Leaders in Pharmaceutical Business Intelligence (LPBI) Group

Vitamin D debates

Larry H. Bernstein, MD, FCAP, Curator

LPBI

Vitamin D: Time for Rational Decision-Making

JoAnn E. Manson, MD, DrPH

References

1. Beveridge LA, Struthers AD, Khan F, et al. D-PRESSURE Collaboration. Effect of vitamin D supplementation on blood pressure: a systematic review and meta-analysis incorporating individual patient data. JAMA Intern Med. 2015;175:745-754. Abstract
2. Hansen KE, Johnson RE, Chambers KR, et al. Treatment of vitamin D insufficiency in postmenopausal women: a randomized clinical trial. JAMA Intern Med. 2015;175:1612-1621. Abstract
3. Baron JA, Barry EL, Mott LA, et al. A trial of calcium and vitamin D for the prevention of colorectal adenomas. N Engl J Med. 2015;373:1519-1530. Abstract
4. Institute of Medicine. Dietary Reference Intakes for Calcium and Vitamin D. Washington, DC: National Academies Press; 2011. http://iom.nationalacademies.org/Reports/2010/Dietary-Reference-Intakes-for-Calcium-and-Vitamin-D.aspx Accessed October 28, 2015.
5. US Preventive Services Task Force. Final Recommendation Statement: Vitamin D Deficiency: Screening, 2014.http://www.uspreventiveservicestaskforce.org/Page/Document/RecommendationStatementFinal/vitamin-d-deficiency-screening Accessed October 28, 2015.

Isn’t there much more to this than the debates entail?

Classic nutritional rickets is caused by the simultaneous deprivation of sunlight exposure and dietary vitamin D. As depicted in Fig. 1, the pathways comprising the metabolic activation of the vitamin to its hormonal form and subsequent functions in target tissues present a number of additional steps where defects elicit vitamin D-resistant rachitic syndromes. Two of these disorders involve the inadequate bioactivation of 25-hydroxy¬ vitamin D3 (25(OH)D3) to 1,25-dihydroxyvitamin D3 (l,25(OH)2D3) by the kidney as catalyzed by the 1-OHase enzyme (Fig. 1).

Figure 1 Bioactivation of vitamin D3 and actions of the 1,25(OH)2D3 hormonal metabolite on intestine, bone and kidney, along with related rachitic syndromes. The production of 1,25(OH)2D3 is depicted in the lowet portion and its functions on mineral ttansport in target cells ate pictured in the upper portion. Defects eliciting rachitic syndromes ate boxed, with the televant mutated gene and chromosomal location denoted where appropriate

Acquired chronic renal failure results in renal rickets and secondary hyperparathyroidism (renal osteodystrophy) when the compromising of renal mass reduces 1-OHase activity (Haussier oc McCain 1977). The etiology of pseudo-vitamin D-deficiency rickets (PDDR) apparently involves a hereditary defect in the gene coding for the 1-OHase enzyme (Labuda et al. 1992). Interestingly, the PDDR locus is resolvable from that of the vitamin D receptor (VDR) but maps very close to it on chromosome 12 in the 12ql3—14 region (Labuda et al. 1992). Recently, a cDNA was cloned for the rat 1-OHase (St-Arnaud et al. 1996) and it is expected that the human renal 1-OHase gene will soon be cloned and its chromosomal location determined. The likelihood that both the gene encoding the enzyme that generates the l,25(OH)2D3 hormone and the cognate hormone receptor gene lie in close proximity on chromosome 12 invites speculation about the evolution of the vitamin D ligand receptor system. The traditional actions of vitamin D, via its l,25(OH)2D3 hormonal metabolite, are to effect calcium and phosphate homeostasis to ensure the deposition of bone mineral on type I collagen matrix (summarized in Fig. 1).

Figure 1 Bioactivation of vitamin D3 and actions of the 1,25(OH)2D3 hormonal metabolite on intestine, bone and kidney, along with related rachitic syndromes. The production of 1,25(OH)2D3 is depicted in the lowet portion and its functions on mineral ttansport in target cells ate pictured in the upper portion. Defects eliciting rachitic syndromes ate boxed, with the televant mutated gene and chromosomal location denoted where appropriate

l,25(OH)2D3 stimulates intestinal calcium and phosphate absorption, bone calcium and phosphate résorption, and renal calcium and phosphate reabsorption, all resulting in a sufficient CaP04 ion product to precipitate hydroxyapatite. Failure to achieve normal bone mineral accretion by these mechanisms leads to rachitic syndromes. Recently, a breakthrough has occurred in our understand¬ ing of what was originally known as hypophosphatemic vitamin D-resistant rickets, a familial disorder of renal phosphate wasting more appropriately referred to as dominant X-linked hypophosphatemic (HYP) rickets (Fig. 1). The gene defect responsible for HYP rickets has been fine mapped in the Xp22T region, harboring a gene identified as PEX, or phosphate regulating gene with homologies to endopeptidases located on the X-chromosome (Francis et al. 1995). One hypothesis is that PEX codes for an endopeptidase that apparently correctly processes a peptide precursor to yield a novel, as yet unidentified, phosphate retaining hormone. The normal function of this hormone may be to oppose the action of parathyroid hormone (PTH) and stimulate phosphate reabsorption by the renal tubule by inducing the Na -phosphate cotransporter. However, the existence of tumor-induced osteomalacia, an acquired disorder that closely resembles the phosphate wasting of HYP rickets and is characterized by low circulating l,25(OH)2D3 (Parker et al. 1981), combined with renal cross-transplantation (Nesbitt et al. 1992) and parabiosis (Meyer et al. 1989) studies in normal and hyp mice, indicates strongly that the HYP phenotype is caused by excessive amounts of a phosphaturic hormone in the circulation. This humoral peptide is distinct from PTH and has been named phosphatonin (Cai et al. 199A, Econs & Drezner 1994). Thus, instead of PEX mutations result¬ ing in insufficient generation of a novel phosphate retaining peptide, they may instead elicit the appearance of abnormally high circulating levels of phosphatonin, with the normal role of the PEX gene product postulated to be the proteolytic inactivation of this phosphaturic principle. Most germane to the vitamin D endocrine system is the fact that serum l,25(OH)2D3 levels are inappropriately low for the prevailing phosphate concentrations in HYP rickets and patients can be cured with a therapeutic combination of phosphate and l,25(OH)2D3 (Harrel et al. 1985). Because it is well known that hypophosphatemia stimulates l,25(OH)2D3 production (Hughes et al. 1975), the PEX/phosphatonin system might constitute yet another regulatory loop in maintaining normal phosphate homeostasis. One could hypothesize that under hypo- phosphatemic conditions, when l,25(OH)2D3 levels are elevated, the sterol hormone not only increases intestinal phosphate absorption (Fig. 1) and suppresses PTH synthesis (DeMay et al. 1992) to conserve phosphate, but also induces the PEX gene product (Rowe et al. 1996) to cleave phosphatonin and further promote renal phosphate reclamation. l,25(OH)2D3 is primarily recognized as a calcémic hormone, perhaps due to the abundance of dietary phosphate, or because calcium homeostasis is more vitamin D-dependent than the regulation of extracellular phos¬ phate. Regardless of the mechanism, traditional vitamin D-deficiency and clinically significant defects in the vitamin D receptor lead invariably to hypocalcemia and secondary hyperparathyroidism, with phosphate being somewhat less affected. As illustrated in Fig. 1, target tissue insensitivity to l,25(OH)2D3 is known as hereditary hypocalcémie vitamin D-resistant rickets (HVDRR) and is caused by defects in the gene on chromosome 12 coding for the VDR. A review of the etiology of HVDRR and the natural mutations in the VDR that confer tissue insensitivity and clinical resistance to l,25(OH)2D3 is particularly instructive in illuminating the physiologic relevance of the l,25(OH)-,D3-VDR hormone-receptor complex as well as structure/function relationships in the receptor itself.

Natural mutations in the nuclear vitamin D receptor Clinically significant hereditary hypocalcémie vitamin D-resistant rickets is an autosomal recessive disorder resulting in a phenotype characterized by severe bowing of the lower extremities, short stature and, often, alopecia (Rut et al. 199A). The serum chemistry in HVDRR includes frank hypocalcemia, secondary hyperpara¬ thyroidism, elevated alkaline phosphatase, variable hypophosphatemia and markedly increased l,25(OH)2D3. The symptoms of HVDRR, with the exception of alopecia, mimic classic vitamin D-deficiency rickets, suggesting that VDR not only mediates the bone mineral homeostatic actions of vitamin D but may also participate in the differentiation of hair follicles in utero. Recently, VDR knockout mice have been created (Yoshizawa et al. 1996), revealing apparently normal hétérozygotes but severely affected homozygotes (VDR-/-), 90% ofwhich die within 8—10 weeks. Surviving mice lose their hair and possess low bone mass, hypocalcemia, hypophosphatemia and 10-fold elevated l,25(OH)2D3 coincident with extremely low 24,25(OH)2D3. All of these parameters in the VDR knockout mouse mimic the phenotype of patients with HVDRR, confirming that VDR normally mediates all of the bone mineral regulating functions of vitamin D. Interestingly, although natural point mutations in other receptors related to VDR, such as thyroid hormone receptor ß (TRß) (Collingwood et al. 1994), are charac¬ terized by dominant negative receptors that generate the thyroid hormone resistant phenotype in the heterozygotic context, no natural, dominant negative mutations have yet been identified in HVDRR patients (Whitfield et al. 1996). Thus, all HVDRR cases studied to date are homozygous for the particular VDR mutation.

Figure 2 Natural mutations in the human vitamin D receptor leading to 1,25(OH)2D¡ hormone resistance. See text for details and citations. N37, K91 and E92 are not sites of VDR natural mutations, but are so designated because they ate heterodimerization contacts that lie within the DNA binding domain (Hsieh et al. 1995, Rastinejad et al. 1995). The eight cysteine residues (C) that tetrahedrally coordinate two zinc atoms in the finger sttucture are also denoted.

Figure 2 illustrates a number of point mutations in VDR that have been detected in HVDRR patients (reviewed in Rut et al. 199A, Haussler et al. 1995). Three of these genetic alterations result in nonsense mutations that introduce stop codons in VDR (K73stop, Q152sfo|> and Y295stop), creating truncated VDRs that lack both hormone- and DNA-binding (heterodimerization) capacities and are associated with unstable mRNAs. More revealing are the series of missense mutations (Fig. 2) that can be classified according to three of the basic molecular functions of VDR: (i) DNA binding/nuclear localization by the N-terminal zinc finger region, (ii) l,25(OH)2D3 hormone binding by the C-terminal domain and (iii) heterodimerization with retinoid X receptors (RXRs) through subregions of the C-terminal domain. As depicted schematically in Fig. 2 and discussed in detail later, VDR is a ligand-dependent transcription factor that controls gene expression by heterodimerizing with RXR and associating specifically with vitamin D responsive elements (VDREs) in target genes. Since VDR is a member of the steroid, retinoid, thyroid hormone receptor superfamily, and belongs to the VDR/retinoic acid receptor (RAR)/TR subfamily of RXR heterodimerizing species (Haussler et al. 1991), it is reasonable to draw from data on RAR and TR for comparison with VDR.

