Retraction: The following article from the Journal of Bone and Mineral Research, “CDP/Cut Is an Osteoblastic Coactivator of the Vitamin D Receptor (VDR)” by Eiji Ochiai, Hirochika Kitagawa, Ichiro Takada, Sally Fujiyama, Shun Sawatsubashi, Mi-sun Kim,Yoshihiro Mezaki, Yu Tsushima, Ken-ichiro Takagi, Yoshiaki Azuma, Ken-ichi Takeyama, Kazuyoshi Yamaoka, Shigeaki Kato, Takashi Kamimura, published online on December 11, 2009 in Wiley Online Library (wileyonlinelibrary.com), has been retracted by agreement between the authors, the journal Editor in Chief, Thomas Clemens, the American Society for Bone and Mineral Research and Wiley Periodicals, Inc. The authors have requested the retraction based on their acknowledgement that several of the figures did not reflect the observations presented.
Vitamin D is a well-described regulator of bone metabolism, calcium homeostasis, and cell differentiation.1–4 1α,25-dihydroxyvitamin D3 (vitamin D3) is the physiologically active form of vitamin D, and it exerts a wide variety of biologic effects by binding to its specific nuclear vitamin D receptor (VDR), resulting in transcriptional control of specific target genes in each tissue. The VDR is a member of the steroid/thyroid hormone nuclear receptor (NR) superfamily.5 Like other NRs, the VDR serves as a ligand-dependent transcription factor. The importance of the VDR to the biologic actions of vitamin D has been confirmed by genetic observations: Disruption of the VDR gene in mice causes a severe rachitic phenotype similar to that seen in humans and animals experiencing vitamin D deficiency.6 Clinical and experimental data demonstrate that the administration of vitamin D3 at appropriate doses effectively increases bone volume without hypercalcemia. Together with a previous report that vitamin D potently stimulates bone formation in vivo and in vitro,7–11 the results implicate that the anabolic effect of vitamin D on bone tissue is mediated by the VDR in osteoblasts.
Like other NR members, the VDR requires distinct classes of coregulators and multiprotein coregulator complexes in order to initiate vitamin D3–mediated chromatin reorganization.12–17 These complexes appear to modify chromatin configuration by controlling nucleosomal rearrangement and enzyme-catalyzed modifications of histone tails. The p160-containing complexes are the best characterized complexes that possess histone acetyltransferase activity.12, 18 After reorganizing chromatin structure, the other coregulator complexes appear to facilitate bridging between NRs and basal transcription factors along with RNA polymerase II, resulting in formation of the transcription initiation complex. Among such complexes, the VDR-interacting protein complex (DRIP, the mediator complex) has been studied extensively.13, 14 We have shown previously that chromatin reconfiguration by the VDR is mediated by a subclass (WINAC) of the SWI/SNF type ATP chromatin remodeling complex. WINAC potentiated VDR function in ligand-induced transactivation as well as transrepression.17 Several coregulators/complexes have been shown to support the function of the VDR. However, those coregulators cannot fully account for all the changes in gene expression and chromatin reorganization mediated by the VDR. Moreover, recent progress in this field has demonstrated that coregulators form functionally distinct complexes with tissue-specific components.19, 20 With this background in mind, we chose to search for and identify new osteoblast-specific VDR coregulators to better understand VDR function during osteoblastogenesis.
