Vitamin D Receptor Interactions with the RatParathyroid Hormone Gene: Synergistic Effects Between

Two Negative Vitamin D Response Elements

JOHN RUSSELL,1 SHEELA ASHOK,1 and NICHOLAS J. KOSZEWSKI2

ABSTRACT

Vitamin D response elements (VDREs) that are required for negative regulation of rat parathyroid hormone(rPTH) gene expression have been characterized. Gel mobility shift assays using DNA restriction enzyme frag-ments and recombinant proteins for vitamin D and retinoic acid X receptors (VDR/RXR) revealed a sequencebetween −793 and −779 that bound a VDR/RXR heterodimer with high affinity (VDRE1). Furthermore, a loweraffinity site (VDRE 2) was detected that acted in combination with VDRE1 to bind a second VDR/RXR complex.As determined by ethylation interference analysis, the nucleotide sequence of VDRE1 consisted of GGTTCA GTGAGGTAC, which is remarkably similar to the sequence of the negative VDRE found in the chicken PTH (cPTH)gene. Using the same technique, VDRE2 was identified between positions −760 and −746 and contained thesequence AGGCTA GCC AGTTCA. Functional analysis was determined by transfection studies with plasmidconstructs that expressed the gene for chloramphenicol acetyl transferase (CAT). The ability of the VDREs toregulate gene expression was tested in their native context with the rPTH promoter as well as when positionedimmediately upstream from the cPTH promoter. With either plasmid construct, exposure to 10−8 M 1,25(OH)2D3

resulted in a 60–70% decrease in CAT gene expression when both VDRE1 and VDRE2 were present. Examinationof the individual VDREs showed that inhibition by 10−8 M 1,25(OH)2D3 was only 35–40% when just VDRE1 waspresent. By itself, VDRE2 was even less effective, as significant inhibition of CAT activity (20%) was observed onlyin the presence of higher concentrations of 1,25(OH)2D3 (10−7 M) or when a plasmid vector that overexpressed theVDR protein was cotransfected. In conclusion, the rPTH gene contains two negative VDREs that act in concert tobind two RXR/VDR heterodimer complexes and that both VDREs are required for maximal inhibition by1,25(OH)2D3. (J Bone Miner Res 1999;14:1828–1837)

INTRODUCTION

BIOSYNTHESIS AND SECRETION of parathyroid hormone(PTH), which plays a central role in maintaining cal-

cium homeostasis, are closely regulated by several fac-tors.(1,2) One of the key elements in this regulatory systemis the active metabolite of vitamin D, 1,25-dihydroxy-vitamin D3 (1,25(OH)2D3).(3–5) Binding of 1,25(OH)2D3 toits nuclear receptor results in an increased affinity of thereceptor for specific DNA sequences in the PTH gene andcoordinated inhibition of PTH gene transcription. Previ-ously we identified a vitamin D response element (VDRE)

in the chicken PTH (cPTH) gene that is responsible fornegative regulation of gene transcription by 1,25(OH)2D3

and capable of binding the vitamin D receptor (VDR) as aheterodimer complex with the retinoic acid accessory pro-tein (RXR).(6) The cPTH VDRE sequence is similar toVDREs that have been characterized in genes that are ac-tivated by 1,25(OH)2D3 and also bind VDR/RXR het-erodimers.(7–14) In agreement with the consensus sequencefor positive VDREs, the cPTH VDRE consists of two hexa-meric repeats separated by a 3 bp spacer. Negative VDREswith similar sequence motifs have been identified in re-lated genes. These include the human PTH gene as well

1Albert Einstein College of Medicine, Department of Medicine, Division of Endocrinology, Bronx, New York, U.S.A.2University of Kentucky Medical School, Department of Internal Medicine, Division of Nephrology, Lexington, Kentucky, U.S.A.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 14, Number 11, 1999Blackwell Science, Inc.© 1999 American Society for Bone and Mineral Research

1828

as the genes for rat and human PTH-related protein(PTHrp).(15–19) To further investigate VDREs that are in-volved in negative regulation of gene transcription, we havecharacterized putative regulatory elements in the rat PTH(rPTH) gene. The rPTH gene was of particular interestbecause the effects of 1,25(OH)2D3 on negative regulationof PTH gene expression in the rat have been well charac-terized(5) and the rat has been used as an in vivo model tostudy the effects of 1,25(OH)2D3 and its analogs on para-thyroid gland function in both normal and uremic ani-mals.(20,21) Furthermore, it was important to determinewhether RXR was involved in binding of VDR to rPTHVDREs. This was of considerable significance becausestudies with VDREs from other mammalian PTH genes(human and bovine), as well as human PTHrp, have indi-cated that a protein distinct from RXR forms the hetero-dimer complex with VDR.(19,22,23) The present study de-scribes the analysis of 58-flanking sequences in the rPTHgene that are capable of binding VDR/RXR heterodimersand able to confer negative regulation of gene transcriptionin the presence of 1,25(OH)2D3. Gel mobility shift assayswere performed on restriction enzyme fragments derivedfrom a 925 bp region of the rPTH gene 58-flanking regionusing recombinant proteins for VDR and RXRa. Two frag-ments were identified that bound the VDR/RXRa complexwith disparate affinities. A high-affinity VDRE was locatedbetween positions −925 and −778, whereas a low-affinityVDRE was found between −777 and −646. Further analysisby gel mobility shift assays and DNA footprint studies con-firmed that the two VDREs act in association to bind twoVDR/RXR heterodimer complexes. Functionality was de-termined by transfection studies that utilized plasmid con-structs in which the two VDREs were present in their na-tive context with the rPTH promoter or where they wereplaced immediately upstream from the cPTH promoter.These studies demonstrate that inhibition of rPTH genetranscription by 1,25(OH)2D3 is regulated by two func-tional VDREs.

