Microbial Tryptophan Metabolites Ameliorate Ovariectomy‐Induced Bone Loss by Repairing Intestinal AhR‐Mediated Gut‐Bone Signaling Pathway

16S rRNA gene sequencing analysis showed that ovariectomy (OVX)‐induced postmenopausal osteoporosis mice exhibited significantly altered microbial community, manifested by marked different diversity and microbiota composition between the two groups (Figure1A–C). Specifically, OVX significantly upregulated Firmicutes and downregulated Bacteroidota in cecal contents of mice, which are dominant bacteria at the phylum level (Figure 1C). At the genus level, comparison between bacterial community characteristics was tabulated using a LEfSe cladogram representing the relative abundance of each genus (Figure 1D). The results suggested that OVX induced profound alterations in the microbiota composition (Figure 1D). Compared with Sham mice, OVX‐induced osteoporosis mice showed significant increases in the gut bacteria belonging to the Muribaculaceae, Bifidobacteriaceae, and Clostridiaceae families (Figure 1D). Especially, the relative abundances of family Lactobacillaceae and genus Lactobacillus were markedly downregulated in the cecal contents of OVX mice in comparison with Sham mice (Figure 1E,F).

To predict the relationship between abundance of gene families and functional pathways of microbial communities in the cecal contents, PICRUSt2 (phylogenetic investigation of communities by reconstruction of unobserved states) was employed based on the 16S rRNA gene sequencing and the Green Genes database (Figure 1G). The results of metagenomic analysis showed that OVX significantly altered many bacterial pathways involving in amino acid, lipid, mineral absorption, and apoptosis and so on. Of particular note was that OVX mice exhibited significantly reduced Trp metabolism that is highly associated with gut microbiota, the downstream metabolites, and metabolic diseases (Figure 1G). Spearman correlation analysis showed that the levels of indole metabolites such as IAA, IPA, and ILA were positively correlated to the relative abundance of Lactobacillus (Figure S1A–C, Supporting Information). To further verify the specific association between Trp metabolites and Lactobacillus, we have also performed direct supplementary experiments in which several Lactobacillus species were colonized to mice with and without antibiotics (ABX) treatment (Figure S1D, Supporting Information). Targeted metabolomics showed that the levels of indole and indole derivatives, especially IAA, in feces and colon of mice upon Lactobacillus supplementation were much higher than those of mice with only antibiotics treatment (Figure S1E, Supporting Information), suggesting that indole metabolites (IAA and IPA) are derived from microbes, especially at least Lactobacillus.

To investigate the relevance of microbial Trp metabolites to metabolic diseases with bone loss, we quantitatively analyzed the concentrations of specific Trp metabolites in fecal samples of several animal models with OVX‐induced postmenopausal osteoporosis (Figure2A; Figure S2, Supporting Information). Significant bone loss was observed in OVX‐induced osteoporosis mouse model (Figure 2B), manifested by marked reduction of bone mineral density (BMD) and bone volume/tissue volume ratio (BV/TV) (Figure 2C,D). Targeted UHPLC‐QQQ‐MS‐based metabolomics analysis indicated that much lower concentrations of AhR ligands such as indole derivatives (IAA and IPA) were observed in fecal samples from animal models with metabolic diseases than those of their corresponding controls (Figure 2E). Interestingly, spearman correlation analysis showed that the levels of indole metabolites, especially IAA and IPA, exhibited markedly positive correlation with parameters of bone loss including BMD, BV/TV, trabecular thickness (Tb.Th), and trabecular number (Tb.N) (Figure 2F). Individuals displaying postmenopausal bone loss, similarly showed lower serum concentrations of IAA and IPA, microbial AhR ligands, than premenopausal individuals (Figure 2G–I). Collectively, these results demonstrate that OVX‐induced bone loss is highly relevant to significantly reduced microbiota‐derived Trp metabolites.

