H3K18la in periodontal ligament fibroblasts regulates immune niches and alveolar bone remodeling under mechanical loads

Abstract

The metabolism and remodeling of alveolar bone are highly active, which dynamically respond to mechanical force. However, the mechanism of how alveolar bone responds to mechanical force remains elusive. Orthodontics is essentially a physiological process of mechanical force–driven alveolar bone remodeling. Here, we demonstrated that orthodontic tension promoted the expression of glycolysis-related genes in periodontal ligament (PDL) fibroblasts. The mechanoactivated glycolysis subsequently enhanced lactylation on histone H3 lysine 18 (H3K18la) in PDL fibroblasts. CUT&Tag analysis revealed that H3K18la was enriched at the promoter region of CXCL1 to upregulate its transcription. 3PO inhibited glycolysis and attenuated stretching force–induced H3K18la and CXCL1 levels in PDL fibroblasts. CXCL1 subsequently augmented oxidative phosphorylation in macrophages and promoted M2 macrophage polarization. Additionally, in vivo inhibition of glycolysis by periodontal injection of 3PO disrupted the CXCL1-mediated CXCR2⁺ macrophage infiltration and inhibited bone formation during orthodontic tooth movement. Furthermore, the blockade of the CXCL1/CXCR2 axis in vivo, using Cxcr2 knockout (Cxcr2^–/^–) mice, impaired tension–induced M2 macrophage polarization in PDL, disrupting the pro-resolving microenvironment conducive to bone formation during orthodontic tooth movement. Overall, the results reveal that the tension–induced glycolysis–H3K18la axis in PDL fibroblasts plays a critical role in creating the pro-resolving periodontal immune niche and facilitating alveolar bone formation during orthodontic tooth movement.

Introduction

Bone architecture adapts to mechanical force according to Wolff’s law.^1,2 Physical loading stimuli, like proper intensity of exercise and muscular activity, will enhance bone mass and bone strength. Alternatively, lack of mechanical stimuli will lead to loss of bone mass and bone strength.³ Therefore, mechanical force plays an essential role in maintaining bone quality and quantity. Mechanical force induces bone remodeling, which involves the interplay of bone formation and bone resorption. Among the skeletal system, the metabolism and remodeling of alveolar bone are highly active due to mechanical stimuli from occlusion.⁴ However, the mechanisms by which mechanical force regulates alveolar bone remodeling remain unclear.

Orthodontics, the main treatment to improve tooth alignment and occlusion, is essentially a physiological process of mechanical force–driven alveolar bone remodeling.⁵ It has been reported that the aseptic immune responses play a critical role in the initiation of alveolar bone remodeling in OTM, which coordinates bone resorption and deposition through the synergistic action of multiple cell types within periodontal microenvironment.^6,7 However, the underlying mechanisms of orthodontic force–induced immune responses and bone remodeling remain obscure.

Periodontal fibroblasts are emerging as key immune sentinel cells that orchestrate periodontal immune responses to modulate alveolar bone remodeling. Numerous studies have highlighted the critical role of periodontal fibroblasts in bacteria-induced inflammation and alveolar bone resorption.^8–11 As the primary mechanosensing cells during orthodontic tooth movement (OTM), periodontal ligament (PDL) fibroblasts also serve as key regulators of mechanical force–induced immune responses by producing inflammatory mediators to initiate bone remodeling.¹² Nevertheless, it remains unclear how PDL fibroblasts sense mechanical forces to induce aseptic inflammation during orthodontic alveolar bone remodeling.

Mechanical stimuli are essential modulators of cellular metabolism.¹³ Recent studies have reported the mechanical regulation of glycolysis to meet the cellular energy demands.^14–16 As the end product of glycolysis, endogenous lactate is the key determinant of histone modification by attaching a lactyl group to lysine residues (referred to as “histone lactylation”), bridging cellular metabolism to epigenetic regulation.^17–19 However, the link between mechanical forces and cellular metabolic–epigenic alterations in PDL fibroblasts remains poorly understood.

Herein, we applied an OTM model in vivo and a mechanical force-loading model in vitro to study the role of PDL fibroblasts in sensing mechanical force and regulating alveolar bone remodeling. We found that tensile force induced metabolic-epigenic reprogramming in PDL fibroblasts during OTM. Tensile force activated glycolysis in PDL fibroblasts and subsequently enhanced lactylation on histone H3 lysine 18 (H3K18la), thereby promoting CXCL1 transcription and secretion, modulating periodontal immune niches and alveolar bone formation. These findings provide novel insights into the mechanisms underlying force–induced immune responses and alveolar bone formation during OTM.

Results

Orthodontic force induces glycolysis in PDL fibroblasts

To investigate the metabolic changes in PDL fibroblasts induced by orthodontic forces, we analyzed single-cell RNA sequencing (scRNA-seq) data of periodontal tissues around the maxillary first molars in mice from sham-operated group (CON) and orthodontic force–treated group (OTM). The gingiva was removed before the collection of single cells to avoid interference of gingival fibroblasts (Fig. 1a). Through the expression pattern of classic cell-type–specific markers, the PDL fibroblast cluster (Lum, Dcn, Col1α1, Postn, Ncam1) was detected and compared between the two groups, comprising 10141 fibroblasts from the CON group and 13865 fibroblasts from the OTM group (Fig. 1b and Fig. S1a, b). Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) revealed enrichment of glycolysis-associated terms in the OTM group compared with the CON group in PDL fibroblasts (Fig. 1c). OTM treatment also upregulated glycolysis-related genes in the PDL fibroblast cluster, such as Pkm, Pfkfb3, Pfkp, Hk2, Pfkm, and Eno1 (Fig. 1d and Fig. S1c). Notably, cell type-specific DEG analysis between the OTM and CON groups revealed prominent enrichment of glycolysis-related GO biological processes in fibroblasts (Fig. S2a). Consistently, representative glycolytic genes showed increased expression in fibroblasts after OTM (Fig. S2b). Consistently, immunofluorescence staining of PDL on the tension side showed significantly increased levels of pyruvate kinase M2 (PKM2) and 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase 3 (PFKFB3) in the OTM group (Fig. 1e, f). In addition, the expression levels of PKM2 and PFKFB3 in PDL on the tension side were higher than that on the compression side (Fig. S1d).

Fig. 1.