MATERIALS AND METHODS

Analysis of DdeI restriction fragments

A nucleotide sequence that spanned the region from−925 to +25 of the 58 flanking sequence of the rPTH genewas prepared by polymerase chain reaction (PCR) from arat genomic DNA preparation obtained from Clontech(San Francisco, CA, U.S.A.). The 950 bp DNA segmentwas digested with DdeI, and the resulting restriction frag-ments were end labeled with 32P by filling in the 58 over-hangs with Klenow enzyme. The fragments were separatedby chromatography using 5% polyacrylamide gels and theposition of the individual fragments were determined byautoradiography. The gel slices containing the individualfragments were placed in buffer containing Tris HCl (20mM, pH 8.0), EDTA (1 mM), and 0.1% SDS and the frag-ments were passively eluted overnight at 37°C. The elutedfragments were precipitated by ethanol and reconstituted in20 mM Tris HCl, pH 8.0. Mobility shift assays with thelabeled restriction fragments and either VDR, RXRa, or a

combination of the two was performed as described in thefollowing text.

Plasmid constructions

The cPTH promoter sequence, −55 to +20, was synthe-sized by PCR and ligated into a commercially availableplasmid chloramphenicol acetyl transferase (pCAT) ex-pression vector (Promega, Madison WI, U.S.A.) using theSalI and XbaI restriction sites present in the multiple clon-ing site. DNA sequences containing the two putativeVDREs (VDRE1 and VDRE2) from the rPTH gene weresynthesized by PCR and placed upstream from the cPTHpromoter using the HindIII and SalI sites that were avail-able. Plasmid constructs that contained just VDRE1 orVDRE2 were prepared from synthetic oligonucleotides thatcontained HindIII and SalI sites at their 58 and 38 ends,respectively. To test the functionality of the putativeVDREs in their native context, a 950 bp DNA sequencefrom the 58 flanking region of the rPTH gene that includedthe rPTH promoter and 25 bp downstream from the tran-scription start site was prepared by PCR and ligated intothe previously described CAT-reporter plasmid using theHindIII and SalI restriction sites. As a control, a secondrPTH plasmid construct was prepared that contained therPTH gene sequence from −745 to +25, which does notinclude either of the putative VDRE sequences.

Mobility shift assays

Recombinant human VDR (hVDR) and hRXRa-containing extracts were prepared from baculovirus-infected Sf9 insect cells as described previously.(21) The cellpellets were homogenized in 6 vol of buffer (20 mM Tris[pH 7.5], 1 mM EDTA, 2 mM DTT, 350 mM KCl, 10 mMNaF, 100 mM Na3VO4, 0.1 mM leupeptin, and 10% glyc-erol) on ice. The cytosols were clarified by centrifugation at30,000g for 60 minutes at 4°C, snap-frozen, and stored at−70°C prior to use. DNA probes were generated by end-labeling linearized and calf intestinal alkaline phosphatase(CIP)-treated plasmids with polynucleotide kinase and[g-32P]ATP. Cytosols of recombinant hVDR and hRXRawere diluted 1:50 in ice-cold KTEDG buffer (400 mM KCl,20 mM Tris (pH 7.5), 1 mM EDTA, 2 mM DTT, and 10%glycerol) prior to use. Samples were kept cold and 2 ml eachof diluted extracts were used in a 20 ml final volume (VDRconcentration ∼1–2 nM) in a buffer that was 120 mM KCl,20 mM Tris (pH 7.5), 1.5 mM EDTA, 2 mM DTT, 5%glycerol, 0.5% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), 10 mM NaF, 100 mMNa3VO4, 1.0 mg dI?dC, and 100 nM 1,25(OH)2D3. Follow-ing incubation on ice for 15 minutes, the radiolabeled probewas added and incubation continued for 1 h. The sampleswere then applied to cooled, prerun 5% polyacrylamidegels (29:1) in 0.5× TBE buffer and electrophoresed at 14V/cm for 3 h. Gels were dried and exposed to X-ray filmovernight at − 80°C with an intensifying screen. Cold com-petition with unlabeled chicken vitellogenin II perfect ERE(top strand 4 GATCCCTGGTCAGCGTG ACCGGAG)or human osteocalcin (hOC) VDRE (top strand 4 GATC-CACCGGGTG AACGGGGGCATTGT) was accom-

VITAMIN D RECEPTOR INTERACTIONS 1829

plished by simultaneous addition with the radiolabeledrPTH probe. Antibodies to VDR have been described pre-viously(24) and antibodies to RXRa were obtained fromSanta Cruz Biotechnologies (Santa Cruz, CA, U.S.A.).Nuclear extracts from opossum kidney (OK) cells were pre-pared according to the method of Schreiber et al.(25)

Interference assay

The ethylation interference footprints were obtained aspreviously described.(26–28) Briefly, 32P end-labeled DNAprobes in 50 mM sodium cacodylate buffer (pH 8.0) weretreated with ethylnitrosourea-saturated ethanol for 20 min-utes at 55°C. Following precipitation with sodium acetate/ethanol and reprecipitation (3×), the pellets were washedwith 70% ethanol, dried, and resuspended in water. Ethyl-ated probes were then used in the gel mobility shift assay asabove, except the amounts of probe were increased to 10–15 fmol. Acrylamide sections corresponding to bound andfree DNA were excised, and the DNA was recovered byelectrochemical elution and precipitation. Cleavage of themodified DNA was accomplished by treating with 100 mMNaOH/0.1 mM EDTA in 10 mM phosphate buffer at 95°Cfor 30 minutes followed by neutralization with 3 M sodiumacetate (pH 5.2) and precipitation with ethanol. DNAsamples were separated through 8% sequencing gels, dried,and autoradiography was performed followed by densito-metric analysis.

