IntroductionOsteoporosis is a highly prevalent systemic skeletal disorder characterized by reduced bone mass, microarchitectural deterioration, and increased fracture risk, particularly in older adults and postmenopausal women.1 A nationwide epidemiological study by Wang et al. reported that the prevalence of osteoporosis among Chinese adults aged ≥40 years was 5.0% in men and 20.6% in women.2 The pathogenesis of osteoporosis is largely driven by impaired bone homeostasis, a tightly regulated process maintained by the dynamic balance between osteoblast-mediated bone formation and osteoclast-mediated bone resorption. As one of the major progenitor sources of osteoblasts, bone marrow mesenchymal stem cells (BMSCs) are essential for osteogenic differentiation and the preservation of skeletal integrity. BMSCs possess multidirectional differentiation potential and can differentiate into osteoblasts, adipocytes, chondrocytes, and other lineages. Notably, the commitment of BMSCs toward the osteogenic lineage is closely linked to cellular metabolic reprogramming. During osteogenic differentiation of BMSCs, the predominant pathway of cellular energy supply shifts from glycolysis to oxidative phosphorylation (OXPHOS), which provides sufficient ATP and metabolic intermediates to support matrix synthesis, mineralization, and osteoblast maturation.3,4 Under aging or pathological conditions, mitochondrial dysfunction in bone tissue cells, including impaired mitochondrial biogenesis, disrupted membrane potential, and reduced OXPHOS activity, contribute to osteoporosis.5,6,7 Emerging evidence indicates that mitochondrial transplantation or pharmacologic activation of mitochondrial biogenesis or OXPHOS can enhance osteogenic differentiation and bone formation.8,9OXPHOS is tightly controlled not only at the transcriptional level but also through epitranscriptomic regulation of mitochondrial (mt) RNA metabolism.10 Beyond transcriptional control, post-transcriptional modifications, particularly on mitochondrial RNAs (mt-RNAs), have emerged as pivotal regulators of mitochondrial genome expression, respiratory chain activity, and overall metabolic homeostasis. RNA N1-methyladenosine (m1A) is a dynamic and reversible post-transcriptional modification that is deposited at the N1 position of adenosine by specific regulatory enzymes. By introducing a positive charge and disrupting the Watson-Crick base-pairing interface, m1A can reshape RNA secondary structure and modulate RNA-protein interactions, thereby influencing multiple post-transcriptional processes, including RNA metabolism and translation.11 m1A modification of mt-RNAs (mt-tRNAs, mt-rRNAs, and mt-mRNAs) has been shown to influence stability, processing, and translation of mt-RNAs, thereby affecting mitochondrial metabolic pathways, particularly OXPHOS.12tRNA methyltransferase 10C (TRMT10C) is a key mediator of mitochondrial RNA regulation and belongs to the family of mitochondrial tRNA methyltransferases. Structurally, TRMT10C consists of an N-terminal mitochondrial targeting sequence and a C-terminal methyltransferase catalytic domain, and has been reported to act on multiple mitochondrial RNA substrates.13 Previous studies have shown that TRMT10C contributes to mitochondrial function by facilitating the 5′ end cleavage of mt-tRNA and catalyzing m1A modification of mitochondrial transcripts, including mt-tRNAs and mt-mRNAs.14,15,16 However, TRMT10C can catalyze m1A modification of mt-tRNA to promote the translation of OXPHOS subunits,13,17,18,19 while it can also catalyze m1A modification of mt-mRNAs to suppress the translation of OXPHOS subunits.20 It remains unclear whether TRMT10C regulates OXPHOS subunit expression through m1A modification of mitochondrial rRNAs (mt-rRNAs). Metodiev et al. reported that the pathogenic mutations in the human TRMT10C gene (c.542G>T) caused a severe mitochondrial disorder characterized by lactic acidosis, dyspnea, and impaired cognitive function, resulting in early lethality within the first five months of life.17,20 These clinical findings highlight the essential role of TRMT10C in maintaining mitochondrial homeostasis and organismal survival. Mechanistically, in fibroblasts from patients with TRMT10C mutations, TRMT10C expression and mitochondrially encoded OXPHOS subunit protein levels are decreased, together with the accumulation of mt-RNA precursors, indicating impaired mt-RNA processing and defective mitochondrial protein synthesis.17 Together, these observations support the notion that TRMT10C is essential for proper mitochondrial RNA metabolism and respiratory chain function. In addition to mitochondrial disease, dysregulated TRMT10C expression is linked to cancer biology. Elevated TRMT10C expression is strongly linked to poor prognosis in various cancers, including hepatocellular carcinoma, glioma, and cervical cancer, and silencing TRMT10C impairs the proliferation, colony formation, and migration of ovarian and cervical cancer cells.21,22,23,24 In contrast, high TRMT10C levels are associated with better prognosis in renal clear cell carcinoma.25 Taken together, these findings indicate that TRMT10C is a critical regulator of mitochondrial RNA metabolism and cellular function.Despite these findings, the physiological role of TRMT10C in bone biology and osteoblast differentiation remains largely unclear. In this study, we generated osteoprogenitor cell-specific Trmt10c mutant mice (Osx-Cre; Trmt10cfl/fl) to investigate the function of TRMT10C in skeletal development and bone homeostasis. We found that loss of Trmt10c in osteoprogenitor cells resulted in growth retardation and bone loss, as evidenced by reduced bone strength, impaired bone formation, increased bone resorption, and multiple long bone fractures. These findings establish TRMT10C as a critical regulator of osteoblast differentiation and bone homeostasis in vivo. Mechanistically, we further demonstrated that, in addition to its previously reported roles in catalyzing m1A modification of mt-tRNA and mt-mRNA, TRMT10C also mediates m1A modification of mt-rRNA, thereby expanding the current understanding of its function in mitochondrial RNA regulation. Most importantly, through multistage virtual screening and subsequent validation, we identified a small-molecule agonist of TRMT10C that enhanced osteoblast bone formation and alleviated osteoporotic phenotypes in ovariectomized (OVX) and aged mice, highlighting the therapeutic potential of targeting TRMT10C for osteoporosis treatment.ResultsTRMT10C promotes osteoblast differentiation and is downregulated in osteoporosisTrmt10c mRNA was highly