Functionally relevant polymorphisms in the human nuclear vitamin D receptor gene

Abstract

The functional significance of two unlinked human vitamin D receptor (hVDR) gene polymorphisms was evaluated in twenty human fibroblast cell lines. Genotypes at both a Fok I restriction site (F/f) in exon II and a singlet (A) repeat in exon IX (L/S) were determined, and relative transcription activities of endogenous hVDR proteins were measured using a transfected, 1,25-dihydroxyvitamin D₃-responsive reporter gene. Observed activities ranged from 2–100-fold induction by hormone, with higher activity being displayed by the F and the L biallelic forms. Only when genotypes at both sites were considered simultaneously did statistically significant differences emerge. Moreover, the correlation between hVDR activity and genotype segregated further into clearly defined high and low activity groups with similar genotypic distributions. These results not only demonstrate functional relevance for both the F/f and L/S common polymorphisms in hVDR, but also provide novel evidence for a third genetic variable impacting receptor potency.

Introduction

The biological actions of 1,25-dihydroxyvitamin D₃ [1,25(OH)₂D₃] are mediated largely, if not entirely, by the vitamin D receptor (VDR), a member of the superfamily of nuclear hormone receptors (Whitfield et al., 1999). This protein is found in tissues known to play a role in calcium homeostasis, and also in numerous other tissues, where it appears to regulate a variety of processes, including cell proliferation and differentiation (Haussler et al., 1998). The significance of the nuclear VDR in calcium homeostasis, as well as in certain differentiation and proliferation processes in skin and uterus, has been confirmed by gene knockout studies in mice (Li et al., 1998, Kato et al., 1999).

A simplified diagram that illustrates how nuclear hVDR mediates transcriptional activation by the 1,25(OH)₂D₃ hormone is shown in Fig. 1A. The key features of this model are: (i) liganding of nuclear VDR by 1,25(OH)₂D₃; (ii) recruitment by 1,25(OH)₂D₃-VDR of its retinoid X receptor (RXR) heteropartner that, in turn, facilitates high-affinity interaction of the dimeric complex with vitamin D responsive elements (VDREs) upstream of target genes; (iii) attraction by VDR of basal transcription factor IIB (TFIIB), the rate-limiting component of the transcription preinitiation complex; and (iv) recruitment by the heterodimer of a number of transcription coactivators, some with histone acetyl transferase (HAT) activity to modify nucleosome/chromatin organization, such as SRC-1 (Gill et al., 1998), and others like the DRIPs (Rachez et al., 1999) that target the VDR supercomplex to the TATA-box/TBP and RNA polymerase II transcription initiation machinery. The net result of this 1,25(OH)₂D₃-triggered response is the regulation of genes coding for proteins that carry out intestinal calcium absorption, bone remodeling, cell differentiation, etc. (Jurutka et al., 2001).

A modular diagram of the functional domains within the hVDR protein is presented in Fig. 1B. The details of the hVDR subdomain arrangement (see figure legend) basically follow the general pattern for the subfamily of nuclear receptors that heterodimerize with RXR, such as the all-trans retinoic acid receptors (RARs) and the thyroid hormone receptors (TRs) (Whitfield et al., 1999). For the purposes of the present communication, the most relevant regions of hVDR are the hormone-binding domain, encoded by exons VI–IX of the human gene (see also Fig. 2), the DNA binding domain/zinc fingers, encoded by exons II–IV, and a set of discontinuous transactivation domains, including regions at the N-terminus (for TFIIB docking) (Jurutka et al., 2000), and in helices 3 and 12 (for coactivator recruitment). Since transactivation is the ultimate biochemical action of the liganded VDR and depends on all of the other capabilities of the receptor (ligand binding, nuclear localization, heterodimerization and VDRE/DNA binding), the present study focuses on this parameter of receptor activity in order to probe for functional significance of hVDR gene polymorphisms.

