Functional assessment of two vitamin D-responsive elements in the rat 25-hydroxyvitamin D3 24-hydroxylase gene.

Two vitamin D-responsive elements (VDRE-1 and VDRE-2) were recently identified in the 5′-upstream region of the rat 25-hydroxyvitamin D3 24-hydroxylase gene at −151/−137 and −259/−245, respectively. We studied the transcriptional regulation of this gene by vitamin D by means of mutational analysis. Introducing mutations into VDRE-1 and VDRE-2 in the native promoter −291/+9 reduced vitamin D-dependent chloramphenicol acetyltransferase activity by 86 and 41%, respectively. Mutation of the direct repeat −169/−155 located at 3 base pairs upstream of VDRE-1 also caused 50% decrease of chloramphenicol acetyltransferase activity. Connection of the element −169/−155 to VDRE-1 enhanced the vitamin D responsiveness of VDRE-1 5-fold through the heterologous β-globin promoter. The fragment −291/−102 containing the two VDREs showed two shifted bands in the presence of the vitamin D receptor and retinoid X receptor in gel retardation analysis, and the appearance of the slower migrating band indicates that two sets of receptor complexes bind to this fragment simultaneously. These results demonstrate that VDRE-1 is a stronger mediator of vitamin D function than VDRE-2 due to the presence of the accessory element −169/−155 located adjacent to VDRE-1, although VDRE-2 exhibits a smaller dissociation constant for the vitamin D receptor-retinoid X receptor complex than VDRE-1.

The vitamin D receptor (VDR) is a member of a steroid and thyroid hormone receptor superfamily (15,16). The VDR occupied with 1,25-(OH) 2 D 3 forms a heterodimer with the retinoid X receptor (RXR) (17,18) and modulates the expression of target genes by binding to the vitamin D-responsive element (VDRE) in the promoter region. Several VDREs have been identified in the genes of human (19) and rat osteocalcin (20 -22) and mouse osteopontin (23). These VDREs consist of a direct repeat of two hexanucleotides separated by 3 base pairs.
Three groups have independently identified two functional VDREs in an antisense orientation in the rat 24-hydroxylase gene promoter; at Ϫ151/Ϫ137 (VDRE-1) (24,25) and Ϫ259/ Ϫ245 (VDRE-2) (26). Zierold et al. (27) examined the interaction between the two VDREs by means of a reporter gene assay using heterologous promoters and by a gel mobility shift assay. They concluded that the two VDREs synergistically increased binding affinity for the receptor complex to elicit powerful promoter activity.
In this study, we assessed the functions of the two VDREs in more detail using a native promoter containing site-specific mutations at the VDREs and determined their affinity for the receptor complex by gel retardation assays. We found that the two VDREs work additively. VDRE-1 coupled with the 3-base upstream segment of Ϫ169/Ϫ155 accounted for most of the vitamin D responsiveness.

EXPERIMENTAL PROCEDURES
All enzymatic manipulations proceeded according to standard procedures (28).
Materials-Restriction and modifying enzymes were purchased from Toyobo Inc. (Osaka, Japan) and Takara Shuzo (Kyoto, Japan).  Fig. 1A) with XhoI and SalI overhangs were synthesized and ligated into the SalI site of a pGCAT vector containing the ␤-globin promoter upstream of the chloramphenicol acetyltransferase (CAT) gene (29). Sequences of the top strands for VDRE-2 and (Ϫ169/Ϫ152)-VDRE-2 shown in Fig. 1B were as follows (VDRE-2 is underlined): VDRE-2, 5Ј-GCAGCGCACCCGCTGAAC-CCTG; and (Ϫ169/Ϫ152)-VDRE-2, 5Ј-CGGTCACCGAGGCCCCGGCG-CACCCGCTGAACCC. These fragments were synthesized with XhoI and SalI overhangs and also ligated into the SalI site of the pGCAT vector. The specific mutations of VDREs in the native promoter were introduced by polymerase chain reaction-based site-directed mutagenesis using p-(Ϫ291/ϩ9) containing the fragment Ϫ291/ϩ9 at the XbaI site of pCAT-basic vector as a template (24). To construct pM-VDRE-2, a set of P1 and P3 primers (P1, 5Ј-TTACGCCAAGCTTGCATGCCT-3Ј, corresponding to the vector sequence at the 5Ј-end of the insert; P3, 5Ј-GGGTCGAGCCCAGGGaaCAGCGGT-3Ј, corresponding to Ϫ256/ Ϫ232; the lower case letters indicate the mutation sites) were used for polymerase chain reaction. The amplified fragment was digested with HindIII and BanII, and the corresponding region of p-(Ϫ291/ϩ9) was replaced. In pM-VDRE-1, a set of P2 and P4 primers (P2, 5Ј-AGCTCA-GATCCTCTAGAGTCAC-3Ј, corresponding to the vector