| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vincent T Lombardi Cancer Center, Department of Biochemistry and Molecular Biology, 3900 Reservoir Road, Georgetown University Medical Center, Washington, DC 20007
(Requests for offprints should be addressed to K R McGaffin, University of Pittsburgh Medical Center, Scaife Hall S559, 200 Lothrop Street, Pittsburgh, Pennsylvania, Email:krmcgaffin{at}aol.com.)
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
1,25-dihydroxyvitamin D3 has been shown to limit proliferation of and promote differentiation in breast cancer cells (Welsh 2003). In earlier studies (McGaffin et al. 2004), we suggested that vitamin D may partially mediate its biologic effect on growth in breast cancer cells by the differential modulation of oncogene products and growth factor receptors such as EGFR. In support of this, we have demonstrated that those breast cancer cell lines with lower levels of EGFR expression, specifically MCF7 and T47D, had the greatest amount of growth inhibition when treated with 1µM of 1,25-dihydroxyvitamin D3 or analog C (a 1,25-dihydroxyvitamin D3 analog with chemical name 1
,25-(OH)2–16-en-23-yn-26,27-F6-vitamin D3). In contrast, those cell lines with higher levels of EGFR expression, such as BT474 and BT549, responded less significantly, or not at all. We observed that downregulation of EGFR expression after vitamin D treatment correlated with growth inhibition only for the MCF7 and T47D cell lines. While growth inhibition was observed in BT474 cells, they unexpectedly showed EGFR upregulation. Further, BT549 cells showed no significant growth inhibition in the face of significant EGFR downregulation. These discordant results suggested to us that the growth inhibitory and EGFR downregulatory effects of vitamin D are cell specific and not necessarily interdependent phenomenon.
Recently, Cordero et al.(2002) and Dusso et al.(2004) have also linked vitamin D to the regulation of EGFR expression and/or cell growth. Specifically, Cordero et al.(2002) examined the repressive effect of vitamin D on EGFR membrane traffcking, EGFR mediated growth signaling, and the ability of nuclear EGFR to bind to cognate DNA sequences in EGFR over-expressing A431 cells. Dusso et al.(2004) extended this work by demonstrating that vitamin D induces growth arrest in rat parathyroid tissue by suppressing EGFR-growth signals from the plasma membrane, as well as by suppression of nuclear EGFR activity. Neither group, however, demonstrated a biochemical mechanism by which vitamin D regulates EGFR expression.
The focus of our present work was therefore to characterize at the molecular level the details of vitamin D mediated EGFR expression in BT549 breast cancer cells. Given that EGFR is a known negative prognostic marker in terms of breast cancer treatment and survival (Bundred 2001), understanding ways in which it can be downregulated has potentially important clinical implications. The purpose of this study using the BT549 cell line, which demonstrates EGFR downregulation in the face of vitamin D treatment, was to characterize the biochemistry involved in this process, and in so doing, help to shed light on whether or not therapy with vitamin D has the potential to improve overall prognosis in breast cancer.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
We have previously reported on the origin of pJFCAT 840 (McGaffin et al. 2004), containing 840 bases of the EGFR promoter between BglII and SacI restriction sites, or between nucleotide positions –860 and –20, respectively. To generate pJFCAT 536, the SacII to SacII fragment of the EGFR promoter spanning positions –536 to –423 was first subcloned into the corresponding site of Bluescript II KS+ in the forward orientation. This clone was then cut at the BamHI polylinker site and the EGFR Bsu36I site (at position –478) to obtain a 83 bp fragment. This 83 bp fragment was then subcloned into the corresponding sites of pJFCAT 840, resulting in the net loss of EGFR sequence between –860 and –536. pJFCAT 478 was obtained by removal of the EGFR promoter sequence between BamHI and Bsu36I sites of pJFCAT 840 and subsequent ligation of Klenow-blunted ends.
For minimal heterologous promoter constructs, Promega’s (Madison, WI, USA) pCAT enhancer vector, which contains a simian virus 40 (SV40) enhancer subcloned downstream of CAT, was used as the backbone plasmid to generate pJFECAT TATA. The HindIII fragment of pJFCAT containing the poly A trimer cassette was inserted into the corresponding polylinker site of the pCAT enhancer vector. In addition to the poly A sequence, this added a BamHI site into the polylinker region of the pCAT enhancer vector, 5' to the SalI site. A synthetic TATA box sequence flanked by a 5' Bsu36I site and a 3' ClaI site was obtained by annealing sense and antisense oligonuclotides of the sequence GCAAGCCCTCAGGTATAAAACCATCG ATGGAAGC. This TATA box sequence was then cut with ClaI and subcloned into the SmaI and ClaI sites of Bluescript II KS+ to allow for confirmation of a single copy insertion and annealing of correct sense and antisense products by sequencing. The BamHI to SalI fragment of this subclone containing the TATA box was then ligated into the corresponding sites of the pCAT enhancer vector containing the poly A trimer cassette, keeping it in the forward orientation, to generate pJFECAT TATA.
Wild type EGFR promoter sequence between the SacII and Bsu36I sites (positions –536 to –478) was isolated from the Bluescript II KS+ subclone containing the SacII to SacII sequence (described above) by BamHI and Bsu36I digestion and subcloned into the corresponding sites of pJFECAT TATA to create the ‘wild type’ pJFECAT TATA construct. pJFECAT TATA containing the Sp1 mutant sequence was generated by removal of wild type sequence in the latter clone with SacII and Bsu36I restriction enzymes and replacement with Sp1 mutant sequence (see below) cut with these same 5' and 3' flanking restriction enzymes. Vitamin D response element (VDRE) and double VDRE/Sp1 mutant containing pJFECAT TATA constructs were generated from ligation of blunt 5' and Bsu36I cut 3' DNA fragments into 3' Bsu36I and 5' klenow blunted NotI sites found within the pJFECAT TATA construct.