Materials and Methods
The GST-VDR (DEF) and GAL4-VDR (DEF) [pM-VDR(DEF)] constructs were prepared as described previously.17, 21 The eukaryotic expression vector for full-length human CDP (pCMV-SPORT6-CDP) was purchased from Open Biosystems (Image ID 5740343). The VP16 activation domain was subcloned into the pM vector (Clontech, Mountain View, CA, USA) (pM-VP16). Five repeats of 17mer GAL4 upstream sequences (UASx5) were inserted into the pGL4 vector (Promega, Madison, WI) (GAL4-UASx5-Luc). The pcDNA3-DRIP205 expression vector was created as described previously.22 siRNA pools for CDP (LQ-011635-00-0002), DRIP205 (LQ-004126-00-0002), and nonspecific control (D-001810-10-05) were purchased from Thermo Fisher Scientific (Fremont CA, USA). CUX1 (CDP) Stealth Select 3 RNAi (HSS102511, HSS102512, HSS102513), MED1 (DRIP205) Stealth Select 3 RNAi (HSS108299, HSS108300, HSS182612), BLOCK-iT Stealth RNAi Control (12935-200), and Lipofectamine RNAiMAX were purchased from Invitrogen (Carlsbad, CA, USA). The following commercially available antibodies were used: anti-CDP and anti-DRIP100 (sc-6327 and sc-5338. respectively, both from Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-VDR (RT-200-P) from Lab Vision (Fremont CA, USA), anti-SRC-1 (clone 1135, 05-522) from Upstate Biotechnology (Lake Placid NY, USA), and anti-FLAG (F7425) from Sigma Chemical (St. Louis, MO, USA).
Purification and separation of VDR-associated proteins
Nuclear extracts (NEs) from HOS and HeLa cells were used to biochemically purify VDR-associated proteins using GST-VDR(DEF) columns, as described previously.17 The proteins that bound to the GST-VDR(DEF) columns were eluted, applied to SDS-PAGE, silver stained, and Western blotted for the indicated proteins. For further purification of CDP-VDR complexes, the eluate obtained from a GST-VDR column was reimmunoprecipitated using anti-CDP antibody or fractionated by glycerol density gradient, as described previously.17, 21
Protein identification by mass spectrometric analysis
The identification of each component was described earlier.17, 21, 23 Briefly, single SDS-PAGE protein bands were excised and digested. The peptides were analyzed by MALDI-TOF/MS (Ultraflex III Mass Spectrometer, Bruker Daltonics, Billerica, MA, USA). After determination of the mass of each peptide fragment, results were analyzed with the MS-Fit program (University of California–San Francisco Mass Spectrometry Facility).
GST pull-down assays
GST pull-down assays were done as described previously.23 Human CDP in a pCMV-SPORT6 vector (Invitrogen) or human DRIP205 in a pcDNA3 vector (Invitrogen) was used to generate [35S]methionine-labeled protein (GE Healthcare, Buckinghamshire, UK).
Immunoprecipitation was performed as described previously.23 HOS cells were transfected with 5 µg of each expression vector using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. After incubation for 48 hours, cells were treated with vitamin D3 or vehicle (ethanol) for 1 hour. The collected cells were resuspended in ice-cold buffer A (10 mM HEPES, pH 7.9, 1.5 mM MgCl2, 10 mM KCl, and 0.5 mM DTT). Cells were passed through a 26G needle and centrifuged for 20 minutes at 10,000g. Nuclear pellets were resuspended in buffer C [20 mM HEPES, pH 7.9, 25% v/v glycerol, 0.3 M NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), and 0.5 mM DTT] and incubated for 30 minutes. The suspension was centrifuged again for 20 minutes at 10,000g. The collected supernatant was diluted with an equal volume of buffer C (lacking NaCl). The solution was centrifuged again for 20 minutes at 10,000g. Immunoprecipitation of the supernatants was performed using anti-FLAG M2 affinity resin (A2220, Sigma), and the bound CDP was detected by Western blotting. For immunoprecipitation of endogenous CDP, cells were treated with vitamin D3 or vehicle (ethanol) for 1 hour, cell lysates were prepared as described earlier, immunoprecipitation was done using anti-CDP antibody, protein G agarose fast flow (16-266; Millipore, Billerica, MA, USA), and the bound VDR was detected by Western blotting.
