Vitamin D Receptor: Key Roles in Bone Mineral Pathophysiology, Molecular Mechanism of Action, and Novel Nutritional Ligands

Authors

  • Peter W Jurutka,

    1. Department of Integrated Natural Sciences, Arizona State University at the West Campus, Glendale, Arizona, USA
    2. Department of Basic Medical Sciences, The University of Arizona College of Medicine-Phoenix in partnership with Arizona State University, Phoenix, Arizona, USA
    Search for more papers by this author
  • Leonid Bartik,

    1. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    Search for more papers by this author
  • G Kerr Whitfield,

    1. Department of Basic Medical Sciences, The University of Arizona College of Medicine-Phoenix in partnership with Arizona State University, Phoenix, Arizona, USA
    2. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    Search for more papers by this author
  • Douglas R Mathern,

    1. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    Search for more papers by this author
  • Thomas K Barthel,

    1. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    Search for more papers by this author
  • Miriam Gurevich,

    1. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    Search for more papers by this author
  • Jui-Cheng Hsieh,

    1. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    Search for more papers by this author
  • Magdalena Kaczmarska,

    1. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    Search for more papers by this author
  • Carol A Haussler,

    1. Department of Basic Medical Sciences, The University of Arizona College of Medicine-Phoenix in partnership with Arizona State University, Phoenix, Arizona, USA
    2. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    Search for more papers by this author
  • Mark R Haussler PhD

    Corresponding author
    1. Department of Basic Medical Sciences, The University of Arizona College of Medicine-Phoenix in partnership with Arizona State University, Phoenix, Arizona, USA
    2. Department of Biochemistry and Molecular Biophysics, College of Medicine, The University of Arizona, Tucson, Arizona, USA
    • Address reprint requests to: Department of Basic Medical Sciences, University of Arizona College of Medicine-Phoenix in partnership with Arizona State University, 425 North Fifth Street, Phoenix, AZ 85004-2157, USA
    Search for more papers by this author

  • The authors state that they have no conflicts of interest.

Abstract

The vitamin D hormone, 1,25-dihydroxyvitamin D3 [1,25(OH)2D3], binds with high affinity to the nuclear vitamin D receptor (VDR), which recruits its retinoid X receptor (RXR) heterodimeric partner to recognize vitamin D responsive elements (VDREs) in target genes. 1,25(OH)2D3 is known primarily as a regulator of calcium, but it also controls phosphate (re)absorption at the intestine and kidney. Fibroblast growth factor 23 (FGF23) is a phosphaturic hormone produced in osteoblasts that, like PTH, lowers serum phosphate by inhibiting renal reabsorption through Npt2a/Npt2c. Real-time PCR and reporter gene transfection assays were used to probe VDR-mediated transcriptional control by 1,25(OH)2D3. Reporter gene and mammalian two-hybrid transfections, plus competitive receptor binding assays, were used to discover novel VDR ligands. 1,25(OH)2D3 induces FGF23 78-fold in osteoblasts, and because FGF23 in turn represses 1,25(OH)2D3 synthesis, a reciprocal relationship is established, with FGF23 indirectly curtailing 1,25(OH)2D3-mediated intestinal absorption and counterbalancing renal reabsorption of phosphate, thereby reversing hyperphosphatemia and preventing ectopic calcification. Therefore, a 1,25(OH)2D3–FGF23 axis regulating phosphate is comparable in importance to the 1,25(OH)2D3–PTH axis that regulates calcium. 1,25(OH)2D3 also elicits regulation of LRP5, Runx2, PHEX, TRPV6, and Npt2c, all anabolic toward bone, and RANKL, which is catabolic. Regulation of mouse RANKL by 1,25(OH)2D3 supports a cloverleaf model, whereby VDR-RXR heterodimers bound to multiple VDREs are juxtapositioned through chromatin looping to form a supercomplex, potentially allowing simultaneous interactions with multiple co-modulators and chromatin remodeling enzymes. VDR also selectively binds certain ω3/ω6 polyunsaturated fatty acids (PUFAs) with low affinity, leading to transcriptionally active VDR-RXR complexes. Moreover, the turmeric-derived polyphenol, curcumin, activates transcription of a VDRE reporter construct in human colon cancer cells. Activation of VDR by PUFAs and curcumin may elicit unique, 1,25(OH)2D3-independent signaling pathways to orchestrate the bioeffects of these lipids in intestine, bone, skin/hair follicle, and other VDR-containing tissues.