Transfection studies

OK cells were cotransfected by the method of lipofectionwith the appropriate VDRE–CAT construct (2.5 mg) andCMV–b-galactosidase (1 mg) plasmids. Analysis of CATand b-galactosidase gene transcripts following treatmentwith hormone (10−8 M) for 24 h were quantitated by ribo-nuclease protection assays using a commercially availablekit (Ambion, Austin TX, U.S.A.). Total RNA from OKcells transfected with the various PTH–CAT and CMV–b-galactosidase plasmid constructs was prepared by extrac-tion with phenol-guanidinium thiocyanate. The RNA ex-tracts were incubated with RNAse-free DNAse I (GIBCOBRL, Bethesda, MD, U.S.A.) for 5 minutes at room tem-perature and ethanol precipitated. The RNA pellets wereincubated with 32P-labeled RNA probes (158 bases forCAT and 370 bases for b-galactosidase) in 20 ml of buffercontaining 80% formamide, 100 mM sodium citrate, pH 6.4,300 mM sodium acetate, pH 6.4, 1 mM EDTA, overnight at43°C. The hybridization mixtures were digested with a com-bination of RNAse A and RNAse T1 at 37°C for 30 min-utes. The protected RNA hybrids were precipitated withpropanol–guanidinium thiocynate and separated on 6%nondenaturing polyacrylamide gels. The gels were driedand exposed to X-ray film for 6 h at −80°C. The resultingautoradiograms were quantitated by densitometric scan-ning, and values for the CAT gene transcripts were normal-ized with respect to the values for b-galactosidase genetranscripts.

UV cross-linking

The 58 or 38 oligonucleotides (58 4 GCAGCAAGCCT-TAGGTTCAG T; 38 4 GAGGTACCATATGCAAAC-CCC) of the rPTH VDRE1 were end-labeled with[g-32P]ATP and T4 polynucleotide kinase. The labeled oli-gos were mixed with an excess of the other unlabeled oli-gonucleotide and annealed to the complementary se-quence. Following hybridization, the double-strandedoligonucleotides were purified by 8% polyacrylamide gelelectrophoresis, and the recovered DNA fragments werethen included in incubations as described above for themobility shift assay. The sample tubes were maintained onice and irradiated for 45 minutes at a wavelength of 254 nm.Following irradiation, the binding reactions were mixedwith 2× SDS-PAGE loading buffer, denatured at 95°C for5 minutes, and the proteins separated by 10% SDS-PAGE.Following electrophoresis, the gels were fixed, dried, andautoradiography was performed.

RESULTS

Identification of sequences that bindVDR/RXR heterodimers

To identify sequences in the rPTH gene that are involvedin negative regulation of gene expression by 1,25(OH)2D3,a 950 bp DNA fragment from the 58 flanking region locatedimmediately upstream from the transcription start site wasdigested with DdeI. The DdeI fragments were tested fortheir ability to bind hVDR, hRXRa, or hVDR/hRXRa het-erodimer complexes. Table 1 shows the results of gel shiftassays using DdeI fragments and recombinant hRXR andhVDR proteins. Using this technique, no fragments werefound that bound VDR or RXR alone; however, two frag-ments were identified that bound VDR/RXR as a hetero-dimer. One fragment bound the heterodimer with a rela-tively high affinity, while the other demonstrated a notablylower attraction for the protein complex. The high-affinityresponse element (VDRE1) was located in the DdeI frag-ment that consisted of the nucleotide sequence −925 to−778. The low-affinity response element (VDRE2) was lo-cated in the adjacent fragment that contained the nucleo-

TABLE 1.

rPTH gene sequence VDR VDR/RXR RXR

−925 to −778 − +++ −−777 to −646 − + −−645 to −617 − − −−619 to −194 − − −−193 to −115 − − −−114 to +25 − − −

Gel mobility shift assays of DNA fragments generated by DdeIdigestion of the 58 flanking sequence of the rat PTH promoter.Fragments were tested for their ability to bind recombinant VDRand RXR proteins either by themselves or as VDR/RXR com-plex. Strong binding is represented by +++, weak binding by +,and no binding by −

RUSSELL ET AL.1830

tide sequence from −777 to −646. Preliminary analysis byethylation interference indicated that the sequence ofVDRE1 was located at the 38 end of the DdeI fragmentbetween positions −793 and −779. The sequence of the mi-nus strand of DNA between −793 and −779 is GGTTCAGTG AGGTAC. This is highly homologous to the se-quence of the negative VDRE present in the cPTH gene,GGGTCA AGG GGGTGT. Homology sequence analysisof the DdeI fragment that contained VDRE2 suggested thatthe minus strand of DNA contained a similar VDRE thatconsisted of two hexameric repeats, AGGATC andAGTTCA, and was also separated by a 3 base pair spacer.Based on this information, two separate oligonucleotidesthat contained the proposed sequences for VDRE1 andVDRE2 were prepared. In addition, a composite oligo-nucleotide that included the sequence from −820 to −700and contained the postulated sequences for both VDRE1

and VDRE2 within the native context of the rPTH genewas prepared by PCR.

Assessment of VDR/RXR affinity for VDRE1

and VDRE2

A comparison of the binding affinities of recombinantVDR/RXR for the separate VDRE1 and VDRE2 se-quences is shown in Fig. 1. Within the range of VDR/RXRconcentration used in this experiment, incremental de-creases in amounts of receptor protein had only a smalleffect on binding to VDRE1. In contrast, not only did thehighest concentration of protein receptor bind with less af-finity to VDRE2, but the same dilution in the receptor pro-tein concentration resulted in a more rapid decrease inbinding to VDRE2 than to VDRE1. Because the oligonu-cleotides used in this study contained essentially just thetwo rPTH VDRE sequences, the disparity in affinity is mostlikely due to differences within the structures of the twohalf sites, as opposed to contributions from surroundingsequences.

Analysis of VDR/RXR binding to the compositeoligonucleotide containing both VDREs

Specificity of the VDR/RXR heterodimer complex forthe composite VDRE oligonucleotide was characterizedfurther by cold competition analysis (Fig. 2) and the abilityof VDR and RXR antibodies to supershift the bound com-plexes. With the composite oligonucleotide, neither VDRnor RXR alone formed bound complexes; however, thecombination of VDR and RXR produced two distinctbound complexes. As will be shown later, the ratio of theslower moving complex (B2) to the faster moving complex(B1) increased with increasing protein concentration.VDR/RXR binding to the composite oligonucleotide wascompletely abolished in the presence of an excess of unla-beled oligonucleotide containing the sequence for eitherthe cPTH or the hOC VDRE. In contrast, no competitionfor the VDR/RXR complex was observed in the presenceof an excess of unlabeled oligonucleotide containing thechicken vitellogenin II perfect estrogen response element.