Chromatin immunoprecipitation (ChIP) assay
HOS cells were treated with vitamin D3 (10−7 M) for 24 hours and immunoprecipitated with antibodies specific for the VDR or CDP. Soluble chromatin was prepared with a ChIP assay kit (Upstate Biotechnology) and immunoprecipitated with antibodies against the indicated proteins. Extracted DNA samples were amplified with primer pairs for CYP24 promoter (5′-GTCCAGGCTGGGGGTATCTG-3′ and 5′-CCAATGAGCACGCAGAGGAGG-3′).24 Optimized PCR conditions for semiquantitative measurement were as follows: 31 cycles of 30 seconds at 95°C, 30 seconds at 58°C, and 40 seconds at 72°C. PCR products were visualized on 2% agarose–Tris-acetate EDTA gels. For quantification of the immunoprecipitated DNA, real-time PCR was performed with a Power SYBR Green PCR Master Mix (Applied Biosystems, Forest City, CA, USA) and a two-step cycling protocol (anneal and elongate at 60°C, denature at 94°C) on a 7900HT fast real-time PCR system (Applied Biosystems). The primer pair for CYP24 promoter was used (5′- CGAAGCACACCCGGTGAACT -3′ and 5′- CCAATGAGCACGCAGAGGAG -3′).25
Human osteosarcoma cell line HOS cells (American Type Culture Collection, Manassas, VA, USA) were cultured with DMEM (Invitrogen) containing 10% fetal bovine serum (FBS) (Invitrogen) at 37°C and 5% CO2. Human osteoblastic SaM-1 cells (kindly provided by Dr Y Koshihara, Tokyo Metropolitan Institute of Gerontology) were cultured with α minimal essential medium (α-MEM) containing 10% FBS (Irvine Scientific, Santa Ana, CA, USA) at 37°C and 5% CO2. For osteoblast differentiation, SaM-1 cells were cultured in α-MEM containing 10% FBS and 2 mM α-glycerophosphate. The culture medium was replaced with fresh medium every other day.26, 27
Transient transfection, luciferase assay
For transient transfection, cells were plated in 48-well plates 1 day before transfection. HOS and SaM-1 cells were transfected with several plasmids using Lipofectamine 2000 reagent. A reporter plasmid (0.1 µg; GAL4-UASx5-Luc) was cotransfected with 0.1 µg of pM(GAL4)-VDR(DEF)28, 29 or pM(GAL4)-VP16 plus several concentrations of pCMV-SPORT6-CDP. As a reference to normalize transfection efficiency, 0.025 µg of pRL-TK plasmid (Promega) was cotransfected in all experiments. Three hours after transfection, the medium was replaced with fresh medium with vehicle or vitamin D3 (10−8 M). Preparation of cell extracts and dual luciferase assays was performed according to the manufacturer's protocols (Promega). All values are mean ± SD of at least six independent experiments. For the RNA interference experiment, siRNA duplexes were transfected with Lipofectamine 2000 in accordance with the manufacturer's protocol.
Alkaline phosphatase (ALP) staining
SaM-1 cells cultured for 7 days were washed twice with PBS and fixed with 3.7% formaldehyde. ALP staining was described previously.22, 30 The area of the ALP+ cells per total area (%) was analyzed by ScionImage (Scion Corporation, Frederick, MO, USA).
Alizarin red S staining
SaM-1 cells cultured for 28 days were washed twice with PBS and fixed with 3.7% formaldehyde. After fixation, cells were stained for calcified nodules with 2% alizarin red S (Wako, Osaka, Japan) for 5 to 10 minutes at room temperature.
Generation of adenovirus
Recombinant adenovirus carrying CDP was constructed using the AdenoX system (Clontech). The parental virus genomes in HEK 293A cells (Invitrogen) were constructed according to the manufacturer's protocol. HOS or SaM-1 cells were infected by incubation with the recombinant adenovirus for 24 hours.
For quantitative real-time reverse-transcriptase PCR (qRT-PCR), 1 µg of total RNA from each sample was reverse transcribed into first-strand cDNA with random hexamers using Superscript III reverse transcriptase (Invitrogen). Gene expression was assessed by qPCR performed using Power SYBR Green PCR Master Mix (Applied Biosystems) and a two-step cycling protocol (anneal and elongate at 60°C, denature at 94°C) with a 7900HT fast real-time PCR system (Applied Biosystems). The primer sets are listed in Table 1. Specificity of primers was verified by dissociation/melting curves for the amplicons. Linear standard curves also were established for each set of gene primers by using the threshold cycle. All transcript levels were normalized to that of β-actin.
