Characterization of the Autocrine/Paracrine Function of Vitamin D in Human Gingival Fibroblasts and Periodontal Ligament Cells

Background We previously demonstrated that 25-hydroxyvitamin D3, the precursor of 1α,25-dihydroxyvitamin D3, is abundant around periodontal soft tissues. Here we investigate whether 25-hydroxyvitamin D3 is converted to 1α,25-dihydroxyvitamin D3 in periodontal soft tissue cells and explore the possibility of an autocrine/paracrine function of 1α,25-dihydroxyvitamin D3 in periodontal soft tissue cells. Methodology/Principal Findings We established primary cultures of human gingival fibroblasts and human periodontal ligament cells from 5 individual donors. We demonstrated that 1α-hydroxylase was expressed in human gingival fibroblasts and periodontal ligament cells, as was cubilin. After incubation with the 1α-hydroxylase substrate 25-hydroxyvitamin D3, human gingival fibroblasts and periodontal ligament cells generated detectable 1α,25-dihydroxyvitamin D3 that resulted in an up-regulation of CYP24A1 and RANKL mRNA. A specific knockdown of 1α-hydroxylase in human gingival fibroblasts and periodontal ligament cells using siRNA resulted in a significant reduction in both 1α,25-dihydroxyvitamin D3 production and mRNA expression of CYP24A1 and RANKL. The classical renal regulators of 1α-hydroxylase (parathyroid hormone, calcium and 1α,25-dihydroxyvitamin D3) and Porphyromonas gingivalis lipopolysaccharide did not influence 1α-hydroxylase expression significantly, however, interleukin-1β and sodium butyrate strongly induced 1α-hydroxylase expression in human gingival fibroblasts and periodontal ligament cells. Conclusions/Significance In this study, the expression, activity and functionality of 1α-hydroxylase were detected in human gingival fibroblasts and periodontal ligament cells, raising the possibility that vitamin D acts in an autocrine/paracrine manner in these cells.


Introduction
Vitamin D 3 is a major component in the regulation of calcium and phosphorus metabolism. The function of vitamin D 3 , via the active hormonal metabolite 1a,25-dihydroxyvitamin D 3 (1,25OH 2 D 3 ), is to regulate the absorption of these essential minerals in the intestine, and their mobilization in bone tissues [1]. 1,25OH 2 D 3 also plays an important role in the regulation of T cell function and immunological reaction [2]. The biological function of 1,25OH 2 D 3 is orchestrated by vitamin D receptor (VDR). After binding with VDR, 1,25OH 2 D 3 acts on the vitamin D responsive element (VDRE) located upstream of its target genes (eg. CYP24A1, RANKL, et al.) and up-regulates their expression which results in the corresponding biological effects [3,4,5,6,7].
Periodontal tissues consist of two hard tissues, bone and cementum, and two soft tissues, gingiva and periodontal ligament. Although bone-derived cells, osteoblast-like cells and osteoblasts can synthesize 1,25OH 2 D 3 [15,16], it is unknown whether this also occurs in periodontal soft tissues. Human gingival fibroblasts (hGF) and human periodontal ligament cells (hPDLC) are two kinds of periodontal fibroblasts and are important components of gingiva and periodontal ligament, respectively. Whether 1,25OH 2 D 3 synthesis occurs in these cells warrants further investigation.
Our previous studies [19,20] demonstrated that levels of plasma 25OHD 3 of patients with aggressive periodontitis were signifi-cantly higher than those of healthy controls; local 25OHD 3 levels in gingival crevicular fluids were 300 times higher than plasma levels in these patients. It is not known why the precursor of 1,25OH 2 D 3 is abundant around periodontal soft tissues. Recently, McMahon et al. reported that 1,25OH 2 D 3 could enhance the antibacterial defense of human gingival epithelial cells [21]. Thus, if 25OHD 3 is converted to 1,25OH 2 D 3 by the cells of periodontal soft tissues, this conversion could enhance the innate immune defense in the oral cavity. Here we hypothesized that hGF and hPDLC have 1a-hydroxylase activity and can convert 25OHD 3 to 1,25OH 2 D 3 . The objective of this study was to test the above hypothesis.