sequence at the 3Ј-end of the insert; P4, 5Ј-GTGTCGGTCACCGAGGCCCCG-GCGCttTCACT-3Ј, corresponding to Ϫ173/Ϫ142; the lower case letters indicate the mutation sites) were used for amplification. The amplified fragment was digested by XbaI and BstEII and the corresponding region of p-(Ϫ291/ϩ9) was replaced. Plasmid pM-(Ϫ169/Ϫ155) was constructed by the same procedure using a set of P2 and P5 primers (P5, 5Ј-GTGTCGGTCACCGAaaCCCCG, corresponding to Ϫ173/Ϫ153; the lower case letters indicate the mutation sites). Plasmid pM-VDRE-1,2 was prepared by polymerase chain reaction of pM-VDRE-1 using the primer set P1 and P3 as described above. Plasmid pM-VDRE-2-(Ϫ169/ Ϫ155) was also constructed by polymerase chain reaction using the primer set P1 and P3 and pM-(Ϫ169/Ϫ155) as a template. A truncated mutant (pM-truncation) was prepared by digesting p-(Ϫ291/ϩ9) with BanII and BstEII, filling in with a Klenow fragment, and self-ligating. All mutation constructs were sequenced, and the base substitutions were confirmed.
Cell Culture and DNA Transfection-LLC-PK1 cells were maintained in medium 199 (Life Technologies, Inc.) containing 5% dextrancoated, charcoal-stripped fetal calf serum and plated at a density of 3 ϫ 10 5 cells/60-mm dish the day before transfection. Transfections were performed by means of calcium phosphate precipitation with 2 g of reporter plasmid (in the experiment in Fig. 2, 10 g of reporter plasmid was used) and 2 g of reference plasmid (pSV-␤-galactosidase control vector from Promega, Madison, WI) as an internal control to correct for variations in transfection efficiency. Four hours after transfection, the cells were treated with 15% glycerol and were then incubated for 2 days with 1,25-(OH) 2 D 3 (or vehicle) at a final concentration of 10 nM. The cells were harvested, and the activities of CAT and ␤-galactosidase were measured by means of a diffusion assay (30) and a ␤-galactosidase assay kit from Promega, respectively. Each set of experiments was repeated at least three times, and the results are presented in terms of -fold induction with the means Ϯ S.E.
Preparation of VDR and RXR-The plasmid p7Xf-hVDR containing a full-length copy of the human VDR cDNA (a gift from Dr. J. W. Pike, Ligand Pharmaceuticals, San Diego, CA) and pSG5RXR-␤ containing the coding region of rat RXR-␤ cDNA (a gift from Dr. S. Kato, Tokyo University of Agriculture) were linearized by enzyme digestion and used to produce capped RNA from the SP6 and T7 promoters, respectively. The resulting RNAs were translated in the reticulocyte lysate system according to the directions of the manufacturer (Promega). The VDR concentration was determined by a ligand binding assay (1-2 fmol/l of lysate).
Gel Mobility Shift Assay-Mobility shift assays were basically performed as described (19). The fragments Ϫ291/Ϫ102, )-M-VDRE-1,2, and (Ϫ291/Ϫ102)-M-truncation were prepared from the corresponding reporter genes by digesting with PstI (at Ϫ291 bp) and BstUI (at Ϫ102 bp), respectively. Ten picomoles of the fragments were end-labeled with [␥-32 P]ATP using T4 polynucleotide kinase. To determine the specific activity of probes, a portion of the reaction mixture was loaded onto an 8% polyacrylamide gel in Tris/ borate buffer to separate it from the [␥-32 P]ATP (normally 5 ϫ 10 6 cpm/pmol of probe). After purifying the remaining reaction mixture using a QIAquick nucleotide removal kit (Qiagen, Hilden, Germany), the probe concentration was calculated by its specific activity. The reticulocyte lysate containing translated VDR (1 l) and RXR (1 l) were incubated in the presence or absence of 500 nM 1,25-(OH) 2 D 3 for 20 min in a buffer containing 25 mM Tris-HCl, 15 mM Hepes, pH 7.9, 40 mM NaCl, 5 mM KCl, 3 mM MgCl 2 , 4.5 mM EDTA, 6% glycerol, 0.08% Tween 20, 1 mM dithiothreitol, and 1 g of poly(dI-dC). One microliter of DNA probe (10 4 cpm/reaction) was then incubated with the reaction mixture (total, 20 l) for 20 min at room temperature, and then the mixture was resolved by electrophoresis on a 5% polyacrylamide gel in 50 mM Tris/ 380 mM glycine buffer, pH 8.5, at 200 V at 4°C. The gels were dried on DE-81 paper (Whatman, Kent, United Kingdom) and visualized by autoradiography overnight. All K d values were determined by the gel retardation assay after changing the probe concentration (0.1-4 nM). The receptor-DNA complex and free probe were quantified with a Fujix 2000 Bio Image Analyzer (Fuji Film, Tokyo, Japan).
