INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The mechanism by which steroid/thyroid nuclear hormone receptors activate gene transcription has been extensively studied (1-8). Among them there have been so many lines of solid evidence showing that vitamin D receptor (VDR),¹ thyroid hormone receptor, and retinoic acid receptor employ a common machinery to exert their ligand-dependent specific effects; all of them utilize a retinoic acid X receptor (RXR) in common as a partner of a heterodimer (2, 4). However, this mechanism seems to be confined only to gene stimulation but not to gene repression. This situation led us to address one of the key points in hormonal biology, negative feedback mechanism. The levels of almost all the hormones synthesized at a specific organ are under rigid control to keep their levels within a very narrow range. One of the representative mechanisms is the so-called end product inhibition. In the cases of the nuclear hormone receptors, negative transcriptional regulation of several pituitary trophic peptide hormone genes by glucocorticoid or sex steroid hormones is a good example. Although individual nuclear hormone receptors might well be involved in such regulation, detailed unified molecular mechanisms such as the manner of dimerization are largely unknown, with a few exceptions (9-14). Likewise, the mechanism of negative gene regulation by VDR and vitamin D is only partially understood (15-21). In this case, an active form of vitamin D, 1,25-dihydroxyvitamin D₃, can be considered an end product of parathyroid hormone (PTH) action, and this metabolite, in turn, inhibits the synthesis of PTH mRNA to keep the blood calcium level constant. Demay et al. (16) first reported that a negative vitamin D response element (nVDRE) exists in the upstream region of the human PTH gene to mediate such gene repression. This element contains a homologous sequence to only one of the two hexameric DNA sequences that form the core sequence of the consensus DNA sequence (VDRE) for positive gene regulation by vitamin D (2, 4, 22-25). In this process, it was shown that VDR, but not RXR, was involved, and the presence of another unknown partner protein(s) of VDR was proposed (17). However, there have been no reports confirming that the nVDRE is conserved among the genes whose expression is negatively regulated by vitamin D. We have found that the human PTH-related peptide (PTHrP) gene, which is assumed to be derived from an ancestoral gene in common with the PTH gene (26), contains a DNA sequence very homologous to the nVDRE of hPTH. Inoue et al. (27) reported that expression of the hPTHrP gene was inhibited by 1,25-dihydroxyvitamin D₃ at the transcriptional level in human adult T cell lymphoma/leukemia virus-infected T cells, MT2 cells. With these cells, the therapeutic potential of vitamin D to decrease the blood calcium level due to the inhibition of PTHrP synthesis was discussed. Unlike parathyroid cells, from which it is hard to establish cultured cell lines, MT2 cells are transfectable cultured cell lines. Here, we examined the molecular mechanism of negative regulation of the PTHrP gene by vitamin D in MT2 cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Homology Search by Computer Analysis-- The DNA sequences homologous to the nVDRE of human PTH (16) were searched in the EMBL gene bank, including the human PTHrP gene, and some of them are shown in Fig. 1. We also compared the nVDRE with other DNA elements reported to be responsible for negative as well as positive gene regulation by vitamin D (16-24).

Synthetic Oligonucleotides and Plasmid Constructions-- All the oligonucleotides used in this report were made by a DNA synthesizer (Biosearch 8700). They were synthesized as follows: nVDRE_hPTHrP, 5'-(GATCC)TGCTATAGATTCATATTTGGTTTATA(T)-3'; 3'-(G)ACGATATCTAAGTATAAACCAAATAT(AGATC)-5'. VDRE_mop, 5'-(GATCC)ACAAGTTCACGAGGTTCACGTCT(T)-3'; 3'-(G)TGTTCAAGTGCTCCAAGTGCAGA(AGATC)-5'. Bases in the parentheses are BamHI and XbaI cohesive ends to facilitate subsequent ligations to a BamHI-XbaI larger fragment of PUTKAT1 (28). The methods to prepare double-stranded DNAs and to construct the Tk promoter-based CAT reporter plasmids were also described (28). The plasmids encoding p70 subunit of Ku antigen in the antisense orientation were described (29).