Results
1a-hydroxylase mRNA was detected in all the cells of the five donors ( Fig. 1 and 2). The 1a-hydroxylase protein was detected with an antibody that detects 1a-hydroxylase expression in human osteoblasts, indicating that the 1a-hydroxylase in hGF and hPDLC could be the same CYP27B1 enzyme as that found in osteoblasts [15,16].
After confirming the expression of 1a-hydroxylase in hGF and hPDLC, the function of 1a-hydroxylase was investigated. While 1000 nM 25OHD 3 did not have a significant cytotoxic effect on any of the cells within 48 h. hGF and hPDLC generated 1,25OH 2 D 3 in response to 25OHD 3 and the amount of 1,25OH 2 D 3 increased with incubation time (Fig. 3A, B). The fact that extra-and intracellular 1,25OH 2 D 3 was generated in the presence of 25OHD 3 provides the most convincing evidence of the existence of 1a-hydroxylase in hGF and hPDLC. Although the total amounts of 1,25OH 2 D 3 synthesized by hGF and hPDLC were not significantly different, more 1,25OH 2 D 3 was released by hPDLC 12 h after adding 25OHD 3 . At all other time points, there was no significant difference in the levels of intracellular and extracellular 1,25OH 2 D 3 between the two cell types. Moreover, CYP27B1 mRNA expression peaked at 1 h and no difference was detected between 0 h and any other time point (Fig. 3C). CYP24A1 mRNA expression increased over time and was significant higher at 24 h and 48 h than at 0 h (Fig. 3D).
In addition, exposure to 25OHD 3 also resulted in an upregulation of the 1,25OH 2 D 3 responsive genes CYP24A1 and RANKL in hGF and hPDLC (Fig. 4A, B), which is further evidence of 1a-hydroxylase activity in hGF and hPDLC.
Based on this direct evidence for 1a-hydroxylase activity in hGF and hPDLC, we examined the effect of 1a-hydroxylase knockdown. The efficiency of RNA interference against CYP27B1 was over 70% (Fig. 5). The generation of 1,25OH 2 D 3 increased with increasing 25OHD 3 concentrations but dropped significantly when CYP27B1 was knocked down using specific siRNA (Fig. 6A, B). Although 1,25OH 2 D 3 is the major ligand of VDR, 25OHD 3 could bind VDR with lower affinity and have a biological effect [22]. To address this issue, we knocked down CYP27B1 and found that the expression of CYP24A1 and RANKL decreased significantly (Fig. 7A, B), which strongly suggests that the effect of 25OHD 3 in hGF and hPDLC occurs after its conversion to 1,25OH 2 D 3 . This is additional evidence for the activity of 1ahydroxylase in hGF and hPDLC.
Despite the detection of the two kinds of DBP receptor expression, megalin mRNA expression was not detected. Both mRNA and protein expression of cubilin were detected in hGF and hPDLC ( Fig. 9, 10).