Both of the VDR/RXR heterodimer complexes were su-pershifted by antibodies specific for VDR or RXRa. In Fig.2, the use of recombinant RXRa rather than other isoformsof RXR in the binding studies was somewhat arbitrary, and

FIG. 1. Comparison of VDR/RXR affinity for VDRE1and VDRE2 using gel mobility shift assays. From left toright the concentration of insect cell extract containing re-combinant VDR/RXR proteins was decreased by 2-fold in-crements. The bottom strand of VDRE1 consisted of thesequence GCTCAGGTTCAGTGAGGTACAGCTT andthe bottom strand of VDRE2 consisted of the sequenceGCATCAGGCTAGCCAGTTCAGCTA.

FIG. 2. Characterization of VDR and RXR binding to thecomposite oligognucleotide that contains both VDRE1 andVDRE2 using gel mobility shift assays. This demonstratesthe binding of VDR and RXR alone as compared with thecombination of VDR/RXR as well as the effects of coldcompetitors and antibodies specific for VDR and RXR.Lane 1, VDR containing cytosol added; lane 2, RXRa con-taining cytosol added; lane 3, VDR/RXRa cytosol addedtogether; lane 4, same as lane 3 plus a 400-fold excess ofcPTH VDRE cold competitor; lane 5, same as lane 3 plus a400-fold excess of chicken vitellogenin II perfect ERE coldcompetitor; lane 6, same as lane 3 plus a 400-fold excess ofhOC VDRE cold competitor; lane 7, same as lane 3 plus 1ml of the anti-VDR N-terminal peptide antiserum Ab192;lane 8, same as lane 3 plus normal rabbit serum; lane 9,same as lane 3 plus 1 ml of anti-RXRa N-terminal peptideantiserum.

VITAMIN D RECEPTOR INTERACTIONS 1831

by no means do we intend to suggest that RXRa is the onlyRXR isoform that augments VDR binding to the rPTH andcPTH VDREs. In fact, we have found that both the a andb isoforms of RXR show a similar ability to enhance thebinding of VDR to the VDREs in both the chicken and rat.Although by immunostaining and Western blotting tech-niques, we have identified the presence of RXRa in bothbovine and rat parathyroids.

As noted, the composite oligonucleotide produced aslower migrating complex (B2) that became more promi-nent as the concentration of VDR/RXR was increased, asshown in Fig. 3. To determine the degree of interactionbetween the two VDREs, the percentage of free probe,complex B1, and complex B2 were plotted as a function ofthe log of the VDR/RXR protein concentration. Based onthe parameters set forth by Senear and Brenowitz(29) forbinding to two dissimilar sites, the heterodimer appearsto interact with the rPTH VDREs in a noncooperativemanner.

DNA footprint analysis of the boundheterodimer complexes

DNA footprint analysis (Fig. 4) of the B1 complexshowed base contacts along the two direct repeats,

GGTTCA and AGGTAC, that previously had been iden-tified as VDRE1. Comparable analysis of the slower mi-grating band (B2) displayed the same base contacts identi-fied in B1 as well as additional contacts with the postulatedsequence for VDRE2, AGGATC GCC AGTTCA. Thetwo different footprints produced with B2 strongly suggestthat B2 results from binding of VDR/RXR complexes toboth VDRE1 and VDRE2, while B1 contains only a singleheterodimer complex bound to VDRE1.

Gel shift analysis using OK cell nuclear extracts

Since OK cells were to be used to test the functionality ofthe rPTH VDREs, nuclear extracts from these cells wereexamined for their ability to bind the rPTH VDREs usinggel shift assays. As shown in Fig. 5, the nuclear extracts, likethe recombinant proteins, produced similar B1 and B2 com-plexes, with the B2 complex becoming more predominantat increased protein concentrations. As with the recombi-nant proteins, both complexes formed by the nuclear ex-tracts were abolished by cold competition with an excess ofeither the hOC or cPTH VDRE.

Functional analysis by transfection in OK cells

To assess the functionality of the two putative VDREs,plasmid constructs were prepared that expressed the genefor chloramphenicol acetyl transferase (CAT) from eitherthe chicken PTH promoter (rPTHcp) or the rat PTH pro-

FIG. 3. Analysis of VDR/RXR concentration on forma-tion of the B1 and B2 complexes. At the top is an autora-diogram of a gel shift assay that has been rotated 90° thatthe top of the gel is on the left and the bottom is on theright. Designations F, B1, and B2 correspond to free probeand bound complexes B1 and B2, respectively. Scatchardanalysis of the data is shown below.

FIG. 4. DNA footprint analysis by ethylation interferenceof complexes B1 and B2 formed by the two rPTH VDREsin the presence of VDR and RXR proteins. F, free probe;Cont, a control digest of unbound probe; and B1 and B2,the two bound complexes that were observed in Fig. 2.

RUSSELL ET AL.1832

moter (rPTHrp) (Fig. 6). The latter was used to verify theeffectiveness of the two VDREs to regulate gene transcrip-tion in their native context. This construct consisted of 925bp of the rPTH gene 58 flanking sequence that is immedi-ately upstream from the transcription start site plus the first25 bp of transcribed sequence. It included both VDREs aswell as the rPTH promoter sequence. As a control, a varia-tion of the rPTHrp construct was prepared (rPTHrp–VDREs) that contained the rPTH gene sequence from−745 to +25. In the rPTH–VDREs construct both of theputative VDREs are absent. The other type of plasmid con-struct used in this study, rPTHcp, contained the cPTH pro-moter gene sequence from −55 to +20. An oligonucleotidethat contained both rPTH VDRE1 and VDRE2 was placedimmediately upstream from the cPTH promoter:CAT genecassette. To determine how the two VDREs functionedindependently, the sequence containing both VDREs wasreplaced by DNA fragments that contained either VDRE1

alone (rPTHcp VDRE1) or just VDRE2 (rPTHcpVDRE2). The various plasmid constructs were used totransfect OK cells and the effects of 10−8 M 1,25(OH)2D3