expressed in the skull and long bones of 12-week-old WT mice (Fig. 1a). Analyses of both whole-cell and mitochondrial lysates revealed that TRMT10C protein levels increased during osteogenic differentiation of BMSCs (Fig. 1a–f). In addition, the expression of mitochondrial mRNAs encoding OXPHOS subunits was uniformly downregulated in osteoblasts during mouse senile osteoporosis (GSE169608) (Supplementary Fig. 1a–o). In line with these findings, IHC staining and IF staining revealed a marked decrease in TRMT10C-positive osteoblasts on trabecular bone surfaces in both OVX and aged mice (Fig. 1g–n). Further analyses revealed that TRMT10C expression was significantly lower in osteoporotic BMSCs than in healthy controls (Fig. 1o–p).Fig. 1The alternative text for this image may have been generated using AI.Full size imageTRMT10C was upregulated during osteogenic differentiation but downregulated in osteoporotic osteoblasts. a Trmt10c mRNA levels in various tissues of 6-week-old mice. n = 4. b–f TRMT10C protein levels in total and mitochondrial lysates during osteogenic differentiation. n = 3. g, h IHC staining for TRMT10C in ovariectomized female mice and the number of TRMT10C-positive cells on the trabeculae. n = 5. i, j IF staining for TRMT10C and ALP in ovariectomized female mice, and TRMT10C-positive osteoblasts on the trabeculae were counted. n = 3. k, l IHC staining for TRMT10C in aged male mice and the number of TRMT10C-positive cells on the trabeculae was determined. n = 5. m, n IF staining for TRMT10C and ALP in aged male mice, and TRMT10C-positive osteoblasts on the trabeculae were counted. n = 3. o, p TRMT10C protein levels in BMSCs derived from healthy individuals and patients with osteoporosis (OP). n = 3. Comparisons were performed using one-way ANOVA in (c, d, f) and unpaired two-tailed t tests in (h, j, l, n, p)The results of functional experiments revealed that TRMT10C overexpression enhanced osteogenic differentiation, whereas TRMT10C knockdown markedly inhibited this process. These effects were evidenced by ALP staining, alizarin red staining, and Western blotting analysis of key osteogenic markers, including Runx2, osterix (OSX), ALP, and osteocalcin (OCN) (Supplementary Fig. 2a–j). Collectively, these data demonstrate that TRMT10C is positively associated with osteoblast function and osteoporosis progression.TRMT10C regulates mitochondrial ribonuclease P activity and catalyzes m1A modification of mt-rRNA/tRNAThe knockdown of Trmt10c resulted in significant decreases in both the mRNA and protein levels of OXPHOS subunits, accompanied by the accumulation of unprocessed mitochondrial transcript precursors (Supplementary Fig. 3a-d).TRMT10C (c.542G>T; p.R176L) mutation disrupts mt-tRNA processing without affecting m1A9 methyltransferase activity,17 we introduced the corresponding mutation (R176L) into mouse Trmt10c plasmid and transfected that into BMSCs from Trmt10c cKO mice. This resulted in a partial recovery of mt-mRNA levels but failed to restore OXPHOS protein expression (Supplementary Fig. 4a-d). Given that the human TRMT10C residue D314 is essential for methyltransferase activity,13 we also generated a mouse D309N point mutation. Compared with the R176L mutation, transfection of the mouse Trmt10cD309N construct into cKO BMSCs led to a more substantial increase in mitochondrial transcript levels but failed to rescue the translation of OXPHOS subunits (Supplementary Fig. 4a-d). Collectively, these findings indicate that both mitochondrial ribonuclease P activity and m1A modification activity on mt-RNA mediated by TRMT10C play a dominant role in regulating the expression of OXPHOS subunits.eCLIP sequencing demonstrated that TRMT10C could bind to mt-rRNA, mt-tRNA and mt-mRNA (Fig. 2a–e). Analysis of the three BMSC samples revealed 123 genes that exhibited common interactions with TRMT10C (Fig. 2b, c). These included 2 mt-rRNAs (4 peaks) and 6 mt-tRNAs (16 peaks) (Fig. 2d). Genomic mapping revealed that the mt-rRNAs contained TRMT10C-binding regions (Fig. 2e). mt-mRNAs mostly lack UTR sequences, and m1A modification of their coding sequence inhibits the posttranscriptional translation process.20 TRMT10C-mutant cells exhibit reduced expression levels of OXPHOS subunits.17 Therefore, TRMT10C promotes the expression of OXPHOS subunits independent of m1A methylation of mt-mRNAs.Fig. 2The alternative text for this image may have been generated using AI.Full size imageTRMT10C catalyzes m1A methylation of mt-rRNA/mt-tRNA. a Schematic illustration of the experimental workflow combining eCLIP-seq and m1A-MAP-seq performed on BMSCs to identify TRMT10C-bound RNA targets and their corresponding m1A modification sites regulated by TRMT10C (Created in BioRender. Zhang, C. (2026) https://BioRender.com/hovpcke). b Bar graph showing the total number of TRMT10C-bound peaks identified by eCLIP-seq in three independent biological replicates of BMSC samples. c Venn diagram illustrating the overlapping and unique numbers of TRMT10C peak-associated genes identified across three independent BMSC samples. d Bar graph quantifying the distribution of TRMT10C-bound regions across different RNA species, including miRNAs, snoRNAs, rRNAs, and tRNAs, as determined by eCLIP-seq. e Representative genome browser track (coverage plot) showing TRMT10C binding to specific mt-rRNAs. f Bar graph indicating the number of differentially m1A-modified sites on tRNA and rRNA transcripts in Trmt10c knockdown BMSCs compared with control BMSCs, as determined by m1A-MAP-seq. g Venn diagram quantifying the number of mitochondrial rRNA (mt-rRNA) sites with significantly decreased m1A modification following TRMT10C knockdown in BMSCs, as analyzed by m1A-MAP-seqTo comprehensively characterize the m1A methylation landscape following TRMT10C knockdown we employed m1A-MAP-seq to profile modifications of mitochondrial RNA species (Supplementary Fig. 5a, b). Notably, while m1A methylation levels on mt-tRNAs remained markedly stable, we observed significant changes in m1A modifications at multiple sites within mt-rRNAs (Fig. 2f, Supplementary Fig. 5c, d). Given the methyltransferase activity of TRMT10C, we integrated eCLIP binding profiles with m1A-MAP data to identify loci that exhibit both TRMT10C association and reduced m1A modifications. TRMT10C knockdown led to the downregulation of m1A modifications across 24 mt-rRNA sites; notably, loci including ChrM:1882, ChrM:1862, and ChrM:1895 displayed particularly profound alterations (Fig. 2g). To our knowledge, these specific positions have not been previously reported. These findings suggest that TRMT10C ablation in bone