Orthodontic force induces glycolysis in periodontal ligament (PDL) fibroblasts during orthodontic tooth movement (OTM). a Flowchart of single-cell RNA sequencing (scRNA-seq) analysis of periodontal tissues during OTM in mice. Gingiva was removed and discarded before single-cell collection. b Uniform Manifold Approximation and Projection (UMAP) visualization of cells from periodontal tissues without gingiva in two groups, sham-operated group (CON, n = 63 385 cells) and 7-day OTM group (n = 16 060 cells). Cell-type annotations were distinguished by different colors, with PDL fibroblasts displayed in red in two groups, sham-operated group (CON, n = 10141 fibroblasts) and 7-day OTM group (n = 13 865 fibroblasts). c Glycolysis-related Gene Ontology (GO) biological process terms of differentially expressed genes (DEGs) enriched in PDL fibroblasts from the OTM group. DEGs were identified using a threshold of P < 0.05 and |log₂(fold change) | > 0.5. The bar chart displays −log₁₀(P-value). d UMAP visualization of PDL fibroblasts and the expression distribution of glycolysis-related genes projected onto the UMAP atlas, along with violin plots of gene expression. e Immunofluorescence staining and quantitative analysis showing the expression of glycolysis-related enzymes, PKM2 and PFKFB3, in the PDL region on the tension side in the CON group and 7-day OTM group (n ≥ 5). Scale bars: 40 μm. AB alveolar bone; R root. Data represent the means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001

Tension-activated glycolysis induces histone lactylation in PDL fibroblasts

To validate the enhanced glycolysis in PDL fibroblasts induced by orthodontic tensile force, we tested the glycolytic capacity of human PDL fibroblasts after cyclic stretch in vitro. The glycolytic proton efflux rates (glycoPER) in real-time were monitored by the extrusion of protons, which were produced by lactate and applied to measure glycolysis. Cyclic stretch applied for 24 h enhanced both basal and compensatory glycolytic rates (Fig. 2a–c). Western blot analysis confirmed gradual upregulation of several key glycolytic enzymes, including PKM2, PFKFB3, and phosphofructokinase, muscle type (PFKM), with the extension of the loading time of cyclic stretch (Fig. 2d, e). Increased expression levels of the glycolytic enzymes were also indicated by immunofluorescent staining in cells treated by stretch for 24 h (Fig. S3a–c). Moreover, lactate accumulation was increased in the PDL fibroblasts after 24 h of cyclic stretch (Fig. 2f).

Fig. 2.

Tension-activated glycolysis induces histone lactylation in human periodontal ligament (PDL) fibroblasts. a Real-time measurements of glycolytic proton efflux rates (glycoPERs) of human PDL fibroblasts measured by a XF Seahorse Analyzer. The glycoPERs were normalized to the total cell number in each group, including control group (CON) and 24-h cyclic stretching force group (Stretch). The final concentrations of rotenone/antimycin A (Rot/AA) and 2-deoxy-D-Glucose (2-DG) were 0.5 μmol/L and 50 mmol/L, respectively. b, c Normalized glycoPERs related to basal glycolysis and compensatory glycolysis (n = 3). d, e Representative immunoblots and quantitative abundance of glycolytic enzymes (PKM2, PFKFB3, PFKM) in human PDL fibroblasts treated with different loading time of cyclic stretching force (n = 3). f Lactate levels of human PDL fibroblasts following 24 h of cyclic stretching force treatment (n = 3). g–i Immunoblots and quantitative data of Pan Kla, H3K18la, and H3K9la protein levels in human PDL fibroblasts treated with different loading time of cyclic stretching force (n = 3). The quantitative data for Pan Kla correspond to the bands appearing near 17 kD. j Immunofluorescence staining of H3K18la level in human PDL fibroblasts after 24 h of cyclic stretching-force treatment. Scale bars: 20 μm. k Immunofluorescence staining and quantitative data of H3K18la levels in the murine PDL region on the tension side in the sham-operated (CON) group and 7-day orthodontic tooth movement (OTM) group (n ≥ 5). Scale bars: 40 μm. AB alveolar bone, R root. Data represent the means ± SD. *P < 0.05, **P < 0.01

We then explored whether lactate-mediated histone lactylation was involved in mechanosensing of PDL fibroblasts. Global lactylation levels in human PDL fibroblasts were significantly elevated after the stretching force was loaded for 12 h and 24 h (Fig. 2g). A predominant band appeared near 17 kDa, which may represent histone H3. We subsequently detected the previously reported histone H3K18 lactylation (H3K18la) and histone H3K9 lactylation (H3K9la),¹⁸ with the level of H3K18la showing a similar trend to global lactylation levels (Fig. 2h, i). Immunofluorescence staining also showed enhanced H3K18la level in the stretching force-treated human PDL fibroblasts in vitro and in the tension-side PDL in vivo (Fig. 2j, k). Additionally, stretching force treatment also enhanced the level of H3K18la in murine PDL fibroblasts (Fig. S4)

H3K18la regulates CXCL1 transcription in PDL fibroblasts

To identify candidate target genes regulated by H3K18la in stretching force-treated PDL fibroblasts, we performed a CUT&Tag analysis using an anti-H3K18la antibody. The heatmap of H3K18la binding peaks showed a significant elevation at the transcriptional start site (TSS) on stretching force-treated human PDL fibroblasts compared with untreated ones (Fig. 3a). The targeted genes of H3K18la binding peaks in the stretch-treated cells exhibited a significant increase in inflammation-related pathways, such as chemokine signaling pathway, inflammatory mediator regulation, and HIF-1 signaling pathway (Fig. 3b). The levels of H3K18la binding at the promoters of chemokine genes, including CXCL1, CCL26, CCL3L3, CXCL2, CXCL14, and CCL7, were significantly increased in cyclic stretching force-treated human PDL fibroblasts compared with the control ones (Fig. 3c). The upregulation of chemokine genes was verified by qRT-PCR analysis, showing that the increase in CXCL1 expression was prominent in the stretching force-treated group (Fig. 3d). CUT&Tag-qPCR validated the enriched abundance of H3K18la on the promoter of CXCL1 after the treatment of stretching force for 24 h (Fig. S5a). The protein level of CXCL1 was also increased in the stretching force-treated human PDL fibroblasts (Fig. 3e). Additionally, according to the scRNA-seq data, the PDL fibroblast clusters in the OTM group displayed an increase in Cxcl1 expression compared with those in the CON group (Fig. 3f).

Fig. 3.

Histone H3 lysine 18 lactylation (H3K18la) regulates CXCL1 transcription in periodontal (PDL) fibroblasts. a Heatmaps showing H3K18la binding peaks in human PDL fibroblasts in the untreated group (CON) and 24-h cyclic stretching force group (Stretch). b Bubble chart showing the enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of upregulated genes with increased H3K18la modification in human PDL fibroblasts in the Tension group compared with the CON group (n = 3). c Enrichment signals at the representative H3K18la target chemokine genes, CXCL1, CCL26, CCL3L3, CCL2, CXCL14, and CCL7, in the CON and Tension group. d Heatmap showing the fold changes of chemokine mRNA expression in human PDL fibroblasts in the CON and Tension group (n = 3). e Immunoblots and quantitative data of CXCL1 protein levels in human PDL fibroblasts treated with different loading time of cyclic stretching force (n = 3). f Uniform Manifold Approximation and Projection (UMAP) visualization of the expression distribution of Cxcl1 projected onto the UMAP atlas and violin plot of gene expression according to the scRNA-seq data of murine PDL fibroblasts. g Immunoblots and quantitative data of CXCL1 and H3K18la protein levels in human PDL fibroblasts treated with exogenous lactate (n = 3). h, i Immunoblots of Pan Kla, H3K18la and CXCL1 protein levels in human PDL fibroblasts treated with cyclic stretching force for 24 h, with or without 3PO (10 μmol/L) pretreatment (n = 3). j Enzyme-linked immunosorbent assay (ELISA) analysis showing CXCL1 secretion by human PDL fibroblasts treated with cyclic stretching force for 24 h, with or without 3PO pretreatment (n = 3). Data represent the means ± SD. *P < 0.05, **P < 0.01

