1,25‐Dihydroxyvitamin D protects against age‐related osteoporosis by a novel VDR‐Ezh2‐p16 signal axis

Abstract To determine whether 1,25‐dihydroxyvitamin D (1,25(OH)2D) can exert an anti‐osteoporosis role through anti‐aging mechanisms, we analyzed the bone phenotype of mice with 1,25(OH)2D deficiency due to deletion of the enzyme, 25‐hydroxyvitamin D 1α‐hydroxylase, while on a rescue diet. 1,25(OH)2D deficiency accelerated age‐related bone loss by activating the p16/p19 senescence signaling pathway, inhibiting osteoblastic bone formation, and stimulating osteoclastic bone resorption, osteocyte senescence, and senescence‐associated secretory phenotype (SASP). Supplementation of exogenous 1,25(OH)2D3 corrected the osteoporotic phenotype caused by 1,25(OH)2D deficiency or natural aging by inhibiting the p16/p19 pathway. The proliferation, osteogenic differentiation, and ectopic bone formation of bone marrow mesenchymal stem cells derived from mice with genetically induced deficiency of the vitamin D receptor (VDR) were significantly reduced by mechanisms including increased oxidative stress, DNA damage, and cellular senescence. We also demonstrated that p16 deletion largely rescued the osteoporotic phenotype caused by 1,25(OH)2D3 deficiency, whereas 1,25(OH)2D3 could up‐regulate the enzyme Ezh2 via VDR‐mediated transcription thereby enriching H3K27me3 and repressing p16/p19 transcription. Finally, we demonstrated that treatment with 1,25(OH)2D3 improved the osteogenic defects of human BM‐MSCs caused by repeated passages by stimulating their proliferation and inhibiting their senescence via the VDR‐Ezh2‐p16 axis. The results of this study therefore indicate that 1,25(OH)2D3 plays a role in preventing age‐related osteoporosis by up‐regulating Ezh2 via VDR‐mediated transcription, increasing H3K27me3 and repressing p16 transcription, thus promoting the proliferation and osteogenesis of BM‐MSCs and inhibiting their senescence, while also stimulating osteoblastic bone formation, and inhibiting osteocyte senescence, SASP, and osteoclastic bone resorption.