INTRODUCTION

As shown in Fig. 1, after its renal production as the hormonal metabolite of vitamin D, 1,25-dihydroxyvitamin D3 [1,25(OH)2D3] functions as the ligand for the vitamin D receptor (VDR), with the hormone–receptor complex inducing calcemic and phosphatemic effects that result in normal bone mineralization and remodeling.math image Intestinal absorption by the calcium transporter TRPV6 (also called ECAC2; Fig. 1, top right) and renal calcium reabsorption through Npt2c (also called SLC34A3; Fig. 1, top left) are triggered by low blood calcium, which stimulates PTH secretion and 1,25(OH)2D3 synthesis by the kidney 1α-hydroxylase (1α-OHase) enzyme (Fig. 1, top left). Intestinal phosphate uptake is also stimulated by 1,25(OH)2D3,math image but phosphate homeostasis additionally involves a newly recognized phosphaturic hormone, fibroblast growth factor 23 (FGF23), which is secreted by osteoblasts to regulate phosphate and calcium metabolism in a reciprocal manner to 1,25(OH)2D3.math image Elevated 1,25(OH)2D3 acts to promote expression of FGF23 in the osteoblastmath image (Fig. 1, bottom). FGF23 has dual actions to (1) suppress phosphate reabsorption from the kidney filtrate and (2) to repress 1,25(OH)2D3 synthesis, thus closing this endocrine regulatory loop. Similarly, elevated Ca+2 and 1,25(OH)2D3 negatively regulate PTH synthesis (Fig. 1, top), a feedback that closes the loop comprising PTH-1,25(OH)2D3–elicited correction of hypocalcemia. Added to its role in controlling calcium and phosphate ion concentrations in the blood, the 1,25(OH)2D3–VDR complex also controls the renal production of 1,25(OH)2D3 by feedback regulatory loops (Fig. 1, center), one of which entails the feedback repression of PTH. The latter point presents a clue to the likely physiological role of FGF23 as a “long acting PTH” in preventing hyperphosphatemia when 1,25(OH)2D3 is in the process of correcting hypocalcemia. Because PTH becomes suppressed rapidly during the action of 1,25(OH)2D3, the sterol hormone marshals a second phosphaturic principle, FGF23, to ensure that phosphate is eliminated while calcium is being acquired. This concept explains the requirement for two phosphaturic hormones, PTH and FGF23, with the former repressed and the latter induced by 1,25(OH)2D3.

Figure FIG. 1..

Calcium and phosphate homeostasis as mediated by 1,25(OH)2D3. Feedback loops to control calcium, involving PTH, and phosphate, including the novel phosphaturic hormone FGF23, are explained in the text. Targets of 1,25(OH)2D3-mediated gene regulation probed in this study (white text on black background) include several anabolic factors for bone (Runx2, LRP5, and osteocalcin [OC]), the respective calcium and phosphate (re)absorbing gene products, TRPV6 and Npt2c, the osteoclast differentiating factor RANKL, and a gene involved in the curtailing of FGF23 levels, PHEX. The inset in the lower right displays the effect of 1,25(OH)2D3 treatment (10−7 M) on the levels of each mRNA after 24 h as monitored by real-time PCR in the following cultured cell lines: MG-63 for human OC and RANKL; Caco-2 for human TRPV6 and Npt2c; UMR-106 for rat FGF23 and PHEX; and MC3T3-E1 for mouse Runx2 and LRP5. Levels of mRNA in each cell line are normalized against the unregulated GAPDH control mRNA, and the fold effects of 1,25(OH)2D3 are expressed as mean ± SE for three to four independent experiments with each experiment performed in triplicate (n = 9–12). All mean values for induction or repression by 1,25(OH)2D3 are significantly different from vehicle controls at p < 0.05.

