Functional Cooperation between Vitamin D Receptor and Runx2 in Vitamin D-Induced Vascular Calcification

The transdifferentiation of vascular smooth muscle cells (VSMCs) into osteoblast-like cells has been implicated in the context of vascular calcification. We investigated the roles of vitamin D receptor (Vdr) and runt-related transcription factor 2 (Runx2) in the osteoblastic differentiation of VSMCs in response to vitamin D3 using in vitro VSMCs cultures and in vivo in Vdr knockout (Vdr -/-) and Runx2 carboxy-terminus truncated heterozygous (Runx2 +/ΔC) mice. Treatment of VSMCs with active vitamin D3 promoted matrix mineral deposition, and increased the expressions of Vdr, Runx2, and of osteoblastic genes but decreased the expression of smooth muscle myosin heavy chain in primary VSMCs cultures. Immunoprecipitation experiments suggested an interaction between Vdr and Runx2. Furthermore, silencing Vdr or Runx2 attenuated the procalcific effects of vitamin D3. Functional cooperation between Vdr and Runx2 in vascular calcification was also confirmed in in vivo mouse models. Vascular calcification induced by high-dose vitamin D3 was completely inhibited in Vdr -/- or Runx2 +/ΔC mice, despite elevated levels of serum calcium or alkaline phosphatase. Collectively, these findings suggest that functional cooperation between Vdr and Runx2 is necessary for vascular calcification in response to vitamin D3.


Introduction
Vascular calcification (VC) is a prevalent and progressive pathological process. VC is caused by a failure to maintain homeostasis of the vasculature in diseases like atherosclerosis, diabetes, chronic kidney disease (CKD), and during normal aging [1][2][3]. Inflammation, reactive oxygen species, and hyperlipidemia are major contributors to the pathogenesis of VC [4][5][6][7]. Interestingly, a number of studies have shown VC recapitulates embryonic osteogenesis, and that vascular smooth muscle cells (VSMCs) are able to transdifferentiate into both osteogenic and chondrogenic cells [1,8,9]. Furthermore, transdifferentiation, also called lineage reprogramming, has been reported to be responsible for the presence of osteoblast-like cells instead of VSMCs during VC [10]. In addition, the transdifferentiation of VSMCs into osteochondroprogenitor-like cells occurs before mineral deposition and is associated with the onset of osteogenic gene expression [10].
Vitamin D receptor (Vdr) is a candidate VC-related gene because it is involved in calcium and phosphate homeostasis and normal bone metabolism [23,24]. Vdr also regulates the expressions of bone modulating factors, such as, Runx2, receptor activator of NF-kappa-B ligand (Rankl), osteocalcin (Ocn), and osteopontin (Opn) [25][26][27]. In transcriptional intermediary factor 1 alpha null mice, VC was induced by upregulating the expressions of Vdr target genes [28], and clinically, VC can be induced by vitamin D deficiency or excess [29,30]. For example, CKD patients show a 'U-shaped' VC development tendency with respect to serum vitamin D levels, which suggests that correct vitamin D dosage is important [31]. Accordingly, the dietary intake allowance of vitamin D should be strictly controlled CKD patients [32].
Although the above-mentioned studies suggest the importance of Vdr and Runx2 in the osteoblastic differentiation of VSMCs, it is not clear whether both are necessary. In the present study, we examined their contributions to VC using Vdr -/-and Runx2 carboxy-terminus deletion mutant (Runx2 +/ΔC ) mice [12,33].