We then manipulated histone lactylation levels in human PDL fibroblasts by treating the cells with exogenous lactate or glycolysis inhibitor. Immunoblot analysis showed increased CXCL1 and H3K18la levels after the addition of exogenous lactate (Fig. 3g). Pretreatment with glycolysis inhibitor, the small molecule 3-(3-pyridinyl)-1-(4-pyridinyl)-2-propen-1-one (3PO), reduced the levels of global lactylation, H3K18la and CXCL1 in human PDL fibroblasts treated with stretching force for 24 h (Fig. 3h, i, Fig. S5b). The expression of CXCL1 was further validated by immunofluorescent staining (Fig. S5c). Enzyme-linked immunosorbent assay (ELISA) also displayed that stretching force promoted the secretion of CXCL1, which was inhibited by 3PO pretreatment (Fig. 3j). The regulation of H3K18la on CXCL1 expression was further validated in murine PDL fibroblasts, in which the cyclic stretching force-enhanced H3K18la and CXCL1 levels were inhibited by pretreatment with 3PO (Fig. S6).

CXCL1 derived from PDL fibroblasts promotes M2 macrophage polarization via mitochondrial respiration

Monocytes/macrophages are essential components of innate immunity in the PDL, contributing to the regulation of alveolar bone remodeling.^20,21 CXCL1 derived from stromal cells plays a critical role in promoting monocyte/macrophage recruitment and activation, thereby modulating tissue inflammation and remodeling.^22,23 Therefore, we investigated the regulatory effects of CXCL1 derived from PDL fibroblasts on macrophage activation. qRT-PCR analysis indicated increased expression levels of M2 macrophage-related genes, including Arg1, Mrc1, Retnla, and Chil3, in murine bone marrow-derived macrophages (BMDMs) treated with IL-4 and CXCL1 compared with those stimulated by IL-4 alone (Fig. S7a). Flow cytometry analysis and immunofluorescent staining showed that addition of CXCL1 promoted the expression of CD206, a marker of pro-resolving macrophages (M2), in murine BMDMs (Fig. 4a, Fig. S7b).

Fig. 4.

CXCL1 derived from periodontal ligament (PDL) fibroblasts promotes M2 macrophage polarization. a Flow cytometry analysis and quantitative data for CD206 and F4/80 expression in bone marrow-derived macrophages (BMDMs) treated with PBS (CON), IL-4 (20 ng/mL), and IL-4 plus CXCL1 (10 ng/mL) (n = 3). b Flow cytometry analysis and quantitative data for CD206 and F4/80 expression in BMDMs under different conditions: cultured alone without PDL fibroblasts (CON group), co-cultured with murine PDL fibroblasts (Co-culture group), co-cultured with stretching force-treated PDL fibroblasts (Force), and pretreated with SB225002 (50 nmol/L) for 2 h followed by co-cultured with stretching force-treated PDL fibroblasts (Force + SB225002) (n = 3). c Heatmap showing differentially expressed genes (DEGs) in BMDMs treated with IL-4 or IL-4 plus CXCL1 (n = 3). d Bubble chart showing the enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) terms of upregulated genes in BMDMs treated with the addition of CXCL1 compared with those in BMDMs treated with IL-4 alone. e Real-time cellular oxygen consumption rates (OCRs) measured by a XF Seahorse Analyzer normalized to the total cell number per group. The final concentrations of oligomycin, FCCP, and rotenone/antimycin A (Rot/AA) were 1.5 μmol/L, 2.5 μmol/L and 0.5 μmol/L, respectively. f ATP content of BMDMs in different treatment groups: IL-4, IL-4 + CXCL1, and SB225002 pretreatment followed by IL-4 + CXCL1 stimulation (n = 3). g–i qRT-PCR results showing mRNA expression levels of genes related with mitochondrial functions in BMDMs (n = 3). Data represent the means ± SD. *P < 0.05, **P < 0.01

We then co-cultured murine PDL fibroblasts with BMDMs to validate the fibroblast-macrophage interaction mediated by the CXCL1/CXCR2 axis. Flow cytometry analysis showed that BMDMs co-cultured with PDL fibroblasts expressed higher levels of CD206 than the control ones, which were further enhanced in BMDMs co-cultured with stretching force-treated fibroblasts (Fig. 4b). Pretreatment with the CXCR2 antagonist SB225002 in BMDMs reduced the expression of CD206 (Fig. 4b), suggesting PDL fibroblasts regulated macrophages via CXCL1/CXCR2 axis.

To elucidate the mechanisms by which CXCL1 regulates macrophage polarization, we conducted RNA sequencing on BMDMs treated with IL-4 alone or in combination with CXCL1. The addition of CXCL1 downregulated several pro-inflammatory genes (Il6, Il8, Ccl4) and upregulated anti-inflammatory genes (Il1r1, Tgfb2, Timd4). Notably, CXCL1 upregulated mitochondrial genes associated with oxidative phosphorylation, such as mt-Atp6, mt-Nd1, mt-Nd2 and Prkag3 (Fig. 4c). Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Set Enrichment Analysis (GSEA) revealed enhanced mitochondrial oxidative phosphorylation in CXCL1-treated BMDMs, a metabolic signature of M2 macrophages (Fig. 4d, Fig. S7c).²⁴ Seahorse analysis monitoring real-time cellular oxygen consumption rates (OCR) demonstrated increased basal and maximal respiration in CXCL1-treated BMDMs (Fig. 4e, Fig. S7d). CXCL1 also enhanced ATP generation in BMDMs, which was abolished by the pretreatment with the CXCR2 antagonist (Fig. 4f). We further validated the increased expression levels of mt-Atp6, mt-Nd1, and mt-Nd2 in CXCL1-treated BMDMs, which were suppressed by the pretreatment with SB225002 (Fig. 4g–i). Therefore, PDL fibroblast-derived CXCL1 could promote M2 macrophage polarization via enhancing macrophage mitochondrial respiration in macrophages.

Inhibition of H3K18la disrupts orthodontic force–induced immune responses

To investigate the role of PDL fibroblast H3K18la in force-guided bone remodeling in vivo, we injected the glycolysis inhibitor 3PO into PDL of mice during OTM. Compared with the PBS-treated group, periodontal injection of 3PO impaired alveolar bone deposition on the tension side between the first and second molar (Fig. 5a). Quantitative analysis of the alveolar bone on the tension side showed that 3PO injection significantly decreased bone volume (BV/TV) and trabecular thickness (Tb.Th), while increasing trabecular separation (Th.Sp) compared with PBS treatment during OTM (Fig. 5b).

