103 knockdown. K, L Expression of circ_0042103 in clinical dental pulp tissues and its correlation with γ-H2AX (n = 10, r = 0.9083, p < 0.001) and IL-6. (n = 10, r = 0.8470, p < 0.001). Data are shown as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001

circ_0042103 binds TAF15 protein to regulate DNA damage and inflammation

circRNA–protein interactions can modulate protein expression and function [31]. Using CIRCAtlas, StarBase, and RBPmap, we identified 10 high-confidence RNA-binding proteins (RBPs) (Fig. 5A). GO and KEGG analyses indicated that these candidate RBPs are involved in inflammation and DNA damage/repair processes (Fig. 5B, C), with DGCR8, TAF15, FUS, and IGF2BP2 showing strong associations with DNA damage–related pathways. Notably, modulation of circ_0042103 had opposite effects on TAF15 protein levels: knockdown reduced TAF15, whereas overexpression increased it, while DGCR8, FUS, and IGF2BP2 remained unchanged (Fig. 5D-F). In contrast, qRT-PCR showed no significant change in TAF15 mRNA after either circ_0042103 knockdown or overexpression (Fig. 5G, H), indicating post-transcriptional regulation. RNA pull-down confirmed a direct interaction between circ_0042103 and TAF15 (Fig. 5I). Immunofluorescence staining further demonstrated that TAF15 was mainly distributed in the nucleus under basal conditions, whereas LPS stimulation markedly reduced its nuclear localization and promoted cytoplasmic accumulation. Notably, silencing circ_0042103 restored nuclear TAF15 levels (Fig. 5J), while overexpression of circ_0042103 further promoted the cytoplasmic accumulation of TAF15 compared with LPS + pcDNA group (Fig. S4). Moreover, nuclear–cytoplasmic fractionation followed by qRT-PCR revealed that circ_0042103 predominantly localized in the cytoplasm, whether affected by LPS stimulation or circ_0042103 knockdown (Fig. 5K). These findings suggest that circ_0042103 interacts with TAF15 and regulates its subcellular localization, thereby influencing DNA damage and inflammatory responses.

Fig. 5.

Identification and validation of TAF15 as a circ_0042103-interacting protein. A circ_0042103 downstream target proteins were predicted by CircAtlas, Starbase, and RBPmap databases. B, C Perform KEGG and GO analysis on the overlapping predicted target proteins. D-F Western blot and quantification of protein levels of candidate RBPs (DGCR8, FUS, TAF15, IGF2BP2) in DPSCs after circ_0042103 knockdown or overexpression. G-H Relative mRNA expression of DGCR8, FUS, TAF15, and IGF2BP2 in DPSCs after circ_0042103 knockdown or overexpression, determined by qRT-PCR. I The binding of circ_0042103 to TAF15 was detected by RNA pull down. J Immunofluorescence showing nuclear and cytoplasmic distribution of TAF15 (red) and nuclei (DAPI, blue) in NC, LPS, si-NC + LPS, and si-circ_0042103 + LPS groups. Scale bars = 20 μm. K The nuclear and cytoplasmic distribution of circ_0042103 in hDPSCs was determined by qRT-PCR. U6 and GAPDH were used as nuclear and cytoplasmic controls, respectively. Data are shown as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001

To further investigate TAF15’s role in LPS-induced DNA damage and inflammation, we established TAF15 knockdown and overexpression models in hDPSCs, with transfection efficiency verified at both mRNA and protein levels (Fig. 6A, B, D). Importantly, TAF15 modulation did not affect circ_0042103 expression (Fig. 6C). Following LPS stimulation (1 µg/mL, 5 h), western blot analysis revealed a significant increase in γ-H2AX expression in the TAF15 knockdown group compared to controls (Fig. 6D, E). qRT-PCR and ELISA showed that TAF15 knockdown did not alter IL-6 expression but significantly increased IL-8 levels (Fig. 6F-I). These findings suggest that TAF15 knockdown exacerbates LPS-induced DNA damage and IL-8–centered inflammatory response in hDPSCs. Conversely, TAF15 overexpression had the opposite effect, indicating a protective role against LPS-induced DNA damage and inflammation (Fig. 6J-O).