on CAT gene transcription were measured by ribonucleaseprotection assay. As shown in Fig. 7A, the ability of1,25(OH)2D3 to inhibit CAT gene expression was similar(60–70%) for both the rPTHcp and the rPTHrp constructs.Furthermore, the inhibitory effect of 1,25(OH)2D3 was en-hanced when a plasmid construct that expressed the hVDRgene was cotransfected with rPTHcp (Fig. 7B). Similar re-sults were also obtained with rPTHrp (data not shown). Asummary of the data obtained from transfection studieswith all of the plasmid constructs is shown in Fig. 8. Cellstransfected with rPTHrp–VDREs, in which both VDREswere absent, showed no significant inhibition of CAT genetranscription in the presence of 1,25(OH)2D3. Addition of1,25(OH)2D3 to cells transfected with rPTH VDRE1 re-sulted in a notable reduction in CAT gene expression (35–40%), but the decrease was significantly less than that ob-served in studies with the rPTHcp construct that contained

both VDREs. In accordance with the protein binding stud-ies, VDRE2 was even less effective than VDRE1 in its abil-ity to repress gene transcription. In cells transfected withrPTHcp VDRE2, inhibition of CAT gene transcription by1,25(OH)2D3 was significant (∼20%) only when higher con-centrations of hormone were used (10−7 M) or the cellswere cotransfected with the VDR expression vector. In ad-dition, the two rat VDREs were more effective in theirability to inhibit gene transcription than the single VDREidentified in the cPTH gene. The degree of transcriptionalinhibition achieved with 10 nmol 1,25(OH)2D3 in the pres-ence of the single cPTH VDRE could be achieved with aslittle as 2 nmol 1,25(OH)2D3 when the two rPTH VDREswere present (data not shown). All transfection studieswere performed in quadruplicate and repeated at leasttwice.

Order of VDR/RXR binding to VDRE1

Previous studies with the cPTH VDRE to resolve thepolarity of VDR/RXR binding demonstrated that VDRbound the 58 half site, while RXR preferred the 38 halfsite.(30) This was in contrast to similar studies with positiveVDREs, where VDR favored the 38 half site and RXR the58 half site.(31,32) Because of the preferred binding of VDRto GGTTCA sequences(33,34) and the striking sequence ho-mology between rPTH VDRE1 and the cPTH VDRE, itwas of interest to determine whether the same polarity ofheterodimer binding was preserved in rPTH VDRE1. Dataobtained from UV cross-linking studies using gapped oli-gonucleotides containing the rPTH VDRE1 half-site se-quences are shown in Fig. 9. Analysis of the 58 half-siterevealed a covalently linked protein–oligonucleotide com-plex that coincided with a molecular weight anticipated ifthe complex contained the VDR protein. Although otherbound complexes were observed, their presence was greatlydiminished by the addition of a nonspecific competitor oli-gonucleotide that contained the sequence for the estrogenresponse element. Conversely, competition with an oligo-nucleotide that contained the sequence for the hOC VDREcompletely eliminated the band that migrated in the posi-tion of VDR. Furthermore, cross-linking to the 38 half-siteclearly showed a single band that migrated with a molecularweight predicted if RXR were present in the bound com-plex.

DISCUSSION

Cooperativity between two positive VDREs was first de-scribed by Liu and Freedman in studies that made use of asynthetic oligonucleotide where two Spp-1 VDRE se-quences were arranged in tandem.(35) More recently, syn-ergism between two natural positive VDREs was charac-terized in the promoter of the 25-hydroxyvitamin D3 24-hydroxylase (CYP24) gene.(36) Unlike the studies with thesynthetic oligonucleotide, synergy was detected betweenthe two VDREs in the CYP24 gene at the level of geneactivation, but not in protein-binding assays.(37,38) Similarly,

FIG. 5. Gel shift analysis using nuclear extracts from OKcells. The oligonucleotide used was the same one used inthe previous studies that contained both rPTH VDREs.Lane 1, 1 mg of OK cell nuclear extract; lane 2, 2.5 mg of OKcell nuclear extract; lane 3, same as lane 2 plus a 400-foldexcess of cPTH VDRE cold competitor; lane 4, same aslane 4 plus a 400-fold excess of hOC VDRE cold competi-tor.

VITAMIN D RECEPTOR INTERACTIONS 1833

the rPTH VDREs demonstrated cooperation only in termsof functional activity. The synergy between the two rPTHVDREs conceivably could create an additional level in thecontrol of rPTH gene transcription by 1,25(OH)2D3. Intransfection studies, the plasmid construct where the tworPTH VDREs were substituted for the single cPTH VDREwas more sensitive to the effects of 1,25(OH)2D3 by nearlyan order of magnitude. Why the rat would require an in-creased sensitivity to 1,25(OH)2D3 is purely speculative atthis point. It may be due to the fact that rats are essentiallynocturnal animals, and limited exposure to sunlight mayresult in low levels of circulating 1,25(OH)2D3. This expla-nation may be oversimplified because studies with the bo-

vine PTH gene have suggested that it may also contain twoVDR elements.(39) However, the precise sequences that areinvolved and whether the two elements act cooperativelyhave yet to be elucidated. Furthermore, it should be notedthat while 1,25(OH)2D3 has an important role in the regu-lation of PTH biosynthesis, calcium remains the dominantplayer. Studies with vitamin D–deficient chickens andrats(40,41) as well as mice in which the genomic effects of theVDR have been ablated,(42,43) have shown that a high cal-cium diet can reverse the resultant hyperparathyroidism inthe absence of regulation by 1,25(OH)2D3. Although thereversal in the vitamin D–deficient animals was greatly aug-mented when 1,25(OH)2D3 was included in the treatment.

