All animal experiments were conducted in accordance with the guidelines established by the Institutional Committees for Animal Experiments at Tokyo Medical and Dental University (currently the Institute of Science, Tokyo, Japan), and all experimental protocols were approved by these committees (#A2023-196C3).

C57BL/6 JJc1 mice (male, 6 weeks old, n = 9) were obtained from CLEA Japan, Inc. (Tokyo, Japan). The protocol for inducing AP has previously been described (20). In brief, the animals were anesthetized via an intraperitoneal injection of ketamine (75 mg/kg) and dexmedetomidine (1 mg/kg) and mounted on a handmade mouth-opening apparatus. Cavity preparation was performed to expose the dental pulp on the left and right mandibular first molars using an electric handpiece (Vivamate G5, Nakanishi, Kanuma, Japan) with a 1/4 round carbide bur (#14820, SS White Dental, Lakewood, NJ, USA) under a stereomicroscope (Zeiss, Oberkochen, Germany). Subsequently, the coronal pulp was removed with a stainless steel K-file #8 (Dentsply Maillefer, Ballaigues, Switzerland) into the mesial and distal canals, and the exposure was left open. Pulp-exposed mice were randomly assigned to the MET administration and phosphate-buffered saline (PBS) control groups using a computer-generated randomization list by an independent researcher blinded to the study protocol (n = 3 in each group). Given its established safety profile and suitability for long-term administration, MET was administered intraperitoneally at a dose of 50 mg/kg/day in this study to investigate its therapeutic effects on AP. This dosage and route of administration were selected based on previous in vivo studies demonstrating their efficacy and relevance in evaluating the pharmacological effects of MET in rodent models (21). The mice received MET in 100 μl PBS or the same volume of PBS intraperitoneally daily for 28 days, starting 1 day before the surgery. The mandible of each mouse was scanned under anesthesia on postoperative days 7, 14, 21, and 28 using in vivo micro-computed tomography (micro-CT; inspeXio SMX-100CT, Shimadzu, Tokyo, Japan; Figure 1A ) with settings of 70 kV, 140 μA, and a 9 μm voxel size. The same animals were euthanized with carbon dioxide on postoperative day 28, and the mandibles were isolated for histological evaluation.

For the negative control, non-treated mice (n = 3) were sacrificed, and their mandibles were subjected to micro-CT to assess the periodontal ligament space in healthy teeth. The volume of apical bone resorption was calculated as previously described (22). In brief, the region of interest was set as the osteolytic and periodontal spaces around the distal root of the first mandibular molar. A blinded observer manually delineated the contours of these regions using CT-analyzer software (Amira, Thermo Fisher Scientific, Waltham, MA, USA) to reconstruct the volume of the periapical lesions. Subsequently, an automated threshold script was used to measure the total volume of bone loss. The volume of true bone loss was determined by subtracting the space of the periodontal ligament in healthy teeth from the volume of total bone loss.

The mandibular samples were fixed in a 4% paraformaldehyde solution at 4°C for 24 h and decalcified with 17% ethylenediaminetetraacetic acid at 4°C for 4 weeks. The tissues were placed in 15% sucrose in PBS for 12 h, transferred to 30% sucrose in PBS until the tissues sank, and embedded in an optimal cutting temperature compound. Frozen sections (5 μm thick) were obtained using a cryostat (Leica CM3050 S, Nussloch, Germany) and subjected to hematoxylin and eosin and to tartrate-resistant acid phosphatase (TRAP) staining. TRAP staining was performed as follows: The TRAP basic incubation medium was prepared by dissolving sodium acetate anhydrous (9.2 g; S-2889, Sigma-Aldrich, Burlington, MA, USA) and L-(+)-tartaric acid (11.4 g; T-6521, Sigma-Aldrich) in 950 mL distilled water, followed by adding 2.8 mL glacial acetic acid. The pH was adjusted to 4.7–5.0 using 5 M sodium hydroxide, and the final volume was made up to 1 L using distilled water. Next, naphthol AS-MX phosphate substrate mix was prepared by dissolving naphthol AS-MX phosphate (20 mg; N-4875, Sigma-Aldrich) in ethylene glycol monoethyl ether (1 mL; E-2632, Sigma-Aldrich) followed by thorough mixing. The working TRAP staining solution was freshly prepared by combining 200 mL of TRAP basic incubation medium, Fast Red Violet LB Salt (120 mg; F-3381, Sigma-Aldrich), and 1 mL naphthol AS-MX phosphate substrate mix, followed by thorough mixing before use. After incubation with TRAP staining solution, the sections were counterstained with 0.08% Fast Green (Santa Cruz Animal Health, Dallas, TX, USA), rinsed with distilled water, air-dried, and mounted in xylene. Images were captured with a light microscope (Nikon ECLIPSE Ci, Tokyo, Japan) and subjected to quantitative analysis using image processing software (ImageJ v2; National Institutes of Health, Bethesda, MD, USA). Tissue sections were examined to identify the lesion area, specifically the periapical region surrounding the distal root of the mandibular first molar. The mandibular first molar was centered in the field of view of the microscope, and the entire periapical region was systematically scanned. TRAP-positive cells were manually counted across the entire root apex. In brief, for the alveolar bone surface, the length along the alveolar bone margin was measured in millimeters (mm), and the number of TRAP-positive cells per unit length (/mm) was quantified. The apical periodontitis lesion area was measured, and the number of TRAP-positive cells (/mm²) was calculated (n = 6 in the PBS group and 5 in the MET group).