Fig. 6.

TAF15 inhibits DNA damage and inflammatory expression. A, B Transfection efficiency of knockdown and overexpression of TAF15 were detected by qRT-PCR. C circ_0042103 expression after TAF15 knockdown or overexpression. D, E Western blot and quantification of protein levels of γ-H2AX in DPSCs after LPS treatment and TAF15 knockdown. F, G IL6 and IL8 mRNA expression were assessed by qRT-PCR following LPS stimulation and TAF15 silencing. H, I IL6 and IL8 protein secretion levels were assessed by ELISA following LPS stimulation and TAF15 silencing. J, K Western blot and quantification of protein levels of γ-H2AX in DPSCs after LPS treatment and TAF15 overexpression. L-O The expression of IL6 and IL8 was assessed by qRT-PCR and ELISA following LPS stimulation and TAF15 overexpression. Data are shown as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001

NER suppression by circ_0042103–TAF15 axis enhances inflammation and DNA damage

Studies have shown that DNA damage tends to accumulate when DNA repair machinery is inhibited [32]. Interestingly, our GSEA analysis of pulpitis revealed significant suppression of the nucleotide excision repair (NER) pathway, a major DNA repair mechanism involved in the DDR, in inflamed pulp tissues (Fig. 7A). Protein–protein interaction (PPI) network analysis further indicated strong associations between TAF15 and core NER components (Fig. 7B), suggesting a potential role for TAF15 in NER regulation. Then we conducted western blot analysis in LPS-stimulated DPSCs to explore the functional role of the circ_0042103/TAF15 axis in the NER. ERCC1 and PCNA, key markers of the NER pathway, were used to evaluate DNA repair activity. Results showed that ERCC1 and PCNA expression was downregulated in both the circ_0042103 overexpression group and the TAF15 knockdown group, whereas expression was increased following circ_0042103 knockdown or TAF15 overexpression (Fig. 7C-F). Furthermore, qRT-PCR analysis further confirmed that TAF15 knockdown significantly reduced, whereas TAF15 overexpression increased, the mRNA levels of ERCC1 and PCNA (Fig. 7G–H). To determine whether TAF15 affects the stability of these transcripts, transcription was blocked using actinomycin D, and mRNA decay was analyzed over time. The results showed that the half-lives of ERCC1 and PCNA mRNAs were markedly shortened upon TAF15 knockdown (Fig. 7I). Given that circ_0042103 binds and functionally antagonizes TAF15, these data indicate that circ_0042103 suppresses NER by destabilizing ERCC1 and PCNA transcripts through TAF15 sequestration, which likely contributes to the accumulation of DNA damage and inflammatory responses in LPS-treated DPSCs.

Fig. 7.

NER suppression by circ_0042103/TAF15 axis. A GSEA of the NER pathway in pulpitis tissue. B Protein-protein interaction (PPI) network of potential TAF15-interacting proteins. C, D Western blot and quantification of NER proteins ERCC1 and PCNA in LPS-treated DPSCs after circ_0042103 knockdown or overexpression. E, F Western blot and quantification of NER proteins ERCC1 and PCNA in LPS-treated DPSCs after TAF15 knockdown or overexpression. G, H qRT-PCR analysis of ERCC1 and PCNA mRNA in LPS-stimulated hDPSCs following TAF15 knockdown (si-TAF15) or overexpression (OE-TAF15). I The ERCC1 and PCNA mRNA stability assay with ActD (n = 3). Data are shown as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001