The greatest number of VDR natural mutations char¬ acterized to date are localized to the DNA binding, zinc finger region (Fig. 2). The first two discovered, G33D and R73Q (Hughes et al. 1988), reside at the ‘tips’ of the fingers and affect charge—charge interactions between VDR and the phosphate backbone of DNA. When viewed in toto, the zinc finger region mutations in HVDRR (Fig. 2) have the following two general prop¬ erties: (i) they occur in residues conserved across the entire nuclear receptor superfamily and (ii) most lie within -helices on the C-terminal side of the first and second fingers which are intimately involved in DNA base recognition and phosphate backbone contacts respectively (Rastinejad et al. 1995). These observations suggest that many of the clinically significant mutations in VDR which are still compatible with life may not greatly perturb the fundamental structure of the DNA binding domain of the receptor, but instead compromise its ability to recog¬ nize DNA with specificity and high affinity. Whether HVDRR cases with mutations in zinc finger region residues unique to VDR will be uncovered depends upon the properties of such alterations, which could range from innocuous to lethal.

Mutations located within the hormone binding domain of VDR also elicit the HVDRR phenotype (Fig. 2), including R274L (Kristjansson et al. 1993) and H305Q (Malloy et al 1995). Transcriptional activation by R274L and H305G VDR is attenuated as a result of inefficient l,25(OH)2D3 binding, ranging from severe in the case of R274L to a modest increase in Kd for H305Q. In both instances, transcriptional activation is restored when the dose of l,25(OH)2D3 is raised to pharmacologie levels (10 m) in transfection experiments (Kristjansson et al. 1993, Malloy et al. 1995). Our laboratory has recently characterized two novel VDR hormone binding domain mutations in HVDRR patients, I314S and R391C, that significantly affect the heterodimerization of VDR with RXR (Whitfield et al. 1996). Both of these C-terminal replacements (Fig. 2), however, do display some degree of what may be a hormone binding deficit, a phenomenon not observable in typical in vitro ligand binding kinetic assays at 4 °C. Thus, only at 37 °C in intact cells do R391C and I314S exhibit apparent slight and significant impairment of l,25(OH)2D3 high affinity retention respectively (Whitfield et al. 1996). Further, the two mutations in question are situated in or adjacent to heptad repeats (Fig. 2), hypothetical coiled-coil-like structures that were originally proposed to participate in the heterodimerization of VDR, RAR, and TR with RXR (Forman & Samuels 1990, Nakajima et al. 1994). Consist¬ ent with this concept, both R391C and I314S VDRs do not bind RXR with normal affinity when assayed in vitro, with the greatest impairment of heterodimerization occur¬ ring with R391C (affinity reduced by one order of magnitude) (Whitfield et al. 1996). Additional evidence supporting blunted RXR heterodimerization by these two mutant VDRs is provided by transfection experiments in restored to that of normal fibroblasts when fibroblasts from patients harboring either the R391C or the I314S mutation are cotransfected with exogenous RXR. Yet this apparent RXR rescue of the mutated VDRs requires approximately 10-fold elevated l,25(OH)2D3 doses com¬ pared with the response to hormone in normal fibroblasts (Whitfield et al. 1996). This latter observation reveals that the hormone binding and heterodimerization functions of VDR are not entirely separable, an aspect which is also apparent from fundamental biochemical analysis of the hormone dependency of VDR-RXR heterodimer binding to VDREs as discussed in detail below.

Understanding the molecular properties of natural VDR mutations in HVDRR allows us to comprehend why the patients respond differentially to therapy with massive doses of l,25(OH)2D3, or suitable analogs. For example, cases with zinc finger region aberrations are unresponsive to the hormone because DNA binding is precluded by the absence of structural complemen¬ tarity between VDR and the VDRE, regardless of the l,25(OH)2D3 liganding or heterodimerization of the receptor in solution. Conversely, patients harboring mutations in the hormone binding/heterodimerization domain can be responsive to pharmacologie doses of l,25(OH)2D3 or analogs, even though the hormone already is increased in the circulation because of the hypocalcemia caused by tissue insensitivity. For example, patient I314S was essentially cured by excess vitamin D metabolite, indicating that compensating for the hormone binding deficit was able to override the milder heterodimerization defect and allow sufficient VDRE binding by the VDR-RXR heterodimer. Conversely, patient R391C responded only modestly to treatment with excess l,25(OH)2D3 analog, presumably because the fundamental heterodimerization defect could not be overcome and therefore normal VDRE binding could not be achieved (Whitfield et al. 1996).

The final insights gained from the natural VDR mutations summarized in Fig. 2 are structural in nature. We have discussed previously that the zinc finger mutations are confined to absolutely conserved residues. In the crystal structure of the DNA binding domain heterodimers of RXRa and TRß (Rastinejad et al. 1995), the lysine and arginine residues corresponding to K45 and R50 in human VDR (hVDR) make direct base contacts with DNA, while the arginines corresponding to R73 and R80 in hVDR make direct DNA phosphate backbone contacts. That mutations in these four residues are clinically important in the etiology of HVDRR argues for structural congruity between the VDR finger region and that of TR. Rastinejad et al. (1995) have extended this assumption to include a modeling of RXR-TR vs RXRVDR bound to DNA which accommodates the fact that TR binds as a heterodimer to a direct hexanucleotide repeat spaced by four nucleotides (DR+4), while VDR binds as a heterodimer to a similar set of half elements spaced by three nucleotides (DR+3). In addition to verifying the common protein-DNA interfaces, their modeling predicts that hVDR residues N37 in the first finger and K91/E92 C-terminal of the second finger (see Fig. 2) engage in heterodimeric contacts with residues in the second zinc finger ofRXR to form effectively a stable, DNA-supported heterodimer. Indeed, recent site-directed mutational studies (Hsieh et al. 1995) indicate that the alteration ofK91 and E92 in hVDR in fact grossly reduces transactivation while moderately attenuating hetero¬ dimerization and DNA binding, thus confirming the importance of K91 and E92. An additional surprising finding was that the K91/E92 double mutant manifested dominant negative characteristics (Hsieh et al. 1995), distinguishing it from the natural HVDRR replacements discussed above. Apparently, the K91/E92 mutant VDR is able to bind DNA sufficiently through its native zinc finger and strong heterodimerization function in the ligand binding domain such that it can block binding by wild type receptor, but is rendered inactive in stimulating transcription because of a presumed conformational per¬ turbation initiated by unstable or improper alignment of the heterodimer on the VDRE.

Based upon recently reported X-ray crystal structures of the ligand binding domains of ligand-occupied hRARy (Renaud et al. 1995), agonist-occupied rat TRa, (Wagner et al. 1995) and unoccupied, but dimeric hRXRa (Bourguet et al. 1995), it is also possible to incorporate the HVDRR mutations in the hormone binding domain (Fig. 2) into a hypothetical structural context. Figure 3 constitutes a schematic compilation of the existing crystallographic data and compares them with natural and artificially generated mutations in hVDR. At the top of Fig. 3, the residue numbers for VDR in the ligand binding domain appear in relation to the older heptad repeat nomenclature (heptads 1—9, dotted boxes). At least some of these heptads, particularly heptads 4 and 9, are thought to facilitate heterodimerization (Nakajima et al. 1994). The El region is a highly conserved area that supports heterodimerization (Whitfield et al. 1995è). The helices depicted schematically in Fig. 3 (open boxes) are those determined for hRARy; this general pattern of -helices and ß-strands (solid boxes) appears to be well conserved across the TR, RAR and RXR members of the subfamily crystallized thus far (Bourguet et al. 1995, Renaud et al. 1995, Wagner et al. 1995). Although the heterodimerization domains have yet to be elucidated by structural analysis, the homodimerization domain of RXR is comprised of helices 7, 9 and 10 (Fig. 3 and Bourguet et al. 1995). Flanking the dimerization region are clusters of ligand binding contacts, shown for RAR and TR in Fig. 3, which paint a picture of hormone binding involving helices 3, 5, 11 and 12 plus portions of helices 6 and 7 along with their intervening loop, as well as the loop between ß-strands 1 and 2.

Figure 3 Hormone binding (R274L and H305Q) and heterodimerization (I314S and R391C) natural mutations in VDR that confer the HVDRR phenotype are positioned in the context of retinoid and thyroid hormone receptor subfamily ligand binding domain structures. See text for details and citations.

As summarized in Fig. 3 and discussed by Whitfield et al. (1995a, 1996), a number of artificially generated mutants in hVDR support the con¬ cept that the dimerization and honnone binding regions in VDR are well aligned with those in RXR, RAR and TR. Of even greater interest and relevance to the present monograph, the four clinically important hVDR mutants under consideration correspond to pertinent locations in the known structures of the retinoid and thyroid hormone receptor ligand binding domains. We postulate that this general structural organization represents that of the VDR ligand binding domain. As shown in Fig. 3, the pure hormone binding mutant hVDRs, namely R274L and H305Q, are located precisely within ligand clusters in helix 5 and in the loop between helix 6 and 7 respectively. I314S, which endows hVDR with combined defects in hormone retention and heterodimerization, lies within helix 7 at a presumed interface of ligand binding and dimerization activities of the receptor (Fig. 3). Finally, R391C is positioned well within the helix 10 dimerization surface, but not far removed from C-terminal ligand binding contacts that are likely influenced by replacement of this amino acid in hVDR. Thus, at least within the context of the assumed structural organization of VDR derived from that of other subfamily members, the I314S and R391C mutations are situated precisely where they would be predicted to lie, given the biological properties of the mutant receptors and the phenotype of the patients. These results not only have profound implications con¬ cerning the putative structure of VDR in relation to its closest relatives, but prove unequivocally that the calcémic actions of l,25(OH)2D3 are mediated by the vitamin D receptor, existing as a l,25(OH)2D3-liganded heterodimer with RXR that is bound to DNA.

Physiology and cellular actions of l,25(OH)2D3

In order to delineate the physiologic roles for the vitamin D hormone, it is appropriate first to place the VDR mediator into the context of vitamin D metabolism and cellular actions. Figure 4 summarizes the integration of vitamin D metabolism and cellular actions introduced in Fig. 1, with physiologic regulatory events now super¬ imposed on the metabolic pathway and the inclusion of an expanded list of physiologic actions for the 1,25( )2 4 hormone. The conversion of vitamin D3 to 25(OH)D3 by the liver is a constitutive metabolic step, followed by the 1-hydroxylation of25(OH)D3 to l,25(OH)2D3, a reaction under exquisite control (Haussler & McCain 1977). When blood calcium is low, activation of this latter step occurs, either as a result of the hypocalcémie state per se, or in response to elevated PTH, each of which serves indepen¬ dently to enhance renal 1-OHase activity. Low phosphate is also capable of separately upregulating the 1-OHase enzyme. To limit activation, the hormonal product, l,25(OH)2D3, effects an ultra-short feedback loop to suppress its own biosynthesis in the kidney and also represses PTH synthesis to remove the peptide hormone stimulus of the 1-OHase via a longer feedback loop (Fig. 4). However, the dominant negative feedback controls of 1-OHase activity appear to result from the concerted actions of l,25(OH)2D3 to stimulate bone mineral résorption and to promote intestinal calcium and phosphate absorption, which together elicit an increase in blood calcium and phosphate levels, each of which down-regulates the 1-OHase.