| INTRODUC TI ON
Vitamin D status influences the overall mineralization of the skeleton, and the rate of bone turnover (Lips & van Schoor, 2011).
Epidemiological studies have shown that vitamin D deficiency is a worldwide health problem (Holick, 2007) and could increase the risk of low bone mineral density, osteoporosis, falls, and fractures (Kuchuk, van Schoor, Pluijm, Chines, & Lips, 2009;van Schoor et al., 2008), whereas long-term supplementation of vitamin D and calcium may be effective in preventing these outcomes (Lips & van Schoor, 2011). Nevertheless, the capacity of calcium and vitamin D supplementation to prevent fractures is still controversial.
In a meta-analysis of 33 randomized clinical trials totaling 51,145 community-dwelling participants over the age of 50, the use of supplements that included calcium, vitamin D, or both, compared with placebo or no treatment, was not associated with a lower risk of fractures (Zhao, Zeng, Wang, & Liu, 2017). The authors concluded that their findings do not support the routine use of these supplements in community-dwelling older people. The reasons for this controversy remain unclear. Vitamin D is converted to 25-hydroxyvitamin D (25OHD) by the action of a liver 25-hydroxylase and is further metabolized into active 1,25-dihydroxyvitamin D [1,25(OH) 2 D] by the action of the kidney 25-hydroxyvitamin D 1α-hydroxylase (1α(OH)ase) enzyme, encoded by CYP27B1.
Active 1,25(OH) 2 D exerts its biological function by binding to the vitamin D receptor (VDR) (Plum & DeLuca, 2010). Production of 1,25(OH) 2 D is influenced by aging and may be reduced by approximately 50% as a result of an age-related decline in renal function (Gallagher, 2013). Thus, it is possible that the effectiveness of vitamin D supplementation, which is dependent on the effectiveness of 1,25(OH) 2 D production, may as a consequence decline with aging.
It is in this context of multiple aging comorbidities that the gerontology community has increasingly recognized the concept that aging itself is the greatest risk factor for most age-related chronic diseases, including atherosclerosis, cancers, dementias, diabetes, and osteoporosis (Tchkonia, Zhu, van Deursen, Campisi, & Kirkland, 2013). Some studies have indicated a close association between low levels of vitamin D and these age-related disorders (Plum & DeLuca, 2010). In the elderly, aging worsens the adverse effects of sex steroid loss on bone by decreasing defenses against oxidative stress, and bone aging is also accompanied by an alteration in the tissue microenvironment with increasing proinflammatory cytokine levels (Farr et al., 2016;McLean, 2009), a critical contributing factor for osteoporosis. In osteoporosis, the amount of bone resorbed is greater than the amount of bone formed leading to a reduction of trabecular and cortical bone volume and density (Chien & Karsenty, 2005;Manolagas & Parfitt, 2010). This decrease in bone mass, as well as a reduction of bone quality, results in decreased bone strength and predisposes the elderly population to an increased risk of fractures. In our previous studies, we found that in mutant mouse models with genetically deleted 1α(OH)ase (1α(OH)ase −/− ) or VDR (VDR −/− ), osteoblast numbers, mineral apposition rate, and bone volume were suppressed below levels seen in wild-type mice (Panda et al., 2004), even when hypocalcemia and secondary hyperparathyroidism were prevented by feeding the animals a high calcium, high phosphorus, lactose-containing "rescue" diet. This suggested that 1,25(OH) 2 D has a direct bone anabolic action besides its role in maintaining calcium and phosphorus balance. We recently demonstrated that 1,25(OH) 2 D exerts an anti-aging role by activation of Nrf2-antioxidant signaling and inactivation of p16/p53 senescence signaling (Chen et al., 2019); however, it is unclear whether 1,25(OH) 2 D plays a role in protection against osteoporosis through its anti-aging mechanisms.
Oxidative stress is an important cause of cellular senescence.
Cellular senescence is the process by which a cell enters a permanent cell cycle block, and senescent cells display a senescence-associated secretory phenotype (SASP) (Coppe et al., 2008). SASP factors which define this phenotype include the production of proinflammatory cytokines, growth factors, chemokines, and matrix remodeling enzymes (Ovadya & Krizhanovsky, 2014).
Senescent cells cause or aggravate the development of aging-related diseases through their growth arrest phenotype and SASP factors. The cell cycle-dependent kinase inhibitor p16 is not only a recognized indicator of cellular senescence, but it also acts as a key effector of cellular senescence (Lopez-Otin, Blasco, Partridge, Serrano, & Kroemer, 2013). During the development of physiological aging and aging-related diseases, the expression level of p16 is gradually increased (Krishnamurthy et al., 2004). Recent Deletion of p16-positive senescent cells not only prolongs the lifespan of premature aging mice, but also prolongs the lifespan of natural aging mice (Baker et al., 2011(Baker et al., , 2016. However, it remains unknown whether p16 deletion can inhibit osteogenic cell senescence by promoting osteogenic cell proliferation and differentiation into osteoblasts. To determine whether 1,25(OH) 2 D exerts anti-osteoporosis action by blocking the p16-cell senescence pathway, wild-type and 1α(OH)ase −/− mice were fed a high calcium/phosphorus rescue diet or aged 1α(OH)ase −/− mice or 18-month-old aged wild-type mice were subcutaneously injected with 1,25(OH) 2 D 3 . As well, 1α(OH) ase and p16 double knockout mouse models (1α(OH)ase −/− p16 −/− ) were generated. The bone phenotypes of the above groups of animals were analyzed using histopathology, cell biology, and molecular biology. Ezh2 is part of the polycomb repressive complex 2 (PRC2 complex) which is responsible for the trimethylation of H3K27 (histone 3) to generate H3K27me3. H3K27me3 is involved in the repression of many genes involved in development and cell differentiation. Down-regulation of Ezh2 and H3K27me3 has been associated with increased senescence although Ezh2 reduction may act, in part, by directly increasing DNA damage (Ito, Teo, Evans, Neretti, & Sedivy, 2018). We also therefore determined whether 1,25(OH) 2 D mediates transcriptional up-regulation of Ezh2-H3K27me3 via the VDR and whether this inhibits the p16 cell senescence pathway.
To assess whether aging-dependent bone loss induced by 1,25(OH) 2 D deficiency was associated with alterations of bone turnover, we analyzed changes of bone formation and resorption in 6-month-old wild-type and 1α(OH)ase −/− mice on the rescue diet by histomorphometric analyses, and calcein and xylenol orange double labeling. We found that the osteoblast number, mineral apposition rate (MAR), bone formation rate (BFR), and long bone strength including maximum load, energy, maximum stress, and elastic modulus were all significantly decreased compared with wild-type mice (Figure 1f-j, Figure S1E-H). However, TRAP-positive osteoclast numbers and C-telopeptide of type 1 collagen (CTX) levels were increased significantly (Figure 1k-m) in 1α(OH)ase −/− mice compared with wild-type mice.
These results demonstrated that 1,25(OH) 2 D deficiency accelerates age-related bone loss by reducing osteoblastic bone formation, increasing osteoclastic bone resorption, increasing p16, and inducing osteocyte senescence and SASP.