Discussion
Interestingly, the 1,25OH 2 D 3 produced by hGF and hPDLC can influence vitamin D-responsive gene expression (Fig. 4, 7), an observation that provides a solid indication of an autocrine/ paracrine action of vitamin D in periodontal fibroblasts. Because metabolism of 25OHD 3 and 1,25OH 2 D 3 by CYP24A1 is very important in the autocrine/paracrine action of vitamin D, we investigated the time course of CYP27B1 and CYP24A1 mRNA expression. CYP27B1 was only up-regulated at 1 h, while CYP24A1 mRNA expression increased over time (Fig. 3C, D). The results indicated that as more 1,25OH 2 D 3 was synthesized, CYP24A1 was more robustly expressed, which resulted in more conversion of 1,25OH 2 D 3 to 1,24,25OH 3 D 3 and 25OHD 3 to 24,25OH 2 D 3 , in turn resulting in less substrate for 1,25OH 2 D 3 synthesis. Thus, hGF and hPDLC could regulate their own 1,25OH 2 D 3 generation; this provided further evidence for autocrine/paracrine action of vitamin D.
Vitamin D binding protein (DBP) is the plasma carrier of vitamin D in humans [23]. Nykjaer et al. [24,25] confirmed that vitamin D enters cells by receptor-mediated endocytosis and via the endocytic receptors megalin and cubilin. In the present study, cubilin mRNA and protein were detected in hGF and hPDLC ( Fig. 9, 10), however, megalin was not detected despite the use of two pairs of primers. Atkins et al. [16] reported that osteosarcoma cell lines and primary osteoblast-like cells express abundant cubilin mRNA but only osteosarcoma cell lines express megalin mRNA. The cells used in our study were primary cells and their expression of DBP receptors was similar to the DBP receptor expression in primary osteoblast-like cells. Due to the presence of DBP receptors, plentiful 25OHD 3 around periodontal fibroblasts [20] could be used as a substrate for 1,25OH 2 D 3 synthesis. In addition, extracellular 1,25OH 2 D 3 (Fig. 3A, 6A), generated by hGF and hPDLC, could enter the surrounding cells through cellular uptake associated with DBP, which suggests the possibility of paracrine action of vitamin D.
In the present study, CYP24A1 and RANKL mRNA expression were chosen to test the downstream biological effects of 25OHD 3 . It is striking that the biological effects caused by the locally produced 1,25OH 2 D 3 after 1000 nM 25OHD 3 treatment were much stronger than those achieved with exogenous 10 nM 1,25OH 2 D 3 (Fig. 4A, B), and those similarly observed in treatment of osteoblasts [15]. After 1000 nM 25OHD 3 treatment, 1,25OH 2 D 3 concentration in the supernatants of hGF and hPDLC were 120 pM-360 pM and 120 pM-600 pM, respectively, much lower than 10 nM, but the biological effects were much stronger. One reason might be that after 25OHD 3 treatment, 1,25OH 2 D 3 is found not only in the supernatant but also in the cell lysates, allowing intracellular 1,25OH 2 D 3 to bind VDR and influence the function of these cells directly. On the other hand, exogenous 1,25OH 2 D 3 should enter the cells before eliciting a response. Thus, the direct availability at the site of action might be of great importance.
IL-1b in gingival crevicular fluids of patients with periodontitis reduced significantly after initial periodontal therapy, indicating that IL-1b is associated with periodontitis [20]. Porphyromonas gingivalis is an important pathogen of periodontitis and butyrate is its metabolite [26]. In our previous studies [27,28], we demonstrated that the butyrate concentrations in gingival crevicular fluids  . Activity of 1a-hydroxylase in hGF and hPDLC. hGF and hPDLC from donors 2, 3 and 5 were incubated with 1000 nM 25OHD 3 for the times indicated and the production of 1,25OH 2 D 3 was determined in (A) supernatants and (B) cell lysates. After prolonged incubation, production of 1,25OH 2 D 3 increased. hPDLC released more 1,25OH 2 D 3 than hGF 12 h after incubation with 25OHD 3 . The data is presented as the mean6SE. ** hGF generated significantly less 1,25OH 2 D 3 than hPDLC at the same time point (p,0.05). The time course of CYP27B1 (C) and CYP24A1 (D) mRNA expression is also presented. CYP27B1 mRNA expression peaked at 1 h and no difference was detected between 0 h and any other time point. CYP24A1 mRNA expression was significantly higher at 24 h and 48 h than 0 h. * CYP27B1 or CYP24A1 mRNA expression at the time point was significantly different from that at 0 h in hGF (p,0.05). # CYP27B1 or CYP24A1 mRNA expression at the time point was significantly different from that at 0 h in hPDLC (p,0.05). doi:10.1371/journal.pone.0039878.g003 of patients with periodontitis were significantly higher than those of healthy controls and butyrate levels in gingival crevicular fluids were significantly correlated with periodontal inflammation. To investigate the regulation of 1a-hydroxylase in hGF and hPDLC, IL-1b, Pg-LPS and sodium butyrate were chosen for the current study. However, although stimuli with periodontal characteristics were used here to simulate a periodontitis condition, this does not properly model a chronic disease situation in vivo and can only help to investigate the regulation of CYP27B1 in hGF and hPDLC. We found the NF-kB activator IL-1b to be a potent upregulator of CYP27B1 mRNA in hGF and hPDLC (Fig. 8), which is in line with the observation in osteoblasts [15]. We also found that sodium butyrate could significantly up-regulate the expression of CYP27B1 while Pg-LPS could not, an observation which requires further study. Classical renal regulators of 1a-hydroxylase are parathyroid hormone, calcium and 1,25OH 2 D 3 [10,29]. Renal 1,25OH 2 D 3 synthesis is primarily stimulated by low serum calcium, and consequently by parathyroid hormone that upregulates CYP27B1 expression while 1,25OH 2 D 3 itself downregulates CYP27B1 by negative feedback [30]. In contrast to this, we found that 1a-hydroxylase expression in hGF and hPDLC was less sensitive to a regulation by these agents (Fig. 8). In addition, 1,25OH 2 D 3 did not significantly change the up-regulation of CYP27B1 by IL-1b and sodium butyrate. A possible explanation might be that the induction of 1a-hydroxylase in hGF and hPDLC involves NF-kB pathways that are different from those involved in renal, cAMP-mediated protein kinase A signaling [31,32] and negative vitamin D response element [33,34]. This concept is similar to that of Hewison [35].