Site directed mutagenesis
Mutations were introduced into putative nuclear factor binding sites by PCR amplification of EGFR wild type SacII to klenow blunted Bsu36I sequence cloned into the SacII and SmaI sites of Bluescript II KS+. Blunting of the Bsu36I site and subsequent ligation into the SmaI site of Bluescript II KS+ regenerated the Bsu36I site. Using this template with a combination of synthetic oligonucleotides incorporating mismatched bases and commercially available universal sequencing primers, the Sp1, VDRE, and double VDRE/Sp1 mutants were obtained. The identity of all PCR products was confirmed by restriction digestion with unique sites inserted into each mutant, as well as sequencing of fragments after subcloning into Bluescript II KS+. To generate the Sp1 mutant, a synthetic oligonucleotide of sequence GCGGTGCCCTGAGGAGTTAATTTCCC GAGAGatGCaTTCCCAGCACTG (mismatched bases in lowercase) was used with the M13 reverse primer. This mutant oligo contains a 3' Bsu36I site to facilitate subcloning and a NsiI site over the Sp1 binding region to facilitate identification from wild type DNA. The VDRE mutant was generated through two separate PCR reactions. In the first, a mutant oligo of sequence GCAAGTCCGCGGCGACCGaagCttGACGGGCAG TGCTG (mismatched bases in lowercase) was used with the T7 sequencing primer. This mutant oligo contains a 5' SacII site to facilitate subcloning and a HindIII site over the 5' VDRE half site to facilitate identification from wild type DNA. After PCR, the product was cut with SacII, subcloned into Bluescript II KS+ SacII and SmaI sites in the reverse orientation, and used as a template in a second PCR reaction with a mutant oligo of sequence GACCGAAGCTTGAtatctAGTGCTGGG AAC (mismatched bases in lowercase) and the M13 reverse primer. This second mutant oligo introduced a EcoRV site over the 3+ VDRE half site to facilitate the identification of the mutant PCR product. To generate the double VDRE/Sp1 mutant, the VDRE mutant sequence was cut with Bsu36I after PCR and subcloned in the forward orientation into the SmaI site of Bluescript II KS+. The mutant Sp1 oligo described above was then used with the T7 sequencing primer to amplify the intervening sequence.
Cell culture and transient transfections
Standard culture conditions consisted of phenol red Improved Minimum Essential Medium, Eagle’s (IMEM) supplemented with 10% heat-inactivated fetal bovine serum, 37 °C humidified atmosphere of 95% air-5% CO2, and media change every 2–4 days. For each transfection of a 78.5 mm2 dish, cells were grown under standard culture conditions to 70% confluence. Further 30 min prior to transfection, 10µg of CsCl banded DNA was incubated with 20µl of lipofectamine (GIBCO/BRL) in 1 ml serum free IMEM at room temperature. Then 2 ml IMEM with 10% FBS and 1 ml serum free IMEM was added per reaction mix, so that the final serum concentration was no more than 5%. Dishes were then incubated under standard conditions for 16–20 h at which time cells were rinsed twice with 1X PBS, trypsinized, pelleted, resuspended, mixed thoroughly, and replated in equal numbers to control for any differences in transfection efficiency from one plate to the next.
Upon replating, cells were either given IMEM with 10% FBS and 1,25-dihydroxyvitamin D3 or analog C, or IMEM with 10% FBS and an equivalent volume of vehicle only (100% ethanol). Treatments were carried out for 24–96 h under standard culture conditions, followed by harvesting. Cell pellets were lysed by freeze/thawing and the concentration of the protein in the lysate determined in duplicate by the Bradford assay. Equal amounts of protein (100µg) were then incubated at 37 °C for 2 h with 0.125µCi 14C-chloramphenicol and 0.5 mM acetyl CoA, and then extracted with ethyl acetate. Samples were spotted onto thin layer chromatography plates, run in a 95:5 chloroform: methanol tank and quantitated by comparison of phosphorimager determination of percent conversion of chloramphenicol to acetylated forms in treated versus untreated samples.
Electrophoretic mobility shift assays
Crude nuclear extract was prepared from BT549 cells treated with either 0.1µM 1,25-dihydroxyvitamin D3 for 48–72 h and/or an equal volume vehicle only (100% ethanol) previously described (McGaffin et al. 2004). Titration experiments looking at the influence of salt concentration, pre-treatment of cells with vitamin D and ligand presence on factor binding were carried out prior to performing in vitro footprint experiments or electrophoretic mobility shift assays (EMSAs) involving competitors and antibodies. For each sample, nonspecific Escherichia coli DNA competitor (0.5µg) was preincubated for 20 min at room temperature with 1,25-dihydroxyvitamin D3 (0.1µM final concentration) or an equal volume of vehicle (100% ethanol), nuclear extract (10 µg) or purified protein (10 ng) and BSA (100–500 ng) in a buffer containing 10 mM HEPES pH7.9, 12% glycerol, 10 mM Tris pH8.0 and 1 mM DTT. Salt concentration ranged from 50 to 150 mM KCl and was titrated by 10 mM increments. Purified VDR was purchased from PanVera Corporation (Madison, WI), purified Sp1 from Promega, and purified retinoid X receptor (RXR)ß from Affinity Bioreagents (Golden, CO, USA). Preincubation of either crude nuclear extract or purified protein was followed by an additional 20-min room temperature incubation with a radiolabeled DNA restriction fragment or commercially available oligonucleotide (10 000 d.p.m./sample). A consensus double stranded Sp1 oligo of sequence ATT CGATCGGGGCGGGGCGAGC was obtained from Promega, a DR3 (VDR) element from the annealing of the sense and antisense strands of a 21 nucleotide sequence GATCGGGTCAGTGAGGTCAGC, an AP-2 oligo of sequence GATCGAACTGACCGCCCG CGGCCCGT from Promega, and an estrogen response element (ERE) oligo from the annealing of the sense and antisense strands of a 53 nucleotide sequence GGGGGTCAGCTGTGCCCCGGTCGCCGAGTGG CGAGGAGGTGACGGTAGCCGCC.