| Supplementation of exogenous 1,25(OH) 2 D 3 prevents bone loss induced by natural aging
We compared the percentage of senescent osteocytes and the osteogenic capacity of 3-and 18-month-old wild-type mice and found that the percentages of β-gal + and p16 + osteocytes were increased from <10% in 3-month-old mice to >40% in 18-month-old mice, comparable to the levels of 6-month-old 1α(OH)ase −/− mice ( Figure S2A, B). We also found that xylenol orange (XO)-positive

| VDR deficiency induces BM-MSC senescence and inhibits their osteogenesis
We next assessed whether 1,25(OH) 2 D acts via the VDR in producing senescence in BM-MSCs. First, the nucleoprotein and total protein expression levels of VDR were compared in BM-MSCs derived from 3-and 18-month-old wild-type mice. Although total protein expression levels of VDR did not decrease significantly in BM-MSCs from 18-month-old wild-type mice compared with those from 3-month-old wild-type mice, the nucleoprotein expression levels of VDR were markedly reduced ( Figure 4a). We then compared the proliferation and senescence of BM-MSCs derived from 6-month-

| Deletion of p16 largely rescues bone aging phenotypes induced by 1,25(OH) 2 D deficiency
We recently reported that p16 deletion can partly postpone aging

| D ISCUSS I ON
We recently reported that 1,25(OH) 2 D 3 can delay aging in large part by maintaining both calcium/phosphorus homeostasis and redox balance (Chen et al., 2019). The role and mechanism of 1,25(OH) 2 D in maintaining calcium, phosphorus, and skeletal homeostasis have been extensively studied (Fleet, 2017;Veldurthy et al., 2016).   in MSCs such as senescence, proliferation, and osteogenic/adipogenic differentiation potential may underlie age-related bone loss (Bergman et al., 1996;Kaneda et al., 2011;Manolagas & Parfitt, 2010;Stenderup, Justesen, Clausen, & Kassem, 2003;Zhou et al., 2008). Previous studies have also demonstrated that VDR deletion leads to premature aging in mice (Keisala et al., 2009) and 1,25(OH) 2 D 3 has been reported to delay cellular senescence with One fundamental aging mechanism that may contribute to multiple age-related morbidities is cellular senescence (Khosla, Farr, & Kirkland, 2018). There is compelling evidence from preclinical models and supportive human data demonstrating an increase in senescent cells in the bone microenvironment with aging. These cells produce a proinflammatory secretome that leads to increased bone resorption and decreased bone formation, and approaches that either eliminate senescent cells or impair the production of their proinflammatory secretome have been shown to prevent age-related bone loss in mice (Farr et al., 2017;Sims, 2016). We recently demonstrated that the aging induced by 1,25(OH) 2 D deficiency was accompanied by increased cellular senescence and SASP in multiple organs, whereas exogenous 1,25(OH) 2 D 3 supplementation could largely rescue the . Representative micrographs of vertebral cortical sections immunostained for (n) β-gal, (p) p16, and (r) IL-6. Quantification for the percentages of (o) β-gal + , (q) p16 + , and (s) IL-6 + osteocytes. Ex vivo primary bone marrow cells cultured for 14 days from vehicle or 1,25(OH) 2 D 3 -treated 18-month-old WT mice and stained with (t) methylene blue to show total CFU-f and a quantitative analysis of CFU-f numbers per well, or (u) cultured for 14 or 21 days and stained with xylenol orange (XO) followed by a quantitative analysis for the percentage of XO + cells, or (v) cultured for 14 day for 5-ethynyl-2′-deoxyuridine (EdU) incorporation (cell proliferation) or (x) stained cytochemically for senescence-associated β-gal (SA-β-gal); quantitative analysis for the percentage of (w) EdU + cells and (y) SA-β-gal + cells. (z) Western blots of bone extracts for the expression of p53, p19, and p16. ß-actin was used as loading control for Western blots. *, p < .05, **, p < .01, ***, p < .001, compared