Fig. 5.

Inhibition of histone H3 lysine 18 lactylation (H3K18la) in periodontal ligament (PDL) fibroblasts disrupts CXCL1/CXCR2 axis and bone remodeling. a, b Micro-CT analysis and quantitative data showing the bone deposition on the tension side in the model of orthodontic tooth movement (OTM) for 7 d (n ≥ 5). Mice received periodontal injections of 3PO (2 mmol/L) or vehicle every two days. BV/TV, bone volume; Tb.Th, trabecular thickness; Th.Sp, trabecular separation. Scale bars: 500 µm. c Immunofluorescence staining of H3K18la and CXCL1 in PDL region on the tension side in the groups of control (CON), OTM, and OTM with 3PO injection (3PO). Scale bars: 40 µm. AB alveolar bone, R root. d Flow cytometry analysis showing the infiltration of F4/80⁺CXCR2⁺ macrophages in cervical lymph nodes (n ≥ 5). e Quantitative data of H3K18la⁺ or CXCL1⁺ cells in the PDL region on the tension side during OTM (n ≥ 5). f Quantitative data of F4/80⁺CXCR2⁺ macrophages in cervical lymph nodes during OTM (n = 6). g, h Immunofluorescent staining and quantitative data of F4/80 and CXCR2 in PDL region on the tension side of orthodontic tooth movement (OTM) model in WT and Cxcr2 KO mice (n = 5). Scale bars: 40 µm. i, j Immunofluorescent staining and quantitative data of osteocalcin (OCN) in PDL region on the tension side of OTM model in WT and Cxcr2 KO mice (n = 5). Scale bars: 40 µm. Data represent the means ± SD. *P < 0.05, **P < 0.01

In line with the increased Cxcl1 expression in PDL fibroblasts in OTM from scRNA-seq data (Fig. 3j), orthodontic force promoted CXCL1 expression in PDL region on the tension side (Fig. 5c). In situ injection of 3PO inhibited force-induced expression of H3K18la and CXCL1 in PDL region (Fig. 5c, e). After OTM, increased CXCL signaling was observed in the cell-cell communication between PDL fibroblasts and monocytes/macrophages (Fig. S8a). Meanwhile, the number of CXCR2⁺ monocytes/macrophages were elevated in periodontal tissue according to scRNA-seq (Fig. S8b). 3PO injection decreased the infiltration of F4/80⁺CXCR2⁺ macrophages in both cervical lymph nodes (Fig. 5d, f) and PDL microenvironment (Fig. 5g, h), suggesting that glycolysis inhibition could disrupt macrophage infiltration mediated by CXCL1/CXCR2 axis. Moreover, 3PO injection inhibited osteocalcin (OCN) expression on the tension side of alveolar bone, indicating impaired bone regeneration during OTM by glycolysis inhibition (Fig. 5i, j).

Blocking CXCL1/CXCR2 axis impairs orthodontic force–guided bone remodeling

To further investigate the in vivo effect of CXCL1 on macrophage polarization and bone remodeling during OTM, we used Cxcr2 knockout (Cxcr2 KO, Cxcr2^–/^–) mice to block CXCL1/CXCR2 signaling in vivo. In the OTM model, Cxcr2 KO mice exhibited impaired force–guided bone deposition on the tension side compared with wild-type (WT) ones (Fig. 6a, b). Meanwhile, immunofluorescent staining and flow cytometry analysis showed a decreased percentage of F4/80⁺CD206⁺ M2 macrophages in PDL compared with WT mice (Fig. 6c, d and Fig. S9), indicating the impaired pro-resolving microenvironment for bone formation. Blocking CXCL1/CXCR2 axis by Cxcr2 KO also decreased the number of OCN⁺ cells and increased the number of tartrate-resistant acid phosphatase (TRAP) positive osteoclasts on the tension side of alveolar bone (Fig. 6e), suggesting impaired orthodontic force–guided bone regeneration. In conclusion, we propose that orthodontic tensile force induces glycolysis–H3K18la axis in PDL fibroblasts, which enhances CXCL1 transcription and subsequently promotes M2 macrophage polarization, creating a pro-resolving immune microenvironment for bone deposition during OTM (Fig. 6f).

Fig. 6.

Blockade of CXCL1/CXCR2 axis disrupts orthodontic force–guided immune responses and bone remodeling. a, b Micro-CT analysis and quantification data showing the bone deposition on the tension side in wild-type (WT) and Cxcr2^–/– (Cxcr2 KO) mice after OTM for 7 d. Scale bars: 500 µm. c, d Immunofluorescence staining and quantitative analysis of F4/80 and CD206 in the PDL region on the tension side in WT and Cxcr2 KO mice after OTM for 7 d (n = 6). Scale bars: 40 µm. e Tartrate-resistant acid phosphatase (TRAP) staining and osteocalcin (OCN) immunofluorescence staining of alveolar bone on the tension side in WT and Cxcr2 KO mice after OTM for 7 d (n = 6). Scale bars: 40 µm. f Schematic illustration summarizing the key findings of this study. Data represent the means ± SD. *P < 0.05, **P < 0.01

Discussion

The mechanisms of mechanotransduction initiating alveolar bone remodeling in OTM remain obscure. Herein, we demonstrate that tension–induced glycolysis and histone lactylation in PDL fibroblasts modulates orthodontic force–guided bone remodeling in mice. The enhanced level of H3K18la in PDL fibroblasts regulates CXCL1 transcription, thereby promoting M2 macrophage polarization to create a pro-resolving immune microenvironment conducive to bone deposition. The current findings provide novel insights into the mechanisms underlying orthodontic force–induced bone remodeling.

As the major stromal component of PDL, PDL fibroblasts serve as the principal mechanosensing cells and critical immunomodulatory regulators as well.⁷ They initiate the mechanical force–induced aseptic inflammatory immune responses, participating in bone turnover in OTM. However, how PDL fibroblasts convert mechanical stimuli into biological signals to initiate the highly coordinated aseptic inflammatory responses remains obscure. The crosstalk between extracellular forces and intracellular metabolic states offers a potential mechanotransduction mechanism. Recent studies have provided evidence that mechanical cues can directly activate or inactivate glycolytic enzymes. For instance, stiff matrix causes the sequester of the ubiquitin ligase of tripartite motif (TRIM)-containing protein 21 (TRIM21), thereby preventing the degradation of phosphofructokinase (PFK) and enhancing cellular glycolysis.¹⁴ In addition, stiff substrate also induces the recruitment of the glycolytic enzyme phosphofructokinase 1 isoform P (PFKP) to focal adhesions, promoting glycolysis via integrin signaling in endothelial cells.¹⁵ Mechanical stimuli can also regulate glycolysis via YAP/TAZ signaling.^25,26 In addition to extracellular matrix stiffness, the mechanical–glycolytic coupling is also reported in the mechanical stimuli of stretching force, modulating the viability and differentiation of Schwann cells and mesenchymal stromal cells.^27,28 In our study, we demonstrate that orthodontic tension activates glycolysis in PDL fibroblasts, as evidenced by both in vivo scRNA-seq data of murine PDL fibroblasts and in vitro experiments of human PDL fibroblasts, providing evidence for potential mechanical–glycolytic crosstalk during OTM.