In the present study, each donor supplied both hGF and hPDLC and the CYP27B1 activity of hGF and hPDLC was compared. We demonstrated that hPDLC released more 1,25OH 2 D 3 than hGF 12 h after adding 25OHD 3 . In the study of van D M et al. [15], osteoblasts incubated with 1000 nM 25OHD 3 also generated 1,25OH 2 D 3 and the concentration of 1,25OH 2 D 3 in the supernatant was about 400-800 pM. In the present study, 1,25OH 2 D 3 concentrations in the supernatants of hGF and hPDLC were 120 pM-360 pM and 120 pM-600 pM, respectively. Thus, osteoblasts might release more 1,25OH 2 D 3 than hGF and hPDLC. hGF and hPDLC are two different types of fibroblasts of periodontal soft tissues. It has previously been demonstrated that hPDLC has a higher degree of similarity with osteoblasts than hGF [36,37,38]. This might be the reason for the higher release of 1,25OH 2 D 3 by hPDLC than by hGF.
In recent years, increasing attention has been paid to the relationship between vitamin D and periodontitis. Dietrich et al. [39] reported that serum 25OHD 3 concentrations are significantly and inversely associated with attachment loss in people aged over 50 years, however, there is no significant association between serum 25OHD 3 concentrations and attachment loss in patients younger than 50 years. Dietrich et al. [40] also reported that subjects with high serum 25OHD 3 concentrations were less likely to have bleeding on probing compared to subjects with low serum  . Efficiency of RNA interference against CYP27B1. All cells were transfected with either a siRNA oligonucleotide for CYP27B1 or a non-silencing control. Using real-time PCR as a measure, the efficiency of RNA interference against CYP27B1 was over 70% in hGF and hPDLC from all 5 donors. Donors are numbered 1-5. The data are presented as the mean6SD. * denotes difference from vehicle (p,0.05). doi:10.1371/journal.pone.0039878.g005 25OHD 3 concentrations. However, periodontally healthy subjects were not distinguishable from patients with either aggressive periodontitis or chronic periodontitis, and local 25OHD 3 levels in gingival crevicular fluids were not detected in these studies. Thus, we designed our previous studies and found that patients with aggressive periodontitis had higher plasma 25OHD 3 concentrations than healthy controls and that after periodontal therapy the higher plasma 25OHD 3 concentrations were significantly reduced, indicating that 25OHD 3 might be involved in periodontal inflammation [19,20]. In addition, local 25OHD 3 levels in gingival crevicular fluids were 300 times higher than plasma levels in patients with aggressive periodontitis [20]. Because 1,25OH 2 D 3 may enhance the antibacterial defense of human gingival epithelial cells [21], the confirmation of 1a-hydroxylase activity in hGF and hPDLC implies that local 25OHD 3 in gingival crevicular fluids might be metabolized to 1,25OH 2 D 3 by hGF and hPDLC and be involved in the innate immune defense in the oral cavity. From this perspective, 1a-hydroxylase activity in hGF and hPDLC may benefit periodontal health. Recently, it was reported that calcium and vitamin D supplementation have a positive effect on periodontal health [41,42]. However, topical application of vitamin D has not been reported. Since hGF and hPDLC have the ability to synthesize 1,25OH 2 D 3 , a topical application of 25OHD 3 might fulfill the function of 1,25OH 2 D 3 . Thus, our data Figure 6. Effect of CYP27B1 silencing on 1,25OH 2 D 3 generation. hGF and hPDLC from donors 2, 3 and 5 were treated with 25OHD 3 at various concentrations indicated in the figure for 48 h after transfection with the siRNA oligonucleotide for CYP27B1 or a non-silencing control and 1,25OH 2 D 3 production was measured in (A) supernatants and (B) cell lysates. When CYP27B1 was not silenced, production of 1,25OH 2 D 3 increased with increasing concentration of 25OHD 3 . When CYP27B1 was silenced, the generation of 1,25OH 2 D 3 decreased significantly compared with when CYP27B1 was not silenced. The data are presented as the mean6SE. * hGF generated significantly less 1,25OH 2 D 3 with the same amount of 25OHD 3 when CYP27B1 was knocked down (p,0.05). # hPDLC generated significantly less 1,25OH 2 D 3 (with the same amount of added 25OHD 3 ) when CYP27B1 was knocked down (p,0.05). doi:10.1371/journal.pone.0039878.g006 suggest a potential benefit of topical application of 25OHD 3 in periodontal therapy.
Because no established cell lines of hGF or hPDLC are available, the cells used in the present study were all obtained from primary culture. Although all data were obtained from cell from at least 3 donors, the existence of individual differences among the cells from primary culture was the limitation of this study.
In conclusion, the results in the current study have identified for the first time hGF and hPDLC as new sites of extrarenal synthesis of 1,25OH 2 D 3 . Similar to other extrarenal 1a-hydroxylases, periodontal 1a-hydroxylase appears to fulfill an autocrine/ paracrine function.