Discussion

Pulpitis is a prevalent oral inflammatory disease characterized by immune activation and tissue damage in response to persistent bacterial infection [33]. Among the hallmarks of chronic inflammation is the overproduction of reactive oxygen species (ROS), which induce cellular stress and DNA damage [13]. Our study, consistent with previous findings by Huang et al. and Qiao et al., confirms the accumulation of DNA damage in inflamed dental pulp tissues, as evidenced by elevated γ-H2AX expression [14, 15]. Moreover, we observed a positive correlation between γ-H2AX and the proinflammatory cytokine IL-6, further supporting the interplay between DNA damage and inflammation in pulpitis. Dental pulp stem cells (DPSCs), due to their regenerative and immunomodulatory potential, are key players in pulp tissue homeostasis and response to injury. While mild inflammation has been reported to promote DPSC proliferation and differentiation, sustained or severe inflammation can result in stem cell dysfunction or apoptosis [34]. Our single-cell RNA sequencing revealed a reduced proportion of DPSCs in inflamed tissues, alongside upregulation of proinflammatory and DNA damage response (DDR)-related genes, suggesting both apoptosis and DDR activation in inflammatory DPSCs. In vitro, we used LPS-stimulated DPSCs, a widely adopted cellular model for studying pulpitis [35], to further investigate the relationship between inflammation and DNA damage. The results showed elevated levels of IL-6, IL-8, and γ-H2AX, confirming activation of the DDR in inflamed DPSCs.

Among the epigenetic regulators of inflammation and DDR, circular RNAs (circRNAs) have attracted increasing interest due to their stability and diverse regulatory roles [28, 36, 37]. Previous studies have shown that circ_0002970 and circ_0003552 respectively participate in inflammatory and DNA damage responses [38, 39]. Our earlier work identified circ_0002456 as a negative regulator of inflammation and DDR in DPSCs [16]. In the present study, we identified circ_0042103 as the most significantly upregulated circRNA in LPS-induced DPSCs. Its overexpression promoted IL-6/IL-8 expression and exacerbated DNA damage, while knockdown reduced IL-8 secretion but had limited impact on IL-6 mRNA levels. This phenomenon may be explained by distinct regulatory architectures of the IL-8 and IL-6 genes. IL-8 has an NF-κB-prioritized promoter, which reportedly renders it rapidly and robustly responsive to inflammatory inputs [40]. In contrast, IL-6 induction exhibits a more complex, threshold-like behavior linked to endotoxin tolerance, often mediated through IL-6R/JAK–STAT signaling [41–43]. Under our acute, single-dose LPS conditions, circ_0042103 knockdown likely promotes an ET-like hyporesponsive state that blunts IL-6 induction [44], while circ_0042103 overexpression intensifies inflammatory signaling, raising the ET threshold and thereby increasing IL-6 [45]. Consistently, inflamed pulp tissues show elevated circ_0042103 with positive correlations to IL-6 and γ-H2AX.

Mechanistically, we found that circ_0042103 exerts its function via direct interaction with TAF15, a member of the FET family of RNA-binding proteins [46]. TAF15 has been implicated in transcription, RNA splicing, and DNA repair processes [47–51]. RNA pull-down assays confirmed the physical interaction between circ_0042103 and TAF15, and circ_0042103 reduced TAF15 protein without altering its mRNA, indicating post-transcriptional regulation. TAF15 possesses intrinsically disordered domains that enable its phase separation into membraneless condensates at DNA damage sites, where it organizes the DNA repair machinery [50]. This process requires TAF15, which is primarily localized in the nucleus. However, TAF15 can shuttle between the nucleus, cytoplasm, and cell surface [49]. Consistent with previous reports, we observed that TAF15 is predominantly nuclear under basal conditions, whereas LPS stimulation drives its translocation to the cytoplasm. Notably, silencing circ_0042103 restores nuclear TAF15, thereby attenuating inflammation and enhancing DNA damage repair. These findings suggest that in LPS-stimulated hDPSCs, circ_0042103 disrupts the DNA damage response by sequestering TAF15 in the cytoplasm, reducing its nuclear availability for the assembly of repair complexes.