A major focus of our current research is the characterization of VDR-regulated genes, particularly those supporting intestinal calcium and phosphate transport, including the molecular mechanisms whereby VDR exerts this primary action, details of which only now are being appreciated. Therefore, we will discuss recent advances in our understanding of 1,25(OH)2D3–VDR signaling and associated pathophysiology. What emerges is a more complete picture of the contribution of VDR signaling to bone mineral homeostasis that involves (1) stimulation of intestinal calcium and phosphate absorption to prevent rickets/osteomalacia, (2) enhancement of bone remodeling through osteoblast-induced osteoclast maturation, and (3) regulation of novel target genes involved in mineral homeostasis that possess anabolic or catabolic activity in terms of their function in either enhancing the formation or degradation of the mineralized skeleton. The 1,25(OH)2D3 ligand may also signal through rapid plasma membrane or cytoplasmic target cell actions that result in Ca2+ influx, MAPK stimulation, and increased PKC activity, events that are either supportive of the genomic functions of 1,25(OH)2D3–VDR or in some cases independent.math image Because the major calcium and bone actions of 1,25(OH)2D3 are ablated by inactivating the DNA binding function of VDR, the molecular details and pathophysiologic significance of rapid actions of the 1,25(OH)2D3 hormone will require further elucidation.

VDR actions in noncalcemic tissues may, in some cases, not involve the endocrine 1,25(OH)2D3 ligand. For example, hair cycling is greatly affected in VDR knockout animalsmath image and in many patients with hereditary hypocalcemic vitamin D–resistant rickets type II (HVDRR-II),math image although this process is not compromised in vitamin D deficiency. In addition, the secondary bile acid, lithocholate, has been recently recognized as a bona fide VDR ligand in the colon, capable of activating VDR to induce detoxifying, cytochrome P450-containing enzymes.math image Accordingly, this study also seeks to characterize binding and activation of VDR by novel ligands, including essential fatty acids and their derivatives, and other nutritionally derived beneficial lipids. These ligands might underlie or support VDR actions in skin, small intestine, and colon, not only to promote normal differentiation and detoxification, but also because they may possess value in cancer prevention in these and other epithelial tissues.

MATERIALS AND METHODS

General procedure for transfecting cultured mammalian cells

All cell lines in this study originated from ATCC and were maintained according to ATCC guidelines. Cells were transfected in 24-well plates using lipofectamine transfection reagent (Invitrogen, Carlsbad, CA, USA) in accordance with the manufacturer's protocol. Each well received 250 ng of pLuc-MCS plasmid (Stratagene, La Jolla, CA, USA) containing two copies of a nuclear receptor responsive element upstream of the firefly (Photinus pyralis) luciferase gene. The VDR element (VDRE) used (designated XDR3) was the distal element from the human cytochrome P450 (CYP) 3A4 gene.math image Cells were also co-transfected with 50 ng of a pSG5-based plasmid for expression of the appropriate nuclear receptor. A vector encoding Renilla luciferase (pRL-null; 20 ng) was included in each experiment to assess the efficiency of transfection and to serve as an internal standard. The cells were treated with a variety of known or potential ligands 18 h after transfection; treatment times ranged from 24 to 30 h. After incubation with ligand, cells were collected and analyzed using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) as described below. All data are reported as either the average of two or more experiments or are representative of two or more independent experiments. Each experimental treatment group was replicated in at least three, and often as many as six, wells.

Mammalian two hybrid transfections

The cells were transfected with components of the Mammalian Two Hybrid system from Stratagene. Retinoid X receptor α (RXRα) and steroid receptor co-activator-1 (SRC-1) were cloned into pCMV-BD (bait), and a human VDR cDNA was inserted into pCMV-AD (prey). Each well received 50 ng of bait and prey vectors along with 500 ng of pFR-luc, a firefly luciferase reporter construct, that was also introduced into the cells.

Analysis of transfected mammalian cells

The amount of reporter gene product (luciferase) produced in the transfected cells was measured using the Dual-Luciferase Reporter Assay System from Promega according to the manufacturer's protocol. The mean ratio of firefly luciferase to Renilla luciferase signal was determined for each experimental group, and the SD was calculated for the replicates (usually three to six wells) using MS Excel (expressed as error bars in figures).