Epigenetic modifications link the metabolic status of cells to their transcriptional output. Recently, lactylation has been identified as a novel epigenetic modification relying on endogenous or exogenous lactate, bridging glucose metabolism to gene expression. Previous studies have demonstrated that histone lactylation regulates gene transcription related to cell fate determination and activation.^29–31 According to the CUT&Tag analysis in human PDL fibroblasts, H3K18la potentially promoted the expression of chemokines in response to cyclic stretching force. Among the candidate chemokines, CXCL1 expression was significantly upregulated in response to cyclic stretching force and could be regulated by glycolytic activity, suggesting that CXCL1 might be a critical transcriptional target of H3K18la. Although our results support a role for H3K18la in the transcriptional activation of CXCL1, H3K18la is unlikely to function as the sole determinant of CXCL1 expression. The coordinated action of chromatin state and transcription factors, such as NF-κB and STAT1, synergistically control the transcription of CXCL1.³²

It is worth noting that histone lactylation is part of the mechanical force-induced epigenetic network. Apart from histone lactylation, mechanical stimulations have been reported to influence other histone modifications, such as histone acetylation and methylation.^33–35 In addition, histone lactylation could interact with histone acetylation through shared enzymes and competition for chromatin sites.³⁶ Therefore, tension might also affect other histone modifications in PDL fibroblasts, which could act together with histone lactylation to regulate transcriptional responses during bone remodeling in OTM.

Chemokines are essential mediators of the complex interactions between tissue-resident fibroblasts and immune cells, regulating immune cell maturation, migration, survival, and activation.³⁷ Our in vivo data demonstrated increased infiltration of CXCR2⁺ macrophages in periodontal tissues along with the enhanced expression of CXCL1 in the stretched PDL during OTM. Additionally, inhibition of glycolysis impaired the infiltration of CXCR2⁺ macrophages in vivo. The functions of chemokines can be complex and context-dependent across different tissues and microenvironments. Several studies have reported pro-inflammatory effects of CXCL1 on macrophages and neutrophils in various inflammatory disorders.^38,39 On the contrary, the promotive effects of CXCL1 on M2 macrophage infiltration or polarization have also been implicated.^40–43 In our study, we found that PDL fibroblast–derived CXCL1 created pro-resolving immune niches by promoting M2 macrophage polarization.

The immune responses in periodontal microenvironment, which are dynamic and time dependent, play critical roles in alveolar bone remodeling during OTM. The early phase of OTM involves acute inflammatory reaction and rapid release of inflammatory mediators to recruit and activate immune cells. The early responses are important for initiating bone resorption and remodeling. As OTM progresses, pro-resolving immune regulation for bone formation start to appear.⁶ Macrophages play an essential role in the resolution phase to clear necrotic tissue debris and initiate tissue repair.⁶ M2 macrophages are known to support bone formation through the secretion of multiple pro-resolving mediators, such as TGF-β, IL-10, VEGF, BMP-2, and PGE₂, which promote osteogenic differentiation, angiogenesis, extracellular matrix remodeling, and resolution of inflammation.^44–46 Therefore, CXCL1/CXCR2-mediated M2 macrophage polarization could facilitate alveolar bone formation by establishing a pro-resolving immune microenvironment.

Notably, the immune responses observed in this study are integrated components of mechanotransduction in orthodontic bone remodeling, rather than alternatives to the established pathways initiated by mechanical loading. Classical mechanisms, including ion-channel activation, cytoskeletal remodeling, and downstream biochemical signaling, remain critical to mechanical force sensing in orthodontic alveolar bone remodeling. Within this framework, immune responses function as a complementary layer that modulates and coordinates bone remodeling.

Apart from immunomodulation, glycolysis has also been recognized as a direct regulator of bone remodeling. Glycolysis can regulate osteoblast differentiation and osteoclast activity, thereby affecting both anabolic and catabolic aspects of bone remodeling.⁴⁷ Glycolysis and its product, lactate, have been reported to promote osteogenesis.^48–50 Meanwhile, glycolysis also fuels osteoclast differentiation and function.^51,52 However, we did not detect significant enrichment of H3K18la in osteogenesis-related pathways in PDL fibroblasts, possibly because we analyzed the relative early cell responses to stretching force in vitro.

From a translational perspective, the present findings raise the possibility that local modulation of the H3K18la–CXCL1 axis could be explored as a potential strategy to promote alveolar bone remodeling during OTM. Enhancing the pathway might improve bone deposition in OTM. In addition, dysregulated immune responses also contribute to side effects such as excessive bone resorption or root resorption, so promotion of a balanced pro-resolving microenvironment through elevating lactate levels or locally applying CXCL1 might also help limit undesirable periodontal tissue damage, which requires further validation.

Nevertheless, the biological effects of glycolysis and H3K18la are highly dependent on the local microenvironment. We found that glycolysis and H3K18la in PDL fibroblasts promoted CXCL1 expression, thereby enhancing pro-resolving macrophage polarization under stretch stimulation. However, glycolysis and H3K18la is also associated with pro-inflammatory responses in macrophages in chronic inflammatory conditions.⁵³ These distinct outcomes might reflect differences in cell type, mechanical context, and inflammatory condition. In patients with chronic inflammation or metabolic disorders, such as periodontitis or diabetes, persistent inflammatory signaling and metabolic dysregulation could induce the glycolysis and H3K18la in macrophages and skew pro-inflammatory phenotype polarization,^54–56 thereby possibly inhibiting alveolar bone formation in OTM. This possibility highlights the need for future studies examining how systemic and local inflammatory conditions impact mechanical force-induced metabolic-epigenic reprogramming in different cell populations for individualized orthodontic strategies.

Our data show the involvement of PDL fibroblast H3K18la in alveolar bone formation via CXCL1/CXCR2-dependent immune modulation of macrophages. However, there are some limitations of our study. First, the two-dimensional Flexcell stretching applies stretching strain on a flat surface, whereas in vivo PDL fibroblasts undergoes heterogeneous, coupled tension–compression, shear, and dynamic viscoelastic loading during OTM. It also lacks native ECM architecture of PDL. Thus, the in vitro model may not fully predict in vivo kinetics. Secondly, though 3PO can decreases glycolytic flux and lactate availability to inhibit histone lactylation, it may cause broad metabolic shifts and off-target–related side effects, such as altered mitochondrial function and stress-response activation. 3PO could also inhibit NF-κB signaling, which directly regulates CXCL1 expression.^32,57 Thus, phenotypic changes may not be attributed exclusively to histone lactylation. Moreover, local periodontal injection of 3PO might also affect glycolytic activity in other cell populations. Given the direct regulations of glycolysis on osteoblasts and osteoclasts,⁴⁷ the in vivo effects of 3PO can’t be interpreted as reflecting definitive fibroblast-specific inhibition. Future studies using cell-type-specific genetic methods are needed to define the contribution of PDL fibroblasts to the mechano-metabolic pathway.