Figure 4 Vitamin D metabolism and cellulat actions, mediated by the VDR-RXR heterodimer binding to a VDRE

The process by which l,25(OH)2D3 causes bone remodeling is complex, involving stimulation of osteoclast differentiation and osteoblastic production of osteopontin, both of which activate résorption in part through the recognition of bone matrix osteopontin by osteoclast surface avß3-integrin. The résorption effect is supported by l,25(OH)2D3-elicited suppression of bone formation via the induction of osteocalcin and the repression of type I collagen. This latter insight that the normal function of osteocalcin is to curtail bone matrix formation arises from the creation of osteocalcin knockout mice (Ducy et al. 1996). In addition to stimulating the transcription of bone-related genes such as osteopontin and osteocalcin, the l,25(OH)2D3 hormone also induces its own eatab¬ olism in kidney as well as other target tissues like bone by enhancing the expression of the vitamin D-24-OHase enzyme. 24-Hydroxylation of l,25(OH)2D3 is the first step in deactivating the hormone, which is eventually metabolized by side chain cleavage to calcitróle acid (Haussler 1986). Thus, the synthesis of l,25(OH),D3 is not only governed by feedback mechanisms that sense l,25(OH)2D3, calcium, PTH and phosphate concentrations, but the hormone induces the termination of its own signal in target tissues, qualifying l,25(OH)2D3 as a bonafide hormone by any definition.

Figure 4 Vitamin D metabolism and cellulat actions, mediated by the VDR-RXR heterodimer binding to a VDRE

As introduced in the section on HVDRR, mediation of the cellular functions of l,25(OH)2D3 requires that VDR bind the hormonal ligand specifically and with high affinity (Fig. 4). Upon such binding, VDR becomes hyperphosphorylated (Jurutka et al. 1993, Haussler et al. 1994) and recruits RXR into a hetero¬ dimeric complex that binds strongly to DNA (Fig. 4). The l,25(OH)2D3-hganded RXR-VDR heterocomplex selectively recognizes VDREs in the promoter regions of positively controlled genes such as osteocalcin (MacDonald et al. 1991), osteopontin (Noda et al. 1990), vitamin D-24-OHase (Ohyama et al. 199A) and ß3-integrin (Cao et al. 1993). Negative VDREs (Haussler et al. 1995) exist in the 5′-regions of the genes for type I collagen (Pavlin et al. 199A), bone sialoprotein (Li & Sodek 1993), PTH (DeMay et al. 1992) and PTH-related peptide (Falzon 1996, Kremer et al. 1996). The mechanisms whereby VDR accomplishes positive and negative control of DNA transcription after VDRE association are not well under¬ stood, although substantial progress has been made in comprehending the stimulation of transcription as detailed in later sections of this article. Moreover, as summarized in Fig. 5, a number of VDREs have been definitively characterized. The prototypical VDBJS is found in the osteocalcin gene, consisting of an imperfect direct repeat of hexanucleotide estrogen responsive element (ERE)-like, half-sites with a spacer of three nucleotides (DR+3). Classic EREs possess a central GT core at positions 3 and 4 of the hexanucleotide, but this feature is only partially conserved in the six natural positive VDREs listed in Fig. 5. There is, however, absolute conservation of the A in position 6 of the 5′ half-element and of the G at position 2 of the 3′ half-element. A preliminary working consensus for the positive VDRE can be derived from these natural VDREs (see boxed sequence in Fig. 5). This generaliz¬ ation is supported, in part, by PCR experiments that were designed to select, from random oligonucleotides, the highest affinity DNA ligand for the RXR-VDR heterodimer (Nishikawa et al. 1994, Colnot et al. 1995).

Figure 5 Natural vitamin D responsive elements (DR+3s) in genes positively tegulated by l,25(OH)2D3. The consensus VDREs are based on either sequence comparisons (boxed) or a selection of random sequences (at bottom).

The random selection process yields an identical VDRE 5′ half-element of GGGTCA (Fig. 5, bottom), which is also a preferred RXR target when RXR homodimers bind to DNA (Yang et al. 1995). This observation is in concert with the conclusion (Jin & Pike 1996) that, with respect to association ofRXR-VDR with VDREs, RXR lies on the 5′ half-element whereas VDR is situated on the 3′ half-element. Examination of both consensus sequences suggests that the G at position 3 of the spacer is important in VDR binding, a deduction consistent with the finding (MacDonald et al. 1991) that this base is partially protected by RXR-VDR in methylation interference assays. How¬ ever, interesting differences arise when one compares the most frequently encountered 3′ half-element bases in natural VDREs, namely the GGGGCA composite which actually occurs in human osteocalcin, with the GGTTCA random consensus selection for the 3′ half-element (Fig. 5). Clearly, GGTTCA represents a potent VDR binding site, a supposition that is bolstered by the fact that osteopontin, which possesses a perfect DR+3 of GGTTCA, is the highest affinity VDRE we have tested (data not shown). Intriguingly, Ts at positions 3 and 4 in the 3′ VDR half-site occur infrequently in the balance of natural VDREs (Fig. 5). The paucity of Ts in the 3′ half-element could be related to a need for varying potency of VDREs in regulated genes, or may even provide for a repertoire of different VDR conformations that could be induced by contact with distinct 3′ half-site core sequences. This postulated range of VDR conforma¬ tions might endow the receptor with the ability to recruit a variety of different coactivators and corepressors, or even to favor the binding of one vitamin D metabolite ligand over another. Irrespective of the above considerations, it is evident that the primary VDRE is a DR+3 recognition site in DNA that directs the VDR to the promoter region of l,25(OH)2D3 regulated genes, ultimately altering the functions of target cells as a result of transcriptional control of gene expression.

Significance of lipophilic ligands in the association of RXR-VDR with DNA

Dimeric complexes are a feature commonly employed in the regulation of eukaryotic transcriptional systems. This process of protein dimerization often will generate novel heterodimeric complexes which display highly cooperative binding to DNA as well as an altered target sequence specificity (Glass 1994). Among the classical steroid hormone receptors, dimerization results in the formation of symmetrical homodimeric protein complexes on palindromic DNA half sites. Dimerization has been shown to be mediated in part by residues within the DNA binding domain of the receptor (Luisi et al. 1991) and is enhanced by residues within the ligand binding domain (Falwell et al. 1990). The other subfamily of nuclear hormone receptors, including VDR, TR and RAR, apparently binds with highest affinity to direct repeat elements either as homodimers or, more commonly, as heterodimers with RXR (Kliewer et al. 1992). In both subgroups of nuclear receptors, protein-protein interactions serve to align the DNA binding domains so that they are optimally positioned to bind to their specific DNA target sequences (Kurokawa et al. 1993, Perlmann et al. 1993, Rastinejad et al. 1995). The ligand binding region of these receptors is multifunctional, in that this domain not only binds the cognate ligand, but also it possesses a dimerization surface as well as the ligand-dependent transactivation function, AF-2 (Gronemeyer 1991, Chambón 1994). The dimerization surface consists of packed helices which are stabilized by hydrophobic heptad repeats interspersed throughout the structure. Ligand apparently can influence different functional components, including the dimerization interface, and the activating AF-2 domain (Renaud et al. 1995, Wagner et al. 1995). Therefore, a likely role for ligand is to regulate the association and dissociation of dimeric protein complexes and hence regulate specific binding to DNA target sequences.

In this regard the following three questions remain regarding l,25(OH)2D3-mediated control of positively regulated genes: (i) does VDR bind as a homodimer (Freedman et al. 1994, Nishikawa et al. 1994) as well as a heterodimer to DR+3 VDREs? (ii) What is the effect of the l,25(OH)2D3 ligand on VDR or VDR-RXR binding to VDREs? (iii) What role does 9-cis retinoic acid, the RXR ligand, play m RXR-VDR binding to VDREs and enhanced transcription of l,25(OH)2D3-responsive genes? It is generally accepted that TR forms homodimers as well as heterodimers with RXR on thyroid hormone responsive elements (TREs), although recent data suggest that the TR homodimer, when unoccupied by thyroid hormone, operates as a repressor of transcription (Chin & Yen 1996, Schulman et al. 1996). Thyroid hormone is proposed to dissociate TR homodimers to facilitate TRRXR heterodimerization on the TRE and stimulate transcription. In contrast, RAR does not appear to be capable of forming homodimers on DR+5 retinoic acid responsive elements (RAREs) (Perlmann et al. 1996), instead cooperating exclusively with RXR in RARE association and vitamin A metabolite-responsive transcrip¬ tion. When present in excess in gel mobility shift DNA binding assays in vitro, both TR and RAR display RXR heterodimeric association with their respective hormone responsive elements (HREs) in the absence of added lipophilic ligand. These in vitro studies are consistent with immunocytochemical data indicating that, unlike classic steroid honnone receptors that reside in the cytoplasm complexed with Hsp-90 and other proteins in their unoccupied state, unliganded TR, RAR and VDR (Clemens et al. 1988) exist in the nucleus in general association with DNA. These findings have led to the dogma that ligand is not required for TR, RAR and VDR to associate with target HREs. Indeed, we have observed that addition of 260 ng baculovirus-expressed hVDR to a gel shift reaction generates weak homodimeric VDR as well as strong VDR-RXR-heterodimeric binding to a rat osteocalcin VDRE probe, both of which are independent of the presence of l,25(OH)2D3 (Nakajima et al. 1994). However, in vivo footprinting experiments (Blanco et al. 1996, Chen et al. 1996) have led to the conclusion that, at least in the case of RAR-RXR heterodimers, RAR ligands are required for RARE binding. We, therefore, sought to devise an in vitro gel shift assay that would more accurately reflect the in vivo situation, primarily consisting of the use of physiologic salt (0-15 m KCl) concentrations and limited amounts of partially purified, baculovirusexpressed VDR and RXRs (Thompson et al. 1997). Utilizing this assay, we have addressed the three questions regarding VDR/RXR listed above, namely heterodimer versus homodimer, the potential role of l,25(OH)2D3 and the effect of 9-cis retinoic acid (9-cis RA).