Alp and von Kossa staining of VSMCs
To detect mineralization, VSMCs were cultured in osteogenic medium for 14 days, washed with PBS, and fixed 2% formaldehyde (Junsei Chemical Co) at room temperature (RT) for 10 minutes. After washing with PBS, cells were treated with Alp staining solution and incubated at 37°C for 30 minutes. Cells were then cultured in osteogenic medium for 21 days and von Kossa stained, as previously described [35]. Briefly, cells were fixed with 0.1% glutaraldehyde (Sigma Aldrich) for 15 minutes at RT, washed with autoclaved/deionized Milli Q water twice, incubated with 5% silver nitrate (Junsei Chemical Co, Tokyo) under an UV lamp (UVP, CA, USA) for 30 minutes at RT, washed twice with water, treated with 2.5% sodium thiosulfate (Sigma Aldrich) to remove nonspecific signals, and processed for scanning (Epson Perfection V10 scanner; Seico Epson Corp, Nagano, Japan).

Real-time RT-PCR analysis
Real-time RT-PCR was used to quantify the relative mRNA levels of various genes, such as, Gapdh, Ocn, Opg, Osx, Rankl, Runx2, Smmhc, and Vdr using a Roche LightCycler 480 (Indianapolis, IN, USA). Primers were designed using ABI Primer Express version 2.0 (Carlsbad, CA, USA). Briefly, total RNA (1-5 μg) was extracted from aortas and reverse transcribed cDNA was used as a template in each PCR. CYBR green PCR master mix (4 μl, Takara Co) and specific primers were combined with templates, and a negative control (no template) was included in each assay. PCR was performed over 50 cycles of 10 minutes at 95°C, 15 seconds at 95°C, 15 seconds at 60°C, and 15 seconds at 72°C. The primers used for real-time RT-PCR are described in Table 1.

Adenovirus expressing Runx2
To overexpress Runx2, we used an adenovirus expressing Runx2, as previously described [37]. Briefly, 70% confluent primary cultured VSMCs from a mouse or rat were plated in 100 mm plates and infected with adenovirus expressing Runx2 (50 moi) in maintaining medium without FBS for 4 hours. Cells were then cultured for 42 hours in experimental medium with/ without 1,25(OH) 2 D 3 .

Co-immunoprecipitation analysis
HEK293 cells (70% confluent) were seeded in 100 mm plates, cultured overnight, and transiently transfected with pCS4-3Flag-Runx2 or empty vector. After 4~6 hours of transfection, cells were cultured in the presence or absence of 10 -7 mol/l 1,25(OH) 2 D 3 for 24 hours and NEs were prepared using the NE-PER kit. Co-immunoprecipitation analysis was performed as previously described [38]. To remove nonspecific binding proteins, protein G Sepharose (Santa Cruz) slurry was added to 130 μg aliquots of NEs and incubated at 4°C for 2 hours with rotation. After a spin down, mouse monoclonal anti-Vdr antibody or mouse normal IgG was added to supernatants at 4°C for 1 hour. Protein G Sepharose beads were then added to mixed supernatants and incubated at 4°C overnight with rotation. The beads were then rinsed five times with wash buffer [10 mmol/l Tris (pH 7.8), 150 mmol/l NaCl, 1 mmol/l EDTA, 1 mmol/l EGTA (pH 8.0), 5 mmol/l NaF, 1 mmol/l Na 3 VO 4 , 1 mmol/l Na 4 P 2 O 7 , 1% Triton X-100, 0.5% NP-40, and protease inhibitor cocktail (Calbiochem, Darmstadt, Germany)]. 6X SDS-sample loading buffer (30 μl) was added to the washed protein G beads which were then heated at 100°C for 5 minutes. Co-immunoprecipitated 3Flag-Runx2 was detected by Western blotting using an anti-Flag monoclonal antibody (Sigma Aldrich).

Animals and ethics statement
C57BL/6N mice were supplied by KOATECH (GyeongGi-Do, Korea). Vdr -/-mice were fed with a high calcium rescue diet as previously described [33,39]. Runx2 +/ΔC mice were fed a normal diet [37]. Animals were maintained under a 12-hour light:12hour darkness cycle at 22~25°C in specific pathogen-free (SPF) conditions and fed standard rodent chow and water ad libitum. All animal experiments were approved by Kyungpook National University (KNU) and all procedures related to animal experimentation were carried out in strict accordance with the guidelines issued by the Institutional Animal Care and Use Committee of KNU (No. KNU-2011-67). Surgery was performed under avertin or isoflurane anesthesia, and all efforts were made to minimize suffering.