Characterization of the Flanking Region of VDRE-1-The region Ϫ169/Ϫ127 has a characteristic structure consisting of five half-sites, which are side by side just 3 bases apart, and a pair of half-sites was identified as a VDRE (VDRE-1). The three other half-sites may be important in enhancing VDRE-1 function (24,27). To evaluate the function of the half-sites flanking VDRE-1, eight fragments were synthesized, and CAT activity was assayed using the heterologous ␤-globin promoter (Fig.  1A). The deletion of a half-site at the 3Ј-end did not affect the responsiveness to 1,25-(OH) 2 D 3 . On the other hand, deletions of one or two of the half-sites from 5Ј-end markedly decreased vitamin D-dependent activation. The two half-sites located upstream of VDRE-1 appear to be a direct repeat 3 motif (16), although this direct repeat Ϫ169/Ϫ155 as such did not mediate any vitamin D responsiveness in LLC-PK1 cells (Fig. 1A). To analyze the function of this direct repeat in more detail, we introduced mutations either into the half-sites or into the spacer (Fig. 1A, (Ϫ169/Ϫ134)-M1, (Ϫ169/Ϫ134)-M2, and (Ϫ169/ Ϫ134)-M3). Mutations in each half-site greatly hampered the inducibility, whereas that in the spacer scarcely affected the induction of CAT activity by 1,25-(OH) 2 D 3 . Connecting the fragment Ϫ169/Ϫ155 to 3 base pairs upstream of VDRE-2 enhanced the vitamin D responsiveness of VDRE-2 about 2.5fold (Fig. 1B). These results suggest that the segment Ϫ169/ Ϫ155 functions as an accessory element that elevates 1,25-(OH) 2 D 3 responsiveness, although it did not respond to vitamin D by itself. Therefore, VDRE-1 coupled with the accessory element Ϫ169/Ϫ155 was more powerful than VDRE-2 for transcriptional activation by 1,25-(OH) 2 D 3 in the heterologous ␤-globin promoter.
Comparison of VDREs in the Native Promoter-To elucidate the induction mechanism regulated by the two VDREs, we introduced specific mutations into VDRE-1, VDRE-2, and the accessory element Ϫ169/Ϫ155 in the native promoter of Ϫ291/ϩ9 by means of polymerase chain reaction-based sitedirected mutagenesis and then assayed CAT activity. As shown in Fig. 2, the mutation in VDRE-1 caused a striking decrease in the vitamin D-dependent CAT induction from 25.6-to 3.7-fold. On the other hand, introduction of the mutation into VDRE-2 retained 15.2-fold induction. The mutation in Ϫ169/Ϫ155 significantly affected the vitamin D-dependent responsiveness (13.0-fold induction). Double mutations in both VDRE-1 and VDRE-2 completely abolished the vitamin D response. The construct of double mutations in VDRE-2 and Ϫ169/Ϫ155 showed only 4.0-fold induction, which was almost identical to that of the construct of pM-VDRE-1. These results indicated that VDRE-1 and VDRE-2 exhibit almost the same responsiveness when measured separately. However, VDRE-1 coupled with the accessory element Ϫ169/Ϫ155 elicited a stronger vitamin D response than VDRE-2. Deleting the region Ϫ243/ Ϫ166, which contains a putative half-site 6 base pairs downstream of VDRE-2 from Ϫ291/ϩ9 did not affect the vitamin D-dependent CAT induction.