We next used the typical OVX‐induced osteoporosis mouse model to explore whether and/or how dietary treatment with microbial Trp metabolites can improve bone loss. µCT imaging and quantitative analysis showed that supplementation with IAA and IPA (20 mg kg⁻¹ body weight) by gavage for 10 weeks markedly improved OVX‐induced bone loss of wild type (WT) mice (Figure3A), shown with significant upregulation of BMD and BV/TV (Figure 3B; Figure S3A, Supporting Information) and Tb. Th (Figure S3B, Supporting Information) together with significant downregulation of trabecular separation (Tb. Sp) (Figure S3C, Supporting Information). Furthermore, immunohistochemical osteocalcin (OCN) staining showed that supplementation with IAA and IPA markedly increased the number of osteoblasts on trabecular bone surface of OVX WT mice (Figure 3D,E). Enzyme‐linked immunosorbent assay (ELISA) revealed that supplementation with IAA and IPA further enhanced serum level of the N‐terminal propeptide of type I procollagen (PINP), a biomarker of osteogenic differentiation (Figure 3F). In addition, quantitative measurement coupled with tartrate‐resistant acid phosphatase (TRAP) staining suggested that treatment with IAA markedly decreased the levels of Trap positive (N. Trap⁺) and C‐telopeptide of type I collagen (CTX‐I), typical biomarkers of osteoclast activity and bone resorption, in OVX WT mice (Figure 3G,H). Most notably, these ameliorative effects of IAA and IPA treatments on bone loss were not significantly observed in OVX mice with intestinal AhR knockout (Villin^CreAhr^fl/fl) (Figure4A–H; Figure S3D–F, Supporting Information), suggesting that supplementation with IAA and IPA has a potential role in ameliorating OVX‐induced bone loss in an intestinal AhR‐dependent manner.

Given that intestinal AhR was found to be closely associated with OVX‐induced bone loss, we next investigated the effects of supplementation with microbial Trp metabolites on intestinal barrier function. Intestinal AB‐PAS staining showed that supplementation with IAA and IPA effectively restored intestinal structural integrity that was altered in OVX mice (Figure 3I), quantitatively manifested by marked elevation in the number of goblet cells (Figure 3J) and mucosa layer thickness (Figure 3K). Similar to previous studies reporting intestinal barrier integrity in OVX mice,^[40, 41^] we found that OVX mice exhibited lower protein and mRNA levels of ZO‐1 and E‐cadherin (Figure 3L–P), junction components at the apical site of epithelial cell, together with lower mRNA level of protein tyrosine phosphatase receptor type H gene (Ptprh), a typical marker of gut permeability relevant to intestinal barrier function (Figure 3M), than those of Sham WT mice. Supplementation with IAA and IPA strikingly restored the levels of intestinal barrier function‐related biomarkers like upregulation of junction proteins between epithelial cells of OVX WT mice (Figure 3L–P). However, supplementation with IAA and IPA did not exhibit intestinal repairing capacity in OVX mice with intestinal AhR knockout, shown with no significant changes in the gut structural integrity‐related parameters including the number of goblet cells (Figure 4J), mucosa layer thickness (Figure 4K), protein and mRNA levels of junction components and gut permeability (Figure 4L–P). These findings suggest that supplementation with IAA and IPA can effectively restore the gut barrier function of OVX mice with bone loss via activation of intestinal AhR.

To explore the molecular mechanisms by which AhR activation regulates epithelial barrier integrity, we analyzed RNA sequencing profiles in intestine obtained from Sham and OVX WT mice with and without IAA treatment. PCA analysis of RNA sequencing revealed clear separation between different groups and robust clustering of samples within each group (Figure S4A, Supporting Information). Differential gene expression analysis showed that supplementation with IAA induced marked upregulation of 477 genes and downregulation of 264 genes in intestine of OVX mice (p < 0.05 and fold change > 1.5) (Figure S4B,C, Supporting Information). Especially, Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis highlighted several signaling pathways including AhR‐related metabolism of xenobiotics by cytochrome, Wnt signaling pathway, cell adhesion molecules, calcium signaling pathway, and innate immunity (Figure5A,B). Transcriptomics analysis further suggested that these differentially expressed genes induced by IAA treatment were mainly originated from activation of immune response, Wnt signaling pathway and cell adhesion molecules (Figure 5C). Notably, gene set enrichment analysis (GSEA) revealed that intestinal Wnt signaling pathway was inhibited in OVX mice and subsequently activated upon IAA treatment (Figure S4D,E, Supporting Information). Based on the analysis of differentially expressed genes, we found that IAA treatment markedly upregulated mRNA levels of Arg1, Il‐10, and β‐catenin (Ctnnb1) and downregulated mRNA levels of CD86 that are involved in innate immunity and Wnt signaling pathway in OVX mice (Figure 5C). Furthermore, qualitative and quantitative analyses with immunohistochemistry and immunofluorescence showed that supplementation with IAA and IPA significantly repaired intestinal epithelium integrity and upregulated the level of β‐catenin in intestine of OVX WT mice (Figure 5D–F). Meanwhile, supplementation with IAA and IPA markedly decreased serum LPS accompanied by significant elevation in the level of serum IL‐10 in OVX WT mice (Figure 5H,I). However, these alterations in the levels of β‐catenin, LPS and IL‐10 were not observed in OVX mice with intestinal AhR knockout (Figure6A–F).