with vehicle   (Baker et al., 2008). Additionally, p21, a target of p53, is also a senescent marker but appears to maintain the viability of senescent cells (Yosef et al., 2017). We therefore used p16 knockout mice as the most appropriate model to assess effects on aging and cell senescence. We demonstrated that p16 deletion not only prolongs the vehicle-treated mice (Farr et al., 2017), suggesting that targeting cellular senescence could prevent age-related bone loss in mice. Our findings indicate that 1,25(OH) 2 D 3 could prevent age-related bone loss by targeting the p16 cellular senescence pathway.
The Ink4a/Arf locus, consisting of the genes p16 Ink4a and p19 Arf , is central to the induction of senescence, and this locus is tightly controlled by PRC family members (Aguilo et al., 2011;Dhawan, Tschen, & Bhushan, 2009). Ezh2 is a PRC family member that can increase H3K27me3 along the Ink4A/Arf locus, thus repressing p16 and p19 transcription (Cakouros et al., 2012;Li et al., 2017). In this study, we therefore examined whether 1,25(OH) 2 D 3 repressed the p16/ to physically bind the Ezh2 promoter was confirmed using ChIP-qPCR. Furthermore, we demonstrated that the putative promoter region containing the predicted VDR binding sites of the Ezh2 gene is sufficient to promote transcription of Ezh2 in the presence of 1,25(OH) 2 D using luciferase assays. We also found that Ezh2 overexpression in VDR-deficient BM-MSCs could down-regulate the F I G U R E 4 VDR deficiency induces BM-MSC senescence and inhibits osteogenesis. (a) Western blot of BM-MSC extracts from young (3 months) and old (18 months) mice for expression of nuclear VDR and total VDR. Histone-H3 was used as nucleoprotein loading control, whereas ß-actin was used as total protein loading control for Western blots. (b) In vitro population doublings of BM-MSCs from 6-monthold WT and VDR −/− mice. (c) Ex vivo primary bone marrow cultures from 3-, 6-and 12-month-old WT mice stained with methylene blue to show total CFU-f. (d) Quantification of the number of CFU-F colonies. Representative micrographs of the second passaged BM-MSCs from 6-month-old WT and VDR −/− mice (e) stained immunocytochemically for EdU, and (g) phase images, or (i) stained cytochemically for SA-β-gal to detect senescence, (k) with 2',7'-dichlorofluorescin diacetate (DCFDA) for reactive oxygen species (ROS) and (m) stained immunocytochemically for γ-H2AX as a marker of DNA damage. Quantification for (f) the percentages of EdU + cells, (h) average cell area, the percentages of (j) SA-β-gal + , (l) DCFDA + , and (n) γ-H2AX + cells. Because many individuals older than 80 have a GFR <50 ml/min, decreased production of 1,25(OH) 2 D in this age group is therefore common (Gallagher, 2013). Our previous results in murine models indicate that the declining 1,25(OH) 2 D levels which occur with aging may in fact contribute to the aging phenotype (Chen et al., 2019). Furthermore, aging-associated comorbidities including cancer, diabetes, and hypertension have been reported to be influenced by 1,25(OH) 2 D deficiency (Bouillon et al., 2008). The results of the current study now provide a model that suggests that 1,25(OH) 2 D protects against age-related osteoporosis by up-regulating Ezh2 via VDR-mediated transcription, thereby increasing H3K27me3, and repressing p16 transcription, thus promoting the proliferation and osteogenesis of BM-MSCs, inhibiting their senescence, stimulating osteoblastic bone formation, and inhibiting osteocyte senescence, SASP, and osteoclastic bone resorption ( Figure 8). Therefore, by targeting a fundamental aging mechanism, 1,25(OH) 2 D may be an effective agent in the treatment and prevention of age-related osteoporosis.
Gender-matched 1α(OH)ase −/− and wild-type (WT) littermates were randomly divided into groups. After weaning, grouped wild-type and 1α(OH)ase −/− mice were weaned onto one of the following