marrow stromal cells (BMSCs) predominantly modulates m1A modifications on mitochondrial ribosomal RNAs, potentially disrupting mitochondrial translational machinery and downstream cellular processes.Here, we demonstrate that TRMT10C can bind to and catalyze the m1A modification of mt-rRNA, in addition to its known role in catalyzing the m1A modification of mt-tRNA and mt-mRNA.14Trmt10c deletion in osteoprogenitor cells resulted in growth retardation and bone lossTo determine the in vivo role of TRMT10C in regulating bone homeostasis in osteoblasts, we generated osteoprogenitor cells with Trmt10c knockout (Osx-Cre; Trmt10cfl/fl, hereafter referred to as Trmt10c cKO), and successful knockout of TRMT10C in BMSCs was confirmed (Supplementary Fig. 6a, b). These BMSCs exhibited markedly reduced proliferation, as evidenced by a lower growth rate, decreased PCNA/Ki67 expression, and decreased EdU incorporation (Supplementary Fig. 6c–g). Trmt10c cKO mice exhibited growth retardation, tooth abnormalities, and multiple fractures (Fig. 3a–c, Supplementary Fig. 6h–l), as well as a significant reduction in bone strength (Supplementary Fig. 7a–c).Fig. 3The alternative text for this image may have been generated using AI.Full size imageTrmt10c deletion in osteoprogenitor cells resulted in growth retardation and reduced bone mass. a Skeletal staining of 2-day-old neonatal mice. b Gross physical pictures of 12-week-old mice. c X-ray images of the trabecular bone of the proximal tibial metaphysis; the red arrow indicates an old fracture line. d The reconstructed images from the μCT analysis of the proximal tibial metaphysis and (e–l) histomorphometric parameters of the cancellous and cortical regions were analyzed. m Histological sections were stained with ABH/OG, and (n–q) the number and area of adipocytes were quantified in male and female mice. n = 8. Comparisons were conducted using two-tailed Student’s t testsWe assessed alterations in bone formation in Trmt10c cKO mice. Osteogenic differentiation was downregulated in the BMSCs from the Trmt10c cKO mice (Supplementary Fig. 7d–f). ELISA analysis confirmed a decrease in the serum level of the bone formation marker PINP in Trmt10c cKO mice (Supplementary Fig. 7g). μCT analyses revealed significant reductions in both trabecular and cortical bone mass, as well as bone mineral density, at the distal metaphysis of the tibia in Trmt10c cKO mice (Fig. 3d–l). Histological staining revealed a reduction in osteoblast number and an increase in both adipocyte number and area in Trmt10c cKO mice (Supplementary Fig. 7h–j, Fig. 3m–q). Additionally, dynamic histomorphometric parameters, including the mineral apposition rate (MAR), mineralizing surface/bone surface ratio (MS/BS), and bone formation rate/bone surface ratio (BFR/BS), were significantly decreased in Trmt10c cKO mice (Supplementary Fig. 7k–o).We assessed alterations in bone resorption in Trmt10c cKO mice. Osteoclast differentiation was weakened in bone marrow cells from Trmt10c cKO mice, as evidenced by TRAP staining and reduced expression of osteoclastic differentiation factors, including nuclear factor of activated T cells 1 (NFATc1), cathepsin K (CTSK), and TRAP (Supplementary Fig. 8a–d). ELISA revealed decreased levels of the bone resorption markers CTX1 and ACP5 in Trmt10c cKO mice (Supplementary Fig. 8e, f). Histological TRAP staining revealed a significant decrease in the number of osteoclasts in Trmt10c cKO mice (Supplementary Fig. 8g–i). In addition, we found that the expression of RANKL and OPG in BMSCs was reduced, with a significantly decreased RANKL/OPG ratio (Supplementary Fig. 8j–l).TRMT10C agonist screening and functional validationTo identify novel TRMT10C agonists, we employed a structure-based virtual screening strategy. Although the precise mechanism for agonizing TRMT10C remains unclear, we hypothesized that allosteric modulators could represent a viable approach to enhance its enzymatic activity. Therefore, we first utilized the PASSer server to systematically predict potential allosteric sites across multiple TRMT10C structures, including the cryo-EM structure (PDB: 7ONU), crystal structure (PDB: 5NFJ) and the AlphaFold-predicted full-length model. After excluding the canonical substrate (SAM) binding site, we identified and selected four high-confidence putative allosteric pockets for further investigation (Fig. 4a).Fig. 4The alternative text for this image may have been generated using AI.Full size imageVirtual screening and functional validation of small-molecule compounds as potential agonists of TRMT10C. a Predicted allosteric binding pockets in the TRMT10C structure. The crystal structure of TRMT10C (PDB: 5NFJ) was used as the template and is shown as a ribbon representation. Predicted allosteric pockets are highlighted as molecular surfaces in distinct colors. b Workflow of the structure-based virtual screening process for identifying TRMT10C agonists. c Western blot analysis of OXPHOS subunit expression levels in BMSCs treated with increasing concentrations of compound TM3. d CETSA results showing the thermal stabilization of the TRMT10C protein in BMSCs after TM3 treatment and the quantification of TRMT10C solubility by CETSA at different temperatures; n = 3. e Surface plasmon resonance (SPR) sensorgrams and fitted binding curves for the interaction between purified TRMT10C and TM3. The data are presented as the mean ± SEM. Statistical significance was determined by two-tailed Student’s t test (d)We subsequently performed a multistage virtual screening campaign against the top-ranked allosteric pockets. A library of more than 1,000,000 commercial compounds was sequentially screened using Glide modules within the Schrödinger suite through high-throughput virtual screening (HTVS), standard precision (SP), and extra precision (XP) docking protocols. The top 100 hits from the docking studies were further refined by calculating their binding free energies using the MM/GBSA method. The final selection of 30 candidate compounds for experimental validation was prioritized according to MM/GBSA scores and structural diversity, followed by manual inspection by experienced medicinal chemists (Fig. 4b).Next, BMSCs were treated with the top 30 candidate compounds individually. Initial assessment based on the increased protein expression level of CYTB (a key mitochondrial-encoded OXPHOS subunit) revealed six initial hits, including compounds TM1, TM3, TM6, TM9, TM15, and TM24 (Supplementary Fig. 9a). These six candidates were further evaluated in BMSCs for their ability to upregulate a broader panel of mitochondrial genome-encoded