Table 1. Primer Sequences for qRT-PCR
Quantitative data are presented as mean ± SD. Data comparisons were made using Student's t test (SAS Software Package, SAS Institute Japan, Ltd., Tokyo, Japan). The difference was considered significant when the p value was less than .05.
Purification and identification of ligand-dependent and osteoblast-specific proteins interacting with the VDR
To isolate osteoblast-specific coregulators for the VDR, NEs from osteoblastic osteosarcoma HOS cells were incubated with a chimeric VDR mutant protein (VDR-DEF) fused to glutathione-S-transferase (GST) in the presence or absence of 10−6 M vitamin D3. Since transcriptional coregulators associate with the C-terminal ligand-binding domain (LBD/DEF) region, a deletion mutant lacking the A/B/C domain was used as purification bait.1 HeLa NEs were used as a nonosteoblastic reference (Fig. 1A). Proteins associating with the VDR were enriched from the NEs as described in previous reports.17, 21 VDR-DEF-interacting proteins were separated by SDS-PAGE and visualized by silver staining (Fig. 1B) for protein identification using MALDI-TOF/MS analysis. Since most of the VDR-interacting proteins were known components of the DRIP complex in HOS cells as well as in HeLa cells, the results appears consistent with the previous reports13, 14, 17, 21, 31 (Fig. 1B). Among them, CCAAT displacement protein (CDP)/Cutl1/Cut32 (Fig. 1B) was identified as an interactant from approximately 200 kDa bands closely positioned to DRIP205 in HOS cells. This protein is a family of homeodomain proteins that are conserved among metazoans and are known as a transcriptional repressor.33 Tissue-specific or cell cycle stage–specific expression of this protein has already been reported.34–36 By Western blotting with specific antibodies, a DRIP coactivator complex component, DRIP100, was detected in the interactants from both cells, confirming that the DRIP complex was present in osteoblastic and nonosteoblastic lineages (Fig. 1C). On the other hand, CDP was detected only in HOS cells (Fig. 1C), suggesting a possibility that CDP is an osteoblast-specific VDR interactant.
CDP forms a protein complex with liganded VDR distinct from DRIP or SRC-1 complexes
The VDR interactants were fractionated by molecular mass using a glycerol density gradient. The result indicated that CDP forms a protein complex with the VDR (Fig. 2A). In a reimmunoprecipitation step using anti-CDP antibody, neither DRIP100 nor SRC-1 was detected in the immunoprecipitants, but both could be seen in the flow-through fraction (Fig. 2B). Together, these results suggested that CDP forms a protein complex with the VDR that is distinct from the DRIP complexes or the p160 family–containing complexes in HOS cells.
CDP interaction with the VDR is ligand-dependent in vivo and in vitro
We then used coimmunoprecipitation to test whether the CDP-VDR complex formed in living cells. By use of anti-FLAG antibody, FLAG fusion of the VDR and ligand-dependent interaction of CDP were detected by Western blotting (Fig. 3A). Next, to clarify whether the interaction of CDP with the VDR was direct, we tested the interaction with a GST pull-down assay. Using a chimeric VDR protein fused to GST as bait, we examined the interaction between VDR and CDP or DRIP205, a known direct interactant with the VDR in the DRIP complex. In this assay, clear ligand-dependent interactions between the VDR and CDP as well as DRIP205 were observed, implying that CDP physically interacts with the VDR in osteoblastic cells (Fig. 3B).