Transfection and CAT Assay-- MT2 cells were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum unless otherwise mentioned. The CAT plasmids or the expression vector encoding p70 subunits of Ku antigen in the antisense orientation (29) were introduced into MT2 cells by the DEAE-dextran method (30). After transfection, cells were equally split into several dishes to avoid differences in transfectional efficiency among dishes. Final amounts of the transfected plasmids per dish were 5 µg. Twelve h later, the medium was changed to RPMI 1640 containing 1% fetal bovine serum with several different concentrations of vitamin D metabolites (purchased from Dupher Co.) or vehicle (100% ethanol) alone. In each case, 40 h after transfection, cells were harvested, and CAT assay was performed. Where indicated, average CAT activity was calculated after three separate transfections using ¹⁴C scintillation counting, and typical results are shown in the figures with a given CAT activity as 100 in each case. Details are shown in each figure.

Preparation of Nuclear Extracts and Gel Shift Assay-- Nuclear extracts were prepared by the method of Schreiber et al. (31) from MT2 cells 40 h after maintaining the cells in the media as indicated. The synthetic oligonucleotides were end-labeled with [-³²P]ATP by T₄ polynucleotide kinase. 10⁴ cpm of the probe (10⁹ cpm/µg) was incubated with 10 µg of the nuclear protein along with 1 µg of poly(dI-dC) in each reaction for 30 min at room temperature. Final KCl concentration was adjusted to 80 mM by the binding buffer containing no KCl (10, 31). The total reaction volume was 25 µl. DNA-protein complexes were resolved on 4% nondenaturing acrylamide gels, dried, and visualized by autoradiography as described (12-14). Where indicated, 0-50-fold molar excesses of nonradiolabeled oligonucleotides were used as competitors. When anti-human VDR rat monoclonal antibody, 9A7 (BIOMOL Research Laboratory Inc.), or anti-human RXR rabbit polyclonal antibody, D-20 (Santa Cruz Biotech. Inc.) was used, each of 1/20 and/or 1/4 dilution and the respective control IgGs (1 µl) supplied by manufacturers were included in the reaction for 15 min before the addition of the radiolabeled probes. We also used 1 µl of anti-Ku antigen human antiserum, which specifically recognizes the p70 subunit of Ku antigen (29) and control serum in the same way in electrophoretic mobility shift assay (EMSA). To explore the effects of vitamin D metabolites on the protein-DNA interaction, cells were treated in the same manner as described in transfection method. Where indicated, 0.1 unit of potato acid phosphatase (Sigma) was directly included in the reaction mixture at 37 C° for the indicated period. Protein concentrations were determined by the Bio-Rad assay kit.

Southwestern Analyses, Immunodepletion, and Immunoblotting-- Thirty µg of the nuclear proteins obtained as described above was mixed with an equal volume of 2× denaturing buffer (5% N-lauroylsarcosine, 5 mM Tris-Cl (pH 6.8), 25% glycerol, 0.05% pyronin Y, and 200 mM dithiothreitol). After a 15-min incubation at room temperature, samples were loaded onto an 8% (or 10%) SDS-polyacrylamide gel. Further details were described elsewhere (28), except that we used a radiolabeled oligo(nVDRE_hPTHrP or VDRE_mop) in this study. Where indicated, we used MT2 nuclear proteins in which VDR was immunodepleted with the 9A7 shown above, according to the method reported previously (28). Immunoblotting with 1 µg/ml of the 9A7 or 1:1000 diluted anti-Ku antigen human antiserum also shown above was performed by the enhanced chemiluminescence method as reported (28).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Putative nVDRE Sequences in PTH and PTHrP Genes from Several Species-- In the upstream region of the human PTHrP gene, we found a DNA sequence homologous to a nVDRE in the hPTH gene (16). In the anti-stranded sequence of this DNA in the hPTHrP gene, 11 out of 13 bases or 10 out of 11 bases were identical to the nVDRE of hPTH (Fig. 1). Both contained one copy of the DNA motif AG(G/A)TTCA. This motif mimics a core sequence (AG(G/A)TCA) for the binding of the specific set of the nuclear hormone receptor including VDR (2). Interestingly, one VDRE found in the mouse osteopontin gene (VDRE_mop; Ref. 22) consists of a direct repeat of the same motif with two (or three) spacers. Furthermore, 4 bases, CTAT, flanking this motif were common between the two. These findings suggest that both might mediate negative gene regulation by 1,25-dihydroxyvitamin D₃. Falzon (18) recently reported that the rat PTHrP gene possessed two distinct DNA elements (nVDREs), both of which were responsible for negative gene regulation by vitamin D. We compared our nVDRE of hPTHrP with Falzon's nVDREs in the rat PTHrP gene (18) and speculated that one of them might be equivalent to our nVDRE because of the similar location and overall sequence similarity (see "Discussion"). Furthermore, we also found a DNA sequence in the chicken TGF2 gene very homologous to nVDREs of both hPTH and hPTHrP. Intriguingly, expression of the chicken TGF2 gene was shown to be inhibited by vitamin D in mesangial cells (20). On the other hand, as shown in Fig. 1, the sequences of other recently proposed nVDREs such as those in the chicken PTH (21) gene and the rat PTHrP gene (19) or positive VDRE in the human osteocalcin gene (23) are considerably different from these three nVDREs.