OXPHOS subunits (including ND1-ND6, CYTB, COX1-COX3, ATP6, and ATP8). From this secondary screen, compound TM3 emerged as the most potent inducer of OXPHOS subunit expression (Supplementary Fig. 9b). A dose‒response assay confirmed that 10 nM was the optimal concentration for TM3 to promote OXPHOS subunit expression in BMSCs (Fig. 4c, Supplementary Fig. 9c).To confirm that TM3 exerts its effects through direct interaction with TRMT10C, we performed cellular thermal shift assays (CETSA) and surface plasmon resonance (SPR) experiments. CETSA results demonstrated that treatment with TM3 significantly stabilized the TRMT10C protein against thermal denaturation in BMSCs, indicating direct intracellular target engagement (Fig. 4d). SPR spectroscopy using the purified recombinant TRMT10C protein further verified this interaction, revealing direct binding with a dissociation constant (Kd) of 2.6 μM (Fig. 4e).We next systematically evaluated the in vivo safety and pharmacokinetic profile of the agonist TM3. Intraperitoneal injection of TM3 in 8-week-old wild-type mice resulted in no significant alterations in body weight, liver function markers (ALT and AST), kidney function markers (UREA and CRE-BD), or histopathology of major organs, as assessed by H&E staining (Supplementary Fig. 10a–e). TM3 also exhibited no acute toxicity in C57BL/6J mice at doses up to 100 mg/kg/day (Supplementary Fig. 11a). All the mice survived the 3-day treatment period, with a 100% survival rate, and no overt signs of toxicity were observed (Supplementary Fig. 11b). Body weight remained stable throughout the study (Supplementary Fig. 11c–e). Biochemical analysis revealed no increase in hepatic enzymes, as ALT/GPT and AST/GOT activities were comparable between the TM3-treated and vehicle groups (Supplementary Fig. 11f, g). Furthermore, comprehensive hematological assessment revealed that TM3 did not affect any blood cell parameters, including WBC, RBC, HGB, HCT, MCV, MCH, MCHC, PLT, PCT, RDW, PDW, or MPV, all of which remained within normal ranges without significant intergroup differences (Supplementary Fig. 11h–s). These findings demonstrate that TM3 is well tolerated acutely, with no hepatotoxicity or hematotoxicity at the tested doses.To establish a dosing regimen with an optimal therapeutic window, we first evaluated the pharmacokinetic profile of TM3 at an exploratory dose of 2 mg/kg. This study confirmed favorable drug-like properties, including a sustained plasma half-life (T1/2 = 1.53 h) and high systemic exposure (Cmax = 7.3 µM) (Supplementary Fig. 10g–h). To prioritize a wide safety margin and minimize potential off-target effects while retaining efficacy, we selected a substantially lower dose of 50 µg/kg for subsequent therapeutic studies (Supplementary Fig. 10i, j). Pharmacokinetic analysis at this efficacy dose revealed a Cmax of 0.19 µM (~44.6 ng/mL), which provides an ~19-fold exposure margin over the in vitro EC50 (10 nM).A TRMT10C agonist promotes osteoblast differentiation and protects against osteoporosisWe then sought to determine whether the TRMT10C agonist TM3 could reduce bone loss induced by aging and ovariectomy. In vitro functional assays confirmed that TM3 treatment potently enhanced osteogenic differentiation of mouse BMSCs, as evidenced by ALP staining, alizarin red staining and elevated protein levels of key osteogenic markers, including Runx2, Osx, ALP, and OCN (Supplementary Fig. 12a-c). However, TM3 failed to promote osteogenic differentiation in the BMSCs derived from the Osx-Cre;Trmt10cfl/fl mice (Supplementary Fig. 12d-f). In addition, TM3 treatment enhanced osteogenic differentiation in human BMSCs (Supplementary Fig. 12g-i).To evaluate its efficacy in vivo, we employed an ovariectomy (OVX)-induced osteoporosis model. Twelve-week-old C57BL/6 wild-type female mice received TM3 or vehicle (50 µg/kg/day) by intraperitoneal injection for 8 weeks, and TM3 treatment had no significant effect on body weight (Fig. 5a). TM3 administration significantly increased the expression levels of m1A and OXPHOS subunits in bone tissue (Fig. 5b, c). μCT analysis of the proximal tibia revealed that TM3 treatment significantly increased trabecular bone mass and bone mineral density in both the sham-operated and OVX mice (Fig. 5d-l). Serum biomarker analysis by ELISA revealed that TM3 treatmentsignificantly increased the levels of the bone formation marker PINP and the bone resorption markers CTX-1 and ACP5 in both the sham and OVX mice (Supplementary Fig. 13a-c). Consistent with the improved bone microarchitecture, histological staining revealed increased numbers of osteoblasts and osteoclasts, accompanied by decreased adipocyte numbers and areas, in both the sham and OVX mice treated with TM3 (Supplementary Fig. 13d-g, Fig. 5m-o). Furthermore, TM3 treatment significantly increased dynamic histomorphometric parameters, including the MAR, MS/BS, and BFR/BS, in both the sham and OVX mice (Supplementary Fig. 13h-l).Fig. 5The alternative text for this image may have been generated using AI.Full size imageTM3 reduced bone loss induced by ovariectomy. TM3 was intraperitoneally injected into OVX mice at a dosage of 50 µg/kg/d for 8 weeks, and μCT analysis was performed on the proximal tibial metaphysis. a Mouse body weight was measured weekly. b, c IHC staining of m1A, ND3 and COX1 in tibial sections; positive cells were counted; n = 8. d The reconstruction images of the trabecular bone were generated, and (e-i) the histomorphometric parameters of the cancellous bone were analyzed; n = 8. j Reconstructed images of the cortical bone were generated, and (k-l) histomorphometric parameters of the cortical bone were analyzed; n = 8. m Histological sections were stained with ABH/OG, and (n-o) the number and area of adipocytes were quantified. n = 8. Comparisons were performed using unpaired two-tailed t tests in (c) and two-way ANOVA followed by Tukey’s test (e-i, k, l, n–o)We further assessed the therapeutic potential of TM3 in a model of age-related osteoporosis. Twelve-month-old male C57BL/6 mice were treated with vehicle or TM3 (50 μg/kg/day, i.p.) for 10 weeks, and compared with control mice, TM3-treated mice exhibited a reduction in body weight (Supplementary Fig. 14a). TM3 administration significantly increased the expression levels of m1A and OXPHOS subunits in bone tissue (Supplementary Fig. 14b, c). TM3 administration significantly ameliorated age-associated bone loss. μCT and histomorphometric analyses confirmed that TM3 treatment led to increased osteoblast and osteoclast numbers and elevated bone mass and mineral density (Supplementary