Radiolabeling of probes was accomplished by 
-32P ATP and T4 polynucleotide kinase, followed by inactivation of the kinase by heating to 65 °C for 1 h and subsequent sepharose column purification of unincorporated nucleotides from the probe. After incubation, samples were electrophoresed on a 6% non-denaturing polyacrylamide gel and visualized by autoradiography. Upon determination of optimum conditions of salt and ligand binding, antibody and DNA competitor treatments were performed using the salt and ligand concentration as indicated with each figure. For these, antibody (1µg) or DNA competitor (10, 50, or 100 fold molar excess) were added to the reaction mix for an additional 20 min at room temperature prior to the addition of probe, followed by electrophoresis and autoradiography. The polyclonal anti-Sp1 antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The rabbit polyclonal anti-VDR antibody (cat# PA1–711 ) and the rat monoclonal 9A7 anti-VDR antibody (cat# MA1–710) were purchased from Affinity BioReagents (Golden, CO, USA). Polyclonal antibodies against , ß, and forms of RXR were obtained from Santa Cruz Biotechnology, while polyclonal antibodies against the corresponding forms of retinoic acid receptor (RAR) were obtained from Dr. Wayne Vedeckis (Louisiana State University Medical Center, New Orleans, LA, USA).
In vitro DNAse I footprinting
Crude nuclear extract was prepared as described above for EMSAs. Generation of a singly end labeled DNA probe was accomplished through PCR amplification of wild type EGFR SacII to Bsu36I sequence subcloned in the reverse orientation into SacII and SmaI sites of the Bluecript II KS+ plasmid. -32P ATP and T4 poly-nucleotide kinase was used to label the T7 sequencing primer, followed by inactivation of the kinase by heating to 65 °C for 1 h. Labeled T7 primer was then added to unlabeled M13 reverse primer to PCR to amplify the Bluescript II KS+ template containing wild type SacII to Bsu36I sequence. Purification of the PCR product from free nucleotides and unincorporated primers was accomplished by running it through a sepharose column after removing excess Bluescript II KS+ polylinker sequence from the unlabeled end by EcoRI digestion. For each sample, preincubation of nuclear extract (10µg), nonspecific E. coli competitor DNA (0.5µg) and 1,25-dihydroxyvitamin D3 (0.1µM final concentration) was performed at room temperature for 20 min in a buffer containing 88 mM KCl, 10 mM HEPES pH7.9, 12% glycerol, 10 mM Tris pH8.0, and 1 mM DTT. Following this, a radiolabeled probe was added (10 000 d.p.m./sample or 1 ng/sample) to the extract and E. coli DNA mix and allowed to incubate at room temperature for an additional 20 min (30 µl total reaction volume). Control reactions were set up in parallel that contained only E. coli DNA, 1,25-dihydroxyvitamin D3 or equal volume vehicle (100% ethanol) and BSA in place of nuclear extract.
For DNAse I digestion, the above reactions containing probe, 1,25-dihydroxyvitamin D3 or vehicle only, protein, and E. coli DNA were incubated with various concentrations of DNAse I (Promega) spanning two orders of magnitude (from 0.1 to 10 units so as to achieve maximum sensitivity) for 2 min at room temperature. The reactions were stopped by adding an excess of SDS and EDTA, extracted using phenol: chloroform, and precipitated using ethanol. Samples were then denatured in 80% formamide by heating to 90°C and run on a 6% denaturing polyacrylamide gel along with Sanger dideoxy sequencing reactions. The sequencing reactions were produced by using the T7 sequencing primer and the same plasmid DNA that was used as a template in PCR to generate the singly end labeled probe.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Our previous transient transfection studies in MCF7, T47D and BT549 breast cancer cells suggested that vitamin D repression of EGFR expression has a transcriptional mechanism that is mediated through promoter sequences located between nucleotide positions –860 and –20 (McGaffin et al. 2004). Examination of this region in the EGFR promoter led to the identification of a putative VDRE between nucleotides –531 and –516 with the sequence GGGTCCACAGGGGCA (half sites underlined). When using the full length pJFCAT 840 construct in MCF7, T47D and BT549 transient transfections, we demonstrated statistically significant reduction in CAT activity (P<0.05 by Student’s t-test) in response to 1µM 1,25- dihydroxyvitamin D3 treatment by 48 h (McGaffin et al. 2004), with most cases only requiring 24 h to see the effect (data not shown). This timeframe correlates with the significant growth inhibition that we also observed in MCF7 and T47D cells when treated with 1µM 1,25-dihydroxyvitamin D3 (McGaffin et al. 2004). Use of 1,25-dihydroxyvitamin D3 at a concentration which was 10 fold lower (0.1µM), however, required between 72–80 h to achieve statistically significant growth inhibition versus vehicle only treated controls (McGaffin et al. 2004). Further, BT549 cells showed no significant growth inhibition at any concentration of vitamin D we used out to 8 days (McGaffin et al. 2004). This is in contrast to the short (within 2 h)and sustained (greater than 72 h) amount of time for which we observed endogenous EGFR repression in response to 1,25-dihydroxyvitamin D3 treatment in MCF7, T47D and BT549 cells (McGaffin et al. 2004). Additionally, while we observed 0.1µM 1,25-dihydroxyvitamin D3 mediated repression of pJFCAT 840 activity by 24 h in BT549 cells, we also observed that transfected CAT protein yields were optimal between 48–96 h after transfection (data not shown). Hence, we chose to treat all transfections with 0.1µM 1,25-dihydroxyvitamin D3 for greater than 72 h to maximize CAT protein yield.