When 20 ng VDR (~ 10 nM) or 20 ng VDR plus 20 ng RXR are incubated with either the rat osteocalcin or mouse osteopontin VDREs (see Fig. 5), no DNA-bound homodimeric VDR species is apparent, but a VDRE complexed VDR-RXR heterodimer occurs that is strik¬ ingly dependent upon the presence of the l,25(OH)2D3 ligand (Thompson et al. 1997). Thus, at receptor levels approaching that in a typical target cell, a VDR liganddependent heterodimer with RXR is the preferred VDRE binding species. Only when VDR or VDR plus RXR levels are raised to 100 ng of each receptor with the mouse osteopontin VDRE (Thompson et al. 1997), or 260 ng with the weaker rat osteocalcin VDRE (Nakajima et al. 1994), can faint homodimers of VDR bound to the probe be visualized. In addition, at these greater amounts ofreceptors, neither the VDR homodimer nor the VDRRXR heterocomplexes are modulated significantly by inclusion of l,25(OH)2D3 in the incubation (Thompson et al. 1997). We, therefore, conclude that higher receptor levels in vitro generate artifactual VDR homodimers as well as attenuate the normal physiological ligand dependence of VDR-RXR binding to the VDRE. To explain seemingly ligand-independent VDR-RXR association with the VDRE, we postulate the existence of a subpopulation of VDR that is unstably activated in the absence of l,25(OH)2D3 (Schulman et al. 1996) and therefore capable of heterodimerization to generate a positive gel mobility shift under conditions of vast receptor excess. In contrast, our physiologically relevant gel shift assay at <10nM receptor levels and 0-15 m KCl reflects the presumed in vivo events of ligand triggered heterodimerization (Blanco et al. 1996, Chen et al. 1996), and extends earlier in vitro data showing that l,25(OH)2D3 enhances VDRRXR complex formation (Sone et al. 1991, MacDonald et al. 1993, Ohyama et al. 1994).

Next, we tested the effect of 9-cis RA in this gel shift assay. A spectrum of data exists on the role of 9-cis RA in l,25(OH)2D3-stimulated transcription, including demon¬ stration of synergistic action with l,25(OH)2D3 (Carlberg et al. 1993, Schrader et al. 1994, Kato et al. 1995, Sasaki et al. 1995), negligible action (Ferrara et al. 1994), or an inhibitory effect (MacDonald et al. 1993, Jin & Pike 1994, Lemon & Freedman 1996). These marked differences likely result from varying transfection and ligand addition protocols, as well as cell and species specificity. Employing the physiological gel shift procedure with biochemically defined components, we obtained clear evidence that 9-cis RA is a potent inhibitor of l,25(OH)2D3-enhanced, VDR-RXR binding to VDREs such as osteocalcin, with dramatic attenuation by the retinoid occurring at concentrations as low as 10 m (Thompson et al. 1997). Previous gel shift data had also hinted at 9-cis RA inhibition (MacDonald et al. 1993, Cheskis & Freedman 1994), even though higher concentrations of 9-cis RA were utilized in these earlier studies. One somewhat puzzling finding, however, was that the suppressive effect of 9-cis RA seemed more pronounced in vitro than in transfected cells, where retinoid inhibition of l,25(OH)2D3-stimulated transcription is significant, but 50% or less in magnitude (MacDonald et al. 1993). This suggested that multiple pathways may exist for the assembly of the RXR-VDR heterocomplex in vivo. To probe for distinct routes of assembly, we varied the order of addition ofVDR, RXR, l,25(OH)2D3 and 9-cis RA in the gel shift assay for VDRE binding (Thompson et al. 1997). The results showed that 9-cis RA is a potent inhibitor of VDR-RXR heterodimerization on the VDRE in all situations except when VDR alone is preincubated with l,25(OH)2D3 followed by addition of RXR (Thompson et al. 1997). To explain these data, we have developed the model depicted in Fig. 6, which hypothesizes two alternative allosteric pathways for the interaction ofVDR-RXR with the VDRE.

Figure 6 Model of two different allosteric pathways for VDR-RXR-1,25(OH)2D3 binding to DNA.

In pathway A (Fig. 6), l,25(OH)2D3 occupies monomeric VDR, altering the conformation of the ligand binding domain such that it recruits RXR for heterodimeric binding to DNA and subsequent VDRE recognition. Importantly, we pos¬ tulate that previously occupied VDR conformationally influences RXR in the resulting heterodimer such that it is incapable of being liganded by 9-cis RA (pathway A, Fig. 6). This action to abolish RXR ligand responsiveness both silences the ability of 9-cis RA spuriously to trigger vitamin D hormone signal transduction, and prevents 9-cis RA from dissociating the RXR-VDR complex in order to divert RXR for retinoid signal transduction. On the other hand, as illustrated in pathway (Fig. 6), we propose that RXR exists in a different, 9-cis RA-receptive, allosteric state in most other circumstances, such as when present as a monomer, in an apoheterodimer with VDR, or even when the apoheterodimer of RXR and VDR is subsequently liganded with l,25(OH)2D3. This latter species of RXR-VDR-l,25(OH)2D3 (pathway B) is hypothesized to be fully competent in VDRE recognition, but the 9-cis RA binding function of the RXR partner has not been conformationally repressed, rendering this form sensitive to dissociation by 9-cis RA, which would then favor the formation of retinoid-occupied RXR homo¬ dimers. Therefore, unless VDR monomers are first occu¬ pied by l,25(OH)2D3 (pathway A), 9-cis RA can operate to divert or dissociate RXR and direct it to form RXR homodimers (pathway B). It is tempting to speculate that the l,25(OH),D3-liganded heterodimer in pathway A is more potent in transcriptional stimulation than the analogous species in pathway B, perhaps because the AF-2 function of the RXR partner is allosterically activated only in the former instance. The l,25(OH)2D3-occupied VDR-RXR in pathway has the advantage of flexible regulation because it is effectively a two-ligand switch. It likely occurs in vivo because, as stated above, the fact that 9-cis RA blunting significant but incomplete suggests that at least two populations of RXR-VDR heterodimers exist. Finally, when our model (Fig. 6) is compared with those for RXR-RAR and RXR-TR (Forman et al. 1995), it is evident that VDR is closer in mechanism of action to the TR, where 9-cis RA inhibits TR signal transduction by diversion of BJÍR (Lehmann et al. 1993). Also analogous is the fact that thyroid hormone occupation of the TR partner abolishes 9-cis RA binding to the RXR counter¬ part (Forman et al. 1995). Finally, the action of RXRPJ\R heterodimers seems to be fundamentally different from that of RXR-VDR in that RAR liganding by a retinoid facilitates RXR occupation by its retinoid ligand, resulting in cooperative stimulation of gene transcription by the repertoire of vitamin A metabolites.

VDR protein-protein interactions that effect gene transcription

Although we now have at least a rudimentary understand¬ ing of ligand-induced VDR binding to a VDRE, the next logical question is how does VDR regulate the machinery for gene transcription? In the basal state ofDNA transcrip¬ tion, the TATA-box binding protein (TBP) and its associated factors (TAFs) are bound to the TATA box at approximately position — 20 in the 5′ region of controlled genes, but the frequency of transcriptional initiations is very low because the RNA polymerase II-basal transcription factor IIB (TFIIB) enzyme complex is not stably associated with TBP-TAFs. The recruitment of the TFIIB-RNA polymerase II complex appears to be the rate limiting step in preinitiation complex formation, and is stimulated dramatically when a transacting factor or factors bind to upstream enhancers. In a process involving DNA looping, transactivators are thought to attract TFIIB and also interact with TAFs, forming a stable preinitiation complex that executes repeated rounds of productive transcription. Recent data indicate that the activation function in the hormone binding domain of the estrogen receptor, AF-2, associates specifically with a TAF known as TAFn30 (Jacq et al. 1994) and that the estrogen receptor (ER) binds to TFIIB in vitro (lng et al. 1992). In collaboration with Ozato and associates and Tsai and O’Malley, we have observed that hVDR also specifically associates with hTFIIB (Blanco et al. 1995). In this work, Blanco et al. (1995) showed that VDR binds to a TFIIB-glutathione S transferase fusion protein linked to glutathione-laden beads. Additionally, it was observed that both TRa and RARa interact with hTFIIB (Blanco et al. 1995), but that RXR does so only very weakly (P W Jurutka, L S Remus and M R Haussler, unpublished results). This last result suggests that, while the ligand binding partners in the VDR/TR/RAR subfamily provide a hard-wired connection to the assembly and en¬ hancement of the transcription machinery, the RXR partner is not primarily engaged in TFIIB contact.

Independent data obtained by MacDonald et al. (1995) using the powerful yeast two-hybrid system to detect protein-protein interactions also revealed that hVDR binds efficiently to TFIIB. Moreover, MacDonald et al. (1995) further exploited the yeast two-hybrid system to prove that, while hVDR and RXR interact, no homodimeric association occurs for hVDR alone, providing further evidence against the existence of physiologically significant VDR homodimers. Utilizing fusion protein technology, they also showed that VDR interacts directly with RXR to form a heterodimer in solution in the absence of DNA and, further, that this process was enhanced 8-fold by the presence of l,25(OH)2D3 hor¬ mone (MacDonald et al. 1995). Because hVDR-TFIIB association is not dependent upon the l,25(OH)2D, ligand (Blanco et al. 1995, MacDonald et al. 1995), the role of l,25(OH)2D3 can now be further resolved to an early participation in conforming VDR such that it attracts RXR followed by the targeting of the resulting RXR-VDR heterodimer to VDREs (see Fig. 6).

Figure 6 Model of two different allosteric pathways for VDR-RXR-1,25(OH)2D3 binding to DNA.

Interestingly, the presence of BJCR further facilitates VDR-TFIIB association, especially in the presence of l,25(OH)2D3 (PW Jurutka, LS Remus and MR Haussler, unpublished results). In fact, because of its capacity to enhance VDR-RXR heterodimerization, the l,25(OH)2D3 ligand is capable ofgenerating high levels of an RXR-VDR-TFIIB ternary complex in solution, sig¬ nificantly in excess ofthat occurring with either RXR and TFIIB or even with VDR and TFIIB (P W Jurutka, L S Remus and M R Haussler, unpublished results). These data not only reaffirm the interaction ofVDR with TFIIB, but also they imply that the l,25(OH)2D3-liganded VDR-RXR complex is the most efficient binder of TFIIB. This latter effect may be the result of positive conformational influences of RXR on liganded VDR, since VDR is the primary attachment moiety for TFIIB.

Because VDR-TFIIB interactions have been detected either in vitro or in the yeast system where certain mammalian cell restrictions may be relaxed, it was import¬ ant to confirm the relevance ofVDR-TFIIB association in mammalian cells. Blanco et al. (1995) have reported functional studies which, for the first time, show the interaction ofTFIIB with a member ofthe steroid receptor superfamily in ligand-dependent activation oftranscription in intact cells. In pluripotent PI9 mouse embryonal carcinoma cells, transfection of hVDR or hTFIIB alone produced no better than a 2-fold induction of VDREluciferase reporter expression by l,25(OH)2D3. However, when transfected together, hVDR and hTFIIB mediated a synergistic transcriptional response of approximately 30-fold when l,25(OH)2D3 was added, an effect which was absolutely dependent on the presence of the VDRE in the luciferase construct. It should be noted that the VDR-TFIIB positive cooperation appears to be cellspecific because similar experiments in contact-inhibited NIH/3T3 Swiss mouse embryo cells resulted in squelching of transcription by TFIIB. Therefore, in more differentiated cells, perhaps including osteoblasts or fibro¬ blasts, accessory coactivators may be present to modulate TFIIB or bridge between VDR and TFIIB.

