MicroRNA-29a Mitigates Osteoblast Senescence and Counteracts Bone Loss through Oxidation Resistance-1 Control of FoxO3 Methylation

Senescent osteoblast overburden accelerates bone mass loss. Little is understood about microRNA control of oxidative stress and osteoblast senescence in osteoporosis. We revealed an association between microRNA-29a (miR-29a) loss, oxidative stress marker 8-hydroxydeoxyguanosine (8-OHdG), DNA hypermethylation marker 5-methylcystosine (5mC), and osteoblast senescence in human osteoporosis. miR-29a knockout mice showed low bone mass, sparse trabecular microstructure, and osteoblast senescence. miR-29a deletion exacerbated bone loss in old mice. Old miR-29a transgenic mice showed fewer osteoporosis signs, less 5mC, and less 8-OHdG formation than age-matched wild-type mice. miR-29a overexpression reversed age-induced senescence and osteogenesis loss in bone-marrow stromal cells. miR-29a promoted transcriptomic landscapes of redox reaction and forkhead box O (FoxO) pathways, preserving oxidation resistance protein-1 (Oxr1) and FoxO3 in old mice. In vitro, miR-29a interrupted DNA methyltransferase 3b (Dnmt3b)-mediated FoxO3 promoter methylation and senescence-associated β-galactosidase activity in aged osteoblasts. Dnmt3b inhibitor 5′-azacytosine, antioxidant N-acetylcysteine, or Oxr1 recombinant protein attenuated loss in miR-29a and FoxO3 to mitigate oxidative stress, senescence, and mineralization matrix underproduction. Taken together, miR-29a promotes Oxr1, compromising oxidative stress and FoxO3 loss to delay osteoblast aging and bone loss. This study sheds light on a new antioxidation mechanism by which miR-29a protects against osteoblast aging and highlights the remedial effects of miR-29a on osteoporosis.


Clinical Specimens
Studies on human bone biopsies were approved by the Chang Gung Medical Foundation Institutional Review Board (Affidavit #202000823B0). Informed consent was obtained from patients preoperatively. Thirteen patients with osteoporosis (5 males and 8 females) and 13 patients without osteoporosis (6 males and 7 females) who required lumbar spine decompression, fixation, discectomy, or vertebroplasty were enrolled in the osteoporosis group and control group, respectively. The BMD of the hip was quantified using dualenergy X-ray absorptiometry before surgery. Leftover bone biopsies and 10 mL peripheral blood were harvested.

miR-29a Knockout Mice
Animal experiments were approved by Institutional Animal Use and Care Committee, Kaohsiung Chang Gung Memorial Hospital (Affidavit #2014120401). The homologous arms of C57BL/B6 mouse miR-29a gene (Chromosome 6, 31029595-31049694) (RP23-173G8) of Bacterial Artificial Chromosome (BAC) library (BACPAC Genomics Resource Center; Emeryville, CA, USA) were cloned using PCR protocols. The loxP site and the neomycin (Neo) resistance cassette (PL452) were inserted into PL253 plasmid (Addgene, Watertown, MA, USA). The Neo gene driven by T7 promoter in mammalian cells was removed by Cre recombinase. The linearized BAC construct was transferred into embryonic stem cells of C57BL/6 mice to produce miR-29a chimera, which further crossed with Sox2-Cre mice (Jackson Laboratory, Bar Harbor, ME, USA) to breed miR-29aKO mice. The genotype of miR-29aKO mice was confirmed using customized primers (Supplementary Table S1) and Antioxidants 2021, 10, 1248 3 of 17 PCR protocols. Mice were raised in a specific pathogen-free laboratory animal facility with feed and drinking water ad libitum.

Age-Mediated Osteoporosis
Three-month-old and 9-month-old male miR-29aTg or WT mice were divided into young and old groups, respectively. To label bone mineralization, 10 mg/kg calcein was injected 9 days and 3 days before euthanasia. Peripheral blood was drawn and centrifuged to harvest serum. Femurs and tibiae specimens were dissected for subsequent experiments.