Gel Mobility Shift Analysis-The human vitamin D receptor and the rat RXR-␤ were prepared by in vitro translation of their cDNAs. As shown in Fig. 3A, the synthetic fragments of VDRE-1 and VDRE-2 migrated as a retarded band in the presence of VDR and RXR, and adding 1,25-(OH) 2 D 3 greatly enhanced the intensity of the shifted band (Fig. 3A, lanes 3 and  6; 10-fold for VDRE-1, 6-fold for VDRE-2). The fragment Ϫ291/ Ϫ102 containing the two VDREs migrated as two retarded bands, and adding 1,25-(OH) 2 D 3 enhanced the formation of the DNA-receptor complex (Fig. 3B, lanes 3 and 4; 2.5-fold). The ligand might have increased the effective VDR⅐RXR complex concentration, since the ligand promotes binding of VDR to RXR (17,32). The enhancement of the band intensity by 1,25-(OH) 2 D 3 suggested that complexes 1 and 2 included the VDR. These bands completely disappeared in the presence of the anti-VDR monoclonal antibody 9A7, which interferes with the DNA binding of VDR (19) (data not shown). When the amounts of VDR and RXR were increased, the intensity of the second band increased. Addition of a large amount of VDR and RXR caused the formation of a third complex (Fig. 3B, lane 6). To determine the binding sites for the VDR⅐RXR complex in fragment Ϫ291/Ϫ102, those containing mutations in each VDRE and/or the accessory element Ϫ169/Ϫ155 were prepared by excising the mutated reporter plasmids illustrated in Fig. 2 with PstI and BstUI. As shown in Fig. 3C, lanes 2-4, mutations in either VDRE-1 or VDRE-2 abolished the slower migrating bands, whereas that in Ϫ169/Ϫ155 did not affect the receptor binding profile. Double mutations of VDRE-1 and VDRE-2 abolished the retarded bands (Fig. 3C, lane 5). The truncated mutant (Fig. 3C, lane 6) migrated as two bands. Complex 1 could be a mixture of binding to VDRE-1 and VDRE-2. Formation of complex 2 showed that two sets of the VDR⅐RXR complex can bind to VDRE-1 and VDRE-2 simultaneously.
Binding Affinity of the VDR⅐RXR Complex to VDREs-The K d was determined by Scatchard analysis of the gel shift data ( Table I). The synthetic fragment of VDRE-2 had higher affinity (K d , 1.07 nM) than that of VDRE-1 (K d , 1.52 nM). The fragment of Ϫ291/Ϫ102 exhibited a much smaller K d value (0.24 nM) for complex 1 than those of the synthetic VDREs. Introducing mutations of either VDRE-1 or the accessory element Ϫ169/Ϫ155 in the fragment Ϫ291/Ϫ102 did not affect the binding affinity, whereas that of VDRE-2 slightly reduced it to 0.43 nM. Since the mutated fragment in VDRE-2 retains the intact VDRE-1 region, the results also indicated that the VDRE-1 region has a relatively lower affinity than VDRE-2 for the VDR⅐RXR complex. The large fragment Ϫ291/Ϫ102 had higher affinity than the synthetic VDREs. One of the important factors affecting the affinity may be the length of the fragments, since the medium length fragment Ϫ291/Ϫ167, which contains only VDRE-2, had a medium K d of 0.53 nM. DISCUSSION There are reportedly two VDREs in the promoter region of rat 25-hydroxyvitamin D 3 24-hydroxylase at Ϫ151/Ϫ137 (VDRE-1) and Ϫ259/Ϫ245 (VDRE-2) (24 -26). Both consist of a direct repeat of hexanucleotides with a 3-base pair spacing. Characteristically, they reside on the antisense strand, which differs from other known VDREs. According to the definition of enhancers, the direction of the response element is not important to elicit its function. However, the preferential binding of RXR to the upstream half-site and VDR to the downstream half-site may influence the trans-activation function depending on the direction of the VDRE (33,34).
In this study, we characterized the responsiveness of these elements to 1,25-(OH) 2 D 3 in the context of homologous and heterologous promoters. VDRE-1 and VDRE-2 conferred almost the same response to 1,25-(OH) 2 D 3 through the heterol- ogous ␤-globin promoter. Double mutations in VDRE-1 and VDRE-2 in the native promoter completely abolished vitamin D responsiveness (Fig. 2, pM-VDRE-1,2). These results indicate that both VDREs are minimal elements that confer vitamin D inducibility. Kahlen and Carlberg (31) reported that the downstream half-sites of VDRE-2 formed a direct repeat 6-type motif located at 6 bases downstream of VDRE-2 and mediated the vitamin D response by a VDR homodimer. The truncation of Ϫ234/Ϫ166 in this study, however, showed that this direct repeat 6 was not responsible for transcriptional activation by vitamin D. The reason for this discrepancy is not clear. It may be related to the concentrations of receptor proteins, since no receptor expression vectors were cotransfected in our experiments, whereas they cotransfected a VDR expression vector.