Competitive binding assays

COS-7 cells were transfected with VDR and RXR expression vectors followed by sonication and centrifugation at 58,000 rpm at 4°C for 30 min. Aliquots of supernatant (10 μL) were mixed with 5 μL of 1,25(OH)2-26,27[3H]dimethyl-cholecalciferol (Amersham, Piscataway, NJ, USA; 183 Ci/mmol; final concentration ∼4.0 × 10−10 M and 51 Ci/mmol). Selections from a panel of known and putative ligands were added to the resulting solution and allowed to equilibrate with radiolabeled 1,25(OH)2D3 for 15 h. The unbound 1,25(OH)2D3 was removed, and the samples were analyzed using a Beckman (Fullerton, CA, USA) LS 5801 scintillation counter.

Real-time PCR

Total RNA was isolated from the indicated cells (treated with ligands for 24 h) using an acid guanidinium-phenol-chloroform method (TRIzol reagent; Invitrogen) according to the manufacturer's protocol, followed by treatment with DNase I (DNA-free; Ambion, Austin, TX, USA). DNase I–treated RNA (2 μg) was reverse transcribed using modified MMLV-derived reverse transcriptase and random hexamer primers (iScript cDNA Synthesis Kit; BioRad, Hercules, CA, USA) according to the manufacturer's protocol. The cDNA synthesized was used as a template in 20 μl PCR reactions containing 10 μl iQ SYBR Green Supermix (Bio-Rad), 1 μl primers, 2 μl cDNA synthesis reaction, and 7 μl molecular grade water. Reactions were performed in 96-well PCR plates and analyzed on a Bio-Rad iCycler iQ Real-Time PCR detection system. Data were analyzed using the comparative Ct method for relative quantitation; all raw values were normalized to an endogenous reference (GAPDH cDNA) and compared with a calibrator (i.e., the normalized Ct value from vehicle-treated cells) and expressed as 2−ΔΔCt according to Applied Biosystems User Bulletin 2: Rev B, “Relative Quantitation of Gene Expression.”

RESULTS AND DISCUSSION

As described in the Introduction and Fig. 1, we now have a more complete understanding of the hormonal control of phosphate and calcium, based on the recent discovery of FGF23 as a new phosphaturic hormone that negatively controls renal 1,25(OH)2D3 biosynthesis and possibly bone mineralization.math image Although FGF23 expression was originally localized to the ventrolateral thalamic nucleus of the brain and thought to be undetectable in bone, more recent data have shown that FGF23 mRNA is expressed in osteoblasts.math image The FGF23 protein is detected in mature osteoblasts,math image and the FGF23 promoter has been shown to respond positively to 1,25(OH)2D3 in transfected ROS 17/2.8 osteoblasts.math image We showed through real-time PCR that endogenous FGF23 mRNA is detectable both in cultured rat UMR-106 osteoblasts, where it is upregulated 78-fold by 1,25(OH)2D3 (Fig. 1, inset at bottom right), and in mouse bone, where it is strikingly upregulated by 1,25(OH)2D3 treatment.math image Concomitantly, FGF23 is elevated 80-fold in the serum of 1,25(OH)2D3-injected mice.math image Therefore, we propose that FGF23 is not only at least one component of the long sought phosphatonin, but that it is an osteoblastic phosphatonin upregulated by 1,25(OH)2D3 (as well as by high phosphate, independently). Aberrations in the FGF23 pathway result in hyper- and hypo-phosphatemic diseases of bone mineral metabolism in which the FGF23 feedback loop is either missing or dysregulated. For example, FGF23-null mice have increased serum 1,25(OH)2D3, hyperphosphatemia, ectopic calcification in soft tissues, and excessive mineralization of bone. In contrast, hypophosphatemic osteomalacia results from malignant cells that constitutively express FGF23, and autosomal dominant hypophosphatemic rickets is caused by FGF23 mutations that prevent its proteolytic inactivation. Taken together, these vital observations allow the construction of a “balanced” mineral homeostasis scheme (Fig. 1), in which renal 1,25(OH)2D3 is central and interplays with both PTH for calcemic control and FGF23 for phosphatemic control, with the final result being the fine tuning of resorption and mineralization to maintain the functional skeleton.