The chromosomal gene for VDR has been cloned (Miyamoto et al., 1997), and several common genetic variants have been described in humans, most of which are identified by a biallelic variation in a restriction endonuclease site (Fig. 2). Genetic variation in the 3′ region of the hVDR gene is observed in specific intronic sites for Bsm I (Morrison et al., 1992) and Apa I (Faraco et al., 1989), a silent Taq I site in exon IX (Morrison et al., 1992), as well as in a singlet(A) repeat in the portion of exon IX encoding the 3′ UTR (Ingles et al., 1997a) (see Fig. 2, right). All of these variations near the 3′ end of the gene are in linkage disequilibrium (Morrison et al., 1992, Verbeek et al., 1997), although this linkage is weaker in some ethnic groups such as African-Americans (Ingles et al., 1997a). Interestingly, none of these polymorphisms affect the structure of the VDR protein itself, although the singlet(A) repeat in the 3′ UTR is expressed in the mature mRNA for hVDR. Singlet(A) variants are classified according to length by the number of consecutive A's in the repeat, with ≥17 A's scored as ‘long’ (L), and ≤15 A's considered ‘short’ (S).

Another polymorphic site has been found in exon II near the center of the hVDR gene (Saijo et al., 1991). This site, which is genetically unlinked to the above Bsm/Apa/Taq/singlet(A) cluster, is unique among common hVDR variants described so far, in that it results in an alteration of the hVDR protein structure (Fig. 2, bottom center). Presence of the Fok I site (designated f) predicts that a 427-residue VDR protein will be produced beginning at Met-1 (M1 according to the numbering scheme of Baker et al. (1988), whereas absence of this site (denoted F) dictates translation from Met-4 (M4), producing a protein of 424 amino acids (Arai et al., 1997).

In an initial report (Morrison et al., 1994), allelic variation in the chromosomal gene for the vitamin D receptor was proposed to represent a major part of the genetic predisposition for low bone mineral density (BMD), and perhaps for osteoporosis and/or skeletal fractures, although these associations have been disputed by other studies [reviewed in (Wood and Fleet, 1998)]. More recently, correlations have been reported between VDR allelic variants and risk of prostate cancer (Ingles et al., 1997b, Watanabe et al., 1999), breast cancer (Ingles et al., 1997c, Ruggiero et al., 1998, Curran et al., 1999), sporadic primary hyperparathyroidism (Correa et al., 1999), and sarcoidosis (Niimi et al., 1999). However, conflicting reports have appeared that minimize or even contradict these associations (Cheng and Tsai, 1999, Correa et al., 1999). Likewise, direct testing of hVDR alleles for activity has yielded somewhat variable results, although, when a difference is found, the b and F hVDR alleles appear to be more active than the B or f alleles (see Section 4).

One caveat in most of the above-cited studies is that correlations were sought between a single, specific polymorphism, or between the Bsm-Apa-Taq linkage group, and the physiological parameter of interest. Very few studies have attempted to control for hVDR genotype at both the Bsm/Apa/Taq/singlet(A) cluster and the Fok I site. In one example (Ferrari et al., 1998), a correlation between Fok I alleles and BMD could not be demonstrated, but ‘cross-genotyping’ with Bsm I alleles revealed a potentially important positive association in prepubertal girls between the ffBB hVDR genotype and low BMD (Ferrari et al., 1998).

Another caveat in the above cited studies is that a direct influence of allelic variation on VDR expression or activity was not demonstrated, leaving open the possibility that the observed correlation might be due to linkage to another nearby site or even to a different gene. In the only two extant studies in which the potential relationship between genotype and activity of the hVDR protein was evaluated (Verbeek et al., 1997, Gross et al., 1998), no functional influence of specific alleles was observed, but again, only a single polymorphic site was examined in isolation.

In the present communication, we report an evaluation of a panel of twenty human fibroblast lines. The current protocol includes simultaneous consideration of the hVDR genotypes at both the singlet(A) and the Fok I loci, which are then correlated with activity of the endogenous VDR in the corresponding cell line. From these data, we conclude that (a) biallelic variants at the Fok I and the singlet(A) sites, in combination, affect transcriptional activation by the endogenous hVDR in the tested human fibroblasts; (b) the singlet(A) L allele is more active than the S allele; and (c) a third, unknown genetic variable appears to influence VDR activity.