Multifunctional Enhancers Regulate Mouse and Human Vitamin D Receptor Gene Transcription

Abstract

The vitamin D receptor (VDR) mediates the endocrine actions of 1,25-dihydroxyvitamin D₃ [1,25(OH)₂D₃] and autoregulates the expression of its own gene in target cells. In studies herein, we used chromatin immunoprecipitation-chip analyses to examine further the activities of 1,25(OH)₂D₃ and to assess the consequences of VDR/retinoid X receptor heterodimer binding at the VDR gene locus. We also explored mechanisms underlying the ability of retinoic acid, dexamethasone, and the protein kinase A activator forskolin to induce VDR up-regulation as well. We confirmed two previously identified intronic 1,25(OH)₂D₃-inducible enhancers and discovered two additional regions, one located 6 kb upstream of the VDR transcription start site. Although RNA polymerase II was present at the transcription start site in the absence of 1,25(OH)₂D₃, it was strikingly up-regulated at both this site and at individual enhancers in its presence. 1,25(OH)₂D₃ also increased basal levels of H4 acetylation at these enhancers as well. Surprisingly, many of these enhancers were targets for CCAAT enhancer-binding protein-β and runt-related transcription factor 2; a subset also bound cAMP response element binding protein, retinoic acid receptor, and glucocorticoid receptor. Unexpectedly, many of these factors were resident at the Vdr gene locus in the absence of inducer, suggesting that they might contribute to basal Vdr gene expression. Indeed, small interfering RNA down-regulation of CCAAT enhancer-binding protein-β suppressed basal VDR expression. These regulatory activities of 1,25(OH)₂D₃, forskolin, and dexamethasone were recapitulated in MC3T3-E1 cells stably transfected with a full-length VDR bacterial artificial chromosome (BAC) clone-luciferase reporter gene. Finally, 1,25(OH)₂D₃ also induced accumulation of VDR and up-regulated H4 acetylation at conserved regions in the human VDR gene. These data provide important new insights into VDR gene regulation in bone cells.

Multiple intronic and upstream enhancers control expression of the vitamin D receptor gene by 1,25(OH)₂D₃, retinoic acid, glucocorticoids, and the PKA activator forskolin.

The biological actions of 1,25-dihydroxyvitamin D₃ [1,25-(OH)₂D₃] are mediated by the vitamin D receptor (VDR), a ligand-activated transcription factor belonging to the steroid receptor family of genes that facilitates the actions of most small molecule hormones (1). Like other members of its class, the VDR forms heterodimers with retinoid X receptor (RXR), binds directly to specific vitamin D response elements (VDREs) located within the vicinity of target gene promoters, and mediates the recruitment of a wide variety of coregulatory complexes, the actions of which are essential for changes in the level of target gene expression (2). These complexes include those involved in modifying the chromatin acetylation or methylation states, in remodeling nucleosomes, and in recruiting RNA polymerase II (RNA pol II) (3). Recent studies have demonstrated that the sites of action of the VDR on target genes can be located not only proximal to the promoter, but at distal upstream and downstream sites as well. In the case of the Rankl gene, for example, at least five enhancers have been identified ranging from 16–76 kb upstream of the Rankl gene’s transcriptional start site (TSS) (4,5). In others, such as the LRP5 gene, a strong enhancer is located not only at the TSS, but in an intron some 30 kb downstream (6). Additional findings suggest that the binding of VDR to these sites increases levels of acetylation on the tails of histone 4 (H4ac) and, in some cases, facilitates the recruitment of RNA pol II to these sites as well (4,6,7,8,9). Thus, these enhancers appear to function as recruitment centers for transcriptional machinery that is essential to changes in gene expression at the TSS, perhaps through a chromatin-looping mechanism (10). Frequently, this activity also produces noncoding RNA transcripts, several of which are known to manifest regulatory function (11,12). It is clear from these studies that the mechanisms of transcriptional activation by steroid hormones are not yet fully elucidated.

The VDR is an absolute determinant of the biological activity of 1,25-(OH)₂D₃. Thus, the expression of this receptor in cells is a requirement for response, and the receptor concentration itself is a key component of cellular sensitivity to the hormone. These issues are highlighted in the VDR null mouse, where the loss of VDR leads to a complete abrogation of transcriptional response in the kidney, intestine, and bone, and thus to deranged calcium and phosphorus homeostasis (13,14). At the physiological level, numerous examples of Vdr gene regulation are evident, such as the induction of VDR during early development of the intestinal tract in rodents (15,16), after B and T cell activation (17,18), and in response to 1,25-(OH)₂D₃ treatment in specific muscle (19) and liver cells (20). Numerous examples of Vdr gene down-regulation occur as well, such as that which results during the differentiation of myeloid precursors into osteoclasts (21) or mature dendritic cells (22), and that which occurs in the parathyroid gland, the kidney, and perhaps in bone during the evolution of chronic kidney disease (23,24). The latter desensitizes the parathyroid gland to negative feedback by the VDR at the PTH gene, thereby exaggerating already elevated levels of PTH that lead to renal osteodystrophy (25). VDR levels can also be modulated both positively and negatively by a variety of steroid and peptide hormones, growth factors, and cytokines. As an example, retinoic acid (RA) prompts an up-regulation of the VDR, particularly in bone cells (26). Glucocorticoids (GCs) are also active in regulating expression of the VDR (27). This regulation is likely important for bone remodeling but can also lead to excessive bone loss as well. Finally, Vdr gene expression can also be regulated by PTH, which is known to affect the levels of VDR mRNA in both the kidney and bone (28). The underlying molecular mechanisms and the factors that are involved in this regulation at the level of the Vdr gene itself remain largely unknown.

Perhaps the most interesting regulation of the Vdr gene is by 1,25-(OH)₂D₃ itself. Indeed, 1,25-(OH)₂D₃ autoregulates VDR levels through both transcriptional (9,29,30) and posttranslational (31,32) mechanisms. In the latter case, the interaction of the ligand with its receptor increases the stability of the VDR protein through a mechanism that remains to be determined. With regard to the former, we have recently used chromatin immunoprecipitation (ChIP) and ChIP DNA microarray (ChIP-chip) analyses to define several regulatory regions within the Vdr gene that mediate the actions of 1,25-(OH)₂D₃ (9). These regions are located in two large introns substantially downstream of the TSS (5) and are direct targets of VDR and RXR binding after treatment of cells with 1,25-(OH)₂D₃. Subsequent mapping studies revealed the presence of several highly conserved VDREs within these regions, at least one of which was functional and responsible for enhancer activity. These and additional studies suggest that autoregulation of the Vdr gene by 1,25-(OH)₂D₃ is mediated directly by several enhancers located within the gene itself. These studies provide an entrée into further investigation of Vdr gene regulation.

In the present studies, we explored further the regulation of the Vdr gene by 1,25-(OH)₂D₃. Using improved ChIP-chip assays, we confirmed the existence of previously identified enhancers, characterized several additional regions, one of which was located upstream of the TSS, and evaluated the properties of each using both isolated enhancer analysis and recombineered Vdr gene bacterial artificial chromosome (BAC) clone reporters. We also explored the role of these enhancers in mediating the up-regulatory activity of RA, GCs (DEX), and forskolin (Fsk), a surrogate for PTH, through the RA receptor (RAR), the GC receptor (GR), and the cAMP response element-binding protein (CREB). These studies provide important new insight into the mechanisms whereby the Vdr gene is regulated in cells and tissues.

Results

1,25-(OH)₂D₃ induces binding of the VDR and RXR to both intronic and upstream regions in the mouse Vdr gene