We next used LPS‐treated Caco‐2 cells acting as in vitro model with intestinal disorder to further explore the roles of AhR and Wnt/β‐catenin signaling pathway in epithelial cells. Interestingly, LPS treatment markedly decreased the protein and mRNA levels of β‐catenin that were significantly reversed by IAA and IPA treatments (Figure 6G–I). IAA treatment also markedly increased mRNA level of Ahr, and ZO‐1 that were decreased by LPS exposure in Caco‐2 cells (Figure S5A,B, Supporting Information). Importantly, we discovered that at least three binding sites of β‐catenin promotor from −2000 to +10 bp interacts with AhR protein via simulation of the JASPAR database (Figure 6J; Figure S6, Supporting Information). To experimentally validate AhR binding to β‐catenin promotor, we constructed a pGL4‐Basic‐β‐catenin‐promotor‐luc plasmid spanning from −2000 to +10 bp of β‐catenin promotor cloned upstream of luciferase in HEK 293T cells (Figure 6J). After transfection of β‐catenin luciferases reporter vectors into HEK 293T cells with and without AhR knockdown (Figure S5C, Supporting Information), we found that only IAA administration induced a marked increase in the luciferase activity of β‐catenin reporter in HEK 293T cells, whereas no significant change was observed in Ahr‐knockdown HEK 293T cells upon IAA supplementation (Figure 6K). Taken together, these findings reveal that activation of AhR by supplementation with IAA repairs intestinal barrier function via regulating Wnt/β‐catenin signaling pathway.

Given that gut‐bone signaling axis highly contributes to bone health, we next explore the crosstalk between the gut and bone. In OVX WT mice, OVX operation induced significant upregulation of M1 macrophages (CD86⁺) and downregulation of M2 macrophages (CD206⁺) due to LPS‐induced inflammatory status in intestine (Figure 5C,G). As a result, mRNA level of Il‐10 secreted by M2 macrophage was consequently decreased in intestine of OVX WT mice compared with Sham mice (Figure 5C). Supplementation with IAA and IPA led to the switching from M1 to M2 activated macrophage phenotypes, manifested by a marked reduction in CD86⁺ macrophages and a substantial increase in CD206⁺ macrophages in intestinal lamina propria of OVX mice (Figure 5C,G). Consequently, supplementation with Trp metabolites, especially IAA, markedly downregulated mRNA level of intestinal Il‐10 (Figure 5C) and upregulated IL‐10 level in serum (Figure 5I), suggesting that supplementation with IAA promoted expansion of IL‐10 from intestine to blood circulation. Interestingly, no significant changes in the switch of macrophage phenotypes and the level of serum IL‐10 were observed in OVX mice with intestinal AhR knockout upon IAA and IPA supplementation (Figure 6D,F).

To verify the direct effects of IL‐10 on osteoblasts and osteoclasts, we further examine cell proliferation and differentiation of osteoblasts and osteoclasts under the condition of IL‐10 administration in vitro. We found that IL‐10 exposure dose‐dependently promoted osteoblast differentiation and formation (Figure7A,E) and inhibited RANKL‐induced osteoclastogenesis (Figure 7C,F). Specifically, IL‐10 exposure induced marked upregulation of alkaline phosphatase (ALP) activity and mineralization nodules formation (Figure 7A,B), suggesting IL‐10‐induced osteoblastogenesis. Supportive evidence can be found from markedly elevated osteogenesis‐related genes including Alp, Opg, and Opn in a dose‐dependent manner (Figure 7E). TRAP staining for osteoclasts also showed that IL‐10 treatment dose‐dependently prevented osteoclasts differentiation (Figure 7C), shown with a marked reduction of the number of TRAP‐positive multinucleated osteoclasts (N. OCs) (Figure 7D), whereas no significant changes were observed in the mRNA levels of osteoclastogenesis‐related genes including Ctsk, Traf6, and Mmp9 (Figure 7F). Collectively, these in vivo and in vitro results suggest that supplementation with IAA and IPA promotes secretion and expansion of IL‐10 from intestinal M2 macrophage to bone marrow, thus simultaneously regulating osteoblast and osteoclast differentiation.