In summary, VDR and TFIIB are hypothesized to exist in a multi-subunit transcription complex which also con¬ tains TAFs and/or coactivators that may be promoter- or tissue-specific. Further characterization of this complex will require the discovery of cell type and promoterspecific components via transfection and biochemical interaction studies. Ultimately, an in vitro transcription system must be devised which utilizes defined components to replicate faithfully l,25(OH)2D3-stimulated gene expression.

One subdomain of VDR that likely interacts with coactivators and/or basal transcription factors is the extreme C-terminus. We have previously shown that 403 hVDR, a truncated receptor that lacks the C-terminal 25 amino acids, binds l,25(OH),D3 ligand with reasonable affinity and heterodimerizes normally with RXR, but is devoid of transcriptional activity (Nakajima et al. 1994). These data suggest that VDR contains a transcriptional activation domain near its C-terminus.

Figure 7 The extreme C-terminal amino acid sequence compared across the nuclear receptor superfamily: VDR appears to share the ligand-dependent transcription activation function (AF-2). AR, androgen receptor; CR, glucocorticoid receptor; PR, progesterone receptor.

Indeed, as illustrated in Fig. 7, the region of VDR from residues 416 to 422 possesses a high degree of similarity to the analogous sequences in the entire nuclear receptor superfamily. One hallmark of this conserved sequence is the glutamic acid residue at position 420 of hVDR (Fig. 7) included in a consensus of (where cp=a hydrophobic amino acid) for this domain (Renaud et al. 1995, Wagner et al. 1995). Allowing for conservative replace¬ ments, it seems virtually certain that hVDR forms an amphipathic helix (corresponding to helix 12 in the other receptors) surrounding glutamic acid-420 that is analogous to the ligand-dependent activation function (AF-2) char¬ acterized for TR (Barettino et al. 199A), RAR (Renaud et al. 1995), RXR (Leng et al. 1995) and ER (Danielian et al. 1992). Although this AF-2 domain is capable of autonomously activating transcription (Leng et al. 1995), that such activity is modest may be because of the fact that the AF-2 region is proposed to operate in a liganddependent fashion, involving a structural rearrangement to reposition the AF-2 for both intramolecular and intermolecular protein—protein interactions. Specifically, based upon the crystal structure of unoccupied RXR (Bourguet et al. 1995) and liganded RAR (Renaud et al. 1995) and TR (Wagner et al. 1995), helix 12/AF-2 appears to protrude outward from the more globular ensemble of helices 1—11 in the absence ofligand, such that it is unable to interact efficiently with coactivator/transcription factor. Upon liganding, a conformational signal is then transmit¬ ted to helix 12 that causes it to fold back on helix 11 and attach to the face of the globular ligand binding domain. The pivoting of helix 12 seemingly accomplishes two feats that mediate ligand-activated transcription by the receptor: (i) closing of a ‘door’ on the channel through which the lipophilic ligand enters the internal binding pocket of the receptor, and (ii) locking helix 12 into a stable confor¬ mation that facilitates its interaction with coactivator/ transcription factor. Ligand binding contacts on or near helix 12 (see Fig. 3) probably are significant in maintaining this active positioning of helix 12, essentially trapping ligand in the binding pocket to effect more sustained transactivation events.

Figure 7 The extreme C-terminal amino acid sequence compared across the nuclear receptor superfamily: VDR appears to share the ligand-dependent transcription activation function (AF-2). AR, androgen receptor; CR, glucocorticoid receptor; PR, progesterone receptor.

In order to evaluate the relevance of the above proposed mechanism for VDR action, we (Jurutka et al. 1997) have altered E-420 and L-417 (see Fig. 7) individually to alanine residues, which preserves the putative -helical character of this region. The altered VDRs bind ligand near-normally, with only a mild increase (about 3-fold) in the Kd for the E420A receptor. Both E420A and L417A hVDRs also heterodimerize efficiently with RXR and associate with VDREs similarly to wild-type hVDR, yet their capacity for l,25(OH)2D3-stimulated transcription is abolished, even at high doses ofligand (Jurutka et al. 1997). These point mutations, therefore, identify a C-terminal AF-2 in VDR which corresponds to similar activation domains in other nuclear receptor superfamily members. Because VDR interacts with TFIIB, one of the first questions we asked was whether the VDR AF-2 consti¬ tutes a contact site for this basal transcription factor. Although some very preliminary evidence existed for an association between TFIIB and the C-terminus of hVDR (MacDonald et al. 1995), we observed that neither the E420A nor the L417A mutant VDRs are impaired in their interaction with TFIIB as probed with glutathione-S transferase—TFIIB fusion protein binding technology (Jurutka et al. 1997). Thus, the domain(s) of VDR that interfaces with TFIIB apparently lies elsewhere in the receptor, possibly in the N-terminal portion of the ligand-binding region (Blanco et al. 1995), in the hinge (MacDonald et al. 1995), or in the vicinity of the DNA-binding zinc fingers.

The present experiments with VDR are in concert with recent insight into the function of AF-2 in other nuclear receptors, which is to recruit coactivators of the type of steroid receptor coactivator-1 (SRC-1) (Oñate et al. 1995). A number of candidate coactivators have been isolated in addition to SRC-1 (Halachmi et al. 199A, Baniahmad et al. 1995, CavaiUes et at. 1995, Lee et al. 1995, Hong et al. 1996) and, in several cases, interaction with nuclear receptors requires intact AF-2 core regions (Baniahmad et al. 1995, CavaiUes et al. 1995). Moreover, AF-2 mutations act as dominant negative receptors, for example in the case of hRARy (Renaud et al. 1995). Indeed, we have observed that VDR AF-2 mutants E420A and L417A exhibit dominant negative properties with respect to transcriptional activation (Jurutka et al. 1997). Such AF-2 altered receptors are inactive transcriptionally, but can bind l,25(OH)2D3 ligand and heterodimerize normally on VDREs, the consequence being competition with wild-type VDR-RXR heterodimers for VDRE binding. These data argue that the AF-2 of the primary VDR partner in an RXR-VDR heterodimer is absolutely required for the mediation of l,25(OH)2D3-activated transcription, not only for its intrinsic activation potential, but also because of its presumed role in stabilizing the retention of l,25(OH)2D3 ligand in the VDR binding pocket.

Figure 6 Model of two different allosteric pathways for VDR-RXR-1,25(OH)2D3 binding to DNA.

What part, if any, is played by the AF-2 domain (Fig. 7) of the RXR ‘silent’ partner in the RXR-VDRl,25(OH)2D3 signal transduction pathway? To investigate this phenomenon, AF-2 truncated mutants of RXRa or RXRß were created and tested for their ability to function as dominant negative modulators of l,25(OH)2D3- stimulated transcription (Blanco et al. 1996). Because previous data with RXR-RAR control of gene expression seemed to indicate that the RXR AF-2 was dispensable (Durand et al. 1994), we were surprised to find that AF-2 truncated RXRs were potent dominant negative effectors of l,25(OH)2D3 action in transfected cells (Blanco et al. 1996). We, therefore, conclude that although the RXR ‘silent’ partner in VDR signaling apparently is not occupied by retinoid ligand (see Fig. 6), its AF-2 does play an active role in transcriptional stimulation. A similar conclusion has also been reached recently by two other groups studying RXR-RAR action (Chen et al. 1996, Schulman et al. 1996), with the use of RAR-specific ligands precluding ligand binding by the RXR partner. However, Schulman et al. (1996) have introduced a caveat to the above theory as they point out that AF-2-truncated RXRs in heterodimers become strong, constitutive binders of corepressors like the silencing mediators of retinoid and thyroid hormone receptors (SMRTs). Thus, an alternative explanation to an active coactivator-binding role for RXR AF-2 in heterodimers is that it plays a more passive role in excluding corepressors. In this latter scenario, truncation or point mutation (Schulman et al. 1996) of RXR AF-2 generates spurious corepressor binding rather than compromising coactivator contact. Only additional research into coactivator and corepressor associations of VDR-RXR heterodimers will resolve this issue.

General mechanism for vitamin D hormone action on transcription

In order to provide a working hypothesis for l,25(OH)2D3 action at the molecular level, we have developed the model illustrated in Fig. 8. It is based primarily on data from our laboratory and others studying 1,25(OH)2D3 and VDR, and it relies on the assumed similarities between VDR action and that of TR and RAR. VDR is proposed to exist in target cell nuclei, perhaps very weakly associ¬ ated with DNA, in a monomeric, inactive conformation with the C-terminal AF-2 domain extended away from the hormone binding cavity. Upon liganding with l,25(OH)2D3, VDR assumes an active conformation, with the AF-2 pivoted into correct position for both ligand retention and coactivator contact. In addition, the hormone facilitates interaction of VDR and RXR through a stabilized heterodimerization interface. In turn, 1,25(OH)2D3-occupied VDR may itselffunction as a kind of allosteric regulator of RXR, perhaps by conveying a confonnational signal through the juxtapositioned dimer¬ ization domains to induce the AF-2 ofRXR into an active conformation for coactivator binding. As discussed above (see Fig. 6), the joining of preliganded VDR and unliganded RXR apparently renders the RXR partner unresponsive to binding and either activation or dissocia¬ tion by 9-cis RA. Alternatively, if 9-cis RA encounters RXR monomer first (Fig. 8), or binds to RXR that is complexed with VDR in an apoheterodimer (Fig. 6), the retinoid is able to divert the RXR to generate homo¬ dimers and effectively blunt l,25(OH)2D3-driven transcription In the primary activation pathway pictured in Fig. 8, the RXR-VDR-l,25(OH)2D3 complex recognizes and targets the genes to be controlled through high affinity association with the VDRE in a gene promoter region. Coactivators that are presumed to bind to VDR and RXR AF-2 s are then postulated to link with TAFs/TBP, thereby looping out DNA 5′ of the TATA box. This series of events positions VDR such that it can independently recruit TFIIB to the promoter complex, a process that initiates the assembly of the RNA polymerase II holoenzyme into the preinitiation complex. Precedents exist for transcription factors independently attracting TFIIB, such as hepatocyte nuclear factor-4 (Malik & Karathanasis 1996), as well as for a sequential, two-step pathway for activator-stimulated transcriptional initiation (Struhl 1996, Stargell & Struhl 1996). Using the latter model as an analogy, the VDR activator would contact both TBP/ TAFs (via – coactivator bridges) and TFIIB in order to initiate RNA polymerase II holoenzyme assembly. The order of attachment of these two ‘arms’ of activation has not been determined but, at least, in the case of acidic activators, recruitment to the TATA element precedes interaction with components of the initiation complex (Stargell & Struhl 1996). It is of interest that the mechan¬ ism of l,25(OH)2D3 action depicted in Fig. 8 is not only essential for induction of bone remodeling and other vitamin D functions, but is also self-limiting via 24-OHase induction. In addition, these actions of l,25(OH),D, would be blunted under conditions within a cell where 9-cis RA concentrations dominate over those of l,25(OH)2D3.