Vitamin D-mediated VC mouse model
This model is commonly used to investigate VC because medial calcification is easily produced [40]. Normal C57BL/6N, Vdr -/-, or Runx2 +/ΔC [12] mice were injected subcutaneously with 6 or 8 x 10 5 IU/kg of vitamin D 3 daily for three days (n=5-8 per group). The vitamin D 3 solution (2.64 x 10 6 IU) was prepared as follows. Vitamin D 3 (66 mg) in 200 μl of absolute ethanol was mixed with 1.4 ml of cremophor (Sigma Aldrich) at RT, and then with 18.4 ml of sterilized water containing 750 mg of dextrose at RT. After injecting vitamin D 3 , body weights were checked daily until sacrifice.

Alizarin red S and von Kossa staining and immunohistochemical analyses
Mice were sacrificed at 9 or 10 days after first vitamin D 3 injection, and aortas were dissected and stained with 0.005% Alizarin red S in 0.5% KOH for 24 hours [41]. For histological analysis, mice were sacrificed at day 9 or 10 after first vitamin D 3 injection, and aortas were fixed in 4% formaldehyde overnight. For paraffin sections, fixed aortas were dehydrated and embedded in paraffin and cut into 3 μm sections. von Kossa staining was performed to detect mineralization as described previously [42]. Deparaffinized sections were washed in Milli-Q water for 30 minutes, incubated in 1% aqueous silver nitrate solution under a UV lamp for 30 minutes, and washed with water three times at RT. Non-specific staining was removed by then treating sections with 2.5% sodium thiosulfate for 5 minutes at RT. Immunohistochemical analyses were carried out as previously described [43] or by following the manufacturer's instructions (InnoGenex) as follows. Briefly, endogenous peroxidase activity was quenched in antigenretrieved sections using 3% H 2 O 2 . After blocking with 1% BSA for 1 hour at RT, sections were incubated with the indicated primary antibodies (Vdr, Runx2, Runx2, Mgp, and Smmhc) diluted at 1:200~1:1000 for 16 hours at 4°C and then with goat anti-rabbit IgG conjugated with HRP (DakoCytomation, Glostrup, Denmark) for 1 hour at RT. To visualize signals, sections were developed using a DAB substrate-chromogen system (DakoCytomation). Rabbit and mouse normal IgG were used as controls at the same dilution as the appropriate primary antibodies. Microscopy was conducted using a light microscope and the AxioVison program (Carl-Zeiss, Jena, Germany).

Serum biochemical analysis
Calcium, phosphate, and Alp levels in mouse sera were analyzed using commercially available kits (Bio Assay Systems). For calcium analysis, 5 μl of serum samples or of standards were transferred into the wells of a clear bottom 96well-plate (Corning Inc, Corning, NY, USA). Working reagent (200 μl) was added, incubated for 3 minutes at RT, and absorbances were then measured at 612 nm using a microplate reader (Tecan, Austria, GmbH, Austria). For phosphate analysis, 2 μl serum samples diluted with 48 μl of autoclaved/deionized Milli-Q water or 50 μl of water (blank) were transferred to the wells of a clear bottom 96-well plate. Reagents (100 μl) were added, incubated for 30 minutes at RT, and absorbances were measured at 620 nm using a microplate reader. For Alp analysis, 10 μl serum samples were transferred to the wells of a clear bottom 96-well plate. Working solution (190 μl) was added and absorbances were measured immediately at 405 nm (time = 0) and again 4 minutes (time = 4 minutes) later using a microplate reader.