Three hexameric half-site-like elements were also found in the flanking region of VDRE-1 (Fig. 1A). We reported that the direct repeats with 3-base spacing at the 5Ј-side of VDRE-1 are important for its function (24). Zierold et al. (27) have claimed that the most proximal half-site is indispensable for VDRE-1 function, although they have not characterized the 5Ј-side. Our systematic analysis of the flanking region of VDRE-1 revealed that the direct repeat (Ϫ169/Ϫ155) at the 5Ј-side of VDRE-1 is more important to enhance the responsiveness of VDRE-1 to 1,25-(OH) 2 D 3 than the 3Ј-side.
The segment Ϫ169/Ϫ155 also greatly increased the responsiveness of VDRE-2 when connected to it. The complex enhancer composed of a consensus hormone response element and a structure called the accessory element is required to elicit a strong hormonal response in the mouse sex-limited protein gene (35). Thus, the segment Ϫ169/Ϫ155 functions as an accessory element, which allows VDRE-1 to exert the maximal effect. Mutational analysis of the native promoter supported this concept, since the introduction of mutations into the direct repeat Ϫ169/Ϫ155 decreased CAT activity to about 50% of that of the native promoter.
Introducing mutations into each half-site of the direct repeat Ϫ169/Ϫ155 strikingly decreased CAT activity, whereas mutations in the gap sequence of the element did not affect it. These results suggested that this accessory element actually possesses a direct repeat 3-type structure. The DR3 structure raises the possibility that the VDR-RXR heterodimer interacts with the element. When we used the fragment Ϫ291/102 and a large amount of in vitro-translated receptors in a gel retardation assay, the third complex migrated just above the second complex (Fig. 3B). However, the VDR neither bound to the   2  region Ϫ169/Ϫ155 nor mediated the vitamin D effect through this region with the heterologous promoter (24). It is under investigation what kind of factor(s) bind(s) to this accessory element.
In this study, we have three enhancer sequences, including two distinct VDREs in the rat 24-hydroxylase gene promoter. This is the first example for the VDRE, although multiple copies of glucocorticoid and estrogen elements have been found in several genes (36). In general, multiple enhancer elements are supposed to permit synergistic gene expression (37). Synergistic interactions of the VDR with SP1, AP-1, and VDR have been found in an artificial arrangement of the elements in the promoter (36). The mechanism of the synergistic interaction in trans-activation may, at least in part, implicate cooperative binding resulting from the specific protein-protein interaction. In the natural promoter, VDRE is juxtaposed to an AP-1 site (19) and the sodium butyrate response element (38) in the human osteocalcin and mouse calbindin D28k promoter, respectively.
In the rat 24-hydroxylase promoter, however, no cooperative binding of receptors was observed in this study. Quantification of the intensity of retarded bands indicates that VDR⅐RXR complexes bind to VDRE-1 and VDRE-2 independently but not cooperatively (Fig. 3B), since the band intensity of complex 2 was not stronger than that of complex 1. The higher affinity of the long fragment Ϫ291/Ϫ102 containing both VDREs was maintained even when the mutations were introduced into VDRE-1 or -2, indicating that there is no interaction between the two VDREs. In concert with the results obtained by binding assay, no obvious synergistic effect of the two VDREs was observed in the trans-activation function of 1,25-(OH) 2 D 3 . Mutants of VDRE-1 and VDRE-2 exhibited 3.7-and 15.2-fold induction, respectively, whereas the intact promoter elicited a 25.6-fold induction in response to 1,25-(OH) 2 D 3 (Fig. 2). Because the sum of the -fold induction (18.9-fold) of individual elements was similar to the -fold induction of the intact promoter, interaction between the two VDREs in the promoter may be interpreted as additive or slightly synergistic. These data are not consistent with the results reported by Zierold et al. (27). They have demonstrated that the tightest binding by receptors was found with the fragment containing two VDREs in a gel shift assay, and it gave maximum trans-activation. The reason of the discrepancy is not clear, but the deference in the promoter context may be a key. We analyzed the function of VDREs by a using homologous promoter, whereas they used a heterologous promoter. In addition, they detected only one band when the fragment Ϫ260/Ϫ134 was applied in the gel shift assay. It is very hard to investigate whether the cooperative binding occurs in that condition.
Chen and DeLuca (39) have described the structure of the human 24-hydroxylase promoter, which also contained two VDREs. The presence of the two VDREs in the promoter of the 24-hydroxylase gene may be useful to induce proper transactivation in response to the circulating levels of 1,25-(OH) 2 D 3 .
In conclusion, we found that the two distinct VDREs function additively in the promoter of the rat 24-hydroxylase gene.
VDRE-1 appears to be a stronger mediator of vitamin D function than VDRE-2 regardless of its lower affinity for the receptor complex. This is due to the presence of the accessory element Ϫ167/Ϫ155 located adjacent to VDRE-1.