Additionally, orthodontic forces not only induce cellular responses, but also lead to vascular changes and local hypoxia in periodontal microenvironment, which could contribute to the activation of glycolysis and histone lactylation. Therefore, the observed metabolic remodeling in vivo might reflect the combined effects of mechanical force and local hypoxia. Moreover, although Cxcr2^–/– mice enabled in vivo evaluation of the functional importance of the CXCL1/CXCR2 axis,^22,23 CXCR2 is expressed by multiple cell populations. Therefore, the observed results could not reflect a macrophage-specific effect. Future studies using Cxcr2 conditional knockout models will be necessary to define the contribution of CXCL1/CXCR2 signaling in macrophages to orthodontic force–guided alveolar bone remodeling. Another limitation of this study is that human PDL fibroblasts were used for mechanistic analyses of stretching force–induced glycolysis, histone lactylation, and CXCL1 transcription, whereas murine PDL fibroblasts were used in macrophage co-culture experiments to maintain compatibility with murine immune cells and the in vivo mouse model. Future studies using a unified single-species platform are required for validation.

In conclusion, we demonstrate that metabolic–epigenetic axis induced by tension in PDL fibroblasts modulates periodontal immune niches and bone formation in OTM. Orthodontic tensile force induces glycolysis and H3K18la in PDL fibroblasts, thereby upregulating CXCL1 expression, which promotes macrophage OXPHOS and M2 macrophage polarization for alveolar bone deposition. These findings provide novel insights into the mechanotransduction mechanisms associated with OTM, which may help in developing therapeutic strategies to enhance alveolar bone regeneration.

Materials and methods

Animals and OTM model

The animal experimental protocol was approved by the Ethics Committee of West China Hospital of Stomatology, Sichuan University (WCHSIRB-D-2023-148). This study conformed with the ARRIVE 2.0 guidelines. The wild-type mice and Cxcr2 KO (Cxcr2^–/^–) mice were purchased from GemPharmatech Co., Ltd (Jiangsu, China). All mice were housed in a pathogen-free environment and fed standard food and water ad libitum with a 12 h day/night light cycle.

OTM models were established on 8-week-old male mice based on a previously described method.⁵⁸ Briefly, mice received anesthesia with ketamine (100 mg/kg) and xylazine (10 mg/kg), and nickel-titanium coil springs (20 g) were placed between the maxillary left first molar and incisors for 7 days. All force application procedures were conducted by the same researcher for consistency. The mice in control group received sham operation. The mice were fed with soft-food diet throughout the experiments.

Cell isolation for scRNA-seq

Periodontal tissues were isolated from mice in the sham-operated group (CON) and orthodontic tooth movement (OTM) group. Periodontal tissues from six mice were pooled into one sample, and each group comprised five samples for scRNA-seq. The gingiva was removed before the collection of single cells to avoid interference of gingival fibroblasts. Briefly, periodontal tissue around the maxillary left first molars were collected and gingiva tissues were discarded. The samples were then rinsed by RPMI 1640 medium and minced on ice. They were digested by collagenase I (2 mg/mL, Gibco) and collagenase II (2 mg/mL, Gibco) for 90 min at 37 °C. Cell suspension was then filtered through a 40 μm strainer, and the supernatant was removed after centrifugation at 300 g for 5 min twice. Cell concentration and viability were analyzed by hemocytometer and Trypan Blue staining. Single cells (800–1 500 cells/μL) were encapsulated into water-in-oil emulsions using the MGISEQ-SEQ series platform (MGI, China). Subsequent reverse transcription and sequencing library preparation were performed according to the manufacturer’s standard protocol for single-cell RNA sequencing (MGI).

Preprocessing of single-cell RNA sequencing data

Single-cell RNA sequencing libraries were prepared using the MGI DNBelab C Series Single-Cell Library Prep Set and sequenced on a DNBSEQ platform. Raw sequencing data were processed with DNBC4tools (v2.1.3). Reference genome indices were generated using the mouse genome (mm10) and corresponding gene annotation files. Read alignment, barcode and UMI processing, and gene quantification were performed using the dnbc4tools rna run pipeline, generating filtered gene–cell count matrices.

Filtered matrices were analyzed in R (v4.5.2) using Seurat (v5.0.1).⁵⁹ Cells were filtered based on the following criteria: 100 < nFeature_RNA < 6 000, nCount_RNA < 50 000, and mitochondrial gene content <10%. Gene expression data were normalized using the LogNormalize method with a scale factor of 10 000.

Doublets were identified and removed using the DoubletFinder package.⁶⁰ The expected doublet rate was estimated based on cell recovery, and optimal parameters were determined using a parameter sweep strategy. Only high-confidence singlets were retained for downstream analysis.

Batch effects were corrected using Harmony on principal component analysis (PCA) embeddings, and the corrected dimensions were used for clustering and UMAP visualization.

Clustering and annotation of cell types

Top variable genes among single cells were identified using the FindVariableGenes function in Seurat (parameters: mean.function = ExpMean, dispersion.function = LogVMR). The mutual nearest neighbors (MNN) method was applied to correct for batch effects within the single-cell RNA sequencing datasets.⁶¹ Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction and visualization were carried out based on the MNN-corrected results. Cells were grouped according to their gene expression profiles through graph-based clustering implemented by the FindClusters function. To identify cluster-specific marker genes and assist in manual cell type annotation, the FindAllMarkers function in Seurat (with test.use = bimod) was employed.

Differential expression analysis

Differentially expressed genes (DEGs) were determined using the FindMarkers function (test.use = MAST) with P < 0.05 and |log₂(fold change) | > 0.5 as the threshold. GO biological pathway enrichment analyses were subsequently conducted using the ggplot2 package in R (P < 0.01) based on the hypergeometric distribution. DEGs were identified by analyzing independent biological replicates separately.

Cell-cell communication analysis

Cell-cell communication analysis between fibroblasts and monocytes/macrophages was performed using CellChat. Fibroblasts and monocytes/macrophages were first subset from the integrated Seurat object according to cell-type annotation. Normalized RNA expression data were extracted using GetAssayData, and the corresponding metadata were used to construct CellChat objects. Cell identities were defined according to the annotated cell population used for the fibroblasts-monocytes/macrophages communication analysis. The CellChat mouse ligand-receptor database was used for interaction inference. For each condition, genes and interactions represented in the CellChat database were retained using subsetData, followed by identification of overexpressed genes and overexpressed ligand-receptor interactions. The expression data were projected onto the mouse protein-protein interaction network, and communication probabilities were calculated using computeCommunProb. Low-confidence interactions were filtered by requiring at least 10 cells per interacting group. Pathway-level communication probabilities were then computed using computeCommunProbPathway, and the overall communication network was aggregated using aggregateNet. Network centrality was further calculated with netAnalysis_computeCentrality.