Statistics
Statistical analyses were performed by unpaired Student's ttest using SigmaPlot 10.0. We considered P values <0.05 as statistically significant.

Effects of 1,25(OH) 2 D 3 on VSMC calcification
To test the osteoblastic differentiation of VSMCs, primary mouse VSMCs were cultured in osteogenic media, and then Alp activity and mineralization were assessed. Osteogenic media promoted Alp activity and bone nodule formation ( Figure  1A), and treatment with 1,25(OH) 2 D 3 further enhanced differentiation of VSMCs into osteoblast-like cells ( Figure 1B). It has been reported that endothelial cells can transdifferentiate into osteoblast-like cells [44], and to test this possibility, primary endothelial cells derived from human umbilical cord vein were cultured in osteogenic media treated with 1,25(OH) 2 D 3 . However, osteoblastic differentiation was not observed in endothelial cells (data not shown).

Reciprocal regulation between Vdr and Runx2 by 1,25(OH) 2 D 3
To determine the relationship between Vdr and Runx2, VSMCs derived from rat aortas were infected with adenovirus expressing Runx2 and treated with 1,25(OH) 2 D 3 for 42 hours. The overexpression of Runx2 was found to increase Vdr expression, and this was further enhanced by treatment with 1,25(OH) 2 D 3 ( Figure 3A). We also performed knockdown experiments to determine whether Runx2 and Vdr have a reciprocal regulatory function. VSMCs derived from mouse aorta were treated with siRNA for Vdr or Runx2 and then their mRNA expressions induced by 1,25(OH) 2 D 3 were assessed. Knocking down Vdr or Runx2 using siRNAs reduced both of their expressions at the mRNA and protein levels ( Figures 3B  and 3C), indicating the presence of reciprocal regulation between Vdr and Runx2 in VSMCs.

Effects of the Vdr/Runx2 interaction on the expressions of osteoblastic genes
Vdr is known to bind to Runx2 in osteoblasts [25], and together these two factors activate osteogenic target genes [25,26]. To determine whether Vdr directly binds Runx2, immunoprecipitation was performed using HEK293 cells. HEK293 cells, transfected with or without 3Flag-Runx2, were treated with 1,25(OH) 2 D 3, immunoprecipitated with anti-Vdr antibody, and probed with anti-Flag antibody. As shown in Figure 4A, Runx2 co-precipitated with Vdr only in 1,25(OH) 2 D 3treated-Runx2-transfected cells. Control immunoprecipitation of normal IgG showed no increase in Runx2 by 1,25(OH) 2 D 3 treatment. These results suggest that Vdr interacts with Runx2.
1,25(OH) 2 D 3 increased the mRNA levels of Vdr, Osx, and Rankl, but decreased Smmhc mRNA levels, whereas Runx2 overexpression increased the mRNA levels of Vdr, Ocn, Osx, and Rankl, but decreased Smmhc mRNA levels ( Figure 4B). Furthermore, Runx2 plus 1,25(OH) 2 D 3 additively increased the expressions of the mRNAs of Vdr, Ocn, and Rankl, but decreased those of Opg and Smmhc, and thus, sharply increased the Rankl/Opg ratio ( Figure 4B). Ocn and Opg    mRNA levels were not changed by 1,25(OH) 2 D 3, and no additive effects were observed in the expression of Osx ( Figure  4B). Taken together, these results indicate that the interaction between Vdr and Runx2 additively induces the expressions of osteoblastic genes.