View larger version (26K): [in this window] [in a new window]   Fig. 1.   DNA sequences homologous to a putative negative vitamin D response element (nVDRE) in the human PTHrP gene (nVDREhPTHrP). The numbers shown above each sequence represent the locations from the respective transcription start sites. The bases similar to the consensus direct repeat, AG(G/A)TTCA (2), are underlined. The identical bases between nVDREhPTHrP and a given sequence are shown as bold letters. The sources of each putative nVDRE are as follows: nVDREhPTH (16), nVDRErPTHrP-1 and nVDRErPTHrP-2 (18), nVDRErPTHrP-3 (19), nVDREcPTH (21), and the nVDRE-like sequence of chicken TGF2 (20). One of the positive VDREs, VDREmop (22), was also included.

Interaction between Nuclear Protein(s) in MT2 Cells and nVDRE_hPTHrP-- We synthesized oligonucleotides corresponding to nVDRE_hPTHrP and VDRE_mop and examined the interactions between each of these oligonucleotides and nuclear protein(s) from MT2 cells by EMSA. As shown in Fig. 2, both of the oligonucleotides formed two protein-DNA complexes. Since each of the lower bands did not consistently appear, we focused on the upper bands in this study. Although the migrating positions of both complexes seemed similar in EMSA, the MT2 protein-nVDRE_hPTHrP complex was competed out only by a 10-50-fold molar excess of the nonradiolabeled nVDRE_hPTHrP. The same amount of VDRE_mop was ineffective as a competitor. Conversely, only a 10-50-fold molar excess of VDRE_mop, but not that of nVDRE_hPTHrP, was able to abolish the formation of MT2 protein-VDRE_mop complex. Nevertheless, a 200-fold molar excess of the reciprocal competitors in each EMSA similarly abolished the binding between MT2 nuclear proteins and each probe (not shown and see "Discussion"). These results indicate that the composition of the nuclear proteins is different, at least in part, between the two protein-DNA complexes.

View larger version (62K): [in this window] [in a new window]   Fig. 2.   EMSA using MT2 nuclear proteins and nVDREhPTHrP (left panel) and VDREmop (right panel) as radiolabeled probes. On each panel, a 10-fold (lanes 2 and 4) and a 50-fold (lanes 3 and 5) molar excess of nonradiolabeled VDREmop (lanes 2 and 3) or nVDREhPTHrP (lanes 4 and 5) were included. The arrows indicate the specific protein DNA complex.