Fig. 14d-p). Consistently, serum analyses revealed increased levels of the bone formation marker PINP and the bone resorption markers CTX-1 and ACP5 (Supplementary Fig. 14q-s). Additionally, a significant reduction in marrow adiposity was observed, as indicated by decreased adipocyte number and area (Supplementary Fig. 14t-v). Collectively, these results demonstrate that the TRMT10C agonist TM3 effectively promotes bone formation and protects against bone loss in both estrogen-deficient and aged murine models of osteoporosis.A schematic model illustrating how TRMT10C promotes osteoblast differentiationTRMT10C facilitates mitochondrial RNA processing and m¹A methylation, thereby increasing the expression of mitochondrial genome-encoded OXPHOS subunits and increasing OXPHOS activity. This process promotes osteogenic differentiation of BMSCs and protects against osteoporosis (Fig. 6).Fig. 6The alternative text for this image may have been generated using AI.Full size imageA schematic model illustrating how TRMT10C promotes osteoblast differentiation. TRMT10C facilitates mitochondrial RNA processing and m¹A methylation, thereby increasing the expression of mitochondrial genome-encoded OXPHOS subunits and increasing OXPHOS activity, which in turn promotes the osteogenic differentiation of BMSCs. Furthermore, virtual screening revealed a TRMT10C agonist (TM3), providing a promising therapeutic compound and target for the treatment of osteoporosisDiscussionMitochondria are dynamic organelles that are essential for mediating energy production, ion homeostasis, reactive oxygen species generation and clearance, and apoptotic signaling. They play an especially critical role in highly energy-demanding cell types, including osteoblasts.26 In undifferentiated BMSCs, OXPHOS is maintained at a low level, and the energy supply relies primarily on glycolysis.3 Wnt7b-LRP5-axis-driven glycolytic reprogramming may play an important role in meeting the energetic and biosynthetic demands of osteoblast precursor cells, such as BMSCs. Overexpression of Wnt7b in osteoblasts (Osx-CreERT; OE-Wnt7b) significantly increased bone mass and bone density in mice by increasing glycolytic reprogramming.27,28 As BMSCs differentiate into osteoblasts, OXPHOS activity is upregulated, and become the main energy source, with a significant increase in both the quantity and the quality of mitochondria to meet the energy demands for bone matrix synthesis.4,29 Therefore, the increase in bone mass in Osx-CreERT; OE-Wnt7b mice may primarily result from the regulation of undifferentiated BMSCs, thereby indirectly promoting osteogenic differentiation. Consistent with these findings, pharmacological agonists of OXPHOS promote osteogenic differentiation of BMSCs.4,30 In contrast, treatment with oligomycin A (an ATP synthase inhibitor), antimycin A (a mitochondrial complex III inhibitor), or siRNA targeting the Atp5a1 gene at the onset of osteogenic induction significantly impaired osteogenic differentiation.31,32,33Mitochondrial transcription factor A (TFAM) is a nuclear DNA-encoded protein that is essential for mtDNA maintenance and regulates mitochondrial transcription and replication. Yoshioka et al. reported that limb mesenchyme-specific Tfam knockout mice (Prx1-Cre; Tfamfl/fl) displayed markedly shortened forelimbs from birth and developed spontaneous postnatal fractures, ultimately leading to severe limb deformities.34 Mutations in Trmt10c (such as c.542G>T) have been linked to severe mitochondrial disorders characterized by OXPHOS deficiency and early-onset mortality.17,35 In this study, we revealed that TRMT10C expression is downregulated in osteoblasts from both OVX mice and aged mice, whereas its expression increases during BMSC differentiation toward osteoblasts. Functionally, we demonstrated that TRMT10C enhances the osteogenic capacity of BMSCs. To elucidate its physiological role, we generated an osteoprogenitor-specific Trmt10c conditional knockout mouse model, which exhibited severe growth retardation, low bone mass, and spontaneous long-bone fractures at 12 weeks of age. Collectively, the similarities observed between Prx1-Cre; Tfamfl/fl mice and Osx-Cre; Trmt10cfl/fl mice suggest that the loss of TRMT10C phenocopies a more general mitochondrial OXPHOS defect.34The accumulation of marrow adipocytes in the tibial medullary cavity of Osx-Cre;Trmt10cfl/fl mice may have arisen from multiple mechanisms. First, the Osx-Cre driver primarily targets osteoprogenitor (Osx-positive) cells rather than mature osteoblasts and can also label a subset of bone marrow mesenchymal stromal/stem cells.36,37 Therefore, a proportion of Osx-positive progenitors may further differentiate into other mesenchymal lineages, particularly adipocytes, thereby contributing to altered marrow adipocyte differentiation and accumulation within the bone marrow cavity. Second, the lineage allocation of BMSCs toward osteoblasts versus adipocytes is intrinsically competitive, and the dynamic balance between these two fates is essential for maintaining the equilibrium between bone formation and marrow adiposity.38,39TRMT10C has been reported to regulate mitochondrial function through facilitating the 5′ end processing of mt-tRNA and mediating m1A modifications on mt-tRNA and mt-mRNA.14,15,16 In this study, we observed that BMSCs from Trmt10c conditional knockout mice exhibited a marked reduction in mature mt-mRNA abundance, accompanied by a concomitant decrease in precursor mt-mRNA transcripts, the latter likely arising from OXPHOS dysfunction caused by the loss of TRMT10C, which in turn indirectly impaired the transcriptional processing of the mitochondrial genome. Mt-mRNAs lack untranslated regions, and m1A modification within their coding sequences has been reported to suppress translation.20 Both the previous literature and our present findings demonstrate that reduced TRMT10C expression leads to the downregulation of the expression of OXPHOS subunits encoded by the mitochondrial genome. Thus, TRMT10C promotes the expression of mitochondrial genome-encoded OXPHOS subunits independent of m1A modification on mt-mRNAs, most likely through its m1A modification activity on mitochondrial mt-tRNAs and mt-rRNAs.We provide novel evidence that TRMT10C catalyzes m1A modification on mt-rRNA, expanding its repertoire beyond mt-tRNA and mt-mRNA modification.14 However, the m1A modification sites on mt-rRNA delineated in the current study exhibit notable divergence from those previously reported. We propose that this disparity is likely attributable to species- and cell-type-specificities (e.g., Safra et al. identified