Effects of vitamin D 3 on osteoblastic differentiation and calcium deposition in mice
VC was induced in mice by administering vitamin D 3 , as previously described in the rat [40]. Injection of vitamin D 3 subcutaneously for 3 consecutive days triggered VC in the aortic medial layer ( Figure 5A). In addition, mineral deposition was observed in kidney and lung ( Figure S1), but not in other tissues, such as, brain, liver, stomach, pancreas, spleen, and small intestine (data not shown). The relative mRNA levels of Vdr, Runx2, Osx, Ocn, Opn, Rankl, and Mgp in aortas were also gradually increased by vitamin D 3 administration ( Figure  5B), whereas the mRNA levels of Smmhc, a smooth muscle cell marker, gradually diminished ( Figure 5B). Serum calcium levels and Alp activities were increased in vitamin D 3 treated mice, but with no change in serum phosphate concentration was observed ( Figure 5C). These results indicate that vitamin D 3 -mediated VC was associated with the upregulations of bone-related genes in calcified sites in our mouse model.

Protection from VC induced by vitamin D 3 in Vdr -/-mice
We also investigated vitamin D 3 -mediated VC in Vdr -/-mice, which have similar calcium, lower phosphate, and higher Alp serum levels than wild-type mice [33]. Unlike their wild type littermates (Vdr +/+ ), mineral deposition was not observed in the aortas of Vdr -/-mice ( Figure 6A). Furthermore, high-dose vitamin D 3 did not increase the protein expressions of Vdr, Runx2, and Mgp in Vdr -/-mice as it did in WT mice ( Figure 6A). In addition, high-dose vitamin D 3 did not increase serum calcium or Alp activity in Vdr -/-mice ( Figure 6B). These results indicate that Vdr deficiency protected mice from the aortic deposition of mineral induced by high-dose vitamin D 3 .

Protection of VC induced by vitamin D 3 in Runx2 +/ΔC mice
Since Runx2 expression is abrogated in Vdr null mice, we examined whether high-dose vitamin D 3 -induced VC was blunted in Runx2 +/ΔC mice. To test this hypothesis, Runx2 +/ΔC mice were injected with vitamin D 3 and matrix mineral deposition and the protein levels of Vdr, Runx2, Mgp, and Smmhc in aortic tissues were assessed by von Kossa staining and immunohistochemical analyses, respectively. Vdr and other bone related markers (Runx2 and Mgp) were not induced by vitamin D 3 in Runx2 +/ΔC mice ( Figure 7A). Furthermore, as was observed in Vdr -/-mice, vitamin D 3 -inudced osteoblastic factors and mineral deposition were inhibited Runx2 +/ΔC mice, despite increased serum calcium levels ( Figures 7A and B). On the other hand, phosphate and Alp levels were unchanged by vitamin D 3 in Runx2 +/ΔC mice ( Figure 7B). The protection from VC observed in Runx2 +/ΔC mice in the presence of elevated levels of calcium suggests that VC is not caused by passive precipitation and that it requires active participation of Runx2 and Vdr.