View larger version (52K): [in this window] [in a new window]   Fig. 3.   VDR, but not RXR, exists in the complex. The effects of anti-human VDR rat monoclonal antibody, 9A7 (A, lanes 2 and 5, 1 to 20 dilution; lanes 3 and 6, 1 to 4 dilution), or anti-human RXR rabbit polyclonal antibody, D-20 (B, lanes 8 and 10, 1 to 4 dilution) on the complex between MT2 nuclear proteins and either nVDREhPTHrP (lanes 1-3 and 7-8) or VDREmop (lanes 4-6 and 9-10) in EMSA. In lanes 1 and 4, 1 µl of rat IgG2, and in lanes 7 and 9, 1 µl of preimmune rabbit IgGs were included. The arrows indicate the specific protein-DNA complex.

nVDRE_hPTHrP Is Necessary for Negative Gene Regulation by Vitamin D in MT2 Cells-- To examine whether nVDRE_hPTHrP is necessary for negative gene regulation by vitamin D in MT2 cells (Fig. 4), we transfected the cells either with the parental TkCAT (28) or with the same CAT plasmid bearing nVDRE_hPTHrP in the sense orientation. As shown in Fig. 4, 10⁸ M 1,25-dihydroxyvitamin D₃ selectively inhibited CAT activity driven by nVDRE_hPTHrP-TkCAT but not TkCAT. Furthermore, less active vitamin D metabolites were ineffective.

View larger version (44K): [in this window] [in a new window]   Fig. 4.   CAT assay in MT2 cells. nVDREhPTHrP-TkCAT (A) or the parental TkCAT (B) was introduced into MT2 cells. Cells were equally split into five conditions after transfection. Forty h after the addition of vehicle (veh) alone (lane 1), 109 M (lane 2) or 108 M (lane 3) 1,25-dihydroxyvitamin D3 (1.25 D3, or 108 M 24,25-dihydroxyvitamin D3 (24.25 D3 (lane 4) or 108 M 25-dihydroxyvitamin D3 (25 D3 (lane 5), cells were harvested, and a CAT assay was performed. Typical results were shown. The average CAT activity after three different transfections is indicated below each condition. CAT activity driven by TkCAT in MT2 cells treated with vehicle alone (B, lane 1) was arbitrarily represented as 100.

Vitamin D Attenuates the Binding between nVDRE_hPTHrP and VDR-- Our next question was what is the effect of vitamin D on the binding between nVDRE_hPTHrP and MT2 proteins. As shown in EMSA (Fig. 5A), 1,25-dihydroxyvitamin D₃ weakened the binding between nVDRE_hPTHrP and MT2 nuclear proteins, although it strengthened the binding between VDRE_mop and these proteins in a dose-dependent manner. In the Southwestern assay, using a radiolabeled nVDRE_hPTHrP as a probe (Fig. 5B), similar dose-dependent effects of 1,25-dihydroxyvitamin D₃ were observed on the band corresponding to about 50 kDa in size but not on the other bands. We then employed MT2 nuclear proteins in which VDR was immunodepleted with the 9A7 shown above in the Southwestern assay. As shown in Fig. 5C, immunodepletion treatment clearly and selectively abolished the binding of this 50-kDa band but not higher molecular mass bands, strongly suggesting this 50-kDa band corresponded to the VDR itself. On the other hand, the radiolabeled VDRE_mop yielded only one band around 96 kDa in size, which was unaffected by the immunodepletion. We assume these higher molecular weight proteins are nonspecific DNA binding proteins. Immunoblotting with 9A7 revealed that treatments with 1,25-dihydroxyvitamin D₃ did not affect the amounts of VDR protein (Fig. 5D). Together, these results suggest that 1,25-dihydroxyvitamin D₃ treatment weakens the binding between nVDRE_hPTHrP and VDR in MT2 cells, thereby exerting its inhibitory effect on the expression of nVDRE_hPTHrP-bearing gene(s) in these cells.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Dose-dependent effects of 1,25-dihydroxyvitamin D3 on VDR or the formation of nVDREhPTHrP-MT2 nuclear protein complex. In A, EMSA using radiolabeled VDREmop (lanes 1-4) or nVDREhPTHrP (lanes 5-8) and nuclear proteins obtained from MT2 cells grown in the presence of 109 M (lanes 2 and 6), 108 M (lanes 3 and 7), or 107 M 1,25-dihydroxyvitamin D3 (1.25(OH)2D3) (lanes 4 and 8) and vehicle alone (v, lanes 1 and 5) is shown. M, molarity. In B, a Southwestern assay using the same set of MT2 protein and radiolabeled nVDREhPTHrP is shown. The arrow indicates a presumptive VDR monomer, which is strongly suggested in C. In C, VDR was immunodepleted with the 9A7 (each, right) or rat IgG2 (each, left) from nuclear proteins obtained from untreated MT2 cells and Southwestern assay with the radiolabeled nVDREhPTHrP (left panel) or VDREmop (right panel) was performed. In D, immunoblotting assay with the 9A7 and MT2 nuclear proteins treated similarly to A and B are shown. Immunoblotting of proteins corresponding to those in lane 1 with rat IgG2 alone is shown in lane 5. The arrow indicates the position of human VDR. The higher molecular weight proteins might be nonspecific proteins. In B, C, and D, molecular weight markers are shown on the left side.