VDREs that are involved in negative regulation of genetranscription have been described for a number of genes. Inaddition to the PTH and PTHrp genes, these include thegenes for calcitonin, atrial natruitic factor, rat bone sia-loprotein, interleukin-2, and beta 3 integrin.(44–48) Al-though negative VDREs have yet to be characterized asextensively as VDREs that activate gene transcription, it isapparent that even at the current level of characterizationthere is a notable correlation. Analogous to the positiveVDREs, most of the negative VDREs described thus farconsist of two direct hexameric repeats separated by a 3 bpspacer and bind the VDR as a heterodimer. Protein bindingstudies with PTH VDREs from both the chicken and therat demonstrated that RXR was capable of forming theheterodimer complex with VDR. This is in contrast to stud-ies with the bovine and human PTH genes, as well as stud-ies with the human PTHrp gene, which suggest that a pro-tein distinct from RXR can also bind with VDR to form theheterodimer complex.

Despite the similarities that exist between VDREs thatactivate transcription and the negative VDREs identified inboth the chicken and the rPTH genes, one of the salientdifferences detected during the course of these studies was

FIG. 6. Plasmid constructs used in transfection studies. All plasmid constructs used expressed the CAT gene. rPTHcpcontained the two rPTH VDRE sequences placed immediately upstream from a 75 bp DNA fragment that contained thecPTH promoter sequence. rPTH contained a 950 bp fragment from the 58 flanking sequence of the rPTH gene thatincluded the rPTH promoter and both VDRE sequences. rPTH–VDREs was the same as rPTH with the exception thatboth VDRE sequences were absent. rPTHcp VDRE1 contained just the high-affinity VDRE sequence and rPTHcpVDRE2 contained only the low-affinity VDRE sequence.

FIG. 7. (A) Analysis of CAT gene transcripts by RNAseprotection assay. In lanes 1 and 2, OK cells were transfectedwith plasmid constructs rPTHcp (lane 1) and rPTHrp (lane2) in the absence of 1,25(OH)2D3. In lanes 3 and 4, OK cellstransfected with rPTHcp (lane 3) and rPTHrp (lane 4) wereexposed to 10−8 M 1,25(OH)2D3. In these transfection stud-ies, OK cells were cotransfected with a plasmid that ex-pressed the b-galactosidase gene as a control for transfec-tion efficiency. (B) OK cells were transfected with rPTHcpand a plasmid vector that overexpressed the gene for thehVDR. In the left lane, OK cells were transfected in theabsence of 1,25(OH)2D3. In the middle and right lanes,transfected cells were exposed to 10−8 M and 10−7 M1,25(OH)2D3, respectively. In these studies, measurementof VDR transcription was used as an internal control.

RUSSELL ET AL.1834

the polarity of VDR/RXR binding to the two VDRE halfsites. Studies with positive VDREs have shown that VDRhas a preference for the 38 half-site and RXR the 58 halfsite. Using the technique of UV cross-linking to determinepolarity of VDR/RXR binding to the rPTH VDRE1 re-vealed an opposite orientation for the heterodimer. In thisinstance, VDR bound to the 58 half site and RXR to the 38half site. This corresponded with earlier studies that showeda similar bias for VDR/RXR heterodimer binding to thecPTH VDRE.(30) Negative regulation of gene transcriptioncan occur by a number of mechanisms. One possibility isthat binding of VDR can preclude the binding of a tran-scription factor that is required for gene activation. Nega-tive regulation by 1,25(OH)2D3 in the genes for calcitonin,bone sialoprotein, beta 3 integrin, and interleukin-2 havebeen adequately explained by this mechanism. This para-digm may not be sufficient to explain transcriptional inhi-bition in the PTH gene because it has been possible tointroduce mutations into the cPTH VDRE sequence that

induced transcriptional activation in the presence of1,25(OH)2D3.(30) Another possibility is that the sequenceof the response element itself can affect the conformationof the bound protein complex. This concept has been sug-gested by studies with both steroid and vitamin D recep-tors.(32,49–51) Because the binding of the VDR heterodimerto both the chicken and rPTH VDREs is intrinsically dif-ferent than the way it binds to positive elements, confor-mational changes in the VDR hetrodimer could result ininteractions that interfere with factors that normally acti-vate transcription or even attract novel factors that are in-volved in silencing of gene transcription.

ACKNOWLEDGMENT

The authors would like to acknowledge Ms. HolliGravatte for her excellent technical assistance. This workwas supported by NIH grants DK-47883 (N.J.K.) and DK-38422 (J.R.).

REFERENCES

1. Jupner H, Brown EM, Kronenberg HM 1999 Parathyroid hor-mone. In: Favus, MJ (ed.) Primer on the Metabolic BoneDiseases and Disorders of Metabolism, 4th ed. Lippincott,Williams, and Wilkins, Philadelphia, PA, U.S.A.

2. Marks KH, Kilav R, Naveh-Many T, Silver J 1996 Calcium,phosphate, vitamin D, and the parathyroid. Pediatr Nephrol10:364–367.

3. Silver J, Russell J, Sherwood LM 1985 Regulation by vitaminD metabolites of messenger ribonucleic acid for prepropara-thyroid hormone in isolated bovine parathyroid cells. Proc NatlAcad Sci USA 82:4270–4273.

4. Russell J, Lettieri D, Sherwood LM 1986 Suppression by1,25(OH)2D3 of transcription of the bovine parathyroid hor-mone gene. Endocrinology 117:2114–2116.

5. Silver J, Naveh-Many T, Meyer H, Schmeizer HJ, PopovtzerMM 1986 Regulation by vitamin D metabolites of parathyroidhormone gene transcription in vivo in the rat. J Clin Invest78:1296–1301.

6. Liu SM, Koszewski N, Lapuz M, Malluche HH, Olivera A,Russell J 1996 Characterization of a response element in the

FIG. 8. Effects of 10−8 M 1,25(OH)2D3 on CAT gene expression in OK cells transfected with the various rPTH plasmidconstructs. *Significant inhibition of CAT gene expression (p < 0.05).

FIG. 9. UV cross-linking with recombinant VDR andRXR to gapped, radiolabeled oligonucleotides containingthe 58 and 38 VDRE1 sequences. Con, control binding con-ditions; hOC, a 200-fold excess of unlabeled hOC probewas added to the binding reaction prior to cross-linking;ERE, a 200-fold excess of chicken vitellogenin II perfectresponse element was added to the binding reaction priorto cross-linking.