Our GSEA analysis also revealed significant downregulation of the nucleotide excision repair (NER) pathway in inflamed pulp tissues, and PPI network analysis identified strong associations between TAF15 and core NER components. Based on these findings, we focused on the NER pathway, which is essential for removing bulky DNA adducts and helix-distorting lesions [52–54]. ERCC1 and PCNA, key NER proteins involved in DNA excision and resynthesis respectively, were used as markers to evaluate NER activity in DPSCs [55]. Our results demonstrated that in LPS-stimulated DPSCs, the circ_0042103/TAF15 axis suppressed NER at the protein level: circ_0042103 overexpression or TAF15 knockdown reduced ERCC1 and PCNA, whereas circ_0042103 silencing or TAF15 overexpression restored their levels. Moreover, actinomycin D chase assays demonstrated a post-transcriptional mechanism whereby TAF15 stabilizes these mRNAs: TAF15 knockdown shortened the half-lives of ERCC1 and PCNA. Together, these findings support that circ_0042103 impairs NER by reducing nuclear TAF15 and by diminishing TAF15-dependent stabilization of ERCC1 and PCNA mRNAs, thereby sustaining DNA damage and amplifying inflammation in LPS-treated DPSCs.

This study used an LPS-stimulated hDPSC model. However, in vitro systems cannot fully recapitulate the native pulp microenvironment, including three-dimensional tissue architecture, vascularization, innervation, and immune–stromal crosstalk. Future studies should employ in vivo pulpitis models to validate the role of circ_0042103 in disease progression. Furthermore, IL-6 induction was less responsive to circ_0042103 than IL-8; future work will quantify NF-κB and p38 MAPK activation kinetics and implement time-course experiments with repeated LPS stimulation to delineate the mechanism.

Conclusions

In summary, circ_0042103 binds TAF15 and restricts its nuclear import, thereby impairing TAF15-mediated nucleotide excision repair, promoting DNA damage, and exacerbating the LPS-induced inflammatory response of hDPSCs. These data implicate the circ_0042103/TAF15/NER axis in pulpitis progression and highlight circ_0042103 as a potential therapeutic target (Fig. 8).

Fig. 8.

Mechanistic diagram of circ_0042103/TAF15 regulation of inflammation and DNA damage in hDPSCs (Created in BioRender. https://BioRender.com/6ai51mm)

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

DPSCs

Dental pulp stem cells

circRNA

Circular RNA

miRNA

MicroRNA

qRT-PCR

Quantitative real-time polymerase chain reaction

γ-H2AX

Phosphorylation of the Ser-139 residue of the histone variant H2AX

LPS

Lipopolysaccharide

DSBs

DNA double-strand breaks

DDR

DNA damage response

ROS

Reactive oxygen species

DGCR8

DGCR8 microprocessor complex subunit

TAF15

TATA-box binding protein associated factor 15

IGF2BP2

Insulin like growth factor 2 mRNA binding protein 2

FUS

FUS RNA binding protein

NER

Nucleotide excision repair

Author contributions

FL and JKZ contributed equally to this work and share first authorship. QZJ designed and conceived the study. FL and LCL critically performed the experiments. FL, LCL, HXD, JKZ and MZL analyzed the data. JKZ, JLZ, YL, WC, XC, YHL and BTW performed the statistical analysis. FL and JKZ drafted the manuscript. QZJ, JKZ, LL, and YGJ critically revised the manuscript. QZJ and YL provided funding. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (no. 82270966), and Guangzhou Municipal Science and Technology Bureau, Basic and Applied Basic Special Research Project (Youth Doctoral “Voyage” Project, 2024A04J3387).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

All experiments involving cells derived from human participants were reviewed and approved by the Institutional Review Board of the Hospital of Stomatology, Guangzhou Medical University (Guangzhou, China) (Approval No. JCYJ2023005; Approved project title: ‘Function and mechanism of circ_dsir (circ_dental stem cell inflammatory related) regulating DNA damage response in pulpitis’; Approval date: August 20, 2023).

Consent for publication

Written informed consent for participation and for the use of biological samples was obtained from all participants.

Competing interests

The authors declare no competing interests.