Ku Antigen, Along with VDR, Mediates Negative Regulation of the Gene(s) Containing nVDRE_hPTHrP by Vitamin D-- In order to find a clue for a probable partner of VDR in the nVDRE_hPTHrP-MT2 protein complex, we reexamined the sequence of nVDRE_hPTHrP and noticed that it contained an octamer ATTTGCAT-like sequence following the core motif. This sequence, ATTTGGTT, is reminiscent of one of the negative calcium-responsive elements, oligo(A). We previously showed that the footprint in the oligo(A) HeLa nuclear protein was localized exactly in this position (15) and demonstrated that the nuclear protein specifically binding to oligo(A) contained Ku antigen (29). Various roles of Ku antigen such as double strand break repair or DNA recombination by virtue of its ability to bind DNA ends nonspecifically have been reported (32-34). However, a recent report demonstrating that Ku antigen recognizes a specific internal DNA element in the mouse mammary tumor virus long terminal repeat to inhibit glucocorticoid receptor-mediated transcriptional stimulation (35) prompted us to examine whether Ku antigen is involved in nVDRE_hPTHrP-mediated gene inhibition by vitamin D. As shown in Fig. 6, inclusion of anti-Ku antigen antibody (29) in EMSA significantly reduced the nVDRE_hPTHrP-MT2 protein complex but not the VDRE_mop-MT2 protein complex. Of note, this treatment seemed to strengthen the binding activity of the faster migrating band when we used nVDRE_hPTHrP as a probe. However, as noted earlier, because the migrating position of the faster migrating band overlaps that of the occasionally observed lower band (Fig. 2), we did not pursue this issue in this manuscript.

View larger version (42K): [in this window] [in a new window]   Fig. 6.   Ku antigen exists in the nVDREhPTHrP-MT2 nuclear protein complex but not in the VDREmop-MT2 protein complex. EMSA using MT2 nuclear proteins and nVDREhPTHrP (lanes 1-2) or VDREmop (lanes 3-4) is shown. One microliter of preimmune human serum (lanes 1 and 3) or anti-Ku antigen antibody (lanes 2 and 4, see Ref. 29) was added in each lane. The solid arrows indicate the specific DNA protein complexes (Fig. 2). The upper dashed arrow in lane 2 indicates the supershifted band.

View larger version (47K): [in this window] [in a new window]   Fig. 7.   Involvement of Ku antigen in nVDREhPTHrP-mediated CAT activity by vitamin D. A, immunoblotting of nuclear proteins obtained from the MT2 cells transiently transfected with an expression vector encoding the p70 subunit in the antisense orientation (right; see Ref. 29) or a pRc-cytomegalovirus empty vector (left, E) with the anti-Ku antigen antiserum, which specifically recognizes the p70 subunit (indicated by an arrow). Molecular mass markers are shown on the left side. B, nVDREhPTHrP-TkCAT (left panel) and VDREmop-TkCAT (right panel) were introduced into MT2 cells transiently transfected with a cytomegalovirus promoter-driven pRc/CMVneo empty vector (Invitrogen) (lanes 1 and 2) or with the same vector encoding the p70 subunit of Ku antigen in its antisense orientation (lanes 3 and 4, see A and Ref. 29). After transfection, cells were treated with vehicle alone (lanes 1 and 3) or 107 M 1,25-dihydroxyvitamin D3 (1.25(OH)2D3) (lanes 2 and 4). Typical results are shown. The average CAT activity after three different transfections is indicated below each condition. To focus on the effect of the antisense-oriented Ku antigen on the CAT activity by vitamin D, CAT activity of each of the first lanes was arbitrarily represented as 100.