Formation of the endogenous CDP-VDR complex on the VDR target gene promoter is cell-type-specific
The cell-type-specific expression of CDP was confirmed by the protein-level analysis in several VDR-expressing cells (Fig. 4A). To confirm the ligand-dependent interaction between endogenous CDP and the VDR in HOS cells, endogenously formed CDP-VDR complex was immunoprecipitated and subjected to Western blotting. Clear ligand-dependent complex formation between CDP and the VDR was detected in HOS cell lysates (Fig. 4B). To determine if the CDP-VDR complex was recruited to the endogenous VDR target gene promoter, a ChIP assay was performed with the CYP24 promoter in HOS cells. A vitamin D3–induced recruitment of CDP was seen (Fig. 4C). Ligand-dependent increase in VDR recruitment also was observed on this promoter, as reported in previous papers.37, 38 From these results, it appears that endogenous CDP and VDR form a complex in a ligand-dependent manner and are recruited to VDR target gene promoters conceivably in osteoblastic cell–specific manner.
CDP is a transcriptional coactivator of the VDR
CDP is a transcriptional repressor for many target genes, directly binding to a specific DNA element.33 However, it remained unclear whether it actually serves as a transcriptional coregulator for the VDR. We therefore examined whether CDP coregulates the transcriptional function of the VDR by coexpression of CDP in HOS cells. A transient expression assay was performed in HOS cells by using pM(GAL4)-VDR(DEF) or pM(GAL4)-VP16 (as a control) and a reporter plasmid containing the luciferase gene and the consensus GAL4 UAS (GAL4-UASx5-Luc). The transactivation of the VDR induced by vitamin D3 (Fig. 5A, left, column 2) was potentiated by cotransfection of CDP or DRIP205 (Fig. 5A, left, columns 3–8). Coactivation was not seen for transactivation of VP16 (Fig. 5A, right). We then asked whether CDP coactivates VDR on a known endogenous VDR target gene, CYP24, in HOS cells by qRT-PCR. In HOS cells overexpressing CDP in an adenovirus system, CYP24 mRNA induction by vitamin D3 was potentiated when compared with cells expressing a mock vector (Fig. 5B). Thus it appears that CDP serves as a coactivator, i.e., a complex component for ligand-dependent transactivation of the VDR on its target gene promoters. Using knockdown strategy with RNAi against CDP or DRIP205, the contribution of endogenously expressed CDP to the transcriptional property of the VDR was observed in SaM-1 cells, preosteoblastic cells derived from human periosteum that are inducible by vitamin D3 in differentiation26, 27 (Fig. 5C). These results indicate that endogenously expressed CDP can act as a coactivator of the VDR in SaM-1 cells.
Vitamin D3–dependent osteoblastic differentiation of human osteoblastic SaM-1 cells
The role of the VDR in osteoblasts is not understood at the molecular level. Thus we examined the physiologic impact of CDP in VDR-mediated biologic actions. Specifically, the role of CDP in vitamin D3–induced osteoblastogenesis was studied with the SaM-1 cells.26, 27 Following 7 days of stimulation with vitamin D3, SaM-1 cells are known to reach an immature osteoblast stage of differentiation and express alkaline phosphatase (ALP). Further treatment with vitamin D3 up to 28 days is known to induce calcification (detected with alizarin red S) characteristic of mature osteoblasts (Fig. 6A). Thus we tested the expression profile of CDP and the VDR (Fig. 6B) during cultivation. The expression levels of both CDP and the VDR were elevated in a time-dependent fashion in concert with vitamin D3–induced cell differentiation.
CDP stimulates vitamin D3–dependent osteoblastic differentiation of SaM-1 cells
When CDP was overexpressed with an adenovirus vector (Fig. 7A), osteoblastogenesis induced by vitamin D3 was potentiated, as indicated by increased ALP expression (Fig. 7B). In contrast, knockdown of CDP by siRNA (Fig. 7C) attenuated the action of vitamin D3 in osteoblastic differentiation (Fig. 7D). Because the knockdown of DRIP205 had little effect on vitamin D3–dependent osteoblastogenesis, we conclude that CDP is a specific regulator of osteoblastogenesis through coregulation of VDR function.
During the differentiation of osteoblasts, several transcription factors are believed to possess distinct, stage-specific roles.39 However, the detailed molecular mechanism of this process is still under investigation. To uncover the molecular details of transcriptional regulation by transcription factors and the linkage to osteoblastic differentiation, we chose a biochemical approach to identify transcriptional coregulators. Although the VDR is considered a key transcription factor responsible for osteoblastogenesis, it appears unlikely that all the known VDR coregulators can account for VDR function during osteoblastic differentiation.13, 14, 40, 41 In this study, CDP was found to be an osteoblast-specific VDR interactant. CDP appears to form a distinct complex that is different from the DRIP complex or p160 family–containing complex. Considering CDP's participation in vitamin D3–dependent osteoblastogenesis, CDP is presumed to be recruited to the VDR on the specific VDR target gene promoters only at an early stage of osteoblastogenesis. However, it is still unclear how CDP coregulates the VDR in terms of histone modifications and chromatin remodeling. In this respect, analyzing the cooperation between general epigenetic regulators and tissue-specific/differentiation stage–specific factors or complexes represents the next challenge in the field.
In previous reports, CDP generally was regarded as a transcriptional repressor.33, 42–44 In proliferating osteoblastic cells, CDP also was reported to be a transcriptional repressor of the osteocalcin gene.45 On the other hand, several reports have implied a role in transcriptional activation.33 Here, we have shown that CDP serves as a transcriptional coactivator for the VDR in a vitamin D3–dependent osteoblast differentiation system of preosteoblastic cells. The function of CDP is believed to be regulated by several posttranslational modification events, including phosphorylation and acetylation.33, 46, 47 Thus stage-specific modification of the CDP protein itself might determine its specific role in osteoblastic differentiation by stage-specific combination of the complex components.
Although there are several reports of vitamin D3–dependent induction of osteoblastogenesis in vitro and in vivo.7, 9–11, 26 the contribution of the VDR to osteoblastogenesis is still controversial. In our experimental system, CDP potentiated the effects of vitamin D3 on osteoblastogenesis of SaM-1 cells (Fig. 7B, D). Together with the findings that CDP is a transcriptional coactivator for the VDR in transcriptional-reporter assays and endogenous gene induction, the VDR appears to be a positive regulator for osteoblastic differentiation at an early stage of osteoblastogenesis. Although in vivo function of CDP in osteoblastogenesis has not been well documented, probably owing to the neonatal death of CDP-deficient models,34, 48 the indicated expression pattern of CDP in bone-related tissues34 is overlapping with the expression pattern of VDR.49–51 Thus we speculate that VDR and CDP should communicate with each other at a specific stage of osteoblastogenesis to work cooperatively for bone formation in vivo. However, to confirm our hypothesis that osteoblastic differentiation is triggered by the VDR and vitamin D3 via CDP, further careful analyses by genetic approaches clearly are required because osteoblastogenesis in intact bones is regulated by complex networks originating from other cell types in the bone. Considering the fact that transcriptional coregulators are functionally shared among different classes of transcription factors, it is possible that CDP transcriptionally coregulates other transcription factors responsible for cytodifferentiation and proliferation of other cell types in bone. Similarly, it is conceivable that activated VDR indirectly controls osteoblastogenesis through regulating functions of cells associating with osteoblasts. This idea is indeed supported by a recent report that activated VDR inhibits osteoblast proliferation.52 Thus a genetic approach will be required to selectively disrupt osteoblast gene expression at specific stages of osteoblastogenesis if we are to fully understand the role of the VDR at each stage of osteoblastogenesis.
EO, YT, K-IT, YA, KY, and TK are employees of Teijin Pharma, Ltd. All the other authors state that they have no conflicts of interest.
We would like to thank Dr Yasuko Koshihara for the SaM-1 cells and technical support. We also thank all the members of our laboratory for their help and discussion. We also express our appreciation for the technical assistance provided by Ms Toshie Jinbo and Ms Kimiko Johnouchi, and we thank Ms Mai Yamaki for manuscript preparation.