RAW264.7 cells, a mouse macrophage cell line, were obtained from the Cell Engineering Division, RIKEN BioResource Research Center (Tsukuba, Japan; Cell No. RCB0535). The cells were cultured in Dulbecco’s modified Eagle’s medium (FUJIFILM Wako Chemicals, Tokyo, Japan) supplemented with heat-inactivated 10% fetal bovine serum (Thermo Fisher Scientific, #1600-500, Waltham, MA, USA) and 1% penicillin-streptomycin (FUJIFILM Wako Chemicals) at 37°C with 95% oxygen and 5% carbon dioxide. The following drugs and reagents were used in this study: metformin hydrochloride (138-18661: FUJIFILM Wako Pure Chemicals), lipopolysaccharide (Escherichia coli O111:B4; Sigma-Aldrich), compound C (171260; Merck Millipore, Burlington, MA, USA), MHY1485 (S7811; Selleck Chemicals, Houston, TX, USA), and BAY 11-7085 (196309-76-9; Cayman Chemical, Ann Arbor, MI, USA).

RAW264.7 macrophages were plated at 2 × 10⁵ cells per well in 12-well plates and allowed to adhere overnight in DMEM. Cells were then pre-treated with MET (0.5 mM) for 6 h, followed by stimulation with LPS (100 ng/mL, 4 h). To activate mTOR, the agonist MHY1485 (MHY, 2 µM) was added for 3 h after MET treatment. To inhibit AMPK, Compound C (CC, 5 µM) was applied for 1 h following MET treatment, and the cells were subsequently stimulated with LPS (100 ng/mL, 4 h). Experimental groups therefore comprised (i) untreated control, (ii) LPS alone, (iii) MET + LPS, (iv) MET + CC + LPS, and (v) MET + MHY. The same treatment schedule was applied for all downstream assays, including RNA-seq, quantitative PCR, western blotting and luciferase reporter analysis.

The effect of MET on cell viability was determined using the Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan), as previously described (23). In brief, RAW264.7 cells were plated in 96-well plates at a density of 7 × 10³ cells/well and incubated overnight. The cells were treated with 0, 0.5, or 1 mM of MET. Cell proliferation was measured at 0, 24, 36, and 72 h using a spectrophotometer (Sunrise, Tecan, Männedorf, Switzerland).

Total RNA was extracted using the QuickGene RNA Cultured Cell Kit S (FUJIFILM Wako Chemicals), and 300 ng/mL of RNA was subjected to cDNA synthesis using PrimeScript™ RT Master Mix (Takara Bio, Kusatsu, Japan). Real-time quantitative polymerase chain reaction (qPCR) was performed using the specific primers outlined in Table 1 and the GoTaq qPCR Master Mix (Promega, Madison, WI). The amplification process was performed on the CFX96 Real-Time qPCR System (Bio-Rad, Kidlington, UK). The gene expression was calculated using the formula 2^−ΔΔCt with β-actin as an internal control.

RNA sequencing was performed by Rhelixa Co. (Tokyo, Japan). Quality control and alignment procedures were applied to FASTQ data to generate an expression matrix. This process entailed using fastp software on a Linux server for data quality control, followed by alignment of the FASTQ reads to the mouse mm10 reference genome using hisat2 software. Quantification and normalization were performed using feature counts and DESeq2, respectively. Principal component analysis was conducted and visualized using the prcomp function in the R package. The plots and graphs generated from the RNA-sequencing analysis were visualized using the R software, version 4.3.3 (R-Project, Vienna, Austria). Gene set enrichment analysis (http://software.broadinstitute.org/gsea/index.jsp) was performed using the hallmark gene sets.