Animal experimental procedures were reviewed and approved by the animal ethics committee of Innovation Academy for Precision Measurement Science and Technology, CAS (APM No: APM20029↗, China). In OVX mice models (animal experiment 1, 4, and 5), female wild‐type C57BL/6 mice (n = 32, 8‐week‐old) and intestinal AhR knockout (Villin^CreAhr^fl/fl) mice (n = 20, 8‐week‐old) were purchased from Charles River Co. Ltd (Beijing, China). All mice were housed under specific pathogen‐free conditions with 12‐h light cycle and fed sterilized food with autoclaved water ad libitum. After acclimation for 2 weeks, mice were supplied with IAA and IPA (20 mg kg⁻¹ body weight) or vehicle by gavage 5 times per week.^[43, 54, 55^] After treatment for 2 weeks, mice were ovariectomy under the condition of tribromoethanol anesthesia. Mice upon sham operation were regarded as control group. Food intake and body weight of mice were monitored and recorded once a week for 10 weeks. In the animal experiment with bacteria colonization (animal experiment 2), a total of 18 male wild‐type C57BL/6 (6‐week‐old) were obtained from Moubaili Biotechnology Co., Ltd. (Wuhan, China) and housed in SPF‐animal room. Mice were separated in three groups (n = 6) and raised with standard normal‐chow diet. Antibiotic treatment was a mixture of antibiotics (ABX: 1 mg mL⁻¹ ampicillin, 0.5 mg mL⁻¹ vancomycin, 1 mg mL⁻¹ metronidazole, and 1 mg mL⁻¹ neomycin) and dissolved in anaerobic PBS. 0.2% tryptophan solutions were prepared freshly and were changed twice a week. For the PBS and ABX group, mice were received PBS or antibiotic mixture solution for 4 weeks, and 0.2% tryptophan supplied in water began at the third week. For the Lactobacillus spp. supplementation groups, Lactobacillus spp. were thawed under anaerobic conditions and diluted with anaerobic PBS to a final concentration of 2 × 10⁸ viable c.f.u. per 0.1 mL. Mice were first treated with an antibiotic mixture solution twice a week for 2 weeks to remove microbiota, and then orally administered with Lactobacillus spp. by gavage for another 2 weeks. All mice were sacrificed under isoflurane anesthesia after fasting for 8 h at the end of corresponding experimental period. Biological samples including bone, cecal contents, plasma, liver, spleen, ileum, and colon were collected and immediately stored at −80 °C for later experiments.

The human cohort studies were approved by the Ethics Committee of Hubei Provincial Hospital of Traditional Chinese Medicine (No: HBZY2020‐C47‐01, China) in accordance with the World Medical Association's Declaration of Helsinki. Human subjects were selected from Hubei Provincial Hospital of Traditional Chinese Medicine (Wuhan, China) and provided informed consent from all participants. Serum of healthy female individuals was collected and categorized into premenopausal (n = 32) and postmenopausal cohorts (n = 22) based on their ages (50 years old) and clinical menopausal diagnosis. For all subjects, fasting was required for 12 h prior to sampling. Targeted quantification of tryptophan metabolites in serum was performed using UHPLC‐QQQ‐MS descripted in detail as below.

The total DNA of gut bacteria was extracted from mouse cecal contents (∼100 mg) for 16S rRNA gene sequencing analysis. The 16S rRNA gene PCR process used 338F_806R primers targeting the V3‐V4 region (Table S2, Supporting Information) and was amplified with a KAPA HiFiHotStart PCR Kit (KAPA Biosystem, USA). Paired‐end 2 × 300 bp reads were sequenced on an Ilumina MiSeq platform by Shanghai Majorbio Bio‐pharm Technology Co., Ltd. Prior to deblurring, sequencing reads were processed using QIIME 2 (version 2020.11). Additionally, R packages were utilized to visualize both α‐ and β‐diversity. The taxonomy of OTUs was determined using the SILVA_134 database. Functional predictions were derived from 16S rRNA sequences employing PICRUSt2^[56^] (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States 2), and subsequently annotated with the KEGG (Kyoto Encyclopedia of Genes and Genomes) database.