To assess the possibility that vitamin D repression of EGFR gene expression is mediated through our putative VDRE, serial deletion promoter CAT constructs were generated based on naturally occurring SacII and Bsu36I restriction sites flanking the –531 to –516 region of the EGFR promoter (Fig. 1
). When BT549 cells were transfected with the pJFCAT 536 construct and treated for 72 h with 1,25-dihydroxyvitamin D3 (n=5), a significant 25–50% (P<0.05 by student’s t-test) decrease in CAT activity was seen when compared with CAT activity in equal volume, vehicle only treated transfected controls (figure 1). In contrast, these same cells transfected with pJFCAT 478 demonstrated no significant change (P>0.05 by student’s t-test) in CAT activity upon 72 h of 1,25-dihydroxyvitamin D3 treatment. Similar results were obtained with MCF7 and T47D cells (data not shown). This suggested to us that the sequence of the EGFR promoter between the SacII and Bsu36I sites, or spanning nucleotide positions –536 and –478, that contains a putative VDRE is important for mediating the 1,25-dihydroxyvitamin D3 repressive effect on EGFR expression.
|
Characterization of factors binding to the EGFR vitamin D responsive region was accomplished by in vitro footprint experiments and EMSAs (Figures 2, 3 and 4). Prior to performing these, optimal conditions of salt and ligand dependence of VDR binding were determined in titration experiments using EMSA. Specifically, it has been shown that proximal binding of VDR to its response element can occur in the absence of ligand (Ross et al., 1993) and is greatly influenced by salt concentration (Kimmel-Jehan et al. 1997). In agreement with the findings of Ross et al., (1993) and Kimmel-Jehan et al., (1997), we found that purified VDR/RXR bound to our nucleotide sequence equally well in the presence or absence of a ligand at salt concentrations between 80–100 mM (data not shown). When salt concentrations approached 150 mM, or were dropped below 50 mM, the complexes formed by purified VDR, RXR and Sp1 proteins, as well as the upper and lower nuclear complexes formed by BT549 nuclear extract, were progressively lost (data not shown). However, at these high and low salt concentrations, the addition of vitamin D to binding reactions resulted in enhanced purified VDR/RXR binding (data not shown). We also noted an increased amount of lower nuclear complex formation compared with upper complex formation when vitamin D was added to binding reactions at any salt concentration spanning the 50 to 150 mM range (Fig. 5). As a result of this, all binding reactions of purified factors and nuclear extracts were carried out in the presence of vitamin D and 88 mM KCl.
|
|
|
|
To ascertain whether or not the putative VDRE sequence identified between nucleotides –531 to –516 binds protein factor(s) present in the nucleus, in vitro DNAse I footprinting was performed using crude BT549 nuclear extract and a PCR generated fragment spanning the SacII to Bsu36I region of the promoter (Fig. 2
). Sanger dideoxy sequencing reactions were run in parallel and delinate the areas of protein binding at the nucleotide level. Similar results were obtained when the nuclear extract was titrated and DNAse I concentration was held constant (data not shown). The regions of protection evident from this assay are labeled I, II, IIIa and IIIb. Region I covers the sequence AACTCCTCA, region II the sequence GAACGCCCCT, region IIIa the sequence GGGGCA, and region IIIb GGGTCC. Based on their similarity to known consensus factor binding sites, region I was postulated to be a potential nuclear receptor half site, region II postulated to be a general transcription factor Sp1 binding site and regions IIIa and IIIb a putative VDRE.
EMSA with crude nuclear extract identifies Sp1 and the VDR as factors that bind to the vitamin D responsive region
Upon addition of the nuclear extract to the labeled probe, an upper and lower complex were seen on EMSA (Fig. 3, lane 2). The upper complex was specifically competed with 50-fold molar excess of a commercially available Sp1 consensus oligonucleotide (lane 3), while the lower complex was specifically competed with a 50 fold molar excess of a consensus VDRE oligonucleotide (lane 4). Neither complex was competed by a 50-fold molar excess of consensus ERE or AP2 oligonucleotides (lanes 5 and 6). Addition of 1µg of a polyclonal antibody directed against transcription factor Sp1 resulted in a supershift of the upper complex, but not the lower (lane 7), while addition of 1µg of a polyclonal antibody directed against the VDR resulted in a supershift of the lower complex, but not the upper (lane 8). Further, use of the anti-VDR 9A7 antibody, which blocks VDR-DNA binding, resulted in less lower complex formation while leaving the upper complex intact (lane 9). As controls, incubation of 1µg of the Sp1 and VDR antibodies with probe alone resulted in no complex formation (data not shown). Altogether, these results suggested to us that the upper complex represents transcription factor Sp1 binding, while the lower represents binding of the VDR.
EMSA with purified proteins confirms VDR and Sp1 binding
Further confirmation of the identity of the binding factors to this region of the EGFR promoter was accomplished through an additional EMSA using nuclear extract and purified factors Sp1, human RXRß and human VDR (Fig. 4). Comparing the upper and lower complexes formed by nuclear extract on the SacII-Bsu36I probe with those complexes formed by purified factors, there is alignment of the upper complex with that formed by mixing 10 ng of purified Sp1 with the probe (lane 3 compared with lane 9). In contrast, incubation of up to 100 ng purified human VDR with the SacII-Bsu36I probe resulted in only a weakly detectable complex with the DNA (possibly a VDR monomer) that migrated to a point below that of the lower complex formed with nuclear extract (compare lane 2 with lane 9). This latter finding strongly suggests that the VDR requires a heterodimeric partner to bind effectively to this sequence of DNA. Indeed, incubation of the probe with 10 ng of purified VDR along with 10 ng of purified RXRß, which has been shown to be a coregulator of VDR binding and transactivation (Barsony & Prufer 2002), resulted in the formation of a strong complex (lane 8). However, since it migrated slightly above that observed for the lower complex formed with nuclear extract (compare lane 8 with lane 9), it suggested to us that the lower nuclear complex is not composed of a VDR/RXRß heterodimer, but rather as indicated by antibody experiments (Fig. 3, lanes 8 and 9), does contain the VDR. As control experiments, incubation of an excess of purified RXRß (>100 ng) with the SacII-Bsu36I probe resulted in no complex formation (Fig. 4, lane 7).