Another factor with implications for the 1,25(OH)2D3–FGF23 axis is phosphate-regulating endopeptidase homolog, X-linked (PHEX). PHEX is synthesized in osteoblasts and participates in a complex cascade of protein degradation, the net effect of which is to reduce FGF23 synthesis and thereby allow kidney phosphate reabsorption to continue.math image The double knockout of PHEX (Hyp mouse) and FGF23math image shows that deletion of FGF23 reverses the Hyp phenotype of hypophosphatemia, inappropriately low 1,25(OH)2D3, and rickets, further confirming the reciprocal relationship of these two factors. Mutations in PHEX lead to the human disorder of X-linked hypophosphatemia (XLH).math image Like FGF23, PHEX is also regulated by 1,25(OH)2D3, but its expression is repressed by the vitamin D hormone.math image We have shown this repression of PHEX by 58% after 1,25(OH)2D3 treatment of osteoblast-like cultured UMR-106 cells (Fig. 1, inset at bottom right). Thus, one key to calcium and phosphate homeostasis is the interplay and feedback control of FGF23, PTH, and PHEX, with 1,25(OH)2D3 acting as the central integration and signaling molecule to appropriately control the blood levels of these ions.

Not only do we theorize that 1,25(OH)2D3 is the master integrator of calcium and phosphate homeostasis through its cross-regulation of FGF23, PHEX, and PTH, but also that 1,25(OH)2D3 impacts the genesis of both osteoclasts and osteoblasts, in part because of the regulation of key genes. Thus, in addition to 1,25(OH)2D3-mediated regulation of FGF23 and PHEX, several other genes have been identified as targets for 1,25(OH)2D3 transcriptional control (Fig. 1, white text on black ovals, and data inset at bottom right). These genes can generally be classified as anabolic or catabolic in terms of their function in either enhancing the formation or degradation of the mineralized skeleton. Bone mineral anabolic genes that we have shown to be regulated by 1,25(OH)2D3 (Fig. 1, inset at lower right) include (1) LRP5, the Wnt co-receptor that is essential for osteoblast proliferation and function, which is induced 6.6-fold; (2) Runx2,math image an osteoblast-specific transcription factor that is absolutely required for bone formation, which is repressed by 50%; (3) TRPV6,math image the intestinal epithelial calcium channel that is vital for absorption of calcium required for proper bone mineralization, which is induced 4.9-fold; and (4) Npt2c,math image a renal sodium-phosphate co-transporter that facilitates the reabsorption of phosphate at the kidney, which is induced 4.7-fold. Catabolic genes regulated by 1,25(OH)2D3 include PTH, which is repressed by 1,25(OH)2D3,math image and RANKL, an osteoblast cell surface protein that stimulates osteoclastogenesis, which is induced 4.8-fold (Fig. 1, inset at bottom right). We hypothesize that 1,25(OH)2D3 is anabolic to bone in physiologic concentrations by inducing LRP5, TRPV6, and Npt2c, but in the face of excess or sustained 1,25(OH)2D3, the hormone is catabolic to bone by inducing RANKL, repressing PHEX, and inducing FGF23 to elicit phosphaturia. Excessive 1,25(OH)2D3 signaling may also delimit bone through repression of Runx2 and induction of FGF23, which would result in decreased bone formation and mineralization, respectively.

Furthermore, we are in the process of elucidating, within the context of bone mineral regulating tissues/cells, the molecular details of signaling events mediated by the 1,25(OH)2D3-liganded VDR in controlling these anabolic and catabolic genes. Although the precise mechanism whereby liganded VDR regulates target gene transcription has yet to be completely characterized, VDR can be considered a typical nuclear hormone receptor,math image comprised of a dual zinc finger-based DNA-binding domain (DBD), nuclear localization signal(s), and a ligand-binding domain (LBD) that binds the cognate ligands.math image Furthermore, VDR is a member of the thyroid hormone, retinoic acid, oxycholesterol, and xenobiotic subfamily of nuclear receptors that primarily heterodimerizes with RXR in response to 1,25(OH)2D3 binding to recognize direct repeat responsive elements in the promoters of regulated genes (Fig. 2A, steps 1 and 2).math image Previous research with VDR-activated genes has indicated that many factors participate in transactivation.math image These include the following (Figs. 2A and 2C): step 3, factors capable of histone acetylation (HATs), such as SRC-1,math image CBP/p300, or pCAF, or of ATP-dependent chromatin remodeling, such as the mammalian homologs of SWI/SNFmath image; step 4, TATA binding protein associated factors (TAFs, especially TAFs 28, 55, and 135math image); step 5, basal transcription factors such as TFIIBmath image; step 6, D-receptor interacting proteins (DRIPs, especially DRIP205, which is a subunit of the mediator complex that couples transactivators to the C-terminal tail of RNA polymerase II)math image; and step 7, NCoA-62, a factor reported to serve as a co-activator for VDR and related nuclear receptors that seems to couple transcription to RNA splicing.math image Step 8, interaction with TRIP1, the mammalian homolog of the yeast SUG factor, results in progressive ubiquitination of VDR and ultimately leads to its recognition and degradation by the proteasome.math image Many of these co-factors interact with a common region of VDR, namely the C-terminal AF-2 motifmath image; thus, it is difficult to conceive of these factors all interacting with VDR to effect transactivation except in a sequential manner (Fig. 2A) or in a complex in which multiple VDR-RXR heterodimers are present (Fig. 2C).