VITAMIN D RECEPTOR INTERACTIONS 1835

58-flanking region of the avian (chicken) PTH gene that medi-ates negative regulation of gene transcription by 1,25-dihydroxyvitamin D3 and binds the vitamin D3 receptor. MolEndocrinol 10:206–215.

7. Kerner SA, Scott RA, Pike JW 1989 Sequence elements in thehuman osteocalcin gene confer basal activation and inducibleresponse to hormonal vitamin D3. Proc Natl Acad Sci USA86:4455–4459.

8. Morrison NA, Shine J, Fragonas J-C, Verkest V, McMenemyMI, Eisman JA 1989 1,25-dihydroxyvitamin D-response ele-ment and glucocorticoid repression in the osteocalcin gene.Science 246:1158–1161.

9. Demay MB, Roth DA, Kronenberg HM 1989 Regions of theosteocalcin gene which mediate the effect of 1,25-dihydroxy-vitamin D3 on gene transcription. J Biol Chem 264:2279–2282.

10. Noda M, Vogel RL, Craig AM, Prahl J, DeLuca HD, DenhardtDT 1990 Identification of a DNA sequence responsible forbinding of the 1,25-dihydroxyvitamin D3 receptor and 1,25-dihydroxyvitamin D3 enhancement of mouse secreted phos-phoprotein 1 (Spp-1 or osteopontin) gene expression. ProcNatl Acad Sci USA 87:9995–9999.

11. Umesono K, Murakami KK, Thompson CC, Evans RM 1991Direct repeats as selective response elements for the thyroidhormone, retinoic acid, and vitamin D3 receptors. Cell65:1255–1266.

12. Darwish HM, DeLuca HF 1992 Identification of a 1,25-dihydroxyvitamin D3 response element in the 58 flanking re-gion of the rat calbindin D-9k gene. Proc Natl Acad Sci USA89:603–607.

13. Zierold C, Darwish HM, DeLuca HF 1995 Two vitamin Dresponse elements function in the rat 1,25-dihydroxyvitamin D24-hydroxylase promoter. J Biol Chem 270:1675–1678.

14. Yu UC, Delsert C, Andersen B, Holloway JM, De Vary OV,Naar AM, Kim SY, Boutin J-M, Glass CK, Rosenfeld MG1991 RXRb: A co-regulator that enhances binding of retinoicacid, thyroid hormone and vitamin D receptors to their cog-nate response elements. Cell 67:1251–1266.

15. Demay MB, Kiernan SM, DeLuca HF, Kronenberg HM 1992Sequences in the human parathyroid hormone gene that bindthe 1,25(OH)2D3 receptor and mediate transcriptional repres-sion in response to 1,25-dihydroxyvitamin D3. Proc Natl AcadSci USA 87:8097–8101.

16. Kremer R, Sebag M, Champigny C, Meerovitch K, Hendy GN,White J, Goltzman D 1996 Identification and characterizationof 1,25-dihydroxyvitamin D3 responsive repressor sequences inthe rat parathyroid hormone-related peptide gene. J BiolChem 271:16310–16316.

17. Falzon M 1996 DNA sequences in the rat parathyroid hor-mone-re1ated peptide gene responsible for 1,25-dihydroxy-vitamin D3-mediated transcriptional repression. Mol Endocri-nol 10:672–681.

18. Darwish HM, DeLuca HF 1996 Analysis of binding of the1,25-dihydroxyvitamin D3 receptor to positive and negative vi-tamin D response elements. Arch Biochem Biophys 334:223–234.

19. Nishishita T, Okazaki T, Ishikawa T, Igarashi T, Hata K, OgataE, Fujita T 1998 A negative vitamin D response DNA elementin the human parathyroid hormone-related peptide gene bindsto vitamin D receptor along with Ku antigen to mediate nega-tive gene regulation by vitamin D. J Biol Chem 273:10901–10907.

20. Brown AJ, Zhong M, Finch J, Ritter C, Slatopolsky E 1995 Theroles of calcium and 1,25-dihydroxyvitamin D3 in the regula-tion of vitamin D receptor expression by rat parathyroidglands. Endocrinology 136:1419–1425.

21. Takahashi F, Finch JL, Denda M, Dusso AS, Brown AJ, Sla-topolsky E 1997 A new analog of 1,25-(OH)2D3, 19-NOR-1,25-(OH)2D2, suppresses serum PTH and parathyroid glandgrowth in uremic rats without elevation of intestinal vitamin Dreceptor content. Am J Kidney Dis 30:105–112.

22. Hawa NS, O’Riordan JL, Farrow SM 1996 Functional analysisof vitamin D response elements in the parathyroid hormone

gene and a comparison with the osteocalcin gene. BiochemBiophys Res Commun 228:352–357.

23. Mackey SL, Heymont JL, Kronenberg HM, Demay MB 1996Vitamin D receptor binding to the negative human parathyroidhormone vitamin D response element does not require theretinoid × receptor. Mol Endocrinol 10:298–305.

24. Koszewski NJ, Reinhardt TA, Langub MC, Malluche HH,Horst RL 1998 Selectivity of a C-terminal peptide antiserumfor different DNA-binding states of the vitamin D receptor.Arch Biochem Biophys 349:388–396.

25. Schreiber E, Matthias P, Miller MM, Schaffner W 1989 Rapiddetection of octamer binding proteins with ‘mini-extracts’ froma small number of cells. Nucleic Acids Res 17:6419–6423.

26. Koszewski NJ, Reinhardt TA, Horst RL 1996 Vitamin D re-ceptor interactions with the murine osteopontin response ele-ment. J Steroid Biochem Mol Biol 59:377–388.

27. Koszewski NJ, Notides AC 1991 Phosphate-sensitive bindingof the estrogen receptor to its response elements. Mol Endo-crinol 5:1129–1136.

28. Siebenlist U, Gilbert W 1980 Contacts between Escherichiacoli RNA polymerase and an early promoter of phage T7. ProcNatl Acad Sci USA 77:122–126.

29. Senear DF, Brenowitz M 1991 Determination of binding con-stants for cooperative site-specific protein–DNA interactionsusing the gel mobility-shift assay. J Biol Chem 266:13661–13671.