Discussion
The results of this suggest functional cooperation between Vdr and Runx2 in the upregulations of their expressions, osteogenic gene expression, and mineral deposition are causally associated with VC. Furthermore, immunoprecipitation revealed that Vdr and Runx2 physically interact with each other. In addition, functional cooperation between Vdr and Runx2 with respect to their expressions and the expressions of osteogenic factors were found to play a key role in the procalcific effects of vitamin D 3 . Finally, our in vitro findings were supported by observation made in vivo our high-dose vitamin D 3 -induced VC mouse model.
The expressions of Vdr and Runx2 additively increased the mRNA expressions of downstream osteogenic genes, and selective siRNA knockdown of Vdr and Runx2 reciprocally inhibited the protein/mRNA expressions of Vdr and Runx2 in VSMCs. Indeed, vitamin D 3 -induced VC and the upregulations of bone related genes were completely diminished in Vdr -/-and in Runx2 +/ΔC mice.
In osteoblasts, it has been shown that Vdr suppresses Runx2 expression by binding to VDRE in Runx2 promoter and that Runx2 autonomously suppresses its own expression [45]. However, the osteoblast-specific deletion of Vdr did not influence Runx2 expression in bone [46]. In the present study, 1,25(OH) 2 D 3 increased Runx2 expression in VSMCs and Vdr played an essential role in vitamin D 3 -induced VC in vivo.
These findings indicate a cell type specificity of Runx2 expression in response to vitamin D 3 treatment. Further studies are needed to dissect the mechanisms responsible for the differential regulation of the Runx2 gene in VSMCs and osteoblasts.
We used a high concentration of 1,25(OH) 2 D 3 (10 -7 mol/l) to determine the molecular mechanism responsible for the transdifferentiation of VSMCs into osteoblast-like cells in cell culture systems because the expressions of Vdr, Runx2, and Smmhc were prominent and comparable to those of in vivo phenotypes at this concentration. In the present study, combined treatment of 1,25(OH) 2 D 3 and Runx2 additively increased the expressions of Vdr, Ocn, and Rankl. Ocn is a skeleton-specific non-collagenous protein and was found recently to act as novel hormone signaling insulin secretion in the pancreas and testosterone synthesis in testes [47,48]. In the present study, Ocn expression was increased in VSMCs by Runx2 and further upregulated by Runx2 and Vdr. The significance of Ocn in the transdifferentiation of VSMCs is unclear, and it needs to be determined whether Ocn is involved in vitamin D 3 -mediated VC. In the present study, we also observed upregulattion of the Rankl/Opg pathway. More specifically, 1,25(OH) 2 D 3 and Runx2 additively increased Rankl and decreased Opg levels, and thus, sharply increased the Rankl/Opg ratio. Reactive oxygen species and inflammationmediated increases in Rankl have been previously reported to be important triggers of VC [4]. Furthermore, it has also been reported that Runx2 increases Rankl and contributes to VC by regulating Bmp2 and bone-related proteins, such as, Mgp [49,50]. These findings suggest that modulation of the   It has been established that VC development shows a Ushaped tendency with respect to serum vitamin D 3 levels [29,30]. Recently, Vdr deficient mice fed a semi-synthetic western diet containing 2% calcium and 1.25% phosphate and LDL receptor deficient mice fed a semi-synthetic diet containing low levels of vitamin D 3 were found to show aortic calcification and upregulations of osteogenic factors [30] and of low-dose Vdr activators that protect against VC [51]. In the present study, VC was clearly induced in mice by high-dose vitamin D 3 injection as has previously been reported in rat models [40]. Furthermore, Vdr -/-and Runx2 +/ΔC mice show strong inhibitory effects on high-dose vitamin D 3 -induced VC, and in these mice, osteogenic markers did not increase in blood vessels. In a previous murine study, smooth muscle cell-specific Runx2 deletion was found to inhibit high fat diet induced VC [52]. In the present study, despite increased serum calcium or Alp levels, the protective effects on VC in Vdr -/-and Runx2 +/ΔC mice indicate that both Vdr and Runx2 act independently of these factors, which suggests that vitamin D daily allowances be strictly adhered to particularly CKD patients [32]. Alp activity was not changed by vitamin D 3 in neither Vdr -/-or Runx2 +/ΔC mice, but an increase was observed in WT mice. Interestingly, basal levels of Alp activity were high in Vdr -/-mice, but similar in Runx2 +/ΔC mice and WT mice. Vdr might be a negative regulator of Alp activity in bone tissues [46]. Although serum Alp activity upregulation by vitamin D 3 reflects VC quite well, further studies are required to clarify the contribution made by Alp activity to the induction of VC.
Collectively, our findings show regulations of Vdr and Runx2 at the gene level and functional interaction between Vdr and Runx2 are required for the regulations of VC-related osteogenic genes and mineral deposition in response to vitamin D 3 in VSMCs. In particular, a deficiency in Vdr or Runx2 completely abrogated vitamin D 3 responses in our in vivo models. This study presents new evidence that targeting the expression levels of Vdr and Runx2 and/or of the functional interaction between Vdr and Runx2 in vascular tissues provides a potential means of reducing cardiovascular morbidity and mortality caused by VC.