Figure FIG. 2..

Chromatin looping model of gene regulation by VDR. (A) Sequential model of gene activation of the mouse osteopontin gene (containing a single VDRE) by the 1,25(OH)2D3-bound VDR-RXR heterodimer. Numbers inside circles refer to discreet stages in the activation of a VDR target gene as discussed in the text. (B) The presence of several potential VDREs in the 5′ flanking region of the mouse RANKL gene, as reported by Kim et al.math image and verified by our experiments (data not shown). (C) How these VDREs might cooperate in a chromatin-looping model. As discussed in the text, multiple VDR-RXR heterodimers may be capable of simultaneously recruiting co-activators and other factors (pictured are the factors from the numbered stages of A) to form a regulatory supercomplex at the promoter.

The RANKL gene promoter (Fig. 2B) is being used as a model system for studying the steps in transcriptional activation by the liganded VDR-RXR heterodimer using in silico analysis as well as chromatin immunoprecipitation (ChIP), gel mobility shift, and transcription assays. Based on our studiesmath image and those of Kim et al.,math image Fig. 2C depicts a postulated chromatin looping model for the mouse RANKL gene. Instead of separate events in which various factors bind to a single VDR-RXR heterodimer in a defined sequence (Fig. 2A),math image we propose that the chromatin looping model (Fig. 2C) allows for simultaneous binding of multiple factors in a supercomplex at the promoter.

Most evidence for a chromatin-looping model in transcriptional control is derived from studies of the globin genesmath image and the T-helper type 2 cytokine locus.math image There is also direct evidence for such looping in transactivation by nuclear receptors. Multiple binding sites for the estrogen receptor have been reported flanking single genes, some as far as 144 kb from the transcriptional start site (TSS).math image Chromosomal looping has also been postulated for androgen-dependent transcription.math image Although direct evidence for chromosomal looping in VDR-mediated transcriptional modulation is not yet available, there are data that multiple, active VDREs are located at considerable distances from the TSS in 1,25(OH)2D3 target genes. Recent results from Kim et al. using ChIP-chip and other techniques, as well as recent data from our laboratory using in silico, ChIP, gel mobility shift, and transcription analyses (data not shown), have provided evidence for active VDREs located anywhere from 76 kb upstream in the RANKL genemath image (Fig. 2B) to 19–29 kb downstream (within the first two introns) in the mouse LRP5math image and VDRmath image genes, as well as 2–4 kb upstream of the TRPV6 gene.math image It therefore seems likely that nuclear receptors, including VDR, also use chromosomal looping in their mechanism of transactivation, at least in some settings.

The existence of multiple enhancer elements, some of varying strengths, may provide an explanation for hitherto unresolved phenomena. For example, the dramatically different patterns of 1,25(OH)2D3-dependent regulation for the RANKL and OPG genes observed in different cell lines can perhaps be attributed to differing abilities to form the chromatin looping complex suggested by Fig. 2C. This is an especially attractive hypothesis if, for example, the complex requires factors other than VDR and RXR that are limiting in certain cell lines. The time that may be required to form such complexes might also explain why, for many hormones, there is a distinctly different regulatory result when the hormone is chronically present versus situations in which there is an acute spike in hormone concentration. There is additional evidence, in the case of VDR-mediated regulation of RANKL, that overexpressing VDR in osteoblasts can eliminate or even reverse the induction of RANKL by 1,25(OH)2D3.math image One could speculate that high concentrations of VDR might lead to formation of supercomplexes in which low-affinity VDREs might attract co-repressors rather than co-activators. Further characterization of the multiple potential VDREs at the RANKL locus will be required to obtain a complete picture of the complex regulation of RANKL by 1,25(OH)2D3-bound VDR, and the current working model depicted in Fig. 2C may serve as a paradigm for other 1,25(OH)2D3–VDR regulated genes that contain multiple VDREs with differential VDR binding affinity.