The cells were lysed using radioimmunoprecipitation assay buffer (Merck Millipore) supplemented with protease and phosphatase inhibitor cocktails (cOmplete and PhosSTOP; Sigma-Aldrich) at 4°C for 15 min. Protein samples were heated to boiling temperature for 5 min in 1× sodium dodecyl sulfate buffer lysates. The proteins were separated by 10–20% e-PAGEL (ATTO, Tokyo, Japan) and transferred onto a polyvinylidene difluoride membrane (Merck Millipore) by semi-dry transfer at 0.15 mA for 1 h using a blotting system (WSE-4040, ATTO). The antibodies used in this study included rabbit monoclonal antibodies (Cell Signaling Technology, Danvers, MA) against p65/RELA (1:1,000; D14E12), phospho-NF-κB p65 (1:1,000; 93H1), phospho-AMPKα (Thr172) (1:2000; D79.5E), AMPK (1:1,000; D5A2), mTOR (1:1,000; 7C10, Cell Signaling Technology), and p-mTOR (1:1,000; D9C2),. Horseradish peroxidase-conjugated anti-glyceraldehyde-3-phosphate dehydrogenase (1:4,000; M171-7, MBL Life Science, Tokyo, Japan) and horseradish peroxidase-labeled anti-rabbit IgG (1:5000, W4011, Promega) were also used. After applying a chemiluminescent substrate (Immobilon, Merck Millipore), protein bands were detected using the LAS-3000 mini-imaging system (FUJIFILM Wako Chemicals). Protein expression levels were measured using the ImageJ software v2 (National Institutes of Health).

The NF-κB signaling activity was evaluated using the pGL4.32 (luc2P/NF-κB-RE/Hygro, Promega) vector, which contains five copies of the NF-κB response element. RAW264.7 cells were transfected with the reporter vector using the Lipofectamine 3000 transfection reagent from Thermo Fisher Scientific. The cells were lysed using a luciferase cell culture lysis reagent (#E1483, Promega). Subsequently, the activity of the light-producing enzyme luciferase was measured using a luciferase assay reagent (#E1483, Promega) and a luminometer (Luminescence PSN, ATTO).

Statistical analyses were performed using the Statistical Package for Social Sciences (version 9.10; SPSS Inc., Chicago, IL, USA) and GraphPad Prism (version 9; GraphPad Software Inc., La Jolla, CA, USA). Normality distributions were assessed using the Shapiro–Wilk test, and Levene’s test was used for variance equality comparisons. For normally distributed variables with equal variance, the unpaired Student’s t-test was applied for comparisons between two groups, and a one-way analysis of variance followed by Tukey’s post hoc test was used for comparisons among three or more groups. For experiments involving two independent variables, a two-way analysis of variance was used to assess the main effects and interactions between factors, followed by Tukey’s multiple comparisons test when appropriate. When the normality assumption was met, but the variances were uneven, the unpaired Welch’s t-test was used for two-group comparisons, and the Brown-Forsythe/Welch one-way analysis of variance with Dunnett’s T3 post hoc test was used for three or more groups. For data with non-normal distribution, the Mann–Whitney U test was employed for two-group comparisons, and comparisons among three or more groups were conducted using the Kruskal–Wallis test with Dunn’s post hoc test. Statistical significance was set at p < 0.05.

The effect of MET on AP development was evaluated using a mouse AP model induced by making unsealed surgical pulp exposure in mandibular first molars. The mice received daily intraperitoneal injections of MET, starting a day before AP induction and continuing until sacrifice ( Figure 1A ). In the control group, periapical bone loss increased progressively over time, with significant bone resorption observed on postoperative days 21 and 28 ( Supplementary Figure S1A ). In contrast, bone resorption was notably suppressed in the MET group on postoperative days 21 and 28 ( Figures 1B, C ). Histological analysis of the periapical lesions showed that inflammatory cell infiltration was suppressed in the MET group compared to the control group ( Figure 2A ). In addition, the MET group showed a significantly lower number of cells positive to TRAP on postoperative day 28 than did the control group ( Figures 2B, C ).