nine m1A sites on human mt-rRNA in HEK293T cells).40 In addition, the identified m1A sites were cross-analyzed with TRMT10C eCLIP data, thereby excluding alterations in m1A modification sites in mt-rRNA that could result from the indirect effects of TRMT10C downregulation. Furthermore, our eCLIP data indicate that TRMT10C interacts with a broad spectrum of other RNAs, including miRNAs, snoRNAs, cytoplasmic rRNAs, and pseudogene transcripts. The functional significance of these interactions represents an important area for future investigations.Using a structure-based virtual screening approach, we identified a series of TRMT10C-targeting compounds. Through subsequent in vitro functional validation, we identified TM3 as a potent TRMT10C agonist that enhances OXPHOS subunit expression and promotes osteogenic differentiation of BMSCs. Most importantly, TM3 administration significantly attenuated bone loss in both OVX and aged mice. However, TM3 treatment did not markedly increase cortical bone mass or bone mineral density in aged mice. This limited response may be attributable to an age-associated decrease in the levels of TRMT10C and OXPHOS activity in bone cells (including osteoblasts and osteoclasts), which could attenuate the pharmacological efficacy of TM3. In addition, although TM3 treatment increased bone mass, it concomitantly enhanced both bone formation and bone resorption, indicating a high-bone-turnover therapeutic profile. Therefore, in subsequent translational studies, bone mass should be closely and longitudinally monitored.In conclusion, our findings establish TRMT10C as an essential regulator of bone acquisition and homeostasis, which enhances mitochondrial RNA processing and m1A-dependent epitranscriptomic regulation, thereby upregulating OXPHOS subunit expression, increasing oxidative phosphorylation and ATP production, and ultimately supporting OXPHOS-dependent cellular programs, including osteogenic differentiation and cell proliferation. Furthermore, we identified the small-molecule agonist TM3 as a promising candidate for mitigating osteoporosis through targeted activation of TRMT10C.Materials and methodsMiceTrmt10c-flox mice (strain S-CKO-11527) were purchased from Cyagen (Suzhou, China). Osx-Cre mice, which target osteoprogenitor cells, were purchased from Biocytogen (#110131; Beijing, China) and crossed with Trmt10cfl/wt mice to obtain Osx-Cre;Trmt10cfl/wt mice. These mice were subsequently further bred with Trmt10cfl/fl mice to generate Osx-Cre;Trmt10cfl/fl mice for bone phenotypic analysis in comparison to Trmt10cfl/fl mice. The primers used for animal genotyping are listed in Supplementary Table 1.An OSX-Cre mouse model obtained from Biocytogen (#110131; Beijing, China) was generated using the CRISPR/Cas9 strategy, and compared with control mice, OSX-Cre mice exhibited no significant differences in tooth or bone development.41 In contrast, the Sp7-tTA–tetO-Cre system (OSX1-Cre) was established using a transgenic approach and has several inherent limitations, including random integration sites, unstable expression levels, and potential toxicity or growth-related defects.42Ten-week-old C57BL/6 wild-type male mice, 12-week-old C57BL/6 wild-type female mice and 12-month-old C57BL/6 wild-type male mice were purchased from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). Female mice underwent ovariectomy followed by intraperitoneal injection of vehicle or a TRMT10C agonist at 50 µg/kg/day for 8 weeks. Twelve-month-old male mice received vehicle or a TRMT10C agonist at 50 µg/kg/day by intraperitoneal injection for 10 weeks.Ten-week-old C57BL/6 wild-type male mice were randomly divided into three groups (n = 5 per group) and received intraperitoneal injections of vehicle (PBS) or TM3 (50 or 100 mg/kg/day) on the first day. Body weight and survival were recorded daily. On day 3, blood samples were collected from the submandibular venous plexus of the mice for biochemical and hematological analyses and complete blood counts.The animal study was conducted in accordance with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines. All the animals were raised in a standard environment with controllable humidity (40–60%) and temperature (22–26 °C) and a light/dark cycle of 12 h, and autoclaved food and water were provided to the mice ad libitum. All animal experimental procedures were approved by the Animal Ethics Committee of The Second Hospital, Cheeloo College of Medicine, Shandong University (number: KYLL2024781).Skeletal stainingGenotyping was performed on postnatal day 2, at which time the mice were euthanized. Their skin and internal organs were removed. The samples were fixed in a 95% ethanol solution for 3 days, followed by immersion in acetone for 2 days to remove fat and keep the sample firm. Alizarin red/Alcian blue staining was then performed for 3 days. After staining, the samples were immersed in 1% KOH solution until the red bone tissue and blue cartilage became clearly visible. Images were then captured for preservation. The staining solution was prepared as follows: 1 mL of 0.1% Alizarin Red (dissolved in 95% ethanol), 1 mL of 0.3% Alcian blue (dissolved in 75% ethanol), 1 mL of glacial acetic acid, and 17 mL of 75% ethanol solution.Mechanical testAfter 12-week-old mice were euthanized, the femurs were immediately harvested and placed on a three-point bending test platform (Bose ElectroForce® 3230, Tianjin Hospital), ensuring that both ends of the femur were in contact with the support points. A gradually increasing force was applied at the midshaft of the femur using the loading device until the bone fractured. Key data were recorded throughout the test, and the mechanical performance parameters of the femur were subsequently analyzed.RNA extraction and real-time RT‒PCRMitochondria were isolated from BMSCs using a cell mitochondria isolation kit (C3601; Beyotime Biotechnology, Wuhan, China). Total RNA and mitochondrial RNA were extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). All reverse transcription was performed using RR037A (TaKaRa Bio, Otsu, Japan), the precursors of the mitochondrial mRNA transcripts were reverse-transcribed with Random 6-mers, and the mRNA was reverse-transcribed with Oligo dT Primer. Quantitative real-time RT-PCR amplification was performed using ChamQ Universal SYBR qPCR Master Mix (Q711-03, VAZYME, Nanjing, China). The primers used are listed in Supplementary Table 2.Cell culturesPrimary BMSCs from 8-week-old mice were isolated and cultured as described previously.43 BMSCs at passages 3–6 were used to perform experiments. Osteogenic differentiation