Phosphatase Treatment of VDR Restores Its Binding to nVDRE_hPTHrP Once Attenuated by Vitamin D-- Recently, Ku antigen was shown to have an inhibitory effect on the glucocorticoid receptor-mediated transcriptional stimulation of the mouse mouse mammary tumor virus long terminal repeat gene (35). It was subsequently shown that the catalytic subunit of Ku antigen (DNA-dependent protein kinase), along with both regulatory subunits of Ku antigen (p70 and p86), phosphorylates glucocorticoid receptor after binding to its specific binding DNA element called negative response element (36). Although the authors did not refer to the direct role of this type of phosphorylation on gene repression (36), their report prompted us to speculate that presumable phosphorylation of VDR by the Ku antigen, which would have been triggered by the treatment with 1,25-dihydroxyvitamin D₃, might weaken its activity to bind to nVDRE_hPTHrP, leading to vitamin D-mediated gene repression. To examine this possibility, we treated nuclear proteins obtained from vehicle- or 1,25-dihydroxyvitamin D₃-administered MT2 cells with potato acid phosphatase for different times and examined their activity to bind nVDRE_hPTHrP by the Southwestern assay as shown in Fig. 8. As expected, the binding of the VDR to nVDRE_hPTHrP, once attenuated by the treatment with 10⁷ µM 1,25-dihydroxyvitamin D₃ (see Fig. 5), but not with vehicle alone, was up-regulated by the phosphatase treatment. On the other hand, the binding of the nonspecific bands above the VDR was not altered by the treatment. These results suggest, albeit indirectly, that phosphorylation of VDR by a certain kinase (e.g. DNA-dependent protein kinase) might decrease its activity to bind to nVDRE_hPTHrP.

View larger version (44K): [in this window] [in a new window]   Fig. 8.   Phosphatase treatment of the nuclear proteins from 1,25-dihydroxyvitamin D3-preadministered MT2 cells up-regulates the binding of VDR to nVDRE>hPTHrP. MT2 nuclear proteins obtained from vehicle alone (V) (first and second lanes) or 1,25-dihydroxyvitamin D3 (1.25(OH)2D3) (107 M)-administered MT2 cells (third and fourth lanes) were treated with 0.1 unit of potato acid phosphatase (PAP) for 0 min (second and fourth lanes) or 30 min (first and third lanes) individually at 37 °C followed by SDS-polyacrylamide gel electrophoresis and Southwestern assay. The radiolabeled nVDREhPTHrP was used as a probe. An arrow indicated the position of VDR (Fig. 5C). Unlike VDR, whose binding activity was up-regulated by phosphatase treatment, several upper nonspecific bands were not affected by the treatment (compare the third and fourth lanes. On the other hand, the intensity of VDR in the first and second lanes did not differ. Molecular mass markers are shown on the left side.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We have shown that the human PTHrP gene contains a DNA element (nVDRE_hPTHrP) very homologous to the negative vitamin D response element found in the human PTH gene (16). Although it is in reverse orientation compared with the nVDRE_hPTH, we demonstrated that nVDRE_hPTHrP functioned similarly to nVDRE_hPTH in terms of transcriptional repression by vitamin D. Sequence comparison between nVDRE_hPTHrP and nVDRE_hPTH revealed 11 identical bases out of 13 nucleotides. Furthermore, it contains a heptameric AG(G/A)TTCA, mimicking a hexameric AG(G/A)TCA motif proposed to be a core element (2) for the binding of the nuclear receptors for thyroid hormone, vitamin D (VDR), or retinoic acid (retinoic acid receptor and RXR). However, unlike VDRE, consisting of two such motifs separated by a three-base spacer (Refs. 2 and 22; see VDRE_mop), nVDRE_hPTHrP and nVDRE_hPTH possess only one motif. This observation led us to speculate that the content of nuclear proteins binding to the nVDREs was different from the usual VDR-RXR heterodimer, as is found in the authentic VDRE (see below). On the other hand, by computer search, we found the nVDRE motif in the 1100-bp upstream region of the mouse TGF2 gene (Fig. 1). 12 out of 13 bases were identical between the two. The only difference lay in the heptameric region. In the chick TGF2 gene, AGATTCA in the hPTHrP gene was changed to AGGTTCA, which was, in turn, identical to the heptamer in the hPTH gene. Very interestingly, expression of this gene in the mesangial cells was reported to be inhibited by 1,25-dihydroxyvitamin D₃ (20). This finding further supports that nVDRE might play a general role in vitamin D-mediated gene inhibition.