MET was applied to LPS-stimulated RAW264.7 cells to investigate the anti-inflammatory effects of MET on pro-inflammatory macrophages. Initially, the cytotoxicity of MET was assessed to determine the appropriate concentration for use. Cell viability was not affected by 0.5 mM MET. However, 1.0 mM MET notably reduced cell viability at 72 h ( Figure 3A ). Based on these results, a concentration of 0.5 mM was selected for subsequent in vitro experiments. Real time-quantitative polymerase chain reaction revealed that MET markedly reduced the mRNA expression of pro-inflammatory cytokines (Il1a, Il1b, Il6, Tnfa, and Monocyte chemotactic protein1(Mcp1)) in LPS-stimulated RAW264.7 cells ( Figure 3B ). Furthermore, phosphorylated-p65 (p-p65) expression, which peaked at 30 min in LPS-stimulated RAW264.7 cells ( Supplementary Figures S2A, B ) was markedly reduced by 0.5 mM MET ( Figures 3C, D ). LPS-induced NF-κB activity was markedly suppressed by MET treatment ( Supplementary Figure S2C ; Figure 3E ).

A comprehensive genetic analysis was performed on the control and MET-treated RAW264.7 cells using next-generation sequencing to elucidate the mechanism by which MET inhibits NF-κB signaling. The principal component analysis scatter plot showed a distinct gene expression profile between the control and MET-treated samples, with a 27.6% proportion of variance in principal component (PC) 1 ( Figure 4A ). Gene set enrichment analysis using hallmark gene sets was conducted to assess the differences in mRNA expression levels between the control and MET-treated RAW264.7 cells ( Figure 4B ). A false discovery rate q-value of < 0.05 is shown in the gene sets in Table 2 . Of these, the mammalian target of rapamycin complex 1 (mTORC1), a signaling pathway associated with inflammation (24), was significantly downregulated in the MET group (P < 0.05), suggesting that MET exerted anti-inflammatory effects on mTOR signaling under the current experimental conditions.

MET significantly reduced the expression of phosphorylated mTOR (p-mTOR) in LPS-stimulated RAW264.7 cells ( Figures 4C, D ). MHY1485 (MHY, 2 µM), an agonist of mTOR, upregulated the expression of p-mTOR ( Figures 4E, F ) and the mRNA expression of pro-inflammatory cytokines (Il1α, Mcp1, and Tnfa), which were notably downregulated by MET treatment ( Figure 4G ). Furthermore, MET markedly downregulated MHY-induced expression of p-p65 ( Figures 4H, I ) and the activity of NF-κB ( Figure 4J ). Next, the effect of BAY11-7085 (BAY, 2 µM), an inhibitor of IκBα phosphorylation was examined (25), because LPS can activate NF-κB signaling through the mTOR pathway (26). BAY11–7085 markedly reduced the mRNA expression of LPS-induced pro-inflammatory cytokines (Il1a, Il1b, and Mcp1) and NF-κB activity, which were restored by MHY ( Supplementary Figures S3A, B ). These findings indicate that MET abrogates NF-κB activation through the mTOR pathway, which is critical in driving inflammatory mediator synthesis in RAW264.7 cells stimulated with LPS.

The precise mechanism by which MET downregulates mTOR-mediated NF-κB activation remains unclear. Given that MET activates AMPK signaling by inhibiting mitochondrial complex I (27), the mechanism by which MET-induced AMPK activation modulates mTOR signaling in RAW264.7 cells was examined. Treating these cells with MET notably enhanced AMPK phosphorylation, and this effect was not influenced by LPS stimulation ( Figures 5A, B ). MET-induced AMPK activation was markedly suppressed in the presence of compound C (CC, 5 μM), an AMPK inhibitor ( Figures 5C, D ). Next-generation sequencing analysis comparing MET-treated RAW264.7 cells and those treated with MET and CC combination revealed distinct gene expression profiles between the two groups (the PC1 accounted for 31.5% of the total variance, Figure 5E ). Notably, gene set enrichment analysis showed the upregulation of mTORC1 signaling in the MET and CC co-treated group compared with that in the MET-treated group ( Figure 5F and Table 3 ), suggesting that MET-mediated suppression of mTORC1 signaling was attenuated by CC. Western blotting confirmed that CC treatment restored p-mTOR expression that was suppressed by MET ( Figures 5G, H ). In addition, CC attenuated the MET-induced suppression of mRNA expression of pro-inflammatory cytokines Il1a, Il1b, IL6, and Tnfa in LPS-stimulated RAW264.7 cells ( Figure 5I ). CC also counteracted the inhibitory effects of MET on NF-κB signaling in LPS-stimulated RAW264.7 cells ( Figures 5J, K ). Similarly, CC markedly downregulated the inhibitory effect of MET on LPS-induced NF-κB signaling ( Figure 5L ). These results indicate that MET-induced AMPK activation suppresses NF-κB signaling by inhibiting the mTOR pathway, exerting an anti-inflammatory effect that may contribute to attenuating periapical bone destruction ( Figure 6 ).