Figure 8 Model for transcriptional activation by 1,25(OH)2D3 on the promoter of a target gene

The above described molecular mechanism whereby the vitamin D hormone controls gene expression requires further experimental evaluation. To advance our under¬ standing of the structure/function relationships in VDR, a physical characterization of the structure of VDR via X-ray crystallography will be required. Furthermore, in order to comprehend the genomic action of vitamin D in calcium homeostatic and other target cells, it will be necessary to elucidate the detailed involvement of various RXR isoforms, specific TAFs and novel coactivators/ corepressors that might influence the regulation of differ¬ ent vitamin D-controlled promoters. This information in its entirety should assist in determining the potential role for VDR and l,25(OH)2D3 in the pathophysiology of osteoporosis and other endocrine-related bone diseases.

References

Baniahmad C, Nawaz Z, Baniahmad A, Gleeson MAG, Tsai M-J & O’Malley BW 1995 Enhancement of human estrogen receptor activity by SPT6: a potential coactivator. Molecular Endocrinology 9 34-43.

Barettino D, Ruiz MdMV & Stunnenberg HG 1994 Characterization of the ligand-dependent transactivation domain of thyroid hormone receptor. EMBOJournal 13 3039-3049.

Blanco JCG, Wang I-M, Tsai SY, Tsai MJ, O’Malley BW, Jurutka PW, Haussler MR & Ozato 1995 Transcription factor TFIIB and the vitamin D receptor cooperatively activate ligand-dependent transcription. Proceedings of the National Academy of Sciences of the USA 92 1535-1539.

Blanco JCG, Dey A, Leid M, Minucci S, Park B-K, Jurutka PW, Haussler MR & Ozato 1996 Inhibition of ligand induced promoter occupancy in vivo by a dominant negative RXR. Genes to Cells 1 209-221.

Bourguet W, Ruff M, Chambón , Gronemeyer & Moras D 1995 Crystal structure of the ligand-binding domain of the human nuclear receptor RXR-a. Nature 375 377-382.

Cai Q, Hodgson SF, Kao PC, Lennon VA, Klee GG, Zinmeister AR & Kumar R 1994 Inhibition of renal phosphate transport by a tumor product in a patient with oncogeneic osteomalacia. New England Journal of Medicine 330 1645-1649.

Cao X, Ross FP, Zhang L, MacDonald PN, Chappel J & Teitelbaum SL 1993 Cloning of the promoter for the avian integrin ß3 subunit gene and its regulation by 1,25-dihydroxyvitamin D3. Journal of Biological Chemistry 268 27371-27380.

Carlberg C, Bendik I, Wyss A, Meier E, Sturzenbecker LJ, Grippo JF & Hunziker W 1993 Two nuclear signalling pathways for vitamin D. Nature 361 657-660.

more…

Interaction of the Retinol/Cellular Retinol-binding Protein Complex with Isolated Nuclei and Nuclear Components
GENE LIAU, DAVID E. ONG, and FRANK CHYTIL
Department of Biochemistry, Vanderbilt University School of Medicine, Nashville, Tennessee 37232
http://jcb.rupress.org/content/91/1/63.full.pdf

Retinol (vitamin A alcohol) is involved in the proper differentiation of epithelia. The mechanism of this involvement is unknown. We have previously reported that purified cellular retinol-binding protein (CRBP) will mediate specific binding of retinol to nuclei isolated from rat liver. We now report that pure CRBP delivers retinol to the specific nuclear binding sites without itself remaining bound. Triton X-100-treated nuclei retain the majority of these binding sites. CRBP is also capable of delivering retinol specifically to isolated chromatin with no apparent loss of binding sites, as compared to whole nuclei . CRBP again does not remain bound after transferring retinol to the chromatin binding sites. When isolated nuclei are incubated with [ 3H]retinol-CRBP, sectioned, and autoradiographed, specifically bound retinol is found distributed throughout the nuclei . Thus, CRBP delivers retinol to the interior of the nucleus, to specific binding sites which are primarily, if not solely, on the chromatin . The binding of retinol to these sites may affect gene expression.

Early histological studies have clearly shown that when animals become vitamin A deficient various epithelial tissues of these animals lose the ability to maintain proper differentiation (1) . However, providing retinol (vitamin A alcohol) to the animal permits tissue repair, with improperly differentiated cells rapidly replaced by normal cells (2) . This indicates that vitamin A has an essential role in cellular differentiation . The action of retinol appears to be mediated by a specific intracellular protein called cellular retinol-binding protein (CRBP). CRBP binds retinol with great avidity and specificity and has been detected in a number oftissues (3, 4) . Recently, CRBP was purified and partially characterized (5, 6) . It is distinct from the well-known serum retinol transport protein called retinol-binding protein (5, 7) . That CRBP plays an important role in the action of vitamin A is suggested by the following observations: It is found complexed with retinol in vivo (4, 8). It binds cis-isomers of retinol with a specificity that parallels the in vivo activity of these isomers (9), Finally, if retinol is first complexed with CRBP, the retinol can bind to the nucleus in a specific and saturable manner (10) . In this study we compare the interaction of the CRBP-retinol complex with isolated nuclei to its interaction with isolated chromatin and follow the fate of both the protein and the ligand . The nuclear binding sites for retinol were localized using autoradiography .

……

The experiments described here were designed to gain insight concerning the still unknown molecular mechanisms by which retinol exerts its effects on the differentiation of epithelia. Alterations in genomic expression appear to be induced in animals fed a retinol-deficient diet, as shown by changes in nuclear RNA synthesis observed in vivo (27-30) as well as in vitro (13) . A working hypothesis has been used that retinol, being a small molecule, might exert its action in a way similar to the accepted model for the mode of action of steroid hormones in differentiation . This model involves binding of the steroid hormone inside the target cell to a specific binding protein called a receptor . The resulting cytoplasmic ligand receptor complex, after undergoing a not fully understood conformational change, translocates to the nucleus . The receptor protein can then be detected in nuclear extracts by its ability to bind specifically the steroid hormone. The receptor steroid complex has been shown to interact with chromatin. Such interaction is believed to lead to an altered expression of the genome, which is the basis for the steroid hormone-induced differentiation (31) .

The steroid hormone model has been used profitably to investigate the mode of retinol action . Indeed, a specific binding protein for retinol, CRBP, was discovered to be present in many tissues (3) . Moreover, after purifying this protein to homogeneity, it was demonstrated that CRBP is able to deliver retinol to the nucleus in a specific manner (10) .

However, we report here a unique feature which appears to be distinct from the steroid hormone model. Using retinol CRBP complex in which the radioactive label is on the protein, we find that CRBP delivers retinol in a specific manner to the nucleus; the retinol associates with chromatin, but the protein itself does not remain bound. This conclusion is based on the observation that the radioactively-labeled protein is still able to deliver retinol inside the nucleus, but it cannot be recovered with the nucleus, in contrast to steroid hormone receptors.

The interaction of the specifically delivered retinol appears to be primarily with chromatin. The outer nuclear envelope is apparently not significantly involved in the interaction as Triton X-100-treated nuclei retain 70% of the retinol binding sites found in intact nuclei. It is still possible that the isolated chromatin and the Triton-treated nuclei contain some of the nuclear matrix and that it is actually the matrix which contains the specific binding sites for retinol. However, preliminary evidence indicates that the specific binding sites remain with a soluble chromatin preparation prepared by mild nuclease digest of nuclei rather than with the nuclear matrix. That the CRBP is necessary for delivering retinol to the nucleus is clearly documented by autoradiography. Free retinol, not bound to CRBP, binds nonspecifically to the nuclei, and to chromatin, and autoradiography shows indiscriminate localization of retinol in the lipid-rich nuclear membrane areas.

The data presented here invite the proposal that the retinolCRBP complex enters the nucleus in some manner which is apparently not dependent on the nuclear membrane. The complex then recognizes a limited number (generally an order of magnitude greater than for steroid hormones) of specific sites on the chromatin where the transfer of retinol from CRBP to these sites takes place. The sites were not detectable and may not be accessible if the retinol is free from CRBP. After the transfer CRBP does not remain associated with the specific sites . The functional significance of the specific interaction between retinol and chromatin remains to be demonstrated .

Folates are water-soluble B vitamins that act as one-carbon donors required for purine biosynthesis and for cellular methylation reactions. They are essential for de novo synthesis of nucleic acids, and thus for production and maintenance of new cells, particularly in rapidly dividing tissues such as bone marrow and intestinal epithelium (Kamen, 1997). Adequate dietary folate availability is especially important during periods of rapid cell division, such as during pregnancy and infancy. Folate deficiency has been associated with reduced erythropoiesis, which can lead to megaloblastic anemia in both children and adults (Ifergan and Assaraf, 2008). Deficiency of folate availability in pregnant women has been linked to neural tube defects, such as spina bifida, in children (Pitkin, 2007). This has prompted the application of folate supplementation schemes either as pills or via fortification of grain products with folates (Eichholzer et al., 2006). Folates have also been proposed to act as protective agents against colorectal neoplasia, although contradictory results have also been reported (Sanderson et al., 2007).

The availability of diet-derived folates is primarily determined by the rate of their uptake into the epithelial cells of the intestine, mediated by the proton-coupled folate transporter (PCFT, gene symbol SLC46A1), localized at the apical brush-border membranes of enterocytes (Subramanian et al., 2008a). PCFT is an electrogenic transporter that functions optimally at a low pH (Qiu et al., 2006;Umapathy et al., 2007). Despite being abundantly expressed in enterocytes, the second folate transporter, termed reduced folate carrier (RFC, gene symbolSLC19A1), has recently been shown not to play an important role in intestinal folate absorption (Zhao et al., 2004; Wang et al., 2005).

The human PCFT gene resides on chromosome 17, contains 5 exons, and is expressed as two prominent mRNA isoforms of 2.1 and 2.7 kilobase pairs (Qiu et al., 2006). Mutations in the PCFT gene have been associated with hereditary folate malabsorption, a rare autosomal recessive disorder (Qiu et al., 2006; Zhao et al., 2007). The PCFT protein is predicted to have a structure harboring 12 transmembrane domains (Qiu et al., 2007; Subramanian et al., 2008a). Although the transport function of PCFT has been studied extensively, relatively little is known about the regulation of PCFT gene expression. PCFT promoter activity has been shown possibly to be epigenetically regulated by its methylation status in human tumor cell lines (Gonen et al., 2008). Furthermore, both the PCFT mRNA expression levels and PCFT promoter activity positively correlate with the level of differentiation of colon-derived Caco-2 cells (Subramanian et al., 2008b).