was induced in the BMSCs supplemented with 10 nM TRMT10C agonist. BMSCs from Osx-Cre;Trmt10cfl/fl and Trmt10cfl/fl mice were utilized to induce osteogenic differentiation. Afterward, the protein expression levels of osteogenic factors were determined at 3 days post-induction, and alkaline phosphatase (ALP) and alizarin red staining were performed at 14 and 21 days. The cell growth rate and proliferation ability were determined using a Cell Counting Kit-8 (PF00004; ProteinTech, Wuhan, China) and a BeyoClick™ EdU Cell Proliferation Kit with Alexa Fluor 488 (C0071S; Beyotime Biotechnology, Wuhan, China), respectively.Mouse bone marrow monocytes (BMMs) from 10 to 14-day-old Osx-Cre; Trmt10cfl/fl and Trmt10cfl/fl mice were isolated and seeded on Petri dishes in the presence of M-CSF (100 ng/mL) (#11792-HNAH; Sino Biological, Beijing, China) for 3 days and then seeded on 48-well plates in the presence of M-CSF (100 ng/mL) and 1 × 10−8 M 1,25(OH)2D3 (5097210001; Merck, USA) for 6–8 days. The cells were then fixed and stained for tartrate-resistant acid phosphatase (TRAP) using a LEUKOCYTE ACID PHOSPHATASE (TRAP) Kit (387A-1KT; Sigma-Aldrich, St. Louis, MO, USA).Constructs, siRNAs and transfectionsThe mouse Trmt10c overexpression plasmid was cloned and inserted into the vector pCMV3 (Sino Biological, Beijing, China), and the mouse Trmt10cD309N mutation plasmid and the mouse Trmt10cR176L mutation plasmid were constructed on the basis of the Trmt10c overexpression plasmid (Sino Biological, Beijing, China). All the vectors or plasmids were transfected into the BMSCs using LipofectamineTM 2000 Reagent (#11668019; Invitrogen, Carlsbad, CA). Mouse Trmt10c siRNAs were synthesized (GENERAL BIOL, Chuzhou, China) and transfected into BMSCs using Lipofectamine RNAiMAX Transfection Reagent (#13778075, Invitrogen, Carlsbad, CA). The sequences of the siRNAs are shown in Supplementary Table 3.Sequencing and analysisSingle-cell RNA sequencing data acquisition and osteoblast identification. Murine bone marrow single-cell RNA sequencing (scRNA-seq) data from 1-, 6-, and 20-month-old mice were retrieved from the Gene Expression Omnibus (GEO) under accession number GSE169608. To identify osteoblasts, we utilized a marker gene strategy. Populations expressing Runx2, Osterix, osteopontin (Opn), and alkaline phosphatase (Alp) were classified as osteoblasts. This marker combination encompassed a range of osteoblast differentiation states, ensuring a comprehensive analysis of Trmt10c expression across the osteoblast lineage.Analysis of Trmt10c expression in osteoblasts. Having defined the osteoblast population, we then investigated the expression dynamics of Trmt10c in these cells across the three age groups. We employed a combination of dimensionality reduction (e.g., t-SNE or UMAP) and clustering algorithms to further refine the osteoblast subpopulations, allowing nuanced comparisons of Trmt10c expression. Differential gene expression analysis was subsequently performed to identify age-dependent changes in Trmt10c levels within each redefined osteoblast subtype. This involved rigorous statistical testing, with p values adjusted for multiple comparisons using the Benjamini‒Hochberg method.Enhanced cross-linking immunoprecipitation–high-throughput sequencing (eCLIP) was performed as follows. When the mouse BMSCs reached 90% confluence, the cells were washed twice with precooled PBS. Immediately irradiate at 254 nm for 400 mJ/cm² using a UVP crosslinker. The cells were collected into centrifuge tubes using a 1.5 ml Eppendorf tube and then centrifuged at 4 °C and 1000 × g for 10 min. The pellet was resuspended in PBS and centrifuged twice, after which eCLIP sequencing was performed by Shanghai Diatre Biological Technology Co., Ltd. (Shanghai, China).The sequence of m1A modification on mt-tRNA/rRNA was determined as follows. Mouse BMSCs were transfected with Trmt10c siRNA using Lipofectamine RNAiMAX Transfection Reagent (#13778075; Invitrogen, Carlsbad, CA) for 48 h. Mitochondria were isolated using a cell mitochondria isolation kit (C3601; Beyotime Biotechnology, Wuhan, China), mitochondrial RNA was then extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA), and m1A modifications on mt-tRNA/rRNA were sequenced by Cloudseq Biotech InC (Shanghai, China).Multistage virtual screening approachThe crystal structure of TRMT10C (PDB: 5NFJ), the cryo-electron microscopy structure (PDB: 7ONU), and the full-length predicted structure from AlphaFold (AF-Q7L0Y3-F1) were used for comprehensive binding site analysis. Potential allosteric pockets on the protein surface were systematically characterized using the PASSer algorithm. The top-ranked predicted allosteric site, which is distinct from the canonical S-adenosylmethionine (SAM) binding site, was selected for subsequent screening. Commercial compound libraries (SPECS and ChemDiv, totaling >1,000,000 molecules) were subjected to a triple-tiered docking protocol using Glide (Schrödinger Suite): HTVs, standard precision (SP), and extra precision (XP). The top-ranked compounds from each stage were advanced to the next stage, resulting in 100 high-scoring hits. These hits were further evaluated by MM/GBSA binding free energy calculations to refine the selection. The final selection of 30 compounds prioritized not only favorable predicted binding affinities but also chemical diversity and drug-like properties, as assessed through manual curation by experienced medicinal chemists.TRMT10C agonist screening and validationPrimary BMSCs from 6 to 8-week-old mice were treated with each of the 30 candidate small molecules (50 nM) for 12 h. The effects on TRMT10C activity were initially assessed by measuring CYTB protein expression via Western blotting, as CYTB is a key downstream subunit regulated by TRMT10C. This initial screen revealed six lead compounds that most potently upregulated CYTB expression. BMSCs were subsequently treated with these six candidate molecules (50 nM, 12 h), and the protein levels of multiple mitochondrial genome-encoded oxidative phosphorylation (OXPHOS) subunits (ND1-6, CYTB, COX1-3, ATP6, and ATP8) were comprehensively analyzed by Western blotting. On the basis of its superior efficacy in enhancing the expression of this panel of OXPHOS proteins, the most potent agonist was selected and designated TRMT10C-ATx.To determine the optimal effective concentration, BMSCs were treated with TRMT10C-ATx at various concentrations (0, 1, 5, 10, 20, and 50 nM) for 12 h. The protein expression levels of OXPHOS subunits and cellular ATP production were measured. The concentration that produced the maximal response