Further binding experiments performed in this series with purified factors included the incubation the VDR and Sp1 together with the probe. In this case, a tertiary complex was formed not previously seen using nuclear extract, presumably representing a VDR/Sp1 complex (Fig. 4, lane 4). When 100-fold molar excess of a consensus VDRE oligonucleotide competitor was used, this tertiary VDR/Sp1 complex was competed down to just a Sp1 complex (lane 5), and when 100-fold molar excess of a Sp1 oligonucleotide competitor was used, this tertiary VDR/Sp1 complex was competed down to just VDR binding (lane 6). We performed additional competition experiments using 100 fold excess competition oligonucleotides against a variety of other known nuclear factors including AP1, AP2, TFIID, NF-kB, NF-Y, E twenty-six avian erythroblastosis virus oncogene (ETS), upstream stimulatory factor (USF), nuclear respriatory factor (NRF-1), Ying Yang (YY1), octomer transcription factor 1 (OCT1), cAMP-response-element-binding protein 1 (CREB1) and the estrogen receptor. None of these demonstrated competition of the Sp1, VDR, Sp1/VDR, or upper or lower nuclear complexes (data not shown).
Specific mutations introduced into footprint protected regions containing putative VDR and Sp1 binding sites abolish VDR and Sp1 factor binding
The results of the footprint experiments, along with a knowledge of consensus sequences for VDR and Sp1 f0actor binding sites, allowed us to design a series of point mutations in footprint protected regions II, IIIa and IIIb. The subsequent wild type, VDRE, Sp1, and double VDRE/Sp1 mutant sequences used as probes in EMSAs are illustrated in Fig. 6A. Footprint protected region II was predicted to be the Sp1 binding site and was mutated by introducing a NsiI site over the center of the ACGCCC motif. Protected regions IIIa and IIIb were predicted to encompass a putative VDRE. These were mutated by introducing HindIII and EcoRV sites over the two putative half sites as shown. The double VDRE/Sp1 mutant was generated by a combinatorial insertion of all three restriction sites over their respective protected regions.
|
). When incubated with nuclear extract, wild type probe demonstrated the formation of the upper and lower complex (lane 2). Use of the VDRE mutant as a probe for nuclear extract resulted in the loss of binding of the lower complex (lane 4), suggesting that the putative VDRE does in fact bind the VDR and its unknown partner. Lane 6, using the Sp1 mutant probe and nuclear extract, demonstrates the loss of upper complex formation, suggesting that region II is in fact a binding site for transcription factor Sp1. Finally, use of the double VDRE/Sp1 mutant as a probe resulted in loss of both lower and upper complex formations. Wild type SacII-Bsu36I sequence confers a vitamin D response upon a heterologous promoter that is lost with mutants, demonstrating the role of ligand, the functionality of the putative VDRE and the involvement of Sp1
To determine whether the sequence between the SacII-Bsu36I sites of the EGFR promoter by itself is able to mediate a repressive vitamin D effect on transcription, a CAT construct driven by a minimal heterologous promoter (a TATA box only) was designed for transient transfection in BT549 cells. The parental construct, named pJFCAT TATA, was generated by subcloning a synthetic TATA box oligonucleotide upstream of the CAT gene in the pJFCAT construct that was the backbone for the reporter constructs used in Fig.1. Initial transfection experiments with pJFCAT TATA containing the SacII to Bsu36I region of the EGFR promoter cloned into it resulted in unmeasurable CAT activity (data not shown). Consequently, to increase basal CAT activity to a level where we could measure any repression resulting from subcloned wild type and mutant sequences in response to 1,25-dihydroxyvitamin D3 treatment, we generated a second CAT vector named pJFECAT TATA by subcloning the pJFCAT poly A cassette and the TATA box oligonucleotide into Promega’s pCAT enhancer vector which contains the SV40 enhancer downstream of CAT (see materials and methods). Subsequent transfections of this construct containing wild type sequence produced measurable basal activity.
To rule out the possibility that 1,25-dihydroxyvitamin D3 may mediate a response through sequences in the pJFECAT TATA vector itself, transfections of this construct were performed in BT549 cells followed by 1µM and 0.1µM 1,25-dihydroxyvitamin D3, analog C, and equal volume vehicle only treatments. The results of three independent experiments in this cell line demonstrated that vitamin D had no effect on CAT activity (P>0.05 by Student’s t-test) when compared with equal volume vehicle only treated transfected controls (data not shown). We then asked whether the wild type, mutant VDRE, mutant Sp1 and mutant VDRE/Sp1 SacII-Bsu36I sequences could mediate a repressive vitamin D response in the context of the pJFECAT TATA vector. A significant (P<0.05 versus vehicle only treated controls by student’s t-test) 0.1µM 1,25-dihydroxyvitamin D3 repressive response was mediated by wild type sequence that was abolished upon specific mutation of either the Sp1 or the VDR binding sites (Fig. 7), suggesting the importance of both in mediating the vitamin D repressive effect. Further, when the Sp1 mutant construct was used, we observed a basal level of ligand independent EGFR repression that was equal in magnitude to that observed for vitamin D mediated repression of the wild type sequence. This latter finding suggests that the VDR is capable of mediating a basal level of EGFR transcriptional repression in the absence of ligand.