Based upon completion of the mouse genome together with multiple improvements in oligonucleotide design, we conducted a new round of ChIP-chip analyses in mouse osteoblastic MC3T3-E1 cells at a 3-h time point (8) and evaluated sites of localization for the VDR and its RXR partner at the Vdr gene locus both in the absence as well as presence of 1,25-(OH)₂D₃. We also conducted these studies using antibodies to RNA polymerase II (RNA pol II). Accordingly, sonicated chromatin samples were selectively immunoprecipitated with the appropriate antibodies, amplified using ligation-mediated PCR, labeled with the fluorescent dyes Cy5 or Cy3 as described in Materials and Methods, and then hybridized overnight to custom microarrays that contained a series of oligonucleotides that spanned the entire Vdr gene locus from 100 kb upstream of the TSS to 100 kb downstream of the final exon. Figure 1 depicts the organization of this gene compared with that of the human, highlighting important differences between the two genes. Figure 2A represents a schematic of the mouse Vdr gene; the exonic locations and the locations of the previously identified enhancer regions S1–S3 are illustrated (9). Figure 2B provides data tracks that represent the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to either VDR or control IgG (VDR_veh vs. IgG_veh) (basal VDR) or from 1,25-(OH)₂D₃-vs. vehicle-treated samples precipitated with an antibody to the VDR (VDR_1,25-(OH)2D3 vs. VDR_veh) (net inducible VDR). A similar set of comparisons is depicted using antibodies to RXR. The data shown are restricted to discontinuous gene segments that demonstrated relevant VDR and RXR interactions, although a more complete view of these data is provided in supplemental Fig. 1 published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org. Statistically valid peaks assessed using NimbleScan 2.4 can be seen in red [false discovery rate (FDR) P < 0.05, see Materials and Methods]. Data obtained on chromosome 15 downstream of nucleotide 97,671,000 or upstream of nucleotide 97,755,000 were unremarkable. Surprisingly, both VDR and RXR were associated with a number of specific sites across the Vdr gene locus before activation by 1,25-(OH)₂D₃. Sites of basal receptor binding were limited for VDR (at S4), but rather extensive for RXR [at S1, S3, S5, TSS, and U1, and at several sites located at the 3′-end of the gene (see supplemental Fig. 1)]. Despite this, treatment with 1,25-(OH)₂D₃ led to a striking appearance of the VDR at S1 and S3, two sites we had identified previously (9), and two new sites at S5 and at U1, the latter located 6 kb upstream of the TSS. Prebound VDR appeared to be lost in the presence of 1,25-(OH)₂D₃, due primarily to the net negative hybridization signals generated at S4. Like VDR, RXR was also induced by 1,25-(OH)₂D₃ at a subset of these sites (see RXR_1,25-(OH)2D3 vs. RXR_veh activity at S1, S2, S3, and S5). It is worth noting here that this net inducible measurement of RXR binding reflects only that RXR component induced to bind after treatment with 1,25-(OH)₂D₃. Indeed, total levels of RXR binding at these inducible sites are actually comprised of the sum of the RXR binding activities observed under both basal and net inducible conditions. These overall findings both confirm our previous identification of 1,25-(OH)₂D₃-responsive enhancers in the Vdr gene locus (9) and at the same time reveal at least two additional sites (S5 and U1). Figure 2C documents the results obtained by ChIP-chip analysis after precipitation with antibody to RNA pol II under treatment conditions as described above. As can be seen, RNA pol II is present at the VDR gene TSS in the absence of ligand (RNA pol II_veh vs. IgG_veh) (basal RNA pol II). This level of RNA pol II likely represents that involved in some way in basal VDR expression, because MC3T3-E1 cells are actively engaged in VDR synthesis irrespective of the presence of 1,25-(OH)₂D₃ (8). Importantly, RNA pol II levels are significantly increased after treatment with 1,25-(OH)₂D₃ (RNA pol II_1,25-(OH)2D3 vs. RNA pol II_veh) (net inducible RNA pol II), both at the TSS and across the upstream half of the Vdr gene as well. Again, as described for RXR above, total RNA pol II levels at the TSS comprise the sum of the enzymes levels detected under both basal as well as net inducible conditions. Interestingly, the induced condition appears to favor an accumulation of RNA pol II immediately downstream of the TSS. RNA pol II accumulation after 1,25-(OH)₂D₃ treatment is also particularly prevalent at the VDR/RXR-regulated enhancer regions and especially at S1, S3, and U1. Although it is unclear why this occurs, it is possible that it is related to unique intronic enhancer function (33). Despite this, these latter observations suggest the presence of a functional consequence to VDR/RXR binding at sites in the Vdr gene locus.

Figure 1.

Organization of the mouse and human VDR genes. The mVDR and hVDR VDR genes are depicted in a standard 5′- to 3′-orientation. Numbered exons are indicated by vertical bars. Exon 10 (mouse) and 9 (human) contain the stop codon and 3′-noncoding sequence. All exons upstream of the translational start site (ATG) represent 5′-noncoding exons. Exons 1 (mouse) and 1a and 1f (human) contain independent promoter activity.

Figure 2.

ChIP-chip analysis of VDR, RXR, and RNA pol II at the mouse Vdr gene. MC3T3-E1 cells were treated with either vehicle (Veh) or 1,25-(OH)₂D₃ for 3 h and then subjected to ChIP-chip analysis using antibodies to VDR, RXR, RNA pol II, or IgG. A, Schematic diagram of the mouse Vdr locus. The nucleotide base pairs indicate location on chromosome 15 (mm8) in megabases (Mb). The arrow at the TSS indicates the direction of transcription on the reverse strand. Exons are indicated by black boxes, and key enhancer regions are defined below the gene. B, Interaction of VDR and RXR with the Vdr gene locus. Data track segments represent the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to either VDR vs. control IgG (VDR_veh vs. IgG_veh) (basal VDR) or from 1,25-(OH)₂D₃- vs. vehicle-treated samples precipitated with an antibody to the VDR (VDR_1,25-(OH)2D3 vs. VDR_veh) (net inducible VDR) (top tracks) or a similar analysis using antibody to RXR (bottom tracks). C, Residual and inducible RNA pol II at the Vdr gene locus. Data tracks represent the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to either RNA pol II vs. control IgG (Pol II_veh vs. IgG_veh) (basal RNA pol II) or from 1,25-(OH)₂D₃-vs. vehicle-treated samples precipitated with an antibody to RNA pol II (RNA Pol II_1,25-(OH)2D3 vs. RNA Pol II_veh). All peaks highlighted in red represent statistically significant peaks [false discovery rate (FDR), P < 0.05]. Vertical blue bands track the locations of enhancers that are depicted below the schematic of the VDR gene. Chr, Chromosome; PP, proximal promoter.

The isolated S1 enhancer is unique in its ability to mediate independent 1,25-(OH)₂D₃ response in MC3T3-E1 cells

We examined the transcriptional capabilities of the enhancer regions identified in Fig. 2 by cloning their corresponding DNA segments into a thymidine kinase (TK)-luciferase reporter vector as documented in supplemental Fig. 2 and assessing their activity in response to 1,25-(OH)₂D₃ treatment after transient transfection into MC3T3-E1 cells. As can be seen in supplemental Fig. 2B, only the S1 enhancer mediated an inducible response to 1,25-(OH)₂D₃. As previously discovered with S2 and S3 (9), none of the newly identified segments were responsive. These results are not particularly surprising, given the context in which they are examined. Thus, despite the presence of VDR binding, these fragments may contain single regulatory elements that are only weakly active, operate exclusively through synergistic interaction with other enhancers, or contain elements that exert a repressive influence on the isolated fragment. Although the presence of a measurable activity such as that observed at S1 is reassuring, the absence of an independent activity does not infer the absence of a potential regulatory element. Indeed, the ability of both VDR and RXR to bind to these regions and to induce focal RNA pol II recruitment (and H4ac, see below) provides compelling evidence that these regions comprise functional enhancers.

Enhancers S1, S3, S4, S5, and U1 contain potential VDREs

Previous analyses in silico have demonstrated potential VDREs in S1, S2, and S3, although only the sequence of the VDRE in S1 appeared to be highly conserved (9). To further explore the presence of potential VDREs, we used the MatInspector program available via Genomatix (34). As can be seen in supplemental Table 1 (Table S1), this analysis revealed the presence of several VDREs in each of the enhancer regions which could represent binding sites for the observed localization of both VDR and RXR. Logos corresponding to the putative VDREs identified in the table are documented in Table S2. Importantly, one of the VDREs found in the S1 region in this in silico analysis (and highlighted by shading in the table) was determined to mediate the activity of 1,25-(OH)₂D₃ at the S1 fragment in our earlier study (9).

A BAC clone construct containing the entire mouse Vdr gene locus produces functional VDR protein and is inducible by 1,25-(OH)₂D₃

The modest results obtained with isolated and cloned DNA fragments of the Vdr gene identified in supplemental Fig. 2 prompted us to construct a full-length Vdr gene BAC clone vector with which to investigate the regulation of VDR by 1,25-(OH)₂D₃ and other hormones. Using standard recombineering techniques, we inserted an internal ribosome entry site (IRES)-luciferase reporter/neomycin-selection cassette into the 3′-noncoding portion of the final Vdr gene exon as documented in Fig. 3A (35,36). As observed in this figure, a hemagglutinin (HA) tag was also introduced at the TSS in exon 3 using the galactokinase (galK) selection system (37). Isolated BAC DNA was linearized and used to prepare collections of G418-resistant MC3T3-E1 cell clones. Figure 3B documents the ability of the stable cell collections (MC-VDR BAC1) to synthesize recombinant VDR protein and to up-regulate this production in response to 1,25-(OH)₂D₃. As can be seen, basal and 1,25-(OH)₂D₃-inducible levels of VDR protein were detected in both wild-type MC3T3-E1 and stably transfected MC-VDR BAC1 cells using an antibody to the VDR. A portion of this receptor protein was derived from the stable BAC clone as supported by the ability of the HA antibody to identify the VDR in the stable MC-VDR BAC1 cells (Fig. 3B) but not in the wild-type MC3T3-E1 cell line (data not shown). To examine whether the recombinant VDR was functionally active, we conducted a ChIP analysis using both the antibody directed toward the VDR and one directed toward the HA tag. As can be seen in Fig. 3C, 1,25-(OH)₂D₃ induced binding of both wild-type and BAC-derived VDR to the VDREs of the endogenous, highly regulated Cyp24a1 gene (8). In a final examination of BAC clone activity, we assessed the ability of 1,25-(OH)₂D₃ to induce expression of the recombineered luciferase reporter located within the final 3′-noncoding exon of the Vdr gene. Stably transformed MC-VDR BAC1 cells were treated with increasing concentrations of 1,25-(OH)₂D₃ as indicated in Fig. 3D, and luciferase activity was assessed 24 h later. As can be seen, luciferase activity was strongly induced by the vitamin D hormone. Because 1,25-(OH)₂D₃ (unlike VDR protein levels) does not affect luciferase protein levels or activity, these studies suggest that the mouse (m)VDR BAC1 clone is capable of recapitulating 1,25-(OH)₂D₃-inducible transcriptional activity at the Vdr gene locus.