Mice femurs were fixed in 4% paraformaldehyde for 2 days and stored in ethanol solution (70%). High‐resolution micro‐computed tomography (µCT) imaging was performed using NanoScan 1172 (Mediso). Skyscan (1276; Bruker). High‐resolution µCT scanner were set to 55 kVp and 200 µA. For the femoral trabecular region, 100 slices were analyzed, beginning 50 slices below the distal growth plate. 3D structure and morphometry were constructed and analyzed for bone volume/tissue volume ratio (BV/TV), bone mineral density (BMD), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular separation (Tb.Sp) in CTvol and Data‐viewer (Bruker Belgium).

Cells were cultured in a humidified 5% CO₂ incubator at 37 °C. Primary calvarial osteoblasts were isolated from newborn C57BL/6 mouse and bone marrow macrophages were isolated from 8‐week‐old C57BL/6 mice as previously published protocol.^[57^] Caco‐2 cells (Procell Life Science & Technology Co, Ltd.) and 293T cells were cultured in MEM basic medium (Gibco) with 10% FBS and 1% penicillin‐streptomycin. For plasmid infection, the cells were selected with puromycin for 2 days with 3 µg mL⁻¹. AhR knockdown in 293T cell was confirmed via qPCR and western blot analysis. In vitro experiments including treatments with LPS, Trp metabolites, and IL‐10 (Sino Biological) were performed.

Targeted quantification of Trp metabolites in feces of OVX mice models were conducted by ultrahigh‐performance liquid chromatograph (Agilent 1290) coupled with a model 6460 triple‐quadrupole mass spectrometer (UHPLC‐QQQ‐MS; Agilent Technologies, Inc.) in multiple‐reaction monitoring (MRM) mode. Data were analyzed via integral peak area by quantitative analysis program (Agilent). The procedures of sample preparation and LC‐MS detection were provided as previous publication.^[58^]

For RNA‐seq, TRIzol Reagent (Plant RNA Purification Reagent for plant tissue) was used to extract the total RNA from colon tissues according to the manufacturer's instructions (Invitrogen). RNA integrity was determined by Bioanalyzer (Agilent 2100) and quantified using the ND‐2000 (NanoDrop Technologies). Only high‐quality RNA samples RIN ≥ 6.5 were used for the subsequent experiments. The libraries were prepared following TruSeq RNA sample preparation Kit (Illumina, San Diego, CA). Next, paired‐end reads were sequenced with Illumina HiSeq Xten/NovaSeq6000 (2 × 150 bp read length). The following data was analyzed on Majorbio according to the standard protocols. Briefly, KEGG pathway enrichment analysis was performed using KOBAS (http://kobas.cbi.pku.edu.cn/home.do↗). Multiple testing was conducted using the Benjamini‐Hochberg (FDR) method, with a p‐value threshold of ≤ 0.05 to define significantly different KEGG pathways. Gene set enrichment analysis (GSEA) was employed to compare vehicle‐ and IAA‐treated mice using GSEA software (http://www.broad.mit.edu/GSEA↗).

293T cells with Ahr‐knockout were seeded in a 24‐well plate (Table S1, Supporting Information). For β‐catenin activity analysis, cells were transfected with either a control vector (pGL4‐basic) or a vector harboring the β–catenin promoter (pGL4‐Basic‐β–catenin‐promoter‐luc). The cells were co‐transfected with a Renilla luciferase vector pRL‐TK as an internal control using Lipofectamine 3000 according to the manufacturer's protocol. The medium was replaced by fresh medium or containing IAA and IPA (50 ng mL⁻¹) after 6 h. After 48 h transfection, the dual‐luciferase activity of β‐catenin was measured using a dual luciferase reporter gene assay kit II (Beyotime Biotechnology). The results were normalized to activity of the pRL‐TK control group.

Colon tissues of mice were immediately fixed in 4% paraformaldehyde solution. Colon tissues were then embedded in paraffin wax and sectioned 4 mm. Following de‐waxing in xylene and re‐hydrating through ethanol gradients, paraffin sections were stained and mounted for Alcian blue‐periodic acid Schiff (AB‐PAS). Images were taken by Olympus BX51 microscope and analyzed by Image J software (1.52v; USA). The number of goblet cells per unit length were counted by Image‐Pro Plus 6.0.