|
Our results suggest the existence of a nuclear factor which must partner with the VDR for strong binding to the region of the EGFR promoter containing the putative VDRE, and is involved in mediating the 1,25-dihydroxyvitamin D3 repression of EGFR expression, at least partially through displacement of Sp1. To provide additional evidence for the existence of this unidentified factor, titration studies were performed using wild type SacII-Bsu36I probe, BT549 nuclear extract, and purified Sp1 and VDR proteins in EMSAs (Fig. 8A and B). In the case where nuclear extract was held at a constant 250 ng (Fig. 8A), adding increasing amounts (1 ng, 2 ng, 3 ng, 4 ng, 6 ng, 8 ng, 10 ng) of purified Sp1 and VDR proteins at a constant 1:1 ratio to the binding reactions (lanes 3–9) caused the gradual disappearance of both the lower and upper nuclear complexes. Further, there was a corresponding gradual appearance of an even slower migrating complex seen previously (compare this to Fig. 4, lane 4), presumably representing the simultaneous binding of Sp1 and VDR to wild type sequence when these proteins are in excess over crude nuclear extract. In the case where purified Sp1 and VDR levels were kept at a constant 10 ng each (Fig. 8B), increasing nuclear extract to a final 10µg (0.25µg, 0.5µg, 1µg, 2µg, 3µg, 4µg, 5µg, 6µg, 8µg, 10µg) resulted in the gradual disappearance of the VDR/Sp1 complex (Fig. 8B, lane 4) and the gradual re-appearance of the upper and lower complexes (lanes 5–14). Taken together, these results suggest the existence of one or more nuclear factors that are required for strong VDR binding to the wild type sequence. They also suggest that simultaneous binding of Sp1 and the VDR to the wild type sequence does not take place in the nucleus of BT549 cells, as the described tertiary complex (Fig. 4, lane 4; Fig. 8A, lane 9; Fig. 8B, lane 4) is never seen with nuclear extract alone.
|
In an attempt to shed light on the mechanism of vitamin D mediated EGFR repression in the BT549 cell line when compared with the induction seen in the BT474 cell line (McGaffin et al. 2004), we examined the pattern of complex formation using the wild type SacII-Bsu36I sequence of the EGFR promoter as a probe and nuclear extracts from each of these cell lines in EMSAs (Fig. 9). While a similar pattern of upper and lower nuclear complex formation was seen in both cases (Fig. 9), when vitamin D was added to reactions containing BT549 extract, the intensity of the lower complex increased consistent with enhanced VDR binding (lane 4 versus 5). In contrast, when vitamin D was added to the binding reaction containing BT474 extract, the intensity of the lower complex was decreased, indicating less VDR binding (lane 2 versus lane 3). The intensity of the upper complexes formed in each case changed in a manner inverse to the lower, such that those samples that demonstrated more intense lower complex formation seemed to have less intense upper formation and vice versa (lanes 2–5).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
). This finding is not entirely unexpected given that the binding sites for the VDR and Sp1 are in such close proximity to one another, being less than 7 nucleotides apart in this stretch of the EGFR promoter (Fig. 6A). Further, mutation of either the VDRE or Sp1 site, which results in loss of corresponding nuclear factor binding as assessed by EMSA (Fig. 6B), causes a complete loss of a vitamin D response (Fig. 7). Taken together, these data strongly suggest that to exert its EGFR repressive effect in BT549 cells, the VDR must first heterodimerize with its unknown partner before it can bind with high enough affinity to the SacII-Bsu36I restriction fragment to displace already bound Sp1. The fact that loss of Sp1 binding, with preservation of heterodimeric VDR binding to the SacII-Bsu36I restriction fragment (Fig. 6A, lane 6), results in suppression of basal CAT activity as well as a loss of the vitamin D response indicates that the mere binding of the VDR is not sufficient for the observed transcriptional repression in response to vitamin D treatment, but is nonetheless sufficient enough to mediate a basal level of transcriptional repression in the absence of ligand.
Our functional data (Fig. 7) supports the assertion that the identified VDRE, by nature of its nucleotide sequence alone, mediates negative rather than positive transcriptional responses by vitamin D in BT549 cells. Specifically, mutation of the Sp1 site, which abolishes Sp1 but not VDR binding, does not subsequently allow the VDR to mediate a positive (rather than negative) transcriptional response. The functional VDRE identified by us in the present study differs from classical upregulatory VDREs by one nucleotide in its 5' half site (underlined): GGGTCC. It has been demonstrated in other reports (Freedman et al. 1994, Colnot et al. 1995) that the last nucleotide in the 5' VDRE half site tends to be an adenine and is found in up to 93% of VDREs that bind the VDR and activate transcription. This suggests that such a difference is not of minimal importance. The 3' portion of our VDRE half site, with sequence GGGGCA, is identical to the 3' half site of the up-regulatory human osteocalcin gene VDRE (Ozono et al. 1990), but once again differs from the canonical DR3 PuGGTCA (Umesono et al. 1991) by one nucleotide. Therefore, it is possible that these two nucleotide differences observed between our VDRE and identified upregulatory VDREs accounts in part for the negative regulation (Koszewski et al. 1999, Koszewski et al. 2000). Other examples of such ‘negative’ VDREs can be found in those genes that encode bone related proteins including the human and avian parathyroid hormone (Liu et al. 1996, Koszewski et al. 2004), the rat PTH-related peptide gene (Kremer et al. 1996), the mouse osteocalcin gene (Lain et al. 1997) and the rat bone sailoprotein gene (Kim et al. 1996, Li et al. 1998).