Figure 3.

Organization and analysis of the activity of mVDR BAC1. A, Schematic organization of mVDR BAC1. The upper panel depicts the location and organization of the mouse Vdr gene. The Vdr gene is located on chromosome 15 between the Hdac7a and the Col2a1 genes. Arrows indicate the locations of the TSSs and the directions of transcription. Black boxes indicate the locations of exons. The Vdr gene is 23 kb upstream of Hdac7a and 67 kb downstream of Col2a1. The lower panel indicates the boundaries of VDR BAC1 and sites of reporter/selection cassette- and HA tag-insertion into the locus. B, Induction of VDR protein expression by 1,25-(OH)₂D₃ in parental MC3T3-E1 and VDR BAC1-stable MC3T3-E1 (MC-VDR BAC1) cell lines. Cells were treated with either vehicle or 1,25-(OH)₂D₃ (10⁻⁷ m) for 24 h, and the lysates were subjected to Western blot analysis using antibodies to VDR (αVDR) or to HA (αHA) as indicated. C, ChIP analysis of endogenous and VDR BAC1-derived VDR at the Cyp24a1 gene locus. MC-VDR BAC1 cells were treated with vehicle or 1,25-(OH)₂D₃ (10⁻⁷ m) for 3 h and then subjected to ChIP analysis using antibodies to VDR (αVDR), the HA tag (αHA), or to IgG. The DNA precipitates were recovered and subjected to qPCR analysis using primers specific to the VDRE-containing region of the mouse Cyp24a1 gene as documented in Materials and Methods. D, Induction of VDR BAC1-derived luciferase (Luc) activity in the MC-VDR BAC1 cell line. Cells were treated with either vehicle or increasing concentrations of 1,25-(OH)₂D₃ as indicated and then evaluated for luciferase activity 24 h later. Each point represents the RLU (relative light units) average normalized to total protein ± sem for a triplicate set of assays. V, vehicle; Neo, neomycin.

Deletion of the VDRE in the S1 region of the VDR BAC reduces, but does not fully abrogate, gene response to 1,25-(OH)₂D₃

Our earlier results suggest activity involving 1,25-(OH)₂D₃ and the VDR at the Vdr gene localizes predominantly upstream of the final Vdr exon and that the S1 region plays a significant role (9). To further explore these activities, we constructed several additional BAC constructs, each of which contained the luciferase reporter cassette at the same site as in mVDR BAC1. As seen in Fig. 4A, mVDR BAC2 is similar to mVDR BAC1 but does not contain the HA tag. mVDR BAC3 is derived from mVDR BAC2 with the exception that sequence downstream of the reporter cassette was deleted. mVDR BAC4 is identical to mVDR BAC3 but contains a 15-bp deletion in S1 that specifically removed the VDRE. Each of these BACs was then introduced into MC3T3-E1 cells, and a series of stable clones was selected using G418 as described earlier. Cloned cell populations were evaluated for luciferase activity in the absence and presence of 1,25-(OH)₂D₃. As can be seen in Fig. 4B, although fold induction levels were slightly reduced relative to MC-VDR BAC1 cell clones with stably integrated mVDR BAC1, those containing parent mVDR BAC2 as well as mVDR BAC3 were both significantly induced by 1,25-(OH)₂D₃. Importantly, however, deletion of the S1 VDRE in mVDR BAC3 strongly reduced, although did not fully abrogate, response to 1,25-(OH)₂D₃. These results suggest that the S1 VDRE contributes significantly to the ability of the vitamin D hormone to induce the Vdr gene and that the additional enhancers may play a contributory role in the induction process as well.

Figure 4.

Deletion of the VDRE sequence in the S1 enhancer of the VDR gene reduces VDR BAC clone response to 1,25-(OH)₂D₃. A, Schematic of mouse Vdr-luciferase constructs with various 3′-deletions. The upper panel depicts the mouse Vdr gene locus as in Fig. 3A. The boundaries and locations of VDR BAC2, VDR BAC3, and VDR BAC4 are illustrated. The location of the luciferase/selection cassette and the site of S1 VDRE deletion are indicated. B, Transcriptional activity of MC3T3-E1 cell lines stably transfected with VDR BAC2, BAC3, and BAC4. Cells were treated with either vehicle or 1,25-(OH)₂D₃ (10⁻⁷ m) and then evaluated for luciferase activity 24 h later. Fold induction was obtained through a comparison of 1,25-(OH)₂D₃-treated activity (relative light units average normalized to total protein ± sem for a triplicate set of assays) to vehicle (V)-treated activity. Luc, Luciferase; Neo, neomycin.

C/EBPβ and runt-related transcription factor 2 (RUNX2) are present at VDR gene enhancers under basal conditions and are induced to bind to a subset of these enhancers in the presence of 1,25-(OH)₂D₃

RUNX2 and C/EBPβ play key roles in the development and function of the bone cell phenotype (38,39). Both are regulated at the transcriptional level by 1,25-(OH)₂D₃ (40,41), interact directly or indirectly with the VDR protein at certain genes (41,42,43), and in the case of C/EBPβ also modulates Vdr gene expression (41). Because basal levels of Vdr expression are significant in MC3T3-E1 cells relative to other cell lines (data not shown), we used ChIP-chip analysis to examine whether C/EBPβ or RUNX2 was present at the Vdr gene locus in the absence of 1,25-(OH)₂D₃ and whether the hormone could up-regulate the level of these factors at these sites as well. MC3T3-E1 cells were treated with either vehicle or 1,25-(OH)₂D₃ and then subjected to ChIP analysis using antibodies to RUNX2, C/EBPβ, or IgG. Figure 5B depicts data tracks that represent the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to either C/EBPβ vs. control IgG (C/EBPβ_veh vs. IgG_veh) (basal C/EBPβ) or from 1,25-(OH)₂D₃- vs. vehicle-treated samples precipitated with an antibody to C/EBPβ (C/EBPβ_1,25-(OH)2D3 vs. C/EBPβ_veh) (net inducible C/EBPβ). A similar set of comparisons using antibodies to RUNX2 is depicted in Fig. 5C. The data shown are again restricted to discontinuous gene segments demonstrating relevant C/EBPβ or RUNX2 interactions; a more complete view of these data is provided in supplemental Fig. 3. Surprisingly, abundant levels of C/EBPβ and RUNX2 were present at the Vdr gene locus in the absence of 1,25-(OH)₂D₃, localized primarily to enhancers previously identified for the VDR/RXR heterodimer complex, but also at a site near the Vdr promoter as well. As can be seen in Fig. 5, B and C, C/EBPβ and RUNX2 binding at the Vdr gene was also up-regulated by 1,25-(OH)₂D₃. Note again that these tracks measure only binding activities that occur above basal levels; total C/EBPβ or RUNX2 activities comprise the sum of either C/EBPβ or RUNX2 activities under both basal and net inducible conditions. Interestingly, the up-regulation of both C/EBPβ and RUNX2 by 1,25-(OH)₂D₃ was largely restricted to the S1 enhancer. These results indicate that both C/EBPβ and RUNX2 localize to the proximal promoter as well as to many additional sites across the Vdr gene locus, suggesting that this binding may play a fundamental role in both basal and inducible expression of the Vdr gene. They also indicate that transcriptional activation by 1,25-(OH)₂D₃ is not mediated exclusively by the VDR but is likely facilitated by least two additional regulatory factors as well.

Figure 5.

ChIP-chip analyses of C/EBPβ and Runx2 at the mouse Vdr gene. MC3T3-E1 cells treated with either vehicle (Veh) or 1,25-(OH)₂D₃ for 3 h and then subjected to ChIP-chip analysis using antibodies to C/EBPβ, Runx2, or IgG. A, Schematic diagram of the mouse Vdr locus, as described in Fig. 2A. B, Interaction of C/EBPβ with the Vdr gene locus. Data track segments represent the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to either C/EBPβ or control IgG (C/EBPβ_veh vs. IgG_veh) (basal C/EBPβ) or from 1,25-(OH)₂D₃-vs. vehicle-treated samples precipitated with an antibody to C/EBPβ (C/EBPβ_1,25-(OH)2D3 vs. C/EBPβ_veh) (net inducible C/EBPβ). C, Interaction of Runx2 with the Vdr gene locus. Data track segments represent the log₂ ratios of fluorescence obtained from cohybridized samples similar to those in panel B but precipitated with antibody to Runx2. All peaks highlighted in red represent statistically significant peaks (FDR, P < 0.05). Vertical blue bands track the locations of enhancers, which are depicted below the Vdr gene schematic. D, siRNA depletion of C/EBPβ mRNA suppresses basal expression of both VDR mRNA and mVDR-BAC1 luciferase activity. Left and center panels, MC3T3-E1 cells, either mock transfected or transfected with nontargeted or C/EBPβ siRNA pools (40 nm), were harvested after 48 h, and the abundance of C/EBPβ (left) or VDR (center) mRNA levels was measured using RT-PCR analysis and normalized to β-actin control levels. The data represent the mean ± sem of three independent experiments. Right panel, MC-VDR BAC1 cells were mock transfected or treated with siRNAs as above, harvested at 48 h, and evaluated in triplicate for mVDR BAC1 luciferase activity. The data are representative of three separate assays each normalized to total protein and depicted as the relative light units average ± sem. *, Significantly different from both mock and nontargeted siRNA controls at P < 0.05. Chr, Chromosome; PP, proximal promoter; Mb, megabases.