In addition to its well known roles in regulating calcium homeostasis and bone mineralization, 1,25-dihydroxyvitamin D₃ (vitamin D₃), the biologically active metabolite of vitamin D, executes many other important functions, particularly in the intestine. For example, vitamin D₃ promotes the integrity of mucosal tight junctions (Kong et al., 2008). Many effects of vitamin D₃ are mediated via its action as a ligand for the vitamin D receptor (VDR; gene symbol NR1I1), a member of the nuclear receptor family of transcription factors (Dusso et al., 2005). VDR typically regulates gene expression by directly interacting with so-called direct repeat-3 (DR-3; a direct repeat of AGGTCA-like hexamers separated by three nucleotides) motifs within the target promoters, as a heterodimer with another nuclear receptor, retinoid X receptor-α (RXRα; gene symbol NR2B1) (Haussler et al., 1997). Genetic variants of VDR have been associated with inflammatory bowel disease (Simmons et al., 2000; Naderi et al., 2008). Similarly to folates, both VDR and its ligand vitamin D₃ have been proposed to be protective against intestinal neoplasia (Ali and Vaidya, 2007). Dietary folate intake has been suggested to regulate gene expression of the components of the vitamin D system, possibly via epigenetic control through the function of folates as methyl donors (Cross et al., 2006). Several intestinally expressed transporter genes, such as those encoding the multidrug resistance protein 1 and multidrug resistance-associated protein 2, have recently been shown to be induced by vitamin D₃ (Fan et al., 2009). We investigated whether vitamin D₃ regulates the expression of the PCFT gene, encoding a transporter crucial for intestinal folate absorption. The human well polarized enterocyte-derived Caco-2 cells exhibit many of the characteristics associated with mature enterocytes and were used here to investigate the effects of vitamin D₃ on PCFT gene expression and folate transport activity.

……..

Vitamin D₃ regulates the expression of its target genes primarily by acting as an agonistic ligand for its DNA-binding nuclear receptor VDR, although nongenomic actions by vitamin D₃ have also been described previously (Christakos et al., 2003;Dusso et al., 2005). VDR, an important regulator of differentiation and proliferation of enterocytes, typically activates gene expression by heterodimerizing with its nuclear receptor partner RXRα. VDR:RXRα heterodimers then directly bind to DR-3-like elements on the target genes. It should be noted that other modes of VDR-mediated regulation, either via direct interaction with other DNA-binding factors or through nongenomic actions, have also been reported (Dusso et al., 2005).

Here we demonstrate that VDR is a ligand-dependent transactivator of the humanPCFT gene, coding for a vital transporter for intestinal absorption of dietary folates. PCFT mRNA is also abundantly expressed in the liver (Qiu et al., 2006). However, VDR is expressed at very low levels in primary human hepatocytes or hepatocyte-derived cell lines (Gascon-Barre et al., 2003; data not shown), suggesting that VDR-mediated regulation of the PCFT gene may not occur in hepatocytes.

Endogenous PCFT mRNA levels were induced by vitamin D₃ in a dose-dependent manner in Caco-2 cells (Fig. 1A). This increase was not further enhanced by cotreatment of cells with the RXRα ligand 9-cis retinoic acid (data not shown), consistent with a previous report that VDR:RXRα heterodimers, at least in some promoter contexts, may not respond to RXRα ligands (Forman et al., 1995). Alternatively, saturating levels of RXRα ligands may already be endogenously present in cells in these experimental conditions. In transient transfection assays, the PCFT promoter fragment −2231/+96 exhibited significant response to exogenous expression of VDR alone in the presence of its ligand (Fig. 2), most probably supported by endogenously expressed RXRα in Caco-2 cells.

Supporting the importance of the VDR:RXRα heterodimer formation for PCFTpromoter regulation, the luciferase values were further significantly elevated upon exogenous expression of RXRα. Exogenous expression of VDR in the absence of vitamin D₃ did not notably influence the activity of the PCFT(−2231/+96) promoter, indicating ligand-dependence of VDR action. In deletional transfection analysis, the strongest induction in response to VDR and RXRα in the presence of their ligands was achieved with the PCFT(−2231/+96) promoter fragment (Fig. 3A). Induction of the shortest deletion variant tested [PCFT(−843/+96)luc] was approximately 50% of that achieved for the PCFT(−2231/+96), indicating that this more proximal region is likely to contain further DNA elements mediating a response to vitamin D₃. However, in our current study, we focused on the distal region between the nucleotides −2231 and −1674 upstream of the transcriptional start site of the human PCFT gene, which confers maximal response to vitamin D₃. In our computational analysis, we identified a putative VDRE within the PCFTpromoter region between nucleotides −1694 and −1680. We have not so far been successful in identifying further binding sites for the VDR:RXRα heterodimer in the more proximal region of the PCFT promoter. It may be that, in addition to direct DNA-binding to the PCFT(−1694/−1680) element identified here, VDR may also affect PCFT promoter activity indirectly, via interactions with other DNA-binding factors. For example, it has been proposed that the p27^Kip1 gene is regulated by VDR via response elements for unrelated DNA-binding transcription factors Sp1 and NF-Y (Huang et al., 2004).

Both endogenously expressed and recombinant VDR and RXRα bound to thePCFT(−1694/−1680) element specifically and as obligate heterodimers (Fig. 4). The interaction between VDR and this region of the PCFT promoter within living cells treated with VDR and RXRα ligands was confirmed by chromatin immunoprecipitation tests (Fig. 5). Heterologous promoter assays proved that thePCFT(−1694/−1680) element can function as an independent VDR response element. The significant decrease in VDR:RXRα-mediated induction upon mutagenesis of the PCFT(−1694/−1680) element confirmed that it is an important functional mediator of the effect (Fig. 6, A and B).

Although we observed vitamin D₃-mediated increase of rat Pcft mRNA expression ex vivo (Fig. 1C), the rat Pcft promoter (chromosome 10; GenBank accession number NW_047336) exhibits no significant overall homology with the humanPCFT promoter over the proximal 3000-bp regions. This suggests that despite the divergence of the promoter sequences between human and rodent PCFT/Pcftgenes, the functional response to vitamin D₃ is conserved.

The activation of PCFT gene transcription by VDR also translates into an increase in PCFT protein function. Vitamin D₃ treatment of Caco-2 cells led to significantly increased uptake of folate across the apical membrane, in a dose-dependent manner (Fig. 7). In keeping with the fact that PCFT strongly prefers an acidic milieu for its transport function (Qiu et al., 2006; Nakai et al., 2007; Unal et al., 2009), we only observed vitamin D₃-stimulated transport activity at pH5.5, but not at neutral pH. These data strongly suggest that vitamin D₃-mediated transcriptional activation of PCFT gene expression leads to an increase of PCFT transport function. Consistent with our model, mRNA expression of the other known folate carrier expressed in Caco-2 cells, RFC, which functions efficiently at neutral pH (Ganapathy et al., 2004; Wang et al., 2004), was not affected by vitamin D₃treatment (Fig. 1B). It has been reported that vitamin D₃-induced gene expression increases as Caco-2 cells differentiate (Cui et al., 2009). Thus, our current findings on VDR-mediated regulation of PCFT expression provide a possible molecular mechanism for a prior observation that folate uptake into Caco-2 cells is enhanced upon confluence-associated differentiation (Subramanian et al., 2008b).

Our results suggest that intestinal folate absorption may be enhanced by an increase in dietary vitamin D₃ intake. Food products are often supplemented with folates, because of their proposed beneficial health effects. Based on our current study, supplementation of vitamin D₃ may enhance the intestinal absorption of folates. PCFT also transports the antifolate drug methotrexate (MTX) (Inoue et al., 2008; Yuasa et al., 2009) widely used in the treatment of autoimmune diseases and cancer. MTX interferes with folate metabolism by competitively inhibiting the enzyme dihydrofolate reductase. Our results may further suggest a potential mechanism to increase intestinal absorption of MTX via simultaneous treatment with vitamin D₃, thereby affecting the bioavailability of MTX. Patients suffering from inflammatory bowel disease are frequently on long-term treatment with calcium and vitamin D₃ as a prophylaxis against osteopenia and osteoporosis (Lichtenstein et al., 2006). This patient group is frequently treated with folates (in the case of folate deficiency) or MTX (as a second-line immunosuppressant) (Rizzello et al., 2002). MTX therapy per se requires prophylactic administration of folates, and these patients often receive additional calcium/vitamin D₃. Our current results may warrant a closer investigation into potential drug-drug interactions between pharmacologically administered vitamin D₃, MTX, and folates. Taking into account the previous report that folates regulate the expression of genes involved in vitamin D₃ metabolism, it may be that folate and vitamin D₃ homeostasis are closely interlinked through such mutual regulatory interactions.

Innate immune response and Th1 inflammation
http://mpkb.org/home/pathogenesis/innate_immunity

Nuclear receptors are a class of proteins found within the interior of cells that are responsible for sensing the presence of hormones and certain other molecules. A unique property of nuclear receptors which differentiate them from other classes of receptors is their ability to directly interact with and control the expression of genomic DNA. Some of the molecules (or ligands) which bind the nuclear receptor activate (agonize) it and some inactivate (antagonize) it.

It is commonly accepted that most ligands, approximately 95% to 98%, inactivate the nuclear receptors. Since the nuclear receptors play a significant role in the immune response, this factor alone may explain why so many drugs and substances found in food and drink are immunosuppressive.

Because the expression of a large number of genes is regulated by nuclear receptors, ligands that activate these receptors can have profound effects on the organism. Many of these regulated genes are associated with various diseases which explains why the molecular targets of approximately 13% of FDA approved drugs are nuclear receptors.⁶

Different cell types have different nuclear receptors. One of the nuclear receptors seen in immune cells is the Vitamin D Receptor (VDR). The VDR has two endogenous or “native” ligands, which are also the two main forms of vitamin D in the human body: 25-hydroxyvitamin D (25-D) and 1,25-dihydroxyvitamin D (1,25-D). Non-native or exogenous ligands can also inactivate or activate a nuclear receptor, depending on its molecular structure.

Ligands compete to dock at nuclear receptors. When is a given kind of ligand such as 25-D as opposed to 1,25-D more likely to bind to the VDR? It depends. 1,25-D tends to be much less common than 25-D – by a factor of 1,000 or more – so it binds to the receptor much more infrequently. A greater concentration of a given molecule can displace competing molecules off the nuclear receptor. Affinity occurs in logarithmic fashion, which is to say that it operates on the basis of a sliding scale. In short, an increase in 1,25-D and a decrease in 25-D can tilt the odds in favor of 1,25-D, and vise versa.

Affinity as well as the question of whether a ligand inactivates or activates a nuclear receptor can all be validated using in silicomodeling. Although less precise, it is also possible to measure these properties in vitro.

Activated by 1,25-D and inactivated by 25-D, the Vitamin D nuclear receptor (VDR) transcribes a number of genes crucial to the function of the innate immune response.