It is known that the RXRs (Barsony & Prufer 2002), and under some circumstances, the RARs (Pemrick et al. 1998), can bind with the VDR to activate or repress corresponding DNA response elements. Since the purified VDR only formed a weakly detectable complex (Fig. 4, lane 2) with the wild type sequence containing our putative VDRE when ligand was present and the amount of purified VDR protein was in excess, it suggested to us that VDR must partner with a factor other than itself to effectively bind to this region of the EGFR promoter. The difference we observed in mobility between the purified VDR/RXRß complex and the lower nuclear complex (figure 4) also suggests that the latter is not composed of a VDR/RXRß dimer, while specific competition and antibody experiments results do at least confirm the presence of the VDR in the lower complex.
The binding properties of the nuclear factor postulated to dimerize with the VDR and bind to our VDRE were explored in titration experiments. While it is interesting that an excess of purified VDR and Sp1 proteins can result in the loss of the nuclear complex containing the VDR and its unknown partner, it is even more intriguing to note the corresponding appearance of an even more slowly migrating complex of both Sp1 and VDR proteins. This Sp1/VDR complex was seen with EMSA using purified factors only, and was found to be disruptable with the addition of increasing amounts of nuclear extract. Initially, we thought that its transactivation properties, if any, may be functionally significant in terms of the BT474 cell line, which in previous studies has been shown to upregulate EGFR in response to vitamin D treatment (McGaffin et al. 2004). However, such a complex was never seen by us in EMSAs using nuclear extract alone. Instead, only the described upper and lower complexes were seen with BT549 and BT474 nuclear extracts (Fig. 9), suggesting that the VDR/sp1 dimer is not functional in these cells through direct binding to the SacII-Bsu36I region of the EGFR promoter.
Examining the results of our EMSA comparing the complexes formed on the wild type probe and extract from BT474 and BT549 cells further (Fig. 9), however, we can speculate on a mechanism of vitamin D mediated induction of EGFR in the BT474 cell line, while causing repression in the BT549 cell line. Specifically, our observed difference in the binding of the VDR to the wild type sequence in the presence and absence of ligand suggests that vitamin D helps to stabilize VDR binding in BT549 cells, resulting in Sp1 displacement. Conversely, it suggests that vitamin D destabilizes VDR binding in BT474 cells, possibly allowing for subsequent productive Sp1 interactions with the rest of the transcriptional machinery. Our previous functional data using the BT474 cell line and a CAT construct containing the wild type sequence of the EGFR promoter in transient transfection experiments lends support to this idea as an inductive response to vitamin D treatment was seen (McGaffin et al. 2004). This interpretation should be looked on with caution, however, as the magnitude of CAT induction we saw in response to vitamin D treatment was less than that observed with endogenous EGFR in response to vitamin D treatment (data not shown), suggesting that additional sequences and/or factors are possibly needed to mediate the full inductive vitamin D response in these cells.
The finding of ligand independent VDR binding to the VDRE identified in the EGFR promoter was somewhat expected given the results of Kimmel-Jehan et al.(1997) who show ligand independent VDR binding to various VDREs, including those in the rat osteocalcin, mouse osteopontin, rat 25-hydroxyvitamin D3 24-hydroxylase, and human parathyroid hormone genes. In our study, we note that the VDR in BT549 extract is capable of binding to the wild type EGFR VDRE sequence in the absence of ligand, albeit much less avidly than when ligand is present (Fig. 5). This suggests that the VDR upon DNA binding may be able to mediate a basal level of transcriptional repression in the absence of ligand, possibly through the displacement of Sp1. The results of transfections using the wild type, VDRE mutant, Sp1 mutant, and VDRE/Sp1 mutant sequences (Fig. 7) lends additional support to this idea. Specifically, the finding of significantly (P<0.05) blunted basal level of CAT activity in the vehicle only treated Sp1 mutant construct when compared with the basal activity of vehicle only treated wild type and VDRE mutant constructs suggests that to exert its repressive effect, the VDR must not only bind to the VDRE, but it must also displace Sp1 in the process. Hence, when Sp1 binding is lost and VDR binding is retained, as in the Sp1 mutant construct, the presence or absence of ligand has no effect on the already blunted level of CAT activity. This functional data puts into perspective our binding data (Fig. 5), and overall suggests that the addition of ligand in fact serves to augment a ligand independent, VDR mediated basal level of transcriptional repression of wild type sequence by increasing VDR binding and the subsequent displacement of Sp1.
It is our assertion that the transcriptional repression we observe in the presence and absence vitamin D in BT549 cells is mediated through positive transcription factor Sp1 displacement by the VDR and its binding partners. This finding of positive transcription factor displacement by the VDR has been characterized for the VDR and other gene promoters as well. Examples include VDR displacement of NF-Y in the human parathyroid hormone promoter (Koszewski et al. 2004), VDR disruption of NFATp/AP-1 binding in the interleukin-2 gene (Alroy et al. 1995), and VDR displacement of AP-1 in the osteocalcin gene (Aslam et al. 1999). Additionally, in the rat bone sialoprotein (BSP) gene promoter, Li et al.(1998) identified a negative VDRE capable of binding to the promoter as a heterodimer at a site that overlapped a unique inverted TATA box. In this case, the authors suggested that competition between the VDR complex and TATA-binding proteins might be a plausible explanation for the vitamin D mediated repression observed. Similar to our findings, Li et al.(1998) note that an upper and lower complex were formed on the BSP VDRE probe when nuclear extract from human bone MG63 cells were used in EMSAs. They characterize this sequence in the BSP promoter as mediating repressive transcriptional responses and binding VDR in the context of an unknown partner. In spite of these similarities, several important differences should be noted when comparing the VDRE identified by us in the EGFR promoter and that in the rat BSP promoter. Specifically, Li et al.(1998) note that the lower and upper complexes are monomer and dimer forms of the VDR, respectively. We conclude from our work that the lower complex that binds to the EGFR VDRE represents a VDR heterodimer, and that the upper complex represents Sp1 binding (figures 3 and 4). Further, the BSP VDRE contains an inverted TATA box which the authors state is crucial for BSP gene expression, likely via interaction with TFIID or other elements of the basal transcription machinery, while our identified EGFR VDRE does not.