Fresh colon tissues of mice were fixed in 4% paraformaldehyde solution for 24 h, and then dehydrated, paraffin‐embedded and sectioned 3–4 µm. The sections were incubated with primary antibodies including ZO1 (1:1000; 21773‐1‐AP; Proteintech), E‐cadherin (1:200; 14472S; Cell Signaling), CD86 (1:500; Invitrogen), CD206 (1:500; Invitrogen), and β‐catenin (1:200; Proteintech). Corresponding secondary antibody was marked with appropriate species‐specific Alexa Fluor 488 antibodies (1:1000; A‐10680; ThermoFisher) and Alexa Fluor 647 Phalloidin (1:500; Solarbio) in 5% BSA in PBS in the dark at room temperature for 1 h. Nuclei were counterstained with DAPI (S2110; Solarbio). Three slices of colonic samples of each mouse were analyzed with 4 points in each slice. Thus, a total of 12 data points was statistically obtained. Nikon A1 confocal microscope was used to image samples.

Femurs of mice were immersed in 4% paraformaldehyde for 48 h and then decalcified with 0.5 m EDTA (Servicebio) at room temperature. Following continuous shaking for 21 days, samples were dehydrated, paraffin‐embedded, and sectioned 4–5 µm. Osteocalcin (OCN) and tartrate‐resistant acid phosphatase (TRAP) stains were obtained to detect OCN⁺ osteoblast and TRAP⁺ osteoclast within the entire ROI areas. OCN primary antibody (Cat. No gb11233; 1:200) and TRAP staining kit (Cat. No 387A‐1KT) were purchased from Servicebio (Wuhan, China) to measure the number of osteoblasts (N. Ocn⁺) and TRAP‐positive multinucleated osteoclasts (N. Trap+). Three slices with five regions in each slice were analyzed for distal femur with immunohistochemistry. Thus, a total of 15 data points was statistically obtained. Images were acquired with an Olympus optical microscope (Tokyo, Japan). The numbers of osteoblasts or osteoclasts were counted by Image Pro Plus 6.0 software (USA) by cells per bone perimeter (B. Pm).

Total RNA isolation was performed using RNAiso Plus reagent (TaKaRa) and RNA concentration was measured with NanoDrop at OD₂₆₀/OD₂₈₀. The cDNA was obtained by RNA reverse transcription with PrimeScriptTMRT Master Mix (TaKaRa) according to the manufacturer's instructions. QPCR reaction was performed using Green QPCR SuperMix (TransGen Biotech) with a QPCR system (ABI StepOne; Applied Biosystems). Mouse and human Gapdh or β–actin was used as the internal housekeeping gene, respectively and the expression levels were calculated with 2^−ΔΔCT method. The primers of genes used for QPCR were shown in Table S2 (Supporting Information).

Total proteins were obtained from colon and bone tissues or cells by RIPA (Beyotime Biotechnology) buffer containing with protease inhibitors (Beyotime Biotechnology). BCA protein assay kit (Sangon Biotech) was used to determine the protein concentration. Polyvinylidene fluoride (PVDF) membrane (Millipore) was used to immobilize and block the separated proteins by wet transfer systems (Bio‐Rad). The primary and secondary antibodies and dilution ratio are as below, E‐cadherin (1:1000; Cell Signaling), ZO‐1 (1:1000; Proteintech), AhR (1:2000; Proteintech), β‐catenin (1:5000; Proteintech), β‐Tublin (1:2000; Proteintech), β‐actin (1:5000; Proteintech), GAPDH (1:50 000; Proteintech) and secondary anti‐mouse/rabbit HRP‐conjugated antibodies (SA00001‐1 and SA00001‐2; Proteintech) were subsequently applied. Blots were analyzed by the enhanced‐chemiluminescence (ECL) HRP substrate (Millipore) and ChemiDoc Imager (Bio‐Rad).

Quantitative measurements of serum CTX, PINP, and IL‐10 were performed according to the manufacturer's instructions (Shanghai Huyu Biotechnology Co., Ltd).

Statistical data analyses were carried out in GraphPad Prism (version 8.0). Correlation network analyses were performed using R software (version 4.2.1) and illustrated with Cytoscape (version 3.7.0). Data were presented as mean ± SD. Unpaired and two‐tailed Student's t‐test were used for comparisons between two groups. One‐way ANOVA followed by Tukey's post hoc test was performed among multiple‐groups. For all experiments, P < 0.05 was set as the statistical significance.

The data that support the findings of this study are available from the corresponding author upon reasonable request.