in both assays was selected for all subsequent in vitro experiments.Cellular thermal shift assay (CETSA)Mouse BMSCs were treated with TRMT10C-ATx or vehicle control for 2 h. The cells were then lysed, and the lysates were heated at different temperatures (47 °C, 52 °C, 57 °C, 62 °C, and 67 °C) for 5 minutes. Precipitated proteins were removed by centrifugation, and the soluble fraction was analyzed by Western blotting using an anti-TRMT10C antibody (1:2000; ProteinTech, 29087-1-AP). The stabilization of the TRMT10C protein, indicated by a higher melting temperature (Tm) shift in the compound-treated group, confirmed target engagement.Surface plasmon resonance (SPR)TRMT10C (residues 203-403) was amplified from human-derived cell line cDNA and inserted into a modified RSFduet-1 vector (Novagen) with an N-terminal 6×His-SUMO tag. The protein was expressed in E. coli strain BL21(DE3) (WeiDi) and purified by two rounds of Ni‒NTA affinity chromatography to remove the 6×His-SUMO tag. The flow-through fraction was further purified on a HiTrap SP FF column and a Superdex 200 Increase 10/300 column (Cytiva).SPR analyses were performed on a Biacore T200 system (GE Healthcare) using series S CM5 sensor chips (Cytiva). The purified TRMT10C proteins were immobilized on the surface of the sensor chip by amine coupling to yield a signal of approximately 12,000 response units. All the Biocore data were collected at 25 °C in PBST running buffer (8 mM Na2HPO4, 2 mM NaH2PO4, 137 mM NaCl, 0.5% Tween 20, pH 7.4). TRMT10C agonist 3 was dissolved in running buffer and injected at concentrations ranging from 0.05 μM to 1 mM. The dissociation constants (KD) of the interactions were calculated by using a steady-state affinity model in Biacore T200 evaluation software.Pharmacokinetic studySix- to eight-week-old female wild-type mice were intraperitoneally injected with a TRMT10C agonist (2 mg/kg/day), and blood samples were then collected from the orbital sinus at various time points: 5 min, 15 min, 30 min, 60 min, 2 h, 4 h, 6 h, 8 h, and 24 h. The samples were stored overnight at 4 °C and then centrifuged at 3000 rpm for 15 min to collect the serum. The concentration of the TRMT10C agonist was determined using LC‒MS/MS, and pharmacokinetic data analysis was performed using WinNonlin version 8.2.Systemic evaluation and tissue distributionSix- to eight-week-old female wild-type mice were intraperitoneally injected with a TRMT10C agonist (2 mg/kg/day) for 8 weeks. The daily body weights of the mice were recorded, and serum was collected to measure serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), blood urea nitrogen (BUN), creatinine (CREA), and uric acid (UA). The hearts, livers, spleens, lungs, and kidneys were fixed, embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E).Six- to eight-week-old female wild-type mice were intraperitoneally injected with a TRMT10C agonist (2 mg/kg). After 10 and 30 min, the mice were anesthetized with isoflurane and perfused with PBS, after which the femur and tibia were dissected and separated. The tissues were homogenized using a homogenate solution (methanol: ddH2O = 1:1), and the concentration of the TRMT10C agonist was determined using LC-MS/MS.Enzyme-linked immunosorbent assay (ELISA)The serum levels of procollagen type I N-terminal peptide (PINP) (EK16107, SAB, College Park, MD), C-terminal telopeptide of type I collagen (CTX-1) (EK13528, SAB, College Park, MD) and acid phosphatase 5, tartrate resistant (EK12408, SAB, College Park, MD) were measured following the manufacturer’s protocols.Radiography and microcomputed tomographyThe mice were anesthetized with isoflurane, and X-ray imaging was performed using a Micro Focus X-ray imaging source (Precision X-ray Inc., North Branford, USA). The tibias were scanned by an in vivo-80 μCT scanner (Scanco Medical, Switzerland), and three-dimensional reconstruction was conducted with associated software.Histology experimentsTibial and femur samples were isolated, fixed with 4% formalin for 2 days, decalcified, embedded in paraffin or OCT, and cut at thicknesses of 5 μm and 10 μm. OCT-embedded tissue sections were subjected to immunofluorescence staining for TRMT10C (1:200; Sigma-Aldrich, HPA036671) and ALP (1:200; HUABIO, M0608-10). Paraffin sections were stained with Alcian blue/hematoxylin/Orange G (ABH/OG), and immunohistochemistry for TRMT10C (1:200, ProteinTech, 29087-1-AP), ND3 (1:200, ABclonal, A17969), COX1 (1:200, HUABIO, HA722838) and ALP (1:200, HUABIO, ET1601-21) with a GTVisionTM III detection system/Mo&Rb (GK500705, GeneTech, Shanghai, China).Dynamic bone histomorphometryOn the seventh and fifth days before euthanasia, the mice were given intraperitoneal injections of calcein (10 mg/kg). After euthanasia, the femurs were collected, fixed, and embedded in methyl methacrylate resin. Sections (8 μm) were then prepared, and dynamic histomorphometric parameters were measured and calculated using OsteoMeasure software (OsteoMetrics, Atlanta, GA, USA). The region of interest begins below the primary spongiosa and spans a length of 1 mm.Western blotting analysisThe BMSCs proteins were extracted with RIPA buffer (P0013; Beyotime Biotechnology, Wuhan, China) and analyzed with a BCA kit (P0010; Beyotime Biotechnology, Wuhan, China). Immunoblotting was performed with primary antibodies against TRMT10C (1:2000, ProteinTech, 29087-1-AP), Osterix (1:2000, Bosterbio, A02077-1), Osteopontin (1:2000, HUABIO, #0806-6), Runx2 (1:2000, ABCAM, ab23981), Alkaline Phosphatase (1:2000, HUABIO, ET1601-21), RANKL (1:2000, ProteinTech, 23408-1-AP), Osteoprotegerin (1:2000, ABCAM, ab183910), Cathepsin K (1:2000, ABCAM, ab187647), NFATc1 (1:2000, ProteinTech, 66963-1-Ig), ND1 (1:2000, ProteinTech, 19703-1-AP), ND2 (1:2000, ProteinTech, 19704-1-AP), ND3 (1:2000, ABclonal, A17969), ND4 (1:2000, ProteinTech, 26736-1-AP), ND4L (1:2000, HUABIO, ER60898), and ND5 (1:2000, GAPDH (1:5000, ProteinTech, 10491-1-AP) or β-actin (1:5000, ProteinTech, 81115-1-RR) was used to normalize total protein, and TOMM20 (1:2000, ProteinTech, 11802-1-AP) was used to normalize MT protein.Statistical analysisAll the data are presented as the mean ± SD. The cell experiments were independently repeated three or five times, as described in the figure or table legends. All animal experiments were independently repeated eight times. Two-tailed unpaired Student’s t test was used for comparisons between two groups. One-way or two-way ANOVA was employed for comparisons among multiple groups, followed by Dunnett’s test or Tukey’s test, respectively, for post hoc comparisons as described in the figure legends. p