In addition to the role of VDR in bone mineral homeostasis, evolutionary considerations indicate that VDR predates mineralized tissue and hair in vertebrates.math image Moreover, VDR is closely related to pregnane X receptor (PXR) and constitutive androstane receptor (CAR), which induce enzymes that detoxify xenobiotics, and to farnesoid X receptor (FXR), the bile acid sensor.math image It is therefore not surprising that VDR also likely functions as a low-affinity sensor for detoxification in certain tissues such as the colon, where VDR has already been shown to bind the carcinogenic secondary bile acid, lithocholate.math image The VDR is also absolutely required for a functional hair cycle as exemplified by the known phenotype of alopecia in HVDDR-II patientsmath image and in VDR knockout mice.math image Furthermore, essential polyunsaturated fatty acid (PUFA) deficiency has adverse effects on skin, including impaired hair growth and dermatitis.math image Administration of PUFAs such as linoleic acid has been shown to reverse these symptoms.math image Intriguingly, the skin phenotype in VDR knockout mice resembles PUFA deficiency, strongly suggesting that VDR plays an important role in the health of skin and hair.math image The notion that VDR could have additional ligands is supported by the aforementioned identification of lithocholate (Fig. 3A, “L”) as a bona fide, low-affinity VDR ligand.math image Thus, novel VDR ligands, likely related to PUFAs, may be important for VDR actions in the skin.

Figure FIG. 3..

Evaluation of potential novel ligands for VDR. (A) Inside dashed area: known ligands for VDR and related mammalian nuclear receptors. The balance of A depicts candidate ligands tested in this study. Ligand abbreviations are given in parentheses. (B) Mammalian two hybrid assay performed in human embryonic kidney-293 cells, measuring ligand-induced heterodimerization between VDR and RXR. Here and in C, the data shown are representative of at least three independent experiments, with each treatment group performed in three to six wells, and the resulting ligand-fold effects (compared with vehicle control) are designated above each bar (the error bars represent SD). The mean values for D, L, and all PUFAs are significantly different from control (vehicle, open bar) at p < 0.05. (C) Ligand-stimulated transcription assay using a VDRE-containing reporter plasmid transfected into human colon cancer cells (Caco-2). Open bars represent ethanol or DMSO vehicle controls. The mean values ± SD for D, L, and CM are all significantly different from their respective vehicle controls (p < 0.05). (D) Competition curves showing that CM and L are able to compete for binding to overexpressed VDR in a COS-7 cellular extract with ≈4.0 × 10−10 M [3H]1,25(OH)2D3. Dexamethasone (Dex) is a lipophilic glucocorticoid receptor ligand and has no appreciable binding to VDR, thus serving as a noncompeting control.

Along with PUFAs, a variety of other compounds listed in Fig. 3A were tested as potential VDR ligands. Figure 3B presents results of a mammalian two hybrid assay performed in human embryonic kidney-293 (HEK-293) cells to evaluate ligand-induced association between VDR and RXR. Figure 3B shows that all of the tested PUFAs significantly induced VDR-RXR heterodimerization (6- to 13-fold). In addition, all PUFAs tested significantly activated VDRE-mediated transcription in HEK-293 cells transfected with a VDRE reporter construct (data not shown). Moreover, these findings were consistent with the results of mammalian two hybrid assays performed in human colon cancer (Caco-2) and osteoblast-like rat osteosarcoma cells (ROS 17/2.8), and VDRE-mediated transcription was also activated by PUFAs in human osteosarcoma cells (TE-85, data not shown). Thus, PUFAs facilitate VDR-RXR heterodimerization and activate VDR-mediated transactivation in renal, intestinal, and skeletal cells. Importantly, at similar concentrations, other compounds, including ligands of PXR, the receptor evolutionarily closest to VDR, failed to stimulate interactions between VDR and RXR (Fig. 3B, right half of panel), thus excluding the possibility that high levels of any small lipophilic molecule can nonspecifically activate VDR.math image