Ethics Statement
The study protocol was approved by the institutional review board of Peking University School and Hospital of Stomatology (PKUSSIRB-2011007) and written informed consent was ob-tained from each participant in accordance with the Declaration of Helsinki.

Cell culture
Primary culture of hGF and hPDLC was carried out according to previously described methods [7,26,36,43]. Briefly, hPDLC were prepared from extracted third molars of 5 young healthy volunteers and hGF were obtained from gingiva of the same 5 donors. The periodontal ligament tissues attached to the middle third of the roots were gently curetted by a surgical scalpel, minced and inoculated into 24-well plates. Gingiva from different donors was also minced and inoculated into 24-well plates. Tissue explants were maintained in Dulbecco's Modified Eagle's Medium (DMEM; Gibco, Grand Island, NY, USA) supplemented with 10% (v/v) fetal bovine serum (FBS; PAA, Coelbe, Germany), 100 U/ml penicillin G and 100 mg/ml streptomycin. Cultures were maintained in a humidified atmosphere of 5% (v/v) CO 2 at 37uC. After reaching 80% confluence, hGF and hPDLC were digested with 0.25% (w/v) trypsin and 0.02% (w/v) EDTA and subcultured at a 1:3 ratio. Cytotoxicity test of 25OHD 3 hGF and hPDLC from three donors were used in the cytotoxicity test. hGF and hPDLC in their logarithmic phase were plated into 96-well plates at a density of 3000 cells/well in DMEM with 10% DCC-FBS. 24 h later, the medium was replaced by DMEM without DCC-FBS. After another 24 h, the medium was changed to DMEM with 10% DCC-FBS and supplemented with 1000 nM 25OHD 3 or vehicle, respectively. The cytotoxicity test was carried out according to the Cell

Detection of CYP27B1 expression
Cells from all five donor were seeded into 6-well plates at a density of 5000/cm 2 in DMEM supplemented with 10% DCC-FBS. 4 days later, a portion of the cells were harvested using Trizol agent (Dongsheng Biotech, Guangzhou, China) for RT-PCR. RNA was extracted using Trizol according to the manufacturer's instructions and was reverse transcribed to cDNA using a reverse transcription kit (Bio-Rad, Hercules, CA, USA). PCR reactions were performed using the Taq PCR MasterMix (Tiangen Biotech, Beijing, China) in a Hybaid PCR Thermal Cycler (Thermo Scientific, Boston, Mass, USA). The primers used are listed in Table 1.
The remaining cells were harvested using lysis buffer [20 mM Tris (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% (v/v) Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM bglycerol phosphate and 2 mM Na 3 VO 4 supplemented with protease inhibitor cocktail (Roche, Mannheim, Germany)] [43] for Western blotting. Protein concentration was determined using the Bicinchoninic Acid Protein Assay Kit (Applygen, Beijing, China). 20 mg of total protein from each sample was loaded onto a gel comprising a 5% (w/v) stacking gel and a 10% (w/v) running gel. At the end of the electrophoresis, samples were transferred onto nitrocellulose blotting membranes (Hybond TM ; Amersham