30. Koszewski NJ, Ashok S, Russell J 1998 Turning a negative intoa positive: Vitamin D receptor interactions with the avian para-thyroid hormone response element. Mol Endocrinol 13:455–465.

31. Schrader M, Muller KM, Nayeri S, Kahien JP, Carlberg C 1994Vitamin D3-thyroid hormone receptor heterodimer polarity di-rects ligand sensitivity of transactivation. Nature 370:382–386.

32. Staal A, van Wijnen AJ, Birkenhager JC, Pols HA, Prahl J,DeLuca H, Gaub MP, Lian JB, Stein GS, van Leeuwen JP,Stein JL 1996 Distinct conformations of vitamin D receptor/retinoid receptor-a heterodimers are specified by dinucleotidedifferences in the vitamin D-responsive elements of the osteo-calcin and osteopontin genes. Mol Endocrinol 10:1444–1456.

33. Nishikawa J, Kitaura M, Matsumoto M, Imagawa M, NishiharaT 1994 Difference and similarity of DNA sequence recognizedby VDR homodimer and VDR/RXR heterodimer. NucleicAcids Res 22:2902–2907.

34. Freedman LP, Arce V, Fernandez RP 1994 DNA sequencesthat act as high affinity targets for the vitamin D3 receptor inthe absence of the retinoid X receptor. Mol Endocrinol 8:265–273.

35. Liu M, Freedman LP 1994 Transcriptional synergism betweenthe vitamin D3 receptor and other nonreceptor transcriptionfactors. Mol Endocrinol 8:1593–1604.

36. Ohyama Y, Ozono K, Uchida M, Shinki T, Kato S, Suda T,Yamamoto O, Noshiro M, Kato Y 1994 Identification of avitamin D-responsive element in the 58-flanking region of therat 25-hydroxyvitamin D3, 24–hydroxylase gene. J Biol Chem269:10545–10550.

37. Ohyama Y, Ozono K, Uchida M, Yoshimura M, Shinki T, SudaT, Yamamoto O 1996 Functional assessment of two vitaminD-responsive elements in the rat 25-hydroxyvitamin D3,24–hydroxylase gene. J Biol Chem 271:30381–30385.

38. Kerry DM, Dwivedi PP, Hahn CN, Morris HA, Omdahl JL,May BK 1996 Transcriptional synergism between vitamin D-responsive elements in the rat 25-hydroxyvitamin D3, 24–hydroxylase (CYP24) promoter. J Biol Chem 271:29715–29721.

39. Farrow SM, Hawa NS, Karmali R, Hewison M, Walters JC,O’Riordan JL 1990 Binding of the receptor for 1,25-dihydroxyvitamin D3 to the 58-flanking region of the bovineparathyroid hormone gene. J Endocrinol 126:55–359.

40. Russell J, Bar A, Sherwood LM, Hurwitz S 1993 Interactionbetween calcium and 1,25-dihydroxyvitamin D3 in the regula-tion of preproparathyroid hormone and vitamin D receptormessenger ribonucleic acid in avian parathyroids. Endocrinol-ogy 132:2639–2644.

RUSSELL ET AL.1836

41. Kollenkirchen U, Fox J, Walters MR 1991 Normocalcemiawithout hyperparathyroidism in vitamin D-deficient rats. JBone Miner Res 6:273–278.

42. Li YC, Amling M, Pirro AE, Priemel M, Meuse J, Baron R,Delling G, Demay MB 1998 Normalization of mineral ion ho-meostasis by dietary means prevents hyperparathyroidism,rickets, and osteomalacia, but not alopecia in vitamin D recep-tor-ablated mice. Endocrinology 139:4391–4396.

43. Li YC, Pirro AE, Amling M, Delling G, Baron R, Bronson R,Demay MB 1997 Targeted ablation of the vitamin D receptor:An animal model of vitamin D-dependent rickets type II withalopecia. Proc Natl Acad Sci USA 94:9831–9835.

44. Silver J, Naveh-Many T 1993 Calcitonin gene regulation invivo. Horm Metab Res 25:470–472.

45. Li Q, Gardner DG 1994 Negative regulation of the humanatrial natriuretic peptide gene by 1,25-dihydroxyvitamin D3. JBiol Chem 269:4934–4939.

46. Alroy I, Towers TL, Freedman LP 1995 Transcriptional repres-sion of the interleukin-2 gene by vitamin D3: Direct inhibitionof NFATp/AP-1 complex formation by a nuclear hormone re-ceptor. Mol Cell Biol 10:5789–5799.

47. Li J, Sodek J 1993 Cloning and characterization of the rat bonesialoprotein gene promoter. Biochem J 289:625–629.

48. Cao X, Teitelbaum SL, Zhu HJ, Zhang L, Feng X, Ross FP1996 Competition for a unique response element mediates ret-inoic acid inhibition of vitamin D3-stimulated transcription. JBiol Chem 271:20650–20654.

49. Sakai DD, Helms S, Carlstedt-Duke J, Gustafsson JA, Rott-man FM, Yamamoto KR 1988 Hormone-mediated repression:A negative glucocorticoid response element from the bovineprolactin gene. Genes Dev 2:1144–1154.

50. Diamond MI, Miner JN, Yoshinaga SK, Yamamoto KR 1990Transcription factor interactions: Selectors of positive or nega-tive regulation from a single DNA element. Science 249:1266–1272.

51. Lian JB, Shaloub V, Aslam F, Frenkel B, Green J, Hamrah M,Stein GS, Stein JL 1997 Species-specific glucocoticoid and 1,25-dihydroxyvitamin D3 reponsiveness in mouse MC3T3–E1 os-teoblasts: Dexamethasone inhibits osteoblast differentiationand vitamin D down-regulates osteocalcin gene expression.Endocrinology 138:2117–2127.

Address reprint requests to:John Russell

Albert Einstein College of MedicineDepartment of Medicine

Division of EndocrinologyBronx, NY 10461 U.S.A.

Received in original form June 9, 1998; in revised form February22, 1999; accepted February 22, 1999.

VITAMIN D RECEPTOR INTERACTIONS 1837