| Mechanical testing
Three-point bending and compression/traction of long bones (femurs and tibias) were performed as described previously (Ren et   after which 5-μm sections were prepared and bone sections were stained with hematoxylin and eosin (H&E), and for total collagen, beta-galactosidase (β-gal) and the osteoclast marker tartrate-resistant acid phosphatase (TRAP) as previously described (Dimri et al., 1995). For assay of dynamic bone formation, mice were intraperitoneally injected with calcein (10 mg/kg; Sigma) 12 days after which they were given an injection of xylenol orange (XO) (90 mg/ kg; Sigma). Nondecalcified bone was embedded in optimum cutting temperature compound (O.C.T), and 7-μm-thick sections were obtained using transparent film kindly provided by Prof. Liu Peng.

| Histology and bone histomorphometry
Analyses of dynamic bone formation parameters were performed using standard software kindly provided by Robert J. Van't Hof.

| Immunocytochemistry and immunohistochemistry staining
For immunocytochemical staining, cultured cells were incubated with primary antibody against histone H2A on Ser139 (γ-H2AX; Cell Signaling Technology) overnight at 4℃, followed by using 594-conjugated goat anti-rabbit secondary antibody to detect immunoreac-

| Western blot
Whole-cell or nuclear lysates were extracted for loading into 10% SDS-PAGE gels, and immunoblotting was performed as previously described (Miao et al., 2008). Primary antibodies against Ezh2 and analyzed by ImageJ.

| Cell cultures and lentiviral infection
Primary mouse BM-MSCs were isolated from grouped WT and VDR −/− mice or from aged WT mice treated with vehicle or 1,25(OH) 2 D 3 , and cultured as described previously (Miao et al., 2001
Briefly, BM-MSCs were isolated from WT mice aged 12 months treated with vehicle or 1,25(OH) 2 D 3 (0.1 μg/kg) for 3 months, and then, cell samples were subjected to immunoprecipitation using either a control IgG or the ChIP-grade VDR (Abcam, ab3508) and H3K27me3 (Cell Signaling Technology) antibody. The coprecipitated chromatin was determined by qPCR for the enrichment of promoter DNA using primers for p16 INK4a , p19 ARF , Ezh2, and Gapdh shown in Table S1.