Measurement of 1,25OH 2 D 3 production
Cells from 3 donors were treated with 1000 nM 25OHD 3 (Sigma, St. Louis, MO, USA) for 1 h, 4 h, 12 h, 24 h or 48 h, after which supernatants were collected and a portion of cells were scraped in PBS containing 0.2% Triton X-100 and stored at 280uC. Prior to use, cell lysates were sonicated on ice in a sonifier cell disrupter for 2615 s. 1,25OH 2 D 3 levels in cell supernatants and cell lysates were detected using a 1,25OH 2 D 3 radioimmunoassay kit (DiaSorin, Stillwater, MN) according to the manufacturer's instructions. The cross-reactivity with 25OHD 3 was less than 0.01%.
In addition, a portion of the cells were harvested using Trizol reagent for the detection of the time course of CYP27B1 and CYP24A1 mRNA expression using real-time PCR. Real-time PCR reactions were performed using SYBRH Premix Ex Taq TM II (TaKaRa Biotechnology, Dalian, China) in an ABI 7500 realtime Thermocycler (Applied Biosystems, Foster city, CA, USA).  The primers used are listed in Table 1. The data were analyzed using the SDS software according to the manufacturer's instructions. Glyceraldehyde-3-phosphate-dehydrogenase (GAPDH) was used as an internal control. The data are presented as relative mRNA levels calculated by the equation 2 2DCt (DCt = Ct of target gene minus Ct of GAPDH) [44].

Detection of CYP24A1 and RANKL mRNA expression
Cells from five donors were incubated with 10 nM 1,25OH 2 D 3 (Sigma, St. Louis, MO, USA), 1000 nM 25OHD 3 or vehicle for 48 h before harvesting. Then RNA and cDNA were obtained and real-time PCR reactions were performed as described previously.

RNA interference of CYP27B1
To confirm the dependence of 25OHD 3 conversion into 1,25OH 2 D 3 on CYP27B1, the highly specific technique of RNA interference was utilized. Cells were seeded at a density of 15000/ cm 2 in 6-well plates. 8 h later, cells were transfected with either CYP27B1 siRNA (10 nM) or non-silencing control siRNA using Hiperfect TM transfection reagent (Qiagen, Duesseldorf, Germany) according to the manufacturer's instructions. The target sequence of CYP27B1 siRNA was 59-CTGGTTTACGGTTTCTTATAA-39 and the non-silencing control was a non-homologous, scrambled sequence equivalent. 60 h after transfection, cells were harvested to confirm the effect of RNAi using real-time PCR. RNA and cDNA were obtained and real-time PCR was performed as described earlier.
After verifying the effect of RNAi, 1,25OH 2 D 3 production after RNAi was determined. Cells were first transfected with CYP27B1 siRNA (10 nM) or non-silencing control siRNA and 12 h after transfection, these cells were treated with 200 nM, 400 nM, 600 nM or 1000 nM 25OHD 3 (Sigma, St. Louis, MO, USA) for another 48 h. Then, the 1,25OH 2 D 3 concentrations in the cell supernatants and cell lysates were determined as described earlier.
CYP24A1 and RANKL mRNA expression was also examined after RNAi. Cells were first transfected with CYP27B1 siRNA (10 nM) or non-silencing control siRNA and 12 h after transfection, these cells were treated with 1000 nM 25OHD 3 for another 48 h. Then, CYP24A1 and RANKL mRNA expression was determined by real-time PCR.

Detection of vitamin D binding protein receptor expression
The mRNA expression of megalin and cubilin were detected by RT-PCR as described earlier and the primers used are listed in Table 1.
Protein expression of cubilin was detected by immunocytochemistry. Cells from five donors were seeded on glass slides at a density of 20000/cm 2 and 8 h later cells on glass slides were incubated with primary antibodies against cubilin (diluted 1:50; Santa Cruz Biotechnology, Santa Cruz, CA, USA). The PV-9000 Polymer Detection System (Zhongshan Golden Bridge Biotechnology, Beijing, China) was used for immunocytochemical staining and a diaminobenzidine (DAB) kit (Zhongshan Golden Bridge Biotechnology, Beijing, China) was used to develop the colour followed by haematoxylin staining. The primary antibody was replaced by PBS for negative controls.

Statistical Methods
The Shapiro-Wilk test was used to determinate the distribution of the variants. The Wilcoxon test and Friedman test were used to compare differences between the mRNA expression levels of CYP24A1 and RANKL in different groups. The effect of RNA interference and the effect of 25OHD 3 exposure on CYP27B1 mRNA expression was analyzed using the paired samples t-test. Comparison of 1,25OH 2 D 3 generation by hGF and hPDLC, comparison of 1,25OH 2 D 3 generation with and without knockdown of CYP27B1 and comparison of CYP27B1 up-regulation in hGF and hPDLC were all performed using a paired samples t-test.
Statistical analyses were carried out using the SPSS 11.5 software package (SPSS Inc., Chicago, IL, USA). A p value,0.05 was considered statistically significant.