Isolation and culture of human PDL fibroblasts

The experimental protocol using human PDL fibroblasts was approved by the Ethics Committee of West China Hospital of Stomatology, Sichuan University (WCHSIRB-D-2023-062). Human PDL tissues were collected from extracted premolars of teenagers (12–14 years). After rinsed with PBS containing 10% penicillin and streptomycin (Gibco), periodontium on the extracted teeth was scraped from the middle 1/3 of the root surface and cut into small pieces, which were resuspended with α-modified Eagle’s medium-α (α-MEM, Gibco, USA) containing 10% fetal bovine serum (FBS, Gibco, USA) and 1% penicillin–streptomycin (Life Technologies). The medium was refreshed weekly until cells reached 80%–90% confluence. Cells at passages 3 to 5 were used for all experiments. Detailed materials and methods are provided in the Supplementary Material.

The human PDL fibroblasts were treated with cyclic stretching forces or gradient exogenous lactate. For the inhibition of glycolysis, cells were pretreated with 3-(3-Pyridinyl)-1-(4-pyridinyl)-2-propen-1-one (3PO) (10 μmol/L) for 2 h before stretching force loading.^15,62

Mechanical force loading method

For cyclic stretching force loading, the Flexcell stretch system device (Flexcell 5000) was used to simulate the orthodontic tension of PDL fibroblasts in vivo. The stretching force-treated group (Force) received cyclic stretching loading with an amplitude of 10% and frequency of 0.5 Hz for up to 24 h.⁶³ Cells in the control group (CON) were placed in the same cell incubator under static conditions.

CUT&Tag

Human PDL fibroblasts from control group and cyclic force-treated group (n = 3 per group) were collected for the CUT&Tag assay, which was performed using the NovoNGS CUT&Tag 4.0 High-Sensitivity Kit (Vazyme, N259-YH01) compatible with Illumina sequencing following the manufacturer’s protocol.⁶⁴ Briefly, 100 000 cells were lysed to prepare nuclei, which were gently washed twice with Wash Buffer. Concanavalin A (10 μL) coated magnetic beads were added per sample and incubated at room temperature for 10 min, followed by the removal of the unbound supernatant. The bead-bound nuclei were resuspended in Primary Antibody Buffer containing anti-H3K18la antibody (1:50, PTM-1427RM, PTMBIO) and incubated overnight at 4 °C. The primary antibody was removed using magnetic stand. Subsequently, cells were incubated with goat anti-rabbit IgG H&L pAb (1:200, PTM-6252) for 60 min at room temperature, followed by three washes using the magnetic stand. The pA-Tn5 adapter complex (1:80) was prepared in ChiTag Buffer and incubated with nuclei at room temperature for 1 h. After additional washes with ChiTag Buffer, the nuclei were resuspended in Tagmentation Buffer and incubated at 37 °C for 1 h. Following cross-link reversal, DNA was then purified using Tagment DNA Extract Beads. For library amplification, 20 μL DNA was then mixed with the barcoded i5 primer (2.5 μL), the uniquely barcoded i7 primer (2.5 μL), and 2× HiFi AmpliMix (25 μL). Library clean-up was performed with NovoNGS DNA Clean Beads.

The BAM files generated from the mapped reads were used as input for peak calling with MACS2 software, applying a significance threshold of q-value < 0.05. Identified peaks were annotated using the ChIPseeker R package. The annotation results were summarized, and their genomic distribution was visualized using R package.

For differential accessibility analysis, peak files from individual samples were merged using BEDTools. Next, read counts overlapping each peak region were quantified with the bedtools multicov function. Differential accessible peaks were assessed using DESeq2. Significantly differential accessible peaks were defined with |log₂(fold change) | > 1 and P < 0.05.

Isolation and culture of murine PDL fibroblasts

Murine PDL fibroblasts were isolated according to a published protocol.⁶⁵ Briefly, maxilla and mandible were harvested and gingiva tissues were removed. Tooth were extracted, transferred to fresh Collagenase P (Col P, Sigma) solution (3 U/mL), and digested in 37 °C for 45 min with shaking. The digestion was terminated by α-MEM culture medium with 10% FBS. After the careful removal of teeth, the cells and digested tissues were centrifuged at 1 200× g for 5 min and resuspended in culture medium. The medium was refreshed weekly until cells reached 80%–90% confluence.

Enzyme-linked immunosorbent assay

The supernatants of human PDL fibroblasts were collected and centrifuged at 1 000× g for 20 min. The supernatants were collected for the enzyme-linked immunosorbent assay (ELISA). Protein expression levels of CXCL1 were examined using an ELISA kit (USEA041Hu, CLOUD-CLONE CORP.) according to the manufacturer’s protocols.

Isolation and culture of BMDMs

BMDM were harvested from long bones of wild type C57BL/6 (6–8 weeks) mice. Cells were rinsed and harvested from the medullary cavity, and resuspended into the culture flask with the culture medium consisted of α-MEM, 10% FBS, 1% penicillin–streptomycin, and 25 ng/mL M-CSF (Peprotech, USA) for 7 days. The BMDMs were treated with 20 ng/mL IL-4 (Peprotech, USA) for 24 h with or without the addition of 10 ng/mL CXCL1 (Peprotech, USA).

Glycolysis and mitochondrial respiration measurements

Glycolysis and mitochondrial function were measured by the Seahorses Biosciences XFe analyzer. Human PDL fibroblasts were treated with cyclic stretching force for 24 h and then seeded in complete culture media at 2 × 10⁴ cells per well in the Seahorse BioSciences XFe assay wells. The glycolytic rate was measured using the Glycolytic Rate Assay Kit (Seahorses Biosciences, 103344). The final concentrations of rotenone/antimycin A (Rot/AA) and 2-Deoxy-D-Glucose (2-DG) were 0.5 μmol/L and 50 mmol/L respectively according to the protocol of the kit. Seahorse XF Glycolytic Rate Assay utilizes both extracellular acidification rate (ECAR) and oxygen consumption rates (OCR) measurements to calculate the glycolytic proton efflux rates (glycoPER) of the cells by the Seahorse XF Glycolytic Rate Assay Report Generators. In terms of mitochondrial respiration, BMDMs were seeded in complete culture media at 4 × 10⁴ cells per well in the Seahorse BioSciences XFe assay wells. After 24 h, different stimuli (IL-4 or IL-4 + CXCL1) were added. OCRs in real time was measured using the Cell Mito Stress Kit (Seahorses Biosciences, 103015). The final concentration of oligomycin was 1.5 μmol/L, the final concentration of FCCP was 2.5 μmol/L, and the final concentration of Rot/AA was 0.5 μM according to the kit. Data were normalized to cell number on culture plates.