A couple of recent reports demonstrated that the rat PTHrP gene contains several DNA sequences responsible for negative gene regulation by vitamin D. Falzon (18) reported two nVDREs in the upstream region of the rat PTHrP gene. He showed that both of them were necessary for vitamin D-mediated gene inhibition. One of them located at the 726-bp upstream region possesses one heptameric sequence, AGGTTCT. This sequence is oriented similar to nVDRE_hPTHrP and in reverse when compared with nVDRE_hPTH. The similarity between nVDRE_hPTHrP and nVDRE_hPTH was found in the sequences adjacent to the heptamer. The five nucleotides, GTGCT, lying 7 to 3 bases preceding the core heptamer were identical between the two nVDREs found in the human and rat PTHrP genes; such identity was not observed in nVDRE_hPTH. Interestingly, this AGGTTCA was repeated with a two-base spacer in the VDRE_mop (22). Like our nVDRE_hPTHrP (Figs. 3 and 5C), the nVDRE in the human PTH gene was reported to bind to VDR as well as another nuclear protein(s) other than RXR (17). On the other hand, Falzon reported another nVDRE at the 805-bp region of the rat PTHrP gene (18). This element resembles a prototype DR3 VDREs, and similar nVDREs mimicking DR3 VDRE were reported to lie at the 1100-bp region of the rat PTHrP gene (19) or at the 80-bp region of the chicken PTH gene (21). VDR and probably RXR could bind to all of them, although it is still unknown why this DR3-type structure was able to mediate negative gene regulation by vitamin D.

Our competition EMSA experiments (Fig. 2) suggested that nVDRE_hPTHrP and VDRE_mop bound to different sets of MT2 nuclear proteins, although each DNA-MT2 nuclear protein complex migrated similarly on a gel. Nonetheless, the use of a 200-fold molar excess of the reciprocal competitors in each EMSA similarly abolished the binding between MT2 nuclear proteins and each probe (not shown), suggesting that each complex might share a common protein(s). The obvious candidate of such a protein is VDR. Our EMSA employing the anti-VDR and anti-RXR antibodies revealed that, unlike the classical interaction between VDR and RXR found in the binding to the authentic VDRE, such as VDRE_mop shown here (Fig. 4), nVDRE_hPTHrP recognized only VDR, and the supershift pattern in EMSA (Fig. 4) predicted that VDR might exhibit a different conformation from that in the binding to RXR, suggesting that another unknown partner (16, 17) might be involved. Although we cannot rule out the possibility that RXRs other than RXR are involved here, the usual partnership composed of VDR and RXR, which is found in the binding to the positive VDRE_mop as shown in Fig. 3, is not predominant in the binding between nVDRE_hPTHrP and MT2 nuclear proteins. This notion was also supported by the Southwestern assay, in which MT2 VDR was immunodepleted by the anti-VDR antibody (Fig. 5C). In this assay, we demonstrated that nVDRE_hPTHrP could not bind to 50-kDa nuclear proteins after immunodepletion, whereas VDRE_mop did not bind to this band irrespective of immunodepletion. This result suggests that VDRE_mop, unlike nVDRE_hPTHrP, can bind to MT2 VDR only when the latter is associated with RXR.