Moreover, the BSP VDRE reported on by Li et al.(1998) binds the VDR equally well as a monomer and homodimer, while our results indicate that VDR when used alone in EMSAs binds to the VDRE in the EGFR promoter only very weakly, requiring a heterodimeric partner for strong binding (Fig. 4). Overall, in the light of these substantial differences, we feel confident that the factor identified by Li et al.(1998) that interacts with the VDRE in the BSP promoter is unique in comparison to that factor which binds to the VDRE we have identified here in the EGFR promoter.
The results of our experiments when compared with those reported by Li et al.(1998) in their work characterizing the VDRE in the BSP promoter provide evidence that the mechanism of vitamin D mediated transcriptional repression of EGFR is sequence specific. The results of Gonzales et al.(2002), when looking at vitamin D effects on EGFR expession in osteoblast-like cells, suggest that the vitamin D effect is also cell type specific. Specifically, Gonzales et al.(2002) focused their attention on the same region of the EGFR promoter that we characterize here. Our present findings are in agreement with those of Gonzales et al.(2002) with regard to VDR/RXR binding to a DNA fragment containing this putative VDRE using purified proteins. However, unlike our study, Gonzales et al.(2002) failed to show functionality of the sequence in a luciferase reporter system. Instead, they noted an increase in EGFR mRNA stability in response to vitamin D, and suggested that it accounts for the up-regulation seen when compared with non-treated cells. In addition to this, there are several other important differences between our experimental observations and those of Gonzales et al.(2002). First, we noted transcriptional repression, not activation, of the EGFR promoter in response to vitamin D in MCF7, T47D, and BT549 cells. Second, in BT474 cells, which in our hands also showed EGFR upregulation in response to vitamin D, we demonstrated that EGFR mRNA stability is not altered with vitamin D treatment (McGaffin et al. 2004). Third, as we used breast cancer cells, not osteoblast-like cells, and as such speculate that there are cell specific co-activating and co-repressive factors which account for the disparate findings.
Given all that has been reported on VDR mediated transcriptional repression, we focused additional effort on identifying the factor present in nuclear extract that dimerizes with the VDR and forms the lower complex. Specifically, we performed additional experiments with polyclonal antibodies directed against known potential partners of VDR, including the RXR, RXRß, RXR, RAR, RARß, and RAR. Despite their ability to supershift their cognizant purified receptors when bound to DNA in our hands, none of these antibodies recognized the lower (or upper) complex, suggesting that it was composed of the VDR and an as of yet unidentified partner (data not shown.) Further, we performed additional EMSA competition experiments using oligonucleotides known to bind transcription factors AP1, AP2, TFIID, NF-kB, NF-Y, ETS, USF, NRF-1, YY1, OCT1, and CREB1, as well as the estrogen receptor (data not shown). None of these consensus DNA sequences were able to compete the nuclear complexes formed on the SacII-Bsu36I restriction fragment containing the putative EGFR VDRE. Having excluded the participation of all common and known VDR binding partners through antibody and oligonucleotide competition experiments listed above, we will direct the focus of future experimentation and publication on de novo identification of this unknown factor.
Prior to performing the experiments presented here, our initial hypothesis was that EGFR repression is initiated through ligand activation of the VDR, followed by subsequent dimerization with an unknown partner and binding to the negative VDRE spanning nucleotides –531 to –516 of the promoter. Based on the binding and functional data we show here in EMSAs and in reported experiments, this model has changed. Data now suggests that the VDR dimerizes with its unknown partner, binds to the VDRE with low affinity, and results in Sp1 displacement from its proximal binding site. This mediates a basal level of transcriptional repression in the absence of ligand. Upon ligand binding, stabilization of VDR-DNA binding occurs, resulting in greater transcriptional repression through enhanced displacement of Sp1 from its proximal binding site directly, or perhaps through intervening co-repressor molecules in the nucleus. In either case, displacement of functional Sp1 interactions with the rest of the transcriptional machinery results in the observed transcriptional repression. While transcriptional repression mediated by vitamin D through competition for binding with other transcription factors has been shown by others (Alroy et al. 1995, Aslam et al. 1999, Koszewski et al. 2004, Li et al. 1998), this mechanism has not been previously reported in terms of the VDR and transcription factor Sp1. Studies that have investigated the relationship between the VDR and Sp1 seem to note a synergistic effect, not antagonistic, with productive transcriptional activity the end result (Huang et al. 2004, Liu & Freedman 1994). In this respect, our current study represents the characterization of a novel mechanism of vitamin D and VDR mediated repression of EGFR expression through interference with the activity of positive transcription factor Sp1.
|
Acknowledgements |
|---|
|
Funding |
|---|
This work was supported by grants from the United States Department of Defense, Army and Materiel Command (K.R.M.), the National Cancer Institute (S.A.C.), the Lombardi Cancer Center Macromolecular Synthesis and Sequencing Shared Resource (US Public Health Service Grant), and the Lombardi Cancer Center Tissue Culture Shared Resource (US Public Health Service Grant). The authors declare that there is no conflict of interest that would prejudice the impartiality of this scientific work.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Aslam F, McCabe L, Frenkel B, van Wijnen AJ, Stein GS, Lian JB & Stein JL. 1999. AP-1 and the vitamin D receptor (VDR) signaling pathways converge at the rat osteocalcin VDR element: requirement for the internal activating protein-1 site for vitamin D-mediated trans-activation. Endocrinology 140:63–70.
Barsony J & Prufer K. 2002. Vitamin D receptor and retinoid X receptor interactions in motion. Vitamins & Hormones, 65:345