Co-culture of PDL fibroblasts and BMDMs

In the co-culture model, murine PDL fibroblasts were cultured in the α-MEM complete medium and seeded at a concentration of 6 × 10⁵ cells per well on the Flexcell plates. PLD fibroblasts received cyclic stretching force stimulation for 24 h. BMDMs were seeded at a density of 2 × 10⁵ cells per well in the upper chamber with the stimulation of 20 ng/mL M-CSF. BMDMs in the SB225002-treated group were pretreated with the CXCR2 antagonist SB225002 (50 nmol/L) for half an hour before being co-incubated with PDL fibroblasts for 48 h. BMDMs were then collected for subsequent analysis.

RNA sequencing

Total RNA was extracted from BMDM samples of two groups (IL-4, IL-4 + CXCL1 group, n = 3 per group) using TRIzol reagent (Takara). Total RNA was then fragmented into RNA fragments and reverse transcribed into cDNA as templates for RNA sequencing (RNA-seq) using the Illumina HiSeq2500 platform (Illumina). The reads were mapped to the reference transcriptome using bioinformatics tools.

Periodontal injection

The periodontal injection of 5 µL 3PO (2 mmol/L in DMSO and saline⁶⁶) or vehicle was accomplished by microinjection through a microsyringe (10 µL syringe with a 32-gauge needle; Hamilton Company, Reno, Nevada, USA) into periodontal tissues around the maxillary left first molars. The periodontal injection was performed every day for 7 doses.

Micro-computed tomography analysis

Maxillary samples were scanned with a micro-computed tomography system (μCT50; SCANCO Medical AG, Switzerland) with a voxel resolution of 5 µm. The SCANCO Visualizer software was applied for the three-dimensional reconstruction of samples. The bone volume per tissue volume (BV/TV), trabecular separation (Tb. Sp), and trabecular thickness (Tb.Th) analyses at the region of interest (ROI) were analyzed by SCANCO Evaluation software. The ROI was defined as a 0.01 mm³ rectangular region of alveolar bone between the maxillary first and second molars.

Flow cytometry analysis

Alexa Fluor® 488 conjugated anti-mouse F4/80 (123119, BioLegend), APC conjugated anti-mouse CD206 (141707, BioLegend), and APC conjugated anti-mouse CXCR2 (149311, BioLegend) were used for the staining. Cells were incubated with anti-mouse CD16/32 antibody (156603, BioLegend) to block Fc receptors and then stained by surface epitopes (anti-F4/80 or anti-CXCR2) for 20 min at 4 °C in staining buffer. Afterward, cells were fixed and permeabilized with the Intracellular Staining Permeabilization Wash Buffer (BioLegend), followed by the staining of intracellular epitopes (anti-CD206) in 1× permeabilization buffer.

Histologic and immunochemistry staining

Maxillary samples were decalcified in 10% ethylenediaminetetraacetic acid (EDTA) and embedded in paraffin. For immunochemistry staining, the sagittal sections were probed with anti-PKM2 (1:200, R1603-5, HUABIO), anti-PFKFB3 (1:200, ET1705-66, HUABIO), anti-H3K18la (1:200, PTM-1427RM, PTMBIO), anti-CXCL1 (1:200, PA5-86508, Thermofisher), anti-OCN (1:200, ER1919-20, HUABIO). Relative quantitative analyses were performed using Image J software. Tartrate-resistant acid phosphatase (TRAP) staining was conducted using the TRAP Kit (387A-1KTF, Sigma-Aldrich; methyl green counterstain). TRAP positive cells were counted as osteoclasts for quantification.

Quantitative real-time polymerase chain reaction assay

Total RNA was extracted by TRIzol (Invitrogen). cDNA from 1 μg total RNA was synthesized with reverse transcriptase (Takara). The quantitative real-time polymerase chain reaction (qRT-PCR) was performed using SYBR Green PCR Master Mix Kit (Takara). Gene expression levels were normalized to β-actin (ACTB). The primer sequences were shown in Table S1.

Western blot analysis

Cells were lysed using RIPA lysis buffer supplemented with protease inhibitors. Equal protein samples were separated by 10% SDS-PAGE (PG222, Epizyme), and transferred onto PVDF membranes (IPVH00010, Millipore). The membranes were blocked with non-fat milk and then incubated with primary antibodies at 4 °C overnight: β-Actin (1:10 000, EM21002, Huabio), PKM2 (1:1 000, R1603-5, HUABIO), PFKFB3 (1:1 000, ET1705-66, HUABIO), PFKM (1:1 000, ET7106-97, HUABIO), Pan Kla (1:2 000, PTM-1401RM, PTMBIO), H3 (1:10 000, ET1701-64, HUABIO) H3K18la (1:1 000, PTM-1427RM, PTMBIO), H3K9la (1:1 000, PTMBIO), CXCL1 (1:1 000, PA5-86508, Thermofisher). Then the membranes were incubated with HRP-conjugated secondary antibodies at room temperature for 1 h. Blots were observed with ECL chemiluminescence reagent (SQ201, Epizyme). Next, membranes were incubated by goat anti-mouse/rabbit secondary antibodies at room temperature for 1 h and exposed under the chemiluminescence imaging system (BioRad).

Detection of ATP content

The ATP production in BMDMs was measured by an ATP assay kit (Beyotime Biotechnology). Following the instructions, the same amounts of BMDMs in different groups were collected with lysate supplied by the kit and centrifuged at 12 000× g for 5 min at 4 °C to collect supernatant. Then, a luminometer (Spark; Tecan, Männedorf, Swiss) was used for detecting ATP concentration in supernatant. The ATP concentration was normalized to total protein concentration in each well.

Statistical analysis

The data were presented as mean ± standard deviation (SD). The unpaired t test was applied for statistical comparison between two groups, and one-way ANOVA followed by Tukey’s multiple analysis was applied for comparisons among multiple groups. Statistical analysis was performed by SPSS 16.0 software. A value of P < 0.05 was considered significant.

Author contributions

B. Zhang, contributed to conception, design, data acquisition and interpretation, and drafted the manuscript; Z.H. Song, contributed to data acquisition and interpretation, and drafted the manuscript; Y.G. Zhang, contributed to data acquisition and interpretation, and drafted the manuscript; J.Y. Cai, contributed to data acquisition and interpretation, and drafted the manuscript; G.Y. Zhu., contributed to conception, design, data acquisition and interpretation, and critically revised the manuscript; Z.H. Zhao, contributed to conception, design, data acquisition and interpretation, and critically revised the manuscript. All authors gave their final approval and agree to be accountable for all aspects of the work.

Data availability

The data supporting the findings of this study are available within the article and its supplementary materials.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41368-026-00465-0.