µCT Analysis of Bone Tissue and Marrow Adipose
The microstructure of femurs and tibiae was visualized using SkyScan 1176 µCT (Bruker, Belgium) and captured 200 radiographs (9-µm voxel size), as previously described [24]. BMD (g/cm 3 ), trabecular volume (BV/TV, %), trabecular thickness (Tb.Th, mm), and trabecular number (Tb.N/mm) of proximal tibiae, were measured using SKYSCAN ® CT-Analysis software, according to the maker's instructions. In a subset experiment, µCT radiography of marrow fat in tibiae was performed upon soaking in OsO 4 , as previously described [25]. The images of OsO 4 -soaked specimens were reconstructed and merged using the software. Marrow fat volume (mm 3 ) and fat surface (mm 2 ) were calculated automatically.

Biomechanical Test
Upon measuring the cross-section area of the middle region of femurs using µCT scanning protocols, the biomechanical properties, including breaking force (N/mm 2 ) and energy (N.mm), of the middle region of femurs upon 3-point bending were quantified using SHIMADZU EZ-SX Material Test System (Shimadzu Corporation, Kyoto, Japan) with TRAPEZIVMX software, according to the maker's instructions.

Ex Vivo Osteogenesis, Adipocyte, and Osteoclast Formation
Mesenchymal cells and macrophage precursor cells in bone marrow were isolated, as previously described [24]. Bone-marrow stromal cells (10 5 cells/well, 24-well plates) were seeded in an osteogenic condition using StemProTM Osteogenesis Differentiation Kits (A1007201 Thermo Fisher Scientific Inc., Waltham, MA, USA) and in an adipogenic condition using StemProTM Adipogenic Differentiation Kits (A1007001) for 21 days and 15 days, respectively. Mineralized matrices and adipocytes were stained using von Kossa stain kits and Nile Red stain kits (Abcam, Cambridge, UK). Bone-marrow macrophage precursor cells (5 × 10 4 cells/well, 48-well plates) were incubated in αMEM with 15 ng/mL M-CSF and 40 ng/mL RANKL (R&D Systems, Minneapolis, MN, USA) for 1 week. Osteoclasts were probed using TRAP stain kits. Von-Koss-stained mineralized matrices in each low-power field (x125 magnification) and Nile-red-stained adipocytes and TRAP-stained osteoclasts in each high-power field were measured using the Zeiss Image Analysis System.

RT-PCR
Total RNA was extracted from WT and miR-29aTg osteoblasts using TRI Reagent™ Solution (Thermo Fisher Scientific Inc., Waltham, MA, USA). High-Capacity cDNA Reverse Transcription Kits (Thermo Fisher Scientific Inc., Waltham, MA, USA) were utilized to reversely transcribe 1 µg of total RNA. PCR was investigated using primers (Supplementary  Table S1), 2× TaqMan ® Universal PCR Master Mix with ABI 7900 Detection System (Applied Biosystems, Foster City, CA, USA), and the threshold value for amplification reaction was probed. Equation 2 −∆∆Ct was used to calculate the fold change of mRNA expression.

Whole Genome Microarray Assay
A total of 10 µg of RNA was amplified using Amino Allyl MessageAmp II aRNA amplification kits (Ambion, Thermo Fisher Scientific Inc., Waltham, MA, USA) and coupled with CyDye using CyDye Post-Labeling Reactive Dye Packs (Cytiva, Marlborough, MA, USA), according to the makers' instructions. Upon purification, the labeled aRNA was hybridized onto Mouse MOA2.0 whole-genome arrays (Phalanx Biotech Inc., San Diego, CA, USA). Data and clustering were processed using Rosetta Resolver Biosoftware ® . Fold change of the gene expression log2 > 1 was considered differentially expressed. Gene ontology and KEGG pathway were clustered using Kobas bioinformatics engine (http://kobas.cbi.pku.edu.cn, accessed on 19 July 2021), as previously described [30].

Methylation-Specific PCR
Genomic DNA in 5 × 10 6 osteoblasts was harvested using MagMAX™-96 DNA Multi-Sample Kits (4413021; Invitrogen, Danvers, MA, USA). Methylated DNA was probed using the EZ DNA Methylation™ Kit (Zymo Research, Irvine, CA, USA), according to the maker's manuals. A total of 10 ng genomic DNA was mixed with 100 µL CT Conversion Reagent with bisulfite at 50°C in the dark for 16 h and incubated at 4°C for 10 min. Upon eluting through Zymo-Spin™ IC Column, the elute was washed with an M-Wash Buffer and mixed with a M-Desulphonation Buffer and an Elution Buffer. PCR analysis of the eluted DNA was performed using the ABI 7900 Detection System together with specific primers, methylated and unmethylated FoxO3 promoter, and miR-29a promoter. PCR analysis of GADPH gene was performed to confirm the equal loading of elute DNA (Supplementary Table S1).

Statistical Analysis
Differences of age, hip BMD, serum miR-29a, and immunostaining in the non-osteoporosis and osteoporosis groups were analyzed using a Wilcoxon test. The patients' gender of both groups was analyzed using a Chi-square test. The investigations of WT and miR-29aKO mice were analyzed using a Student's t-test. The investigations of young and old WT and miR-29aTg mice and in vitro models were analyzed using an ANOVA test and a Bonferroni post hoc test. p value < 0.05 resembled significant difference.

miR-29a Loss, Senescent Osteoblast Overburden, and Oxidative Stress in Human Osteoporosis
We examined if oxidative stress or osteoblast senescence or miR-29a expression correlated with human osteoporosis. We collected bone biopsies and serum from 13 patients with Antioxidants 2021, 10, 1248 6 of 17 osteoporosis (5 males and 8 females; 75.8 ± 1.1 years old) and 13 patients without osteoporosis (6 males and 7 females; 36.6 ± 1.9 years old) who required lumbar spine decompression, fixation, discectomy, or vertebroplasty. Serum miR-29a levels ( Figure 1a) and bone mineral density of hips (Figure 1b) in patients with osteoporosis were less than in patients without osteoporosis. Serum miR-29a level significantly correlated with the development of osteoporosis (Figure 1c). Plenty of osteoblasts in osteoporotic bone specimens showed strong cellular senescence markers, including β-galactosidase and p16 Ink4a immunostaining (Figure 1d), as compared to the control group. Bone cells in the osteoporosis group also displayed strong DNA methylation marker 5-methylcytosine (5mC) immunoreaction and oxidative damage marker 8-hydroxydeoxyguanosine (8-OHdG) (Figure 1e) immunoreactivity, suggesting that DNA hypermethylation and oxidative stress were present in the development of osteoporosis. and a Bonferroni post hoc test. p value < 0.05 resembled significant difference.

miR-29a Loss, Senescent Osteoblast Overburden, and Oxidative Stress in Human Osteoporosis
We examined if oxidative stress or osteoblast senescence or miR-29a expression correlated with human osteoporosis. We collected bone biopsies and serum from 13 patients with osteoporosis (5 males and 8 females; 75.8 ± 1.1 years old) and 13 patients without osteoporosis (6 males and 7 females; 36.6 ± 1.9 years old) who required lumbar spine decompression, fixation, discectomy, or vertebroplasty. Serum miR-29a levels ( Figure 1a) and bone mineral density of hips (Figure 1b) in patients with osteoporosis were less than in patients without osteoporosis. Serum miR-29a level significantly correlated with the development of osteoporosis (Figure 1c). Plenty of osteoblasts in osteoporotic bone specimens showed strong cellular senescence markers, including β-galactosidase and p16 Ink4a immunostaining (Figure 1d), as compared to the control group. Bone cells in the osteoporosis group also displayed strong DNA methylation marker 5-methylcytosine (5mC) immunoreaction and oxidative damage marker 8-hydroxydeoxyguanosine (8-OHdG) (Figure 1e) immunoreactivity, suggesting that DNA hypermethylation and oxidative stress were present in the development of osteoporosis. Figure 1. Analysis of miR-29a, osteoblast senescence, oxidative stress, and DNA methylation in human osteoporosis. Decreased serum miR-29a levels (a) and hip BMD (b) in the osteoporosis group. Serum miR-29a levels correlated with osteoporosis (c). Data are mean ± standard error calculated from 13 patients without osteoporosis and 13 patients with osteoporosis who required spine surgery. Osteoblasts in age-mediated osteoporotic bone showed strong β-galactosidase, p16 (d), 5mC, and 8-OHdG (e) immunostaining (arrows) (scale bar, 20 μm). Immunohistochemical data (mean ± standard error) are calculated from 7-8 bone biopsies. Significant difference (asterisks *) was analyzed using a Wilcoxon test. Figure 1. Analysis of miR-29a, osteoblast senescence, oxidative stress, and DNA methylation in human osteoporosis. Decreased serum miR-29a levels (a) and hip BMD (b) in the osteoporosis group. Serum miR-29a levels correlated with osteoporosis (c). Data are mean ± standard error calculated from 13 patients without osteoporosis and 13 patients with osteoporosis who required spine surgery. Osteoblasts in age-mediated osteoporotic bone showed strong β-galactosidase, p16 (d), 5mC, and 8-OHdG (e) immunostaining (arrows) (scale bar, 20 µm). Immunohistochemical data (mean ± standard error) are calculated from 7-8 bone biopsies. Significant difference (asterisks *) was analyzed using a Wilcoxon test.

Transgenic Overexpression of miR-29a Slowed Bone Loss in Old Mice
Given that miR-29a loss accelerated bone mass loss and microstructure deterioration, we asked whether transgenic overexpression of miR-29a in bone tissue changed the skeletal integrity of old animals. We bred osteoblast-specific miR-29a transgenic mice under the control of the Bgp promoter (miR-29aTg) [24] (Figure 3a). We divided 3-and 9-month-old mice into young and old mouse groups, respectively. miR-29a expression in bone tissue ( Figure 3a) and serum bone formation marker Bgp levels were downregulated in old WT animals, whereas serum resorption marker C-telopeptide of collagen I (CTX-1) levels were increased (Figure 3b). miR-29a loss and serum bone marker alteration were compromised in old miR-29aTg mice. Old WT mice developed sparse bone microstructure (Figure 3c mised in old miR-29aTg mice. Old WT mice developed sparse bone microstructure ( Figure  3c) as evident from decreased BMD, BV/TV, and Tb.N (Figure 2d). Old miR-29aTg mice showed mild BMD loss and a well-connected trabecular bone network. Old WT bone tissue developed marrow adiposity as evidenced in increases in marrow fat volume and surface (Figure 3e). These animals had less bone biomechanical strength, including breaking force and energy (Figure 3f), than young WT mice. miR-29a overexpression compromised marrow fat overproduction and biomechanics loss in old mice. Figure 3. Effects of miR-29a overexpression on bone mass, mechanics, and histomorphometry of young and aged mice. Schematic drawing for the linear gene construct for the generation of osteoblast-specific miR-29aTg mice. miR-29a overexpression attenuated miR-29a loss in bone tissue (a) and serum Bgp and CTX-1 level alteration in old mice (b). μCT revealing sparse trabecular microstructure in old WT mice and well-woven bone architecture in old miR-29aTg mice (c); scale bar, 50 μm in upper panels; 90 μm in lower panels. miR-29a overexpression improved age-induced loss in BMD, BV/TV, and Tb.N (d). miR-29a compromised fatty marrow development (scale bar, 5 mm) (e) and biomechanics loss (f) and reversed mineral acquisition (scale bar, 40 μm) (g) and von-Koss-stained trabecular bone volume (scale bar, 40 μm) (h), as well as reduced TRAP-stained osteoclast formation (scale bar, 20 μm) (i). miR-29a overexpression attenuated βgalactosidase, 5mC, and 8-OHdG immunoreactivity (j); scale bar, 15 μm. The significant difference (asterisks *) of investigations (mean ± standard error, n = 5-6 mice) was analyzed using an ANOVA test and Bonferroni post hoc test. Schematic drawing for the linear gene construct for the generation of osteoblast-specific miR-29aTg mice. miR-29a overexpression attenuated miR-29a loss in bone tissue (a) and serum Bgp and CTX-1 level alteration in old mice (b). µCT revealing sparse trabecular microstructure in old WT mice and well-woven bone architecture in old miR-29aTg mice (c); scale bar, 50 µm in upper panels; 90 µm in lower panels. miR-29a overexpression improved age-induced loss in BMD, BV/TV, and Tb.N (d). miR-29a compromised fatty marrow development (scale bar, 5 mm) (e) and biomechanics loss (f) and reversed mineral acquisition (scale bar, 40 µm) (g) and von-Koss-stained trabecular bone volume (scale bar, 40 µm) (h), as well as reduced TRAP-stained osteoclast formation (scale bar, 20 µm) (i). miR-29a overexpression attenuated β-galactosidase, 5mC, and 8-OHdG immunoreactivity (j); scale bar, 15 µm. The significant difference (asterisks *) of investigations (mean ± standard error, n = 5-6 mice) was analyzed using an ANOVA test and Bonferroni post hoc test.
Old WT bone tissue showed less von-Kossa-stained trabecular bone morphology (BV/TV; Figure 3g) and decreased bone mineral acquisition (MAR; Figure 3h) together with plenty of tartrate-resistant acid-phosphate-stained osteoclasts developed along trabecular bone (Oc.N/mm; Figure 3i) as compared to young WT mice. miR-29Tg bone revealed a mild loss in static and dynamic bone formation and osteoclast overburden. A high number of osteoblasts in old WT bone showed strong β-galactosidase, 5mC, and 8-OHdG immunoreactivity. Mild senescence, DNA methylation, and oxidative stress were present in osteoblasts in old miR-29aTg mice (Figure 3j).

miR-29a Attenuated Senescence and Osteogenesis Loss in Old Mice
Consistent with the analysis of senescent osteoblast overburden in osteoporotic bone, senescence program, including p16 Ink4a , p21 CIP expression (Figure 4a), inflammatory cytokine IL-6 expression (Figure 4b), and SA-β-gal activity (Figure 4c) were significantly increased in bone-marrow mesenchymal cells of old WT mice. However, the osteogenesis of bone-marrow stromal cells as evidenced in von-Kossa-stained mineralized matrix production (Figure 4d), Runx2, and Bgp expression (Figure 4e) was reduced in old WT mice. Senescence program, inflammation, and osteogenesis loss were improved in old Antioxidants 2021, 10, 1248 9 of 17 miR-29aTg mice. On the other hand, osteoclast factor RANKL expression (Figure 4f) and fluorescent Nile-red-stained adipocyte formation of bone-marrow stromal cells (Figure 4g), as well as TRAP-stained osteoclast formation of bone-marrow macrophage precursor cells (Figure 4h), were upregulated in old WT mice. The ex vivo adipocytic activity and osteoclastogenic differentiation capacity were compromised in old miR-29aTg mice.

miR-29a Attenuated Senescence and Osteogenesis Loss in Old Mice
Consistent with the analysis of senescent osteoblast overburden in osteoporotic bone, senescence program, including p16 Ink4a , p21 CIP expression (Figure 4a), inflammatory cytokine IL-6 expression (Figure 4b), and SA-β-gal activity (Figure 4c) were significantly increased in bone-marrow mesenchymal cells of old WT mice. However, the osteogenesis of bone-marrow stromal cells as evidenced in von-Kossa-stained mineralized matrix production (Figure 4d), Runx2, and Bgp expression (Figure 4e) was reduced in old WT mice. Senescence program, inflammation, and osteogenesis loss were improved in old miR-29aTg mice. On the other hand, osteoclast factor RANKL expression (Figure 4f) and fluorescent Nile-red-stained adipocyte formation of bone-marrow stromal cells (Figure 4g), as well as TRAP-stained osteoclast formation of bone-marrow macrophage precursor cells (Figure 4h), were upregulated in old WT mice. The ex vivo adipocytic activity and osteoclastogenic differentiation capacity were compromised in old miR-29aTg mice.

miR-29a Repressed DNMT3b-Mediated Foxo3 Methylation
The analysis of less DNA methylation and mild FoxO3 loss in an old miR-29aTg bone microenvironment was used to verify if the microRNA affected FoxO3 methylation in aged osteoblasts. Multiple passaged osteoblasts (passage 10, P10) ( Figure 6a) were incubated as a model of osteoblast aging [26][27][28]. miR-29a expression ( Figure 6b) and FoxO3 level were reduced upon multiple passages, whereas 5mC levels were increased ( Figure  6c). Forced miR-29a expression by miR-29a mimetic downregulated DNA methylation, reversing FoxO3 signaling in multiple-passaged osteoblasts (Figure 6c). The methylation of CpG dinucleotide around −4057~−2058 bp in the transcription start site of the FoxO3 promoter (Figure 6d) was significantly upregulated in multiple-passaged cells. miR-29a attenuated FoxO3 promoter hypermethylation in aged osteoblasts (Figure 6d), as is evident from the bisulfite conversion of DNA and methylation-specific PCR analysis. Transcriptomic profiles of miR-29aTg and wild-type osteoblasts were analyzed in 3 mice. Orx1, Glrx, Gsr, Cyc1, Prdx5, Runx2, and Bgp mRNA expression was increased in miR-29aTg osteoblasts, as evidenced in RT-PCR analysis (e). Significant difference of investigations (mean ± standard error, n = 4 mice) was analyzed using a Student's t-test. Orx1 and FoxO3 loss in bone-marrow mesenchymal cells was compromised in old miR-29aTg mice (f). Significant differences (asterisks *) of investigations (n = 5 mice) were analyzed using an ANOVA test and a Bonferroni post hoc test.

Oxr1 Attenuated Oxidative Stress-Induced miR-29a Loss and Senescence
Given that miR-29a and Orx1 loss were present in senescent osteoblasts, we investigated what role antioxidant protein Oxr1 may play in miR-29a control of osteoblast senescence. Osteoblasts were incubated in H 2 O 2 as an in vitro model of oxidative stress 2induced cellular senescence [29]. Murine MC3T3-E1 osteoblasts were incubated in an osteogenic medium with or without H 2 O 2 or Oxr1 recombinant protein for 15 days.

Discussion
Osteogenic cell senescence involves postnatal bone homeostasis, microstructure integrity, or osteoporosis development [33]. Programmed senescence in osteogenic progenitor cells sustains long bone development in late pubertal mice [34]; however, senescent osteoblast overburden in a skeletal microenvironment correlates with the development of age [3], chemotherapy [35], and irradiation [36]-induced bone mass loss and architecture damage. Accumulating studies reveal that oxidative stress [37] and the epigenetic pathway [38], including DNA methylation and microRNA signaling, regulate cell fate and metabolism. Little is known about the epigenetic pathway control of oxidative stress in osteoblast aging and bone loss. This study uncovered that miR-29a slowed osteoblast senescence, preserving extracellular mineralized matrix production to protect from bone loss in old mice. miR-29a attenuated FoxO3 loss and oxidative stress through repressing Dnmt3b-mediated antioxidant protein Oxr1 loss and DNA hypermethylation in aged osteoblasts. Collective evidence conveys a new insight into the antioxidation and anti-aging potential of miR-29a against bone loss.
Investigations of clinical specimens uncovered a plethora of reactions, such as miR-29a loss, DNA hypermethylation, osteoblast senescence, and oxidative stress in human osteoporosis. Expanding studies show increased p16 expression [39] or DNA methylome alteration [13] in bone biopsies of human postmenopausal osteoporosis. This analysis prompted us to investigate what role miR-29a may play in DNA methylation and oxidative stress in osteoporosis. of investigations (mean ± standard error, n = 3 experiments) was analyzed using an ANOVA test and a Bonferroni post hoc test. Schematic graphs showing miR-29a protection against oxidative stress, osteoblast senescence, and osteoporosis. miR-29a targets Dnmt3b-mediated Foxo3 promoter hypermethylation and increased antioxidant protein Oxr1, compromising age-induced Foxo3 loss, oxidative stress, osteoblast senescence, and osteoporosis development (f).

Discussion
Osteogenic cell senescence involves postnatal bone homeostasis, microstructure integrity, or osteoporosis development [33]. Programmed senescence in osteogenic progenitor cells sustains long bone development in late pubertal mice [34]; however, senescent osteoblast overburden in a skeletal microenvironment correlates with the development of age [3], chemotherapy [35], and irradiation [36]-induced bone mass loss and architecture damage. Accumulating studies reveal that oxidative stress [37] and the epigenetic pathway [38], including DNA methylation and microRNA signaling, regulate cell fate and metabolism. Little is known about the epigenetic pathway control of oxidative stress in osteoblast aging and bone loss. This study uncovered that miR-29a slowed osteoblast senescence, preserving extracellular mineralized matrix production to protect from bone loss in old mice. miR-29a attenuated FoxO3 loss and oxidative stress through repressing Dnmt3b-mediated antioxidant protein Oxr1 loss and DNA hypermethylation in aged osteoblasts. Collective evidence conveys a new insight into the antioxidation and anti-aging potential of miR-29a against bone loss.
Investigations of clinical specimens uncovered a plethora of reactions, such as miR-29a loss, DNA hypermethylation, osteoblast senescence, and oxidative stress in human osteoporosis. Expanding studies show increased p16 expression [39] or DNA methylome alteration [13] in bone biopsies of human postmenopausal osteoporosis. This analysis prompted us to investigate what role miR-29a may play in DNA methylation and oxidative stress in osteoporosis.
We, for the first time, have revealed phenotypes of low bone mass in miR-29a knockout mice, suggesting that this molecule was required to maintain skeletal tissue integrity. The adverse effect of miR-29a loss on bone homeostasis is also pointed out by a study manifesting that transgenic overexpression of miR-29-3p tough decoy in mice inhibits bone mineral accretion and trabecular bone volume [40]. Nine-month-old mice develop osteoporosis, including low bone mass or trabecular volume [26][27][28][29]. In this study, the extent of skeletal deterioration was worsened in 9-month-old miR-29aKO mice. miR-29a deletion induced DNA hypermethylation and osteoblast senescence. The impact of miR-29a loss on osteoblast fate and bone integrity in this murine model further explained the histological features of senescent osteoblast overburden in human osteoporotic bone.
Striking findings were that old miR-29Tg mice showed fewer osteoporosis signs, including mild MAR loss and osteoclast formation, than age-matched WT mice. miR-29a appeared to attenuate excessive bone turnover by preserving the osteogenic differentiation capacity of mesenchymal progenitor cells and reducing RANKL expression to downregulate osteoclastic activity. The senescence-associated secretory phenotype, like IL-6 expression [41], was also repressed in mesenchymal progenitor cells of old miR-29aTg mice. These investigations underpinned the important role of miR-29a in delaying age-induced skeletal deterioration.
We revealed that miR-29a, at least in part, disrupted the Dnmt3b catalysis of FoxO3 promoter hypermethylation to compromise FoxO3 loss, reversing senescence and mineralized matrix underproduction in aged osteoblasts. Transcription factor Foxo3 is important to slow cellular aging, promoting tissue homeostasis and function [35]. FoxO3 interference induces reactive oxygen radical overproduction, inhibiting the osteogenic differentiation capacity of mesenchymal stem cells [42]. The methylation status of the FoxO3 gene correlates with human tissue aging [43]. In addition, a reciprocal regulation was present in miR-29a and Dnmt3b. Dnmt3b inhibition mitigated miR-29a promoter methylation, in turn reversing miR-29a and FoxO3 functions to drive osteoblasts away from senescence. The inhibition of DNA methyltransferase in bone-forming cells promotes osteogenic activity. DNMT3b knockdown or 5-aza-induced DNA hypomethylation promotes the osteogenic differentiation of human [17] and murine [44]. The array of analysis in this study throws a new light upon an epigenetic crosstalk through which microRNA and methyl DNA regulated osteoblast senescence in the development of osteoporosis.
Oxidative stress is a prominent deleterious reaction, dysregulating protein stability or epigenetic pathways to hinder survival or biological activity in senescent cells [45]. Of note, miR-29a repressed 8-OHdG formation in aged osteoblasts. A transcriptomic landscape was that miR-29a affected redox reactions by promoting a plethora of antioxidant proteins in osteoblasts. Expanding studies reveal the biological roles of Oxr1, Glrx, and Gsr in cellular senescence. For example, forced Gsr expression downregulates Klotho loss-mediated renal cell aging and dysfunction [46]. Antioxidant protein Oxr1 promotes senolytic activity to sustain the viability of aged fibroblasts. Oxr1 knockdown aggravates oxidative stress and apoptosis [31]. In this study, Orx1 was advantageous to fend off oxidative stress, FoxO3 loss, and senescence in H 2 O 2 -treated osteoblasts and aged skeleton. Oxidative stress dysregulates epigenetic signatures, affecting the differentiation capacity of mesenchymal progenitor cells [47]. miR-29 involves the oxidative stress and viability of human mesenchymal stem cells upon DGCR8 interference [48]. The in vitro analysis of this study consolidated the importance of miR-29a and Oxr1 in repressing oxidative stress to downregulate DNA hypermethylation-induced osteoblast dysfunction. In this study, patients in the osteoporosis group were significantly older than the control group. However, the limitation of the experiment for human bone biopsies is the absence of osteoporosis indications in age-matched young patients who required spinal surgery. The possibility cannot be ruled out that ubiquitous miR-29a knockout may change other tissues' integrity or function, which may influence osteoblast function and bone homeostasis. miR-29a signaling may affect other DNA methylation enzymes or redox regulators to maintain anabolic activity in aged osteoblasts. The investigations related to the interactions of miR-29a, oxidative stress, and epigenetic pathways reveal a new molecular mechanism underlying bone mass loss.

Conclusions
Taken together, profound evidence revealed that miR-29a loss, oxidative stress, and DNA hypermethylation correlated with osteoblast aging in human osteoporosis. miR-29a knockout accelerated osteoblast senescence and bone loss. miR-29a reversed FoxO3 loss to improve senescence program through targeting Dnmt3b-mediated FoxO3 methylation and increased antioxidant proteins, including Oxr1, to downregulate oxidative stress and DNA methylation, compromising age-induced osteoblast loss and osteoporosis (Figure 8f). This study highlights a new epigenetic mechanism underlying oxidative stress-mediated osteoblast senescence and senile osteoporosis, as well as the anti-aging and antioxidation effects of miR-29a, DNA methylation inhibitor, and Oxr1 on age-induced osteoblast dysfunction and bone loss.