0n the other hand, our experiments using anti-Ku antigen antibody, which recognizes one of its subunits, p70, demonstrated that it supershifted the complex between MT2 nuclear proteins and nVDRE_hPTHrP but not VDRE_mop (Fig. 3). Furthermore, our experiments using the antisense-oriented Ku antigen expression vector (Fig. 6) demonstrated that negative regulation by vitamin D via the nVDRE_hPTHrP was abrogated by the introduction of such a vector. Because such an abrogation was not observed in the case of the VDRE_mop-TkCAT, we concluded that Ku antigen was crucially involved in negative gene regulation by vitamin D in MT2 cells. Although the mode of action of Ku antigen has been currently under extensive study (32-34), only recently has its role as a transcription factor been established (29, 33). Ku antigen consists of two regulatory subunits, p70 and p86, as well as the catalytic unit of another large protein of DNA-dependent protein kinase, DNA-dependent protein kinase (32). Its ability to bind DNA ends nonspecifically could explain its function such as double strand break repair or DNA recombination. We (29) recently demonstrated that p70 and p86 subunits of Ku antigen could interact in association with another nuclear protein, redox factor 1 (1), with one of the negative calcium-responsive elements in a sequence-specific manner. We showed that such an interaction led to extracellular calcium-mediated transcriptional inhibition of the genes bearing negative calcium-responsive elements. Furthermore, of particular interest, the DNA sequence following the heptamer in the nVDRE_hPTHrP, ATTTGGTT, was similar to the partial sequence of one of the negative calcium-responsive elements, oligo(A) (15, 28), ATTTGTGT. The latter sequence was protected from DNase digestion in HeLa cells (15), and we had proposed that Ku antigen along with ref1 protein (29) bound to oligo(A) in a sequence-specific manner. Therefore, we hypothesized that two different portions within the nVDRE_hPTHrP bound to two different nuclear proteins, VDR and Ku antigen, and that protein-protein interaction between VDR and Ku antigen might not occur. Such an independent binding of Ku antigen and another protein(s) to one DNA fragment was reported in mouse mammary tumor virus long terminal repeat where Ku antigen recognized one specific internal DNA sequence 20-30 bases away from the glucocorticoid receptor binding site (35). However, since the examination of the exact DNA binding manner of Ku antigen requires very careful preparation of DNA samples due to the the Ku antigen nature of binding DNA ends nonspecifically, proving this hypothesis has so far been unsuccessful.

The underlying mechanism by which vitamin D weakens the binding between nVDRE_hPTHrP and MT2 VDR as shown in Fig. 5 is currently unclear. Although there have been several reports describing up-regulatory effects of vitamin D on the amount of VDR protein or on its binding activity to VDRE (37), the mechanism by which vitamin D inhibits the binding of VDR to nVDRE has not been addressed. One attractive hypothesis is that Ku antigen might play some role in such negative regulation as shown in this manuscript. Particularly, the recent report suggesting that protein kinase activity of DNA-dependent protein kinase, the catalytic subunit of Ku antigen, would modulate glucocorticoid receptor activity after the specific binding to DNA (35, 36) prompted us to carry out the Southwestern assay in which the VDR from 1,25-dihydroxyvitamin D₃-administered MT2 cells was dephosphorylated by potato acid phosphatase (Fig. 8). Although dephosphorylation of VDR did not affect its binding to nVDRE_hPTHrP when VDR was obtained from MT2 cells treated with vehicle alone, it reversed VDR binding to nVDRE_hPTHrP pre-attenuated by the treatment with 1,25-dihydroxyvitamin D₃. Therefore, we raise the possibility that phosphorylation of VDR by the Ku antigen, which would have been triggered by the treatment with 1,25dihydroxyvitamin D₃, might weaken its activity to bind to nVDRE_hPTHrP followed by vitamin D-mediated gene repression, although direct evidence to support this contention is still lacking.

If vitamin D mediates conditional inhibition of PTHrP gene as suggested here, what is the significance of the constitutively active binding between nVDRE_hPTHrP and MT2 nuclear protein(s) in the unliganded state? We speculate that such a binding might be somewhat related to constitutively active production of PTHrP in such cell lines as MT2 cells, leading to humoral hypercalcemia of malignancy, the most common cause of human hypercalcemia (26). We suggest here that a therapeutic approach to targeting to the transcriptional apparatus would be of potential benefit.

In this paper, we focused on the difference between positive and negative VDREs as described above. Future investigation is directed to the detailed analyses of the interaction between nVDRE_hPTHrP and VDR including mutagenesis of nVDRE_hPTHrP as well as characterization of VDR domain involved here, all of which are now in progress in our laboratory.