Role of Vitamin D Receptor in innate immunity

Th2 inflammation

Role of Dihydroxyvitamin D3 and Its Nuclear Receptor in Novel Directed Therapies for Cancer

S. Ondková, D. Macejová and J. Brtko
Gen. Physiol. Biophys. (2006), 25, 339—353   http://www.gpb.sav.sk/2006_04_339.pdf

Dihydroxyvitamin D3 is known to affect broad spectrum of various biochemical and molecular biological reactions in organisms. Research on the role and function of nuclear vitamin D receptors (VDR) playing a role as dihydroxyvitamin D3 inducible transcription factor belongs to dynamically developing branches of molecular endocrinology. In higher organisms, full functionality of VDR in the form of heterodimer with nuclear 9-cis retinoic acid receptor is essential for biological effects of dihydroxyvitamin D3. This article summarizes selected effects of biologically active vitamin D3 acting through their cognate nuclear receptors, and also its potential use in therapy and prevention of various types of cancer.

……

Vitamin D family consists of 9,10-secosteroids which differ in their side-chain structures. They are classified into five forms: D2, ergocalciferol; D3, cholecalciferol; D4, 22,23-dihydroergocalciferol; D5, sitosterol (24-ethylcholecalciferol) and D6, stigmasterol (Napoli et al. 1979). The main forms are vitamin D2 (ergocalciferol: plant origin) and vitamin D3 (cholecalciferol: animal origin). Both 25-hydroxyvitamin D2 and 1α,25-dihydroxyvitamin D2 have been evaluated for their biological functions. Vitamin D itself is a prohormone that is metabolically converted to the biologically active metabolite, 1,25-dihydroxyvitamin D3 in kidney. This vitamin D3, currently considered a steroid hormone, activates its cognate nuclear receptor (vitamin D receptor or VDR) which alter transcription rates of the target genes responsible for its biological responses. In general, vitamin D is essential for mineral homeostasis, for absorption and utilization of both calcium and phosphate and it aids in the mobilization of bone calcium and maintenance of serum calcium concentrations. Through these function, it plays an important role in ensuring proper functioning of muscles, nerves, blood clotting, cell growth and energy utilization. It has been proposed that vitamin D is also important for insulin and prolactin secretion, immune and stress responses, melanin synthesis and for differentiation of skin and blood cells (Lips 2006). Vitamin D metabolites also play a role in the prevention of auto-immune diseases and cancer (Pinette et al. 2003; Dusso et al. 2005). The steroid hormone 1α,25-dihydroxyvitamin D3 (calcitriol) exerts biological responses by interaction with both the well-characterized nuclear receptor (VDRnuc) responsible for activation gene transcription and not fully characterized membrane-associated protein/receptor (VDRmem) involved in generating a variety of rapid, non-genotropic responses (Evans 1988; Norman et al. 2002).

Vitamin D metabolism

Vitamin D, the “sunshine” vitamin, is synthesized under the influence of ultraviolet light in the skin. Many mammals have provitamin D (7-dehydrocholesterol) which is converted to provitamin D3 in their skin. When human skin is exposed to sunlight, the UV-B photons (wavelengths 290–315 nm) interact with 7-dehydrocholesterol causing photolysis and cleavage of the B-ring of the steroid structure, which upon thermoisomerization yields a secosteroid. Thus, provitamin D3 which is inherently unstable rapidly converts by a temperature-dependent process to vitamin D3 (MacLaughlin et al. 1982; Holick 1994). Vitamin D3 enters the blood circulation and binds to vitamin D binding protein (DBP) (Haddad et al. 1993) which carries vitamin D3 to liver and kidney for bioactivation (Wikvall 2001). In the first activation step, vitamin D3 is hydroxylated by the enzyme 25-hydroxylase to 25- hydroxyvitamin D3 mainly in the liver. This metabolite is present in the circulation at the concentration of more than 0.05 µmol/l (20 ng/ml). In the second step, the biologically active hormone 1α,25-dihydroxyvitamin D3 is generated by hydroxylation of 25-hydroxyvitamin D3 at 1α-position in kidney. The enzyme 1α-hydroxylase has been shown to be also present in keratinocytes and prostate epithelial cells, suggesting that those organs may also be able to generate 1α,25-dihydroxyvitamin D3 from 25-dihydroxyvitamin D3 (Schwartz et al. 1998). The activity of 1α-hydroxylase in the kidney serves as the major control point in production of the active hormone. The active metabolite 1α,25-dihydroxyvitamin D3 is present in human plasma at the concentration ranging from 0.05 to 0.15 nmol/l (20–60 pg/ml) (Hartwell et al. 1987; Gross et al. 1996). In general, 90 to 100% of the most human being vitamin D requirement comes from exposure to sunlight (Holick 2003) and the rest of the vitamin D3 content is obtained from diet (Malloy and Feldman 1999). The catabolism of vitamin D occurs by further hydroxylation of 25-dihydroxyvitamin D3 by 24-hydroxylase to yield 24,25-dihydroxyvitamin D3. The 24-hydroxylase is ubiquitous enzyme and is expressed in all the cells expressing VDR. This enzyme is regulated by parathyroid hormone and 1α,25-dihydroxyvitamin D3. The major significance of 24-hydroxylation is inactivation of vitamin D (Nishimura et al. 1994; Brenza and DeLuca 2000). The combinations of 1,25-dihydroxyvitamin D3 with inhibitors of 24-hydroxylase such as ketoconazole or liarozole may enhance its antitumour effects in prostate cancer therapy.

Vitamin D3 receptor

More than 2000 synthetic analogues of the biological active form of vitamin D, 1α,25-dihydroxyvitamin D3, are presently known. Basically, all of them interfere with the molecular switch of nuclear 1α,25-dihydroxyvitamin D3 signalling, which is the complex of the VDR, the retinoid X receptor (RXR), and a 1α,25-dihydroxyvitamin D3 response element (VDRE) (Carlberg 2003).

VDR is the only nuclear protein that binds the biologically most active vitamin D metabolite, 1α,25-dihydroxyvitamin D3, with high affinity (Kd = 0.1 nmol/l). This classifies the VDR into the classical endocrine receptor subgroup of the nuclear receptor superfamily, which also contains the nuclear receptors for hormones as retinoic acid, thyroid hormone, estradiol, progesterone, testosterone, cortisol, and aldosterol (Carlberg 1995). Similarly, like other biologically active ligand for nuclear hormone receptors, 1,25-dihydroxyvitamin D3 can modulate expression of selected ion transport protein genes (Van Baal et al. 1996; Hudecova et al. 2004).

The VDR was first isolated after trancfection of COS-1 cells with cloned sequences of complementary DNA that was isolated from human intestine (Baker et al. 1988). VDR has been found in more than 30 tissues including intestine, colon, breast, lung, ovary, bone, kidney, parathyroid gland, pancreatic β-cells, monocytes, keratinocytes, and many cancer cells, suggesting that the vitamin D endocrine system may also be involved in regulating the immune systems, cellular growth, differentiation and apoptosis (Jones et al. 1998). The active form of vitamin D binds to intracellular receptors that then function as transcription factors to modulate gene expression. Like the receptors for other steroid hormones and thyroid hormones, the VDR has specific hormone-binding and DNA-binding domains. It contains two zinc finger structures forming a characteristic DNA-binding domain (DBD) of 66 amino acids and a carboxy-terminal ligand-binding domain (LBD) of approximately 300 amino acids, which is formed by 12 α-helices. Ligand binding causes a conformational change within the LBD, in which helix 12, the most carboxy-terminal α-helix, closes the ligand-binding pocket via a “mouse-trap like” intramolecular folding (Moras and Gronemeyer 1998). Moreover, the LBD is involved in a variety of interactions with nuclear proteins, such as other nuclear receptors, corepressor and coactivator proteins. These ligand-triggered protein-protein interactions are the central molecular event of nuclear 1α,25-dihydroxyvitamin D3 signalling.

….

Role of vitamin D3 in cancer

Some of biologically active ligands for nuclear receptors exert tumour-suppressive activity, and they have therapeutical exploitation due to their antiproliferative and apoptosis-inducing effects (Brtko and Thalhamer 2003).

…..

During the last decade, evidence for vitamin D3 effects has been accumulating not only for prostate cancer (Feldman et al. 1995; Ma et al. 2004) but also for colon cancer (Cross et al. 1997; Bischof et al. 1998). 1α-hydroxylase was found to be Vitamin D and Cancer Treatment 347 expressed and active in colorectal cancer (Bareis et al. 2001; Cross et al. 2001; Tangpricha et al. 2001; Ogunkolade et al. 2002) and ovarian cancer (Miettinen et al. 2004). In both colon and also lung tumours, CYP24A1 mRNA was significantly up-regulated, while VDR mRNA was generally down-regulated when compared to respective normal tissues. When the level of VDR in 12 malignant colonic tumours was compared with that of adjacent normal tissue, in 9 cases out of 12, expression of VDR in tumours was decreased. However, in that study, the expression of CYP24A1 was not assessed. It has also been shown that, at least in human colon cancer cell lines, the level of VDR correlates with the degree of cell differentiation (Shabahang et al. 1993; Anderson et al. 2006).

Recently, it has been suggested that actually 20–30% of colorectal cancer incidence might be due to insufficient exposure to sunlight. This fact was strengthen by correlation between reduced colorectal cancer incidence and sunlight exposure, low skin pigmentation, nutritional vitamin D intake and high serum levels of 25- hydroxyvitamin D3 (Grant and Garland 2003). In the colon at least, CYP27B1 and VDR expression was described to be actually elevated during early tumour progression and that described dual positivity was found in many, but not all the tumour cells. In human colon tumours, CYP24 mRNA is quite highly expressed and the studies also demonstrated that with the exception of differentiated Caco-2 cells, CYP24 activity is constitutively present or can be induced by 1α,25- dihydroxyvitamin D3. During tumour progression in the colon, not only VDR but CYP27B1 and CYP24 expression were found to be increased in tumour tissues (Bareis et al. 2001; Bises et al. 2004).

Androgens, retinoids, glucocorticoids, estrogens and agonists of peroxisome proliferator-activated receptor directly or indirectly have reasonable impact on vitamin D signalling pathways, and vice versa. It was proposed that sex hormones might reduce colorectal cancer risk (McMichael and Potter 1980). The studies suggested that current and long-term use of estrogens is associated with a substantial decrease in risk of fatal colon cancer. The mechanism, however, by which estrogens could inhibit colonic tumour growth, remains an enigma. There are at least two distinct estrogen receptors in the human body: ERα and ERβ. In the normal human colon, ERβ is widely regarded to be the predominant subtype (CampbellThompson et al. 2001). In a recently terminated pilot study together with Strang Cancer Prevention Centre at Rockefeller University (NY, USA), tissues from postmenopausal women receiving 17β-estradiol for expression of CYP27B1 by real time RT-PCR were examined. CYP27B1 was found to be elevated significantly in all subjects after receiving 17β-estradiol for 4 weeks.

Amplification of chromosomal region 20q12q13 containing the CYP24A1 gene has been reported in ovarian cancer, as well (Tanner et al. 2000). Although inhibition of ovarian cancer cell growth by 1α,25-dihydroxyvitamin D3 has been reported (Saunders et al. 1992, 1995), a clinical trial testing the efficacy of 1α,25- dihydroxyvitamin D3 combined with isotretinoin in treating 22 epithelial ovarian cancer patients for 74 weeks has not produced positive results (Rustin et al. 1996).

Read Full Post »