Another compound evaluated as a potential VDR ligand is the turmeric-derived polyphenol, curcumin (Fig. 3A, “CM”). Chemoprevention by CM has been documented in the intestine and skin, two major VDR target tissues.math image There is also evidence that CM potentiates 1,25(OH)2D3-stimulated differentiation in human leukemia cells (HL-60).math image Therefore, we hypothesized that VDR may be a direct mediator of curcumin bioactions. Figure 3C shows the results of an experiment performed in human colon cancer cells (Caco-2) transfected with a VDRE-containing reporter plasmid. Cells treated with 6.7 × 10−6 and 10−5 M CM showed a dose-dependent increase (2.1- and 5.0-fold, respectively) in the level of transcription of the reporter plasmid, the latter of which is comparable to stimulations by 10−8 M 1,25(OH)2D3 and 10−4 M lithocholate (L); CM also facilitates VDR-RXR heterodimerization in a mammalian two hybrid system (Caco-2 cells; data not shown) similar to the actions of PUFAs depicted in Fig. 3B. Some of the tested compounds (Fig. 3A) also stimulated RXR-mediated transcription, but neither PUFAs nor CM activated the glucocorticoid receptor to a significant extent, indicating a specific effect on VDR and/or RXR. To show that curcumin can bind directly to VDR and does not activate transcription solely by binding to RXR or some other co-factor, in vitro competition VDR-binding assays were performed. As shown in Fig. 3D, CM successfully competed with radiolabeled 1,25(OH)2D3 for binding to overexpressed VDR present in a COS-7 cellular extract. At 10−4 M, CM effectively competed >50% of the amount of radiolabeled 1,25(OH)2D3 bound to VDR compared with negligible competition by 10−4 M Dex, a lipophilic ligand without appreciable binding to VDR. Lithocholate was included for comparison with CM and was only slightly more effective than CM in competing with 1,25(OH)2D3 for VDR binding. The PUFAs evaluated in this study were likewise capable of competing with labeled 1,25(OH)2D3 for binding to VDR (data not shown), indicating that both PUFAs and curcumin represent low affinity activating ligands of VDR, effective in concentrations that, like lithocholate, actually occur naturally at select VDR targets such as colon and skin.

Overall, these new findings indicate that CM and PUFAs not only directly bind to VDR, but also induce RXR recruitment and activate transcription through a VDRE. The evidence presented herein supports the hypothesis that curcumin and PUFAs are bona fide ligands for VDR and suggest that at least some of their bioeffects, particularly in the colon and skin, are mediated by this nuclear receptor. Because all tested PUFAs activate VDR at very high concentrations, it is possible that a specific PUFA metabolite generated enzymatically in a select VDR target site, such as the hair follicle, is a high-affinity VDR ligand that actually functions to drive the mammalian hair cycle through VDR mediation of transcriptional control. In addition, these observations provide a novel explanation for beneficial effects of certain PUFAs on bone, which may result from direct activation of VDR.math image Thus, this study implicates VDR as a sensor of circulating essential fatty acids and other beneficial dietary lipids, such as curcumin.

The complete spectrum of VDR bioactions is therefore expanding to include responses to the presence of alternative lipophilic ligands to effect 1,25(OH)2D3-independent actions in nontraditional target tissues, with likely implications for nutritional and pharmacological interventions to promote epithelial tissue health and potential cancer prevention. As for the mechanisms by which the liganded VDR-RXR heterodimer influences gene transcription, we are beginning to appreciate the true complexity of these actions at a single promoter, which in many cases seems to involve multiple, widely separated VDREs that may cooperate in ways that remain to be elucidated. This complexity, combined with the likelihood of many alternative (and largely uncharacterized) mechanisms for negative regulation of transcription by liganded VDR, underscores the need for further elucidation of VDR signaling to reveal the full repertoire of this intriguing nuclear hormone receptor and its array of physiologically and pharmacologically relevant ligands.

Acknowledgements

This study was funded by National Institutes of Health Grants DK33351 and DK063930 to MRH.

Ancillary