C/EBPβ influences basal levels of Vdr gene expression

Previous studies have suggested that C/EBPβ plays a role in 1,25-(OH)₂D₃-mediated induction of the human Vdr gene (41). To explore the hypothesis that C/EBPβ might also influence basal levels of VDR expression, we suppressed the level of C/EBPβ mRNA using selective small interfering (siRNA) treatment and assessed the consequence of this action on basal VDR mRNA levels. We also examined whether this treatment could suppress basal luciferase activity expressed from mVDR BAC1 in the MC-VDR BAC1 stable cell clones. As seen in Fig. 5D (left and center panels), a 48-h treatment of MC3T3-E1 cells with a C/EBPβ siRNA pool suppressed C/EBPβ mRNA levels by more than 50% as compared with a nontarget siRNA control or a mock transfection. Importantly, this reduction in C/EBPβ levels also reduced basal VDR mRNA expression by approximately 40%. As also seen in Fig. 5D (right panel), a similar effect of C/EBPβ down-regulation (data not shown) was also observed on basal luciferase activity produced from the mVDR BAC1 stable cell clones. These observations support our suggestion that C/EBPβ may contribute to the basal level of expression of the Vdr gene in these bone cell lines.

Both basal and inducible H4ac are observed at the Vdr gene locus

Histone acetylation represents a genome-wide epigenetic mark at gene promoters that can also be found highlighting functional enhancers for specific genes (44,45). Moreover, the induction of transcription from many promoters is often associated with an increase in either H3 or H4 acetylation or the combination (46). In view of the high levels of basal expression of the Vdr gene in MC3T3-E1 cells and the presence of C/EBPβ and RUNX2 at the gene in both the absence and presence of 1,25-(OH)₂D₃, we next explored the status of H4ac at this locus after treatment with either vehicle or 1,25-(OH)₂D₃. ChIP-chip analysis was therefore used to contrast levels of tetra-acetylated H4 across the VDR locus in the absence (H4ac_veh vs. IgG_veh) (basal H4ac) and presence of 1,25-(OH)₂D₃ (H4ac_1,25-(OH)2D3 vs. H4ac_veh) (net inducible H4ac). The results in Fig. 6B reveal that under basal conditions histone H4 was strongly acetylated at several of the enhancer regions including S3, S4, the TSS, and U1. These regions were those occupied by C/EBPβ and RUNX2, providing further support for the idea that these factors might indeed be involved in basal Vdr gene expression. 1,25-(OH)₂D₃ treatment prompted a striking increase in acetylation at both these and additional sites, including those at S1/S2, S3, S5, the TSS, and U1. Thus, the overall levels of H4ac at these sites and across the Vdr gene locus, as measured by the sum of both basal and net inducible activities, were substantially increased after treatment with 1,25-(OH)₂D₃. Interestingly, as seen through ChIP-chip analysis in Fig. 6C, the boundaries of this hormone-stimulated increase in the general level of H4ac appear to be defined by the presence of strong CCCTC-binding factor (CTCF) binding sites (47,48) located considerably upstream and downstream of the Vdr gene locus. Whereas some H4ac activity does appear to be induced by 1,25-(OH)₂D₃ immediately outside these boundaries, virtually all inducible H4ac activity was rapidly lost as a function of distance both upstream and downstream of the schematic boundaries shown in the figure. These data suggest that strong basal and inducible H4ac defines the Vdr locus and may be largely limited to this region through constitutive CTCF binding activity. These studies, therefore, add support to the existence of multiple enhancers within a defined Vdr gene locus that may contribute to both basal and 1,25-(OH)₂D₃-induced gene expression.

Figure 6.

CTCF sites define 1,25-(OH)₂D₃-induced tetraacetylation at H4. MC3T3-E1 cells were treated with either vehicle (Veh) or 1,25-(OH)₂D₃ for 3 h and then subjected to ChIP-chip analysis using antibodies to tetraacetylated H4, CTCF, or IgG. A, Schematic diagram of the mouse Vdr locus extending from 97.664 Mb to 97.800 Mb as described in Fig. 2A. Arrows indicate TSSs for the VDR gene and a putative upstream neighbor. B, H4ac across the Vdr gene locus. Data tracks represent the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to either H4ac vs. control IgG (H4a _veh vs. IgG_veh) (basal H4ac) or from 1,25-(OH)₂D₃-vs. vehicle-treated samples precipitated with an antibody to the H4ac (H4ac_1,25-(OH)2D3 vs. H4ac_veh) (net inducible H4ac). C, Detection of CTCF binding within the Vdr gene locus. The data track represents the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to CTCF vs. control IgG (CTCF_veh vs. IgG_veh) (basal CTCF). All peaks highlighted in red are statistically significant (FDR, P < 0.05) above baseline activity. Vertical blue bands track the locations of enhancers, the locations of which are depicted below the Vdr gene schematic. Chr, Chromosome; PP, proximal promoter; Mb, megabases.

The PTH surrogate Fsk, RA, and DEX also induce expression of the Vdr gene

PTH, RA, and DEX induce expression of the Vdr gene in bone cells, although the actions of the latter two regulators can also be inhibitory as well, depending upon the nature and maturity of the target cell (26,28,29). As can be seen in Fig. 7A, the up-regulation of VDR protein can be observed in MC3T3-E1 cells with the protein kinase A (PKA) activator Fsk and more modestly with DEX, observations that have been made previously in this cell line. Surprisingly, RA was not effective in inducing VDR mRNA in this experiment. All three inducers were active, however, in up-regulating VDR in the MC-VDR BAC1 stable cell clones containing mVDR BAC1, as seen in Fig. 7A. The results using the HA antibody indicate that at least a portion of the newly expressed and up-regulated protein is derived directly from the integrated BAC clone(s). To examine this more fully, we treated the stable MC-VDR BAC1 cell clones with increasing concentrations of Fsk, RA, or DEX and assessed luciferase activity 24 h later. As can be seen in Fig. 7B, each was effective at inducing, in a dose-dependent fashion, an up-regulation of the reporter. mVDR BAC2 and mVDR BAC3 as well as mVDR BAC4, which contained mutation of the S1 VDRE, all retained inducibility with RA and Fsk (data not shown). These results suggest that all of the VDR BAC clones contain sufficient regulatory information to mediate up-regulation of Vdr gene expression by not only 1,25-(OH)₂D₃, but the above inducers as well. The lack of RA activity at the endogenous VDR level is unclear.

Figure 7.

Transcriptional activity of mVDR BAC1. A, Induction of the VDR protein expression by Fsk, RA, and DEX in parental MC3T3-E1 and mVDR BAC1-stable MC3T3-E1 (MC-VDR BAC1) cell lines. Cells were treated with either vehicle, Fsk, RA (AtRA), or DEX for 24 h, and the lysates were subjected to Western blot analysis using antibodies to VDR (αVDR) or to HA (αHA) as indicated. B, Induction of VDR BAC1-derived luciferase activity in the MC-VDR BAC1 cell line. Cells were treated with either vehicle or increasing concentrations of Fsk, RA (AtRA), or DEX and then evaluated for luciferase activity 24 h later. Each point represents the relative light units average normalized to total protein ± sem for a triplicate set of assays. Veh, Vehicle.

Fsk, RA, and DEX induce localization of target transcription factors CREB, RAR, and glucocorticoid receptor (GR) at subsets of VDR gene enhancers

PKA activators such as Fsk are strong inducers of CREB (49,50). This protein is believed to mediate the up-regulation of the Vdr gene in bone cells via a CRE located near the TSS (51). To explore this mechanism further, we treated MC3T3-E1 cells with either vehicle or Fsk and conducted a ChIP-chip analysis using antibodies to phospho-CREB (pCREB). We examined both basal levels of pCREB across the Vdr gene locus in the absence of inducer (pCREB_veh vs. IgG_veh) (basal pCREB) and levels of pCREB after induction by Fsk vs. vehicle (pCREB_fsk vs. pCREB_veh) (net inducible pCREB). Because C/EBPβ can also be induced through PKA activation (52), we examined the effect of Fsk on C/EBPβ binding at the Vdr gene as well. Surprisingly, the results in Fig. 8B (and supplemental Fig. 4B) reveal that like C/EBPβ, pCREB also occupies enhancer sites in the absence of the inducing agent. For pCREB, these sites included the previously identified region U1 and a significant site immediately downstream of the TSS. Additional novel sites for prebound CREB and C/EBPβ were present as well, including several located at the 3′-end of the gene (supplemental Fig. 4B). The presence of pCREB at these sites in the absence of Fsk and its phosphorylation state (53) suggest that this transcription factor could be activated and thus involved in the basal expression of the Vdr in MC3T3-E1 cells. Thus, like C/EBPβ and perhaps RUNX2 (Fig. 5), CREB could contribute to the residual levels of H4ac found at the Vdr locus. Interestingly, pCREB levels were induced by Fsk at S3 and at U1. Thus, substantial levels of pCREB could be found at S3 (induced), the TSS (basal), and particularly at U1 (basal plus induced) after treatment with Fsk. C/EBPβ binding, on the other hand, was only modestly increased at all sites by this inducer. These findings suggest that both pCREB and C/EBPβ occupy regulatory sites at the Vdr gene in the absence of Fsk and, at least in the case of pCREB, are also induced to bind to the gene in a site-selective manner in its presence. It is therefore possible that in addition to C/EBPβ and perhaps RUNX2, CREB and its upstream signaling pathways might also be involved in the basal expression of the Vdr gene. Figure 8C (and supplemental Fig. 4C) documents a similar ChIP-chip analysis wherein we assessed the distribution of RAR across the Vdr gene in the absence (RAR_veh vs. IgG_veh) (basal RAR) and presence (RAR_RA vs. RAR_veh) (net inducible RAR) of RA. As can be seen, RAR also occupied several sites at the VDR locus in the absence of RA, although it was up-regulated at S5 and U1 upon treatment with RA. Finally, Fig. 8D (and supplemental Fig. 4D) documents a similar ChIP-chip analysis wherein we assessed the distribution of GR at the Vdr gene in the absence (GR_veh vs. IgG_veh) (basal GR) and then in the presence (GR_Dex vs. GR_veh) (net inducible GR) of DEX. As can be see, GR binding was largely inducible and localized to S1/S2, S3, U1, and to a novel and more distal region termed “U2.” The ability of each of these transcription factors to target not only subsets of enhancers but novel sites such as the TSS and U2 as well suggests that, although many of the Vdr gene enhancers we have identified may be modular, some display unique properties of activation.

Figure 8.

ChIP-chip analysis of pCREB, C/EBPβ, RAR, and GR at the mouse Vdr gene. MC3T3-E1 cells were treated with vehicle, Fsk, RA, or DEX as indicated for 3 h and then subjected to ChIP-chip analysis. A, Schematic diagram of the mouse Vdr locus, as described in Fig. 2A. B (top), Interaction of pCREB at the Vdr gene locus. Data track segments represent the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to either pCREB vs. control IgG (pCREB_veh vs. IgG_veh) (basal pCREB) or from Fsk-vs. vehicle-treated samples precipitated with an antibody to pCREB (pCREB_fsk vs. pCREB_veh) (net inducible pCREB). B (bottom), Interaction of C/EBPβ at the Vdr gene locus. Data tracks represent the log₂ ratios of fluorescence obtained from cohybridized samples similar to those in panel B but precipitated with antibody to C/EBPβ. C, Interaction of RAR at the Vdr gene locus. Data track segments represent the log₂ ratios of fluorescence obtained from cohybridized samples similar to those in panel B but precipitated with antibody to RAR. D, Interaction of GR at the Vdr gene locus. Data track segments represent the log₂ ratios of fluorescence obtained from cohybridized samples similar to those in panel B but precipitated with antibody to GR. All peaks highlighted in red represent statistically significant peak (FDR, P < 0.05). Vertical blue bands track the locations of enhancers, which are depicted below the Vdr gene schematic. Chr, Chromosome.

The regulatory regions S3 and U1 mediate the activities of RA and DEX in an independent context

To explore further the activities of Fsk, RA, and DEX, we examined their ability to induce transcription from each of the regulatory regions identified above. The individual DNA fragments previously cloned into the TK-luciferase vector and listed in supplemental Fig. 2A and the additional U2 region were transfected into MC3T3-E1 cells, the cells were treated with Fsk, RA, or DEX, and their activities were measured. As can be seen in supplemental Fig. 5A, Fsk was unable to induce any of the enhancer regions independent of the Vdr gene locus. The proximal promoter for the mouse Vdr gene was not examined. RA, on the other hand, actively induced transcriptional response at S3, a general region we had identified previously in the human VDR gene (54). DEX also actively induced transcription, as observed in supplemental Fig. 5B. Again, however, this activity was restricted to the U1 enhancer and was not detected at any of the other sites to which GR bound. We conclude that although independent activity is manifested at S1, S3, and U1, the bulk of these enhancer regions may contribute to the regulation of Vdr gene expression only in the context of the natural gene locus and/or in an additive or perhaps synergistic fashion with each other.

Enhancer regions contain potential regulatory elements for CREB, C/EBPβ, GR, RAR, and RUNX2

The above studies revealed multiple enhancers at the Vdr gene locus that are responsible for its up-regulation. We therefore used the MatInspector program (Genomatix) (34) to explore these regulatory regions further for potential sites of binding for CREB, C/EBPβ, RAR, RUNX2, and GR. Supplemental Table 3 documents putative C/EBPβ binding sites and supplemental Table 4 documents putative CREB, RAR, RUNX2, and GR binding sites that were contained in each of these regions. The position weight matrix logos for each of these response elements are documented in supplemental Table 5. Importantly, several of the CREB and C/EBPβ sites near the proximal promoter may be analogous to those identified in the human Vdr gene promoter by Christakos and colleagues (41,51) through classic transfection analyses.

Enhancer regions in the human VDR gene are conserved

The human VDR gene is highly conserved relative to the mouse gene (supplemental Fig. 6) with the exception that its expression is believed to be controlled by not only the primary promoter at exon 1a, which corresponds to exon 1 of the mouse gene, but several additional promoters as well. One, termed “exon 1f,” is located more than 30 kb upstream of exon 1a (55). We therefore contrasted the ability of 1,25-(OH)₂D₃ to induce VDR binding to the VDR gene in human MG63 cells and to stimulate H4ac across the VDR gene locus using ChIP-chip analysis. Accordingly, MG63 cells were treated with either vehicle or 1,25-(OH)₂D₃ and then subjected to Chip-chip analysis using antibodies to either the VDR (VDR_1,25-(OH)2D3 vs. VDR_veh) (net inducible VDR), H4ac and IgG (Ac-H4_veh vs. IgG_veh) (basal H4ac), or (H4ac_1,25-(OH)2D3 vs. IgG_veh) (induced H4ac) or H4ac (H4ac_1,25-(OH)2D3 vs. H4ac_veh) (net inducible H4ac). We also searched for CTCF sites in a similar fashion as well. The results documented in Fig. 9 reveal an overall enhancer organization that is similar, although not identical, to that of the mouse gene. Thus, the VDR, as seen in Fig. 9B, was induced to bind to the human VDR gene at S1 and S3 as well as at a newly identified region immediately adjacent to S1. The VDR also bound to the primary TSS (exon 1a) and to a site(s) (U3) immediately upstream of the more distal promoter located at exon 1f. Although basal H4ac was prevalent at the region surrounding the primary promoter at exon 1a and immediately upstream of the promoter 1f, it was substantially increased upon cellular treatment with 1,25-(OH)₂D₃ in both of these regions and at S3, as observed in Fig. 9C. Although H4ac was not up-regulated at all of the VDR-responsive enhancers, these data suggest a functional consequence to 1,25-(OH)₂D₃-induced VDR binding at the VDR gene. As with the mouse gene, CTCF-binding sites (47,48) were identified at conserved locations upstream and downstream of the VDR gene locus as well as at exon 1a, as documented in Fig. 9D. None of these residual CTCF activities were influenced by 1,25-(OH)₂D₃. Additional sites were also evident within the locus itself. These differences suggest that although the locations of the enhancers and their overall organization relative to the mouse gene appear to be generally conserved, an additional level of complexity is apparent within the human homolog.

Figure 9.

ChIP-chip analysis of VDR, H4ac, and CTCF localization at the human VDR gene. Human MG63 osteosarcoma cells were treated with either vehicle or 1,25-(OH)₂D₃ for 3 h and then subjected to ChIP-chip analysis. A, Schematic diagram of the human VDR gene locus. The nucleotide base pairs indicate the location in megabases (Mb) on chromosome 12. Exons are numerically indicated. The arrows indicate the VDR TSSs at exons 1d and 1f and the direction of transcription (on the reverse strand). The arrow at 46.644 Mb represents the TSS of the putative neighboring gene on the forward strand. Key enhancer regions are defined below the gene. B, Interaction of VDR with the human VDR gene locus. The data track represents the log₂ ratios of fluorescence obtained from 1,25-(OH)₂D₃-vs. vehicle-treated samples precipitated with an antibody to the VDR (VDR_1,25-(OH)2D3 vs. VDR_veh) (net inducible VDR). C, H4ac across the human VDR gene locus. Data tracks represent the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibodies to either H4ac vs. control IgG (H4as_veh vs. IgG_veh) (basal H4ac), 1,25-(OH)₂D₃-treated samples precipitated with antibodies to either H4ac vs. control IgG (H4ac_1,25-(OH)2D3 vs. IgG_veh) (inducible H4ac), or from 1,25-(OH)₂D₃- vs. vehicle-treated samples precipitated with an antibody to the H4ac (H4ac_1,25-(OH)2D3 vs. H4ac_veh) (net inducible H4ac). D, Detection of CTCF binding within the human VDR gene locus. The data track represents the log₂ ratios of fluorescence obtained from vehicle-treated samples precipitated with antibody to CTCF vs. input control (CTCF_veh vs. Input_veh) (basal CTCF) or 1,25-(OH)₂D₃-treated samples precipitated with antibody to CTCF vs. input control (CTCF_1,25-(OH)2D3 vs. Input_1,25-(OH)2D3) (induced CTCF). All peaks highlighted in red represent statistically significant peak (FDR, P < 0.05). Vertical red bands track the locations of enhancers, which are depicted below the human VDR gene schematic. U3, Novel upstream enhancer; I1–I3, CTCF-binding sites. Chr, Chromosome; PP, proximal promoter; Mb, megabases.

Discussion

The current studies employed DNA microarrays containing synthetic oligomers that were derived from the completed mouse genome, manifested higher resolution, and exhibited better temperature balance than previous studies to identify not only the 1,25-(OH)₂D₃-inducible Vdr gene enhancers discovered earlier, but several additional peaks of VDR-binding activity as well. One was intronic and the other was uniquely located −6 kb upstream of the gene’s TSS. These studies also identified novel enhancers that mediated the up-regulating actions of RA, DEX, and Fsk. In several cases, these regulatory regions overlapped those that mediated 1,25-(OH)₂D₃ action, suggesting true modularity; in others, the enhancers were novel. Our results also document the consequences of transcription factor binding on the recruitment of RNA pol II and the up-regulation of H4ac as well. Importantly, although many of these regions did not mediate transcriptional up-regulation independently when cloned upstream of a heterologous reporter, all contained potential binding sites as assessed through in silico examination. Indeed, these analyses identified not only novel elements, but also the VDREs we had previously described in the S1, S2, and S3 enhancers (9) and regulatory sites for CREB and C/EBPβ perhaps analogous to those identified by Christakos and colleagues (41,51) at the proximal human VDR promoter. The capacity of isolated S3- and U1-containing DNA fragments to mediate RA and DEX response after transfection will enable functional identification of both the retinoic acid response elements (RAREs) and the glucocorticoid response elements that are likely contained within. The modularity of these enhancers together with their distance from the Vdr TSS highlight the utility of the ChIP-chip assay for exploring large, highly complex genes.

By conducting ChIP-chip analyses under both uninduced and net 1,25-(OH)₂D₃-induced conditions, we were able to assess both the individual characteristics of VDR and RXR binding in the absence of ligand and the net consequences of ligand on the subsequent binding of both receptors. In the absence of 1,25-(OH)₂D₃, VDR was found localized to at least one region (S4). The addition of 1,25-(OH)₂D₃, however, resulted in apparent loss of prebound VDR at S4 and the unique localization of this receptor to S1/S2, S3, S5, and U1. RXR, made up predominantly of RXRα and to a lesser extent RXRβ in MC3T3-E1 cells (based upon mRNA expression analysis, data not shown), was found associated with multiple regions of the Vdr gene in the absence of 1,25-(OH)₂D₃. All of these regions, with the exception of the TSS, comprise those to which the VDR bound after treatment with 1,25-(OH)₂D₃. RXR binding at several of these sites also increased in response to the hormone as well. Surprisingly, however, the VDR did not localize to the TSS despite the presence of basal levels of RXR at that site. Collectively, we suggest the possibility that prebound RXR located at some sites within the Vdr gene locus may serve as markers for potential VDR binding. The molecular status of prebound RXR is unclear, however, because this receptor can bind to DNA as a homodimer and as a heterodimer with RAR and peroxisome proliferator-activator receptor isoforms as well as with other receptors in this class (2,56,57,58). The activation state(s) of these potential homo- or heterodimers is equally unclear, because many members of this class of nuclear receptor can be activated by metabolically derived intracellular ligands (58). If activated, RXR heterodimers could contribute, like other factors, to the high residual levels of H4ac observed at S3 and the TSS.

ChIP-chip scans revealed that although RNA pol II was present at the Vdr gene TSS in the absence of 1,25-(OH)₂D₃, the ligand strongly induced the recruitment of additional RNA pol II at the TSS and at enhancers such as S1/S2, S3, and U1 as well. The accumulation of basal RNA pol II at the TSS suggests the enzyme complex could be poised for activation (59). The Vdr gene is highly expressed in MC3T3-E1 cells, however, suggesting that RNA pol II accumulation at this site is more likely the result of significant basal transcription. The mechanistic basis for the recruitment of RNA pol II to active enhancers, as shown here, is unclear, although this phenomenon has been observed rather frequently by ChIP-chip analyses (33,60,61). In this context, it is possible that these enhancers may function as recruitment centers for RNA pol II, which is delivered subsequently to the TSS (10).

The authenticity of the regulatory regions we identified here is strengthened by the discovery that many of these regions exhibit high levels of H4ac in the absence and strikingly elevated levels in the presence of 1,25-(OH)₂D₃. Indeed, 1,25-(OH)₂D₃ induces a particularly striking increase in H4ac at the TSS and at virtually all the enhancers to which the VDR becomes bound. This could reflect contact between the TSS and individual enhancers via looping mechanisms that are now considered to be essential to the functional activity of distal enhancers (10). Peaks of H4ac are also rather broad, indicating the possible spreading of this epigenetic mark well beyond the site of modification initiated by the receptor and also of extensive remodeling activity (62). Given significant basal enhancer-specific levels of H4ac in the Vdr gene, we explored whether these levels might be due to the activity of cell-specific transcription factors responsible for basal VDR expression in bone cells. Two candidates include RUNX2, a master regulator of osteoblast differentiation and function (38), and C/EBPβ, a regulatory factor involved in the up-regulation of differentiated bone cell genes (39). Interestingly, we found that both of these transcription factors were indeed bound to regions of the Vdr gene including the TSS that aligned similarly with those that exhibited increased levels of H4ac. Previous studies have demonstrated the presence of C/EBPβ sites within the proximal exon 1a promoter of the human VDR gene (41). Potential RUNX2 sites were also identified in silico in the human gene as well (Pike, J. W., unpublished data). Thus, it seems likely that both of these factors may contribute to basal as well as cell-specific expression of the VDR in the MC3T3-E1 cell line, an hypothesis supported, in part, by our studies which demonstrate that siRNA-mediated suppression of C/EBPβ reduced the basal level of Vdr gene expression and the recombinant BAC clone in MC3T3-E1 cells. Importantly, the contribution of these transcription factors to the basal expression of the Vdr gene could provide clues as to the signaling pathways essential for its expression in bone cells. Interestingly, the central role of C/EBPβ in the regulation of Vdr gene expression takes on a special relevance because the up-regulation of liver-enriched inhibitory protein, a natural inhibitor of C/EBPβ, appears to play a dominate role in suppressing VDR expression in the parathyroid gland and perhaps the kidney during chronic kidney disease (23,24). Surprisingly, 1,25-(OH)₂D₃ also induced an up-regulation of both C/EBPβ- and RUNX2-binding activity in the Vdr gene locus at several target sites. The mechanism of this up-regulation is unclear at present, although 1,25-(OH)₂D₃ is known to up-regulate C/EBPβ expression (41). This mechanism does not provide an explanation for RUNX2 induction, however, because 1,25-(OH)₂D₃ suppresses the expression of this factor (40). Alternatively, it is possible that the reported ability of 1,25-(OH)₂D₃ to induce membrane-signaling pathways and various protein kinase cascades could be involved in the mechanism of induction (63). Regardless of these possibilities, the data strongly suggest that localization of the VDR and its RXR partner at multiple sites across the Vdr gene provokes functional responses that are likely to be important to the gene’s up-regulation.

RAR and GR were also observed at many of the enhancers in the Vdr gene locus. RAR was found both in the absence and presence of endogenously added RA, as noted earlier, whereas GR was found only upon induction by DEX. It seems likely that the RA response at S3 in the mouse Vdr gene corresponds to a region we previously identified earlier in the human VDR gene (54). Whereas RAREs remain to be identified in this region of both genes, a potential RARE candidate does appear in the S3 region of the mouse gene in our in silico analysis (supplemental Table 4). Finally, phosphorylated CREB and C/EBPβ (as discussed earlier) were also observed at the Vdr locus both in the absence and presence of Fsk. Many of the enhancer sites with which CREB (and C/EBPβ) interacts overlap with those noted for VDR, RAR, and GR, thereby strengthening the validity of these enhancers as sites of complex regulation. Like C/EBPβ, CREB-binding sites have also been previously described in the human VDR proximal promoter 1a (51). The observations made here therefore confirm that particular activity but also demonstrate that several additional sites may also contribute as well. CREB is a noted transcription factor target of the PKA-signaling pathway that is triggered by numerous extracellular factors and hormones including PTH (64), a hormone known to up-regulate the VDR in bone and kidney cells (28). C/EBPβ can also be induced by activators of the PKA pathway (52). Although both pCREB and C/EBPβ were present at the enhancers in the absence of Fsk, however, only pCREB was significantly up-regulated after treatment. Thus, it seems likely that activation of the PKA pathway targets primarily the CREB transcription factor in MC3T3-E1 cells. Regardless, the presence of CREB, C/EBPβ, and RUNX2 at the Vdr gene has significant implications for the control of basal Vdr gene expression, particularly in bone cell lines such as MC3T3-E1. Interestingly, each of these transcription factors represents a target for multiple signaling pathways integral to bone cell activity. RUNX2, for example, represents a master switch for osteoblast differentiation and mature cell gene regulation and is activated by a number of signaling pathways essential to bone development including both the bone morphogenetic protein/SMAD and the Wnt/β-catenin pathways (65,66,67). CREB, on the other hand, can be activated by not only the PKA but by other calcium-regulated pathways as well (68). The ability of all of these signaling pathways to converge on a set of transcription factors that are key to Vdr expression is likely important. The involvement of these tissue-specific regulatory factors could also facilitate the identification of enhancers with tissue-specific features.

We used large BAC clones together with enhancer DNA fragment analyses to evaluate both collective as well as independent transcriptional activity of the Vdr gene enhancers identified in these studies. Accordingly, a BAC clone comprising the entire mouse Vdr locus was engineered to contain an IRES-luciferase insert in the final 3′-noncoding Vdr exon and used to prepare a series of stable MC3T3-E1 cell clones. This collection of stable cell clones, as well as collections containing additionally modified BAC clones, produced recombinant VDR protein functionally similar to that expressed from the endogenous Vdr gene. Recombinant VDR proteins were also up-regulated in these stable cell lines by 1,25-(OH)₂D₃, RA, DEX, and Fsk. Thus, with the exception of RA activity, the BAC clone recapitulated the Vdr gene induction seen at the endogenous gene in MC3T3-E1 cells. Importantly, the regulation of the Vdr gene was also detected via the luciferase reporter inserted in the 3′-noncoding exon. Accordingly, 1,25-(OH)₂D₃, RA, DEX, and Fsk all induced a dose-dependent increase in luciferase activity that was maximal at the concentrations expected for these inducers. Deletion of DNA sequence downstream of the inserted IRES-luciferase cassette did not appeared to manifest a striking effect of the overall basal activity or on the capacity of 1,25-(OH)₂D₃, RA, or Fsk to promote an up-regulation, although elimination of the primary VDRE located in the S1 enhancer strikingly reduced the ability of 1,25-(OH)₂D₃ to promote up-regulation. These observations support the concept that Vdr gene regulation is mediated primarily through enhancers located upstream of exon 4. They also support the idea that although important to induction by 1,25-(OH)₂D₃, the VDRE in S1 does not represent the exclusive means whereby the Vdr gene is up-regulated in response to 1,25-(OH)₂D₃. Thus, the additional enhancers identified in this study likely play a role. These large BAC DNA segments are currently being used to prepare transgenic mice capable of expressing both the VDR and the luciferase reporter such that tissue-specific regulatory features of the Vdr locus can be defined.

Surprisingly, only three of the isolated DNA fragments were capable of mediating up-regulation by 1,25-(OH)₂D₃, RA, or DEX. This frequency is perhaps lower than we have seen in our earlier investigations. In the Rankl gene, for example, three of five fragments exhibited sensitivity to 1,25-(OH)₂D₃ (7,69), although only one was identified initially (7). Previous studies identified a functional VDRE in S1, an observation supported by data presented herein (9). With respect to RA, however, only the S3 region mediated response to this hormone. Finally, although the GR appeared to bind to several regions of the VDR gene, only U1 was responsive. Surprisingly, none of the fragments assessed were capable of mediating Fsk response, even when analyzed in the presence of coexpressed exogenous CREB or C/EBPβ (data not shown). This absence of responsiveness is particularly interesting in view of the fact that the Vdr gene BAC clone itself is highly sensitive to Fsk. There are many possible reasons for this absence of response especially because most of the fragments contain potential regulatory elements, as identified through in silico analyses. We conclude that, in contrast to the individual properties of S1, S3, and U1, the bulk of these enhancers do not function independently, but rather cooperatively and only in the context of the native gene. Thus, these data do not detract from the preponderance of both binding and activity measurements that suggest the importance of each of these regulatory regions.

Although the human gene for the VDR is more complex, our data indicate that its overall regulatory features are similar to those found in the mouse homolog. Perhaps most interesting is the activity observed at exon 1f. Previous studies suggested the presence of RNA transcripts containing this exon in the human gene, although the expression levels of these transcripts were low and emanated largely from kidney tissue (55). The current analysis provides additional support for the presence of this upstream promoter and suggests it may be regulated by 1,25-(OH)₂D₃ as well. The data also confirm the presence of CTCF at sites located in regions highly conserved relative to the mouse genome. These sites, as in the mouse gene, appear to define functional gene boundaries and speak to a potential role for CTCF as a gene insulator (47,48). The absence of a more robust induction of H4ac by 1,25-(OH)₂D₃ in the human gene limits this interpretation, however. It is clear, nevertheless, that as with the organization of the VDR gene-coding regions, overall regulatory features for both mouse and human genes are similar.

The results described here have particular implications with respect to previously defined restriction fragment-length polymorphisms as well as single-nucleotide polymorphisms in the human VDR gene locus (70). Certain of these, e.g. the Fok1 polymorphism located at the VDR translational start site, directly affects the structure and activity of the VDR protein itself (71,72). Others, such as those in the promoter at GATA binding factor (73) and CDX2 (74) transcription factor-binding sites, have the potential to alter Vdr gene expression. However, the altered activities of the VDR associated with most polymorphisms are unexplained. Our results suggest that VDR regulation is exceedingly complex and is mediated by multiple enhancers located in unexpected sites across the VDR gene locus. Thus, although we have not identified enhancers in the downstream half of the VDR gene, it is possible that non-bone-specific components may exist in these regions that could affect VDR expression levels. These polymorphisms could also reside in the upstream 1f promoter as well. The existence of single-nucleotide polymorphisms in this area strongly supports this possibility (75). Studies are ongoing to identify additional regions of regulation in the human gene and to determine how these might impact VDR expression and disease.

In summary, we have used ChIP-chip analysis together with BAC DNA clones to explore the expression and regulation of the mouse and human VDR. We identified a series of enhancers located both upstream of the VDR gene TSS and within downstream enhancers. The functionality of these enhancers is supported by multiple transcription factor binding and strengthened by the inducible presence of RNA pol II and by increased H4ac activity that is generally predominant at these sites. Finally, these regions are conserved in the human gene as well. The presence and locations of these regulatory regions provide important new insights into how expression of the VDR gene is controlled in bone cells.

Materials and Methods

Reagents

1,25-(OH)₂D₃ was obtained from Tetrionics, Inc. (Madison, WI). Fsk and DEX were obtained from Sigma Chemical Co. (St. Louis, MO) and all-trans-retinoic acid was obtained from Spectrum, Inc. (Gardena, CA). Antibodies to VDR (C-20), RXR (N-197), RAR (M-454), GR (M-20), and C/EBPβ were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Antitetraacetyl-H4 antibody (06-866) and antiphospho-CREB antibody (06-519) were acquired from Upstate Biotechnology, Millipore Corporation (Charlottesville, VA). Anti-CTCF antibody was obtained from (Billerica, MA). Anti-RNA polymerase II antibody (8WG16) and anti-HA antibody (HA.11) were obtained from Covance Laboratories, Inc. (Emeryville, CA). An siRNA pool designed to suppress C/EBPβ mRNA expression (and corresponding nontargeted control siRNA) was obtained from Dharmacon RNA Technologies (Lafayette, CO).

Cell culture

Mouse MC3T3-E1 osteoblastic cells and human MG-63 osteosarcoma cells were obtained from American Type Culture Collection (Manassas, VA). MC3T3-E1 cells were cultured in α-MEM and MG-63 cells were cultured in DMEM. Medium was supplemented with 10% fetal bovine serum obtained from Hyclone Laboratories, Inc. (Logan, UT). Upon confluence 48 h after plating and immediately before the experiment, culture medium was removed, the cells were briefly washed with PBS, and fresh medium was reintroduced. 1,25-(OH)₂D₃ and other ligands were then added to the cells in ethanol or dimethylsulfoxide (1% or 0.1% final concentration, respectively) or PBS for the periods indicated.

ChIP assay

ChIP was performed as described previously (4,6,7,8,9). Briefly, MC3T3-E1 or MG-63 cells were treated for 3 h with vehicle or the indicated experimental inducer. Maximal VDR binding was observed at this time point (8). The treated cells were washed several times with PBS and then subjected to a 15-min cross-linking reaction with 1.5% formaldehyde. After the cross-linking reaction, isolated cell extracts were sonicated to prepare chromatin fragments (average DNA size of 500–600 bp DNA as assessed by agarose gel electrophoresis) using a Fisher model 100 Sonic Dismembranator (Fisher Scientific, Pittsburgh, PA) at a power setting of 1. Precleared samples were subjected to immunoprecipitation using either a control IgG antibody or the indicated experimental antibody. The precipitated DNA was then washed, the cross-links were reversed, and the DNA fragments were purified using QIAquick PCR Purification Kits (QIAGEN, Valencia, CA). The isolated DNA [or DNA acquired before precipitation (input)] was then subjected to quantitative real-time PCR (qPCR) using primers designed to amplify fragments of the murine or human VDR gene promoter and/or for ChIP-chip analysis.

ChIP-chip analysis

Output DNA from the ChIP assays were further amplified using a ligation-mediated PCR method optimized by Oberley and Farnham (76) and then reassayed for retention of differential signals at a subset of specific sites on the Vdr gene by qPCR. These samples were then labeled according to an established protocol. Briefly, experimental or reference DNA samples (1.5 μg) were labeled with 1 OD of 5′-Cy5 or 5′-Cy3 tagged random nonamer wobble primers (TriLink BioTechnologies, San Diego, CA), respectively. Six micrograms of each labeled sample was combined with 2× Hybe Buffer and Hybe component A (Roche NimbleGen, Inc., Madison, WI), denatured at 95 C for 5 min, and then hybridized to custom NimbleGen DNA microarrays overnight at 42 C using a MAUI hybridization system (BioMicro, Salt Lake City, UT). Custom arrays were designed to include 100 kb upstream and downstream of vitamin D-responsive genes of interest including either the mouse or human VDR gene loci on a 385k feature platform. All oligonucleotide sequences for the array were generated using NimbleGen’s design standard, Tm-balanced 50–75 oligomers with an approximate spacing of 70 bp. Low complexity and repeat regions were excluded from tiling. After hybridization, arrays were washed and scanned at 5 μm using the GenePix 4000B scanner (Axon/Molecular Devices, Sunnyvale, CA). Data and peaks were extracted and analyzed using the NimbleScan (version 2.4) software (NimbleGen Systems). The log₂-ratios [log₂(n) = ln(n)/ln(₂) = log(n)/log(₂)] of test vs. experimental data were calculated for each point, and peaks were called at an FDR of P < 0.05 using a NimbleScan algorithm that employed a sliding window of eight probes spanning 700–800 bp wherein at least five adjacent probes were required to generate fluorescent signals above a normalized baseline. The data shown are representative of two or more ChIP-chip analyses performed for each experimental setting.

Plasmids

The pCH110-β-galactosidase reporter plasmid and the pcDNA-human (h)VDR vector were previously described (4,7). pRST7-GRα was kindly provided by Donald McDonnell (Duke University, Durham, NC). pTK-mVDR S1-luc and pTK-mVDR S2-luc were prepared as previously described (7). pTK-mVDR S3-luc, pTK-mVDR S4-luc, pTK-mVDR S5-luc, pTK-mVDR U1-luc, and pTK-mVDR U2-luc were constructed by cloning the appropriate mouse Vdr DNA fragments obtained through DNA amplification of mouse BAC DNA (RP23–136G8, see below) into the pTK-luc vector using BamHI (or HindIII) and SalI restriction sites. The pIRESluc plasmid was provided by Charles A. O’Brien (University of Arkansas for Medical Sciences), and the pgalK plasmid and appropriate cell lines were provided by the National Cancer Institute (see BAC clone reporters).

BAC clone reporters

A series of large-scale mouse Vdr luciferase reporter constructs was generated via recombinogenic targeting (35,36). BAC clone RP23-136G8 containing the mouse Vdr gene, 63 kb of 5′-flanking-region and 88 kb of 3′-flanking region, was obtained from the BACPAC Resource Center (Oakland, CA) and introduced into Escherichia coli strain SW106 cells by electroporation. The integrity of the DNA was verified by restriction enzyme digestion and pulse-field gel electrophoresis. The targeting construct used to prepare mVDR BAC2 was generated by PCR amplification of an IRES-luciferase-TK-neomycin cassette from the pIRESluc plasmid (77) using primers that contain 50 bp of homology to the 3′-untranslated region and 20 bp of homology to the end of the cassette. The amplified DNA was digested with DpnI, gel purified and introduced into SW106 cells harboring the RP23–136G8 BAC clone as described by Lee et al. (36). Recombinants were selected on LB plates containing kanamycin and chloramphenicol. Colonies were screened for correct targeting using 2 μl of bacterial culture as a template for PCR amplification with primers that flanked the recombination site. BAC DNA was prepared from the bacterial culture of a positive clone, linearized by NotI digestion and used for generating stable cell lines (see below). An internal NotI site in the Vdr sequence is located 23 kb downstream of the Vdr gene. Therefore, the 3′-flanking region of the mVDR BAC2 construct is shortened upon digestion from approximately 88 kb to 23 kb.

The mVDR BAC3 construct was prepared as described above except that the primers used to amplify the IRES-luciferase-TK-neomycin cassette differed. Accordingly, one primer had a 50-bp homology to a region located 88 kb downstream of the Vdr gene and the other 20 bp homology to the 3′-untranslated region. As a result, all of the sequence downstream of the final 3′-noncoding Vdr exon was removed when the mVDR BAC3 cassette was transformed into SW106 cells.

The mVDR BAC4 (ΔS1 VDRE) construct was prepared as a modification of mVDR BAC3 using the galK positive/negative selection as described by Warming et al. (37). The targeting construct was generated from the pgalK plasmid by PCR amplification of the galK open reading frame using primers that contain 50 bp of homology to the regions flanking the transcriptionally active S1 VDRE, the location of which was previously described (9). The amplified DNA was digested with DpnI for 1 h, gel purified, and introduced into SW106 cells harboring the mVDR BAC2 construct (37). The recombinants were selected first on M63 minimal media plates containing galactose, leucine, biotin, and chloramphenicol. Several emerging colonies were further selected on MacConkey agar plates to remove Gal− contaminants. Gal+ colonies were selected and cultured for use in the second stage of the recombineering effort. A double-stranded DNA oligo with homology to the regions flanking the galK site and containing only the sequence flanking the S1 VDRE was used for transformation. Thus, removal of the galK cassette resulted in the seamless deletion of the VDRE from S1. The recombinants were selected on M63 minimal media plates containing 2-deoxy-galactose, leucine, biotin, and chloramphenicol. Colonies were analyzed by restriction enzyme digestion and pulsed-field gel electrophoresis. A colony with a digestion pattern matching the parent construct was confirmed by PCR analysis and sequencing. DNA for the mVDR BAC3 construct was prepared, linearized by NotI digestion, and then used to generate stable cell lines (see below).

An HA tag was introduced into the translational start site of mVDR BAC2 to generate mVDR BAC1 using the galK system as described above. The targeting construct was generated using primers that contain 50 bp of homology to the translational start site of VDR and 20 bp of homology to the end of the cassette. A double-stranded DNA oligo with homology to the regions flanking the galK site and containing the HA tag was transformed into the culture. Thus, the galK cassette was replaced with the HA tag in a seamless exchange. A colony with a digestion pattern matching mVDR BAC2 was confirmed by PCR analysis and sequencing, expanded, linearized by Not1 digestion, and then used to prepare stable cell lines (see below).

Preparation of stable cell lines

Stable cell lines were prepared as previously described (77). MC3T3-E1 cells were seeded into six-well plates at a concentration of 1.5 × 10⁵ cells per well in α-MEM containing 10% fetal bovine serum (FBS). Cells were transfected 24 h later with 4 μg of BAC-luciferase reporter vector in serum- and antibiotic-free medium using Lipofectamine (Invitrogen, Carlsbad, CA). After transfection, the cells were cultured in medium supplemented with 20% FBS for 24 h. The cells were then collected by trypsinization, replated into two 10-cm dishes, and subjected to positive selection 24 h later using G418 (200 μg/ml). Colonies emerging from the three plates after 10–14 d were harvested, pooled, and examined.

Transient transfection luciferase assay

MC3T3-E1 cells were seeded into 24-well plates in α-MEM containing 10% FBS at a concentration of 5.0 × 10⁴ cells per well and transfected 24 h later with Lipofectamine PLUS (Invitrogen, Carlsbad, CA) in serum- and antibiotic-free medium. Individual wells were cotransfected with 250 ng of a luciferase reporter vector, 50 ng of pCH110-βgal, and 50 ng of pcDNA-hVDR or pRS-GRα After transfection, the cells were cultured in medium supplemented with 20% FBS with or without 1,25-(OH)₂D₃, all-trans-retinoic acid, DEX, or Fsk. Cells were harvested 24 h after treatment, and the lysates were assayed for luciferase and β-galactosidase activities as previously described (78). Luciferase activity was normalized to β-galactosidase activity in all cases.

Analysis of luciferase activity in stable cell lines

Stable MC3T3-E1 cell collections were seeded into 24-well plates at a density of 7.5 × 10⁴ cells per well, treated with the indicated concentrations of 1,25-(OH)₂D₃, RA, DEX, or Fsk for 24 h, and then harvested for analysis. Lysates were evaluated for luciferase activity and the activities were normalized to total protein.

siRNA-mediated mRNA depletion of C/EBPβ

MC3T3-E1 or MC-VDR BAC1 cells were seeded into six-well plates in α-MEM containing 10% FBS at a concentration of 1.5 × 10⁵ cells per well and transfected 24 h later with either Lipofectamine PLUS (Invitrogen) in serum- and antibiotic-free medium (mock) or with Lipofectamine PLUS containing either nontargeted or C/EBPβ siRNA pools (40 nm). MC3T3-E1 cells were harvested 48 h later, and the total RNA was isolated using Tri-Reagent and then subjected to real time RT-PCR analysis using primers to C/EBPβ, VDR, or β-actin mRNAs. Real-time PCR was performed on an Eppendorf Realplex using Power SYBR green PCR Master Mix with standard cycling conditions. Standard curves were created through serial dilutions of PCR-amplified cDNA. Primer sequences for these analyses are available upon request. MC-VDR BAC1 cells were harvested at 48 h, and lysates were evaluated for luciferase activity and normalized to total protein.

Western blot

MC3T3-E1 cells were plated in six-well dishes (1 × 10⁵ cells/ml) and treated the following day with either vehicle, 1,25-(OH)₂D₃, RA, DEX, or Fsk for 24 h. Cells were washed twice with PBS and then dissolved directly on the plates using 10 mm Tris-HCl, pH 7.4, 0.3 m NaCl, 1 mm dithiothreitol, and 1% Nonidet P-40. Cell lysates were harvested and centrifuged briefly, and the supernatants were evaluated for protein content. Samples (30 μg) were subjected to SDS-PAGE on 4–20% gradient gels. Proteins were transferred to Immunoblot polyvinylidene difluoride membranes (Millipore Corp., Billerica, MA) and subjected to Western blot analysis using the anti-VDR monoclonal antibody 9A7 as previously described (78) or an anti-HA antibody (HA.11). Images were developed using an enhanced chemiluminescence kit (GE Healthcare, Piscataway, NJ).