| Luciferase reporter assay
To generate the Ezh2 promoter-activated luciferase reporter, −1702 to +52 bp of Ezh2 promoter was cloned into PGL3 basic. Luciferase reporter assay was performed as previously described (Chen et al., 2019). Briefly, BM-MSCs in 12-well plates were transfected with Ezh2promoter or Ezh2-promoter mutant luciferase reporter plasmid. 12 hr later, the medium was changed and the indicated concentration of vehicle or 1,25(OH) 2 D 3 was added. After another 36 hr, luciferase activity was measured using the Dual-Luciferase Assay Kit (Promega). pRL-TK was co-transfected to normalize transfection efficiency. Each experiment was performed in triplicate and repeated at least three times. F I G U R E 7 1,25(OH) 2 D 3 reduces cellular senescence and promotes ectopic bone formation of human BM-MSCs. (a) Representative micrographs of early (4th) and late (12th) passaged human BM-MSC cultured for 21 days in osteogenic differentiation medium and stained with xylenol orange (XO) to quantify the percentage of XO + area and (b) were subcutaneously transplanted, After 6 weeks, transplants were harvested and stained with HE (upper part of panel b) or Masson trichrome (bottom part of panel b). (c) The analysis of bone volume based on H＆E staining. (d) Western blot of BM-MSC extracts from the cultures as (a) for expression of nuclear VDR and total VDR, and for Ezh2 and p16 protein levels. Histone-H3 was used as nucleoprotein loading control, whereas ß-actin was used as total protein loading control for Western blots. Representative micrographs of the 4th and 12th passaged BM-MSCs as (a) stained with (e) H&E, (f) immunocytochemically for EdU, and (g) cytochemically for SA-β-gal. Quantification for (h) average cell area, the percentages of (i) EdU + , and (j) SA-β-gal + cells. (k) Human BM-MSCs pretreated with vehicle or 1,25(OH) 2 D 3 and were subcutaneously transplanted into recipient SCID mice. 6 weeks later, the implants were collected and prepared sections were stained (l) with XO, (m) H＆E, and (n) Masson trichrome. Green arrows indicate osteoblasts, and red arrows indicate osteocytes. Bone histomorphometric analysis of (o) mineralization (XO + area, %) and (p) bone volume. *, p < .05, **, p < .01, ***, p < .001, compared to the 4th passaged or vehicle-treated human BM-MSCs F I G U R E 8 Model of mechanisms used by 1,25(OH) 2 D to protect against age-related osteoporosis via the VDR-Ezh2-p16 signaling axis. 1,25(OH) 2 D binds to the VDR, and the VDR-RXR heterodimer then binds to the VDRE on Ezh2 up-regulating Ezh2 and increasing H3K27met. This results in repression of p16/p19 transcription, thus promoting the proliferation of bone marrow MSCs and inhibiting their senescence and SASP production. Senescence and SASP production of osteocytes is also inhibited. As a consequence, osteoblastic bone formation is stimulated, and osteoclastic bone resorption is inhibited. Bone quantity and microarchitecture are thus improved as shown by the μCT images

| Ectopic bone formation assay
Mouse BM-MSCs were cultured to 4th passage, after which they were treated with 1,25(OH) 2 D 3 or vehicle for 72 hr before transplantation. The 8th to 12th passaged human BM-MSCs were cultured in the presence or absence of 1,25(OH) 2 D 3 prior to transplantation. The ectopic bone formation model was described previously (Yin et al., 2009). Briefly, the gelfoam used is the sponge mixed with 100% gelatin, which is isolated from pigskin. The gelfoam was then cut into small pieces (5 mm diameter × 5 mm height) for loading the cells at a total of 2 × 10 6 mouse BM-MSCs or 1 × 10 6 human BM-MSCs. Cells were resuspended in 40 μl α-MEM and transplanted into the dorsal subcutaneous tissue of 5-week-old NOD-SCID mice. Mice were sacrificed at 6 weeks after transplantation, and implants were collected and processed for staining with H&E and Masson trichrome.

| RNA isolation and real-time RT-PCR
Total RNA was extracted from cultured MSCs and lumbar vertebrae using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Complementary DNA (cDNA) was synthesized using Synthesis SuperMix (Invitrogen). The real-time RT-PCR was carried out using an Agilent Real-time System. Gapdh was amplified at the same time to normalize gene expression. Groups at least six mice were examined, and each experiment was repeated three times to determine relative gene expression differences.
The PCR primer sequences used in this study are shown in the Table S2.

| Statistical analysis
All analyses were performed using SPSS software (Version 16.0; SPSS Inc.). Measured data were described as mean ± SEM fold change over control and analyzed by Student's t test, one-way, or two-way ANOVA to compare differences between groups. Qualitative data were described as percentages and analyzed using a chi-square test as indicated. p Values were two-sided, and p < .05 was considered statistically significant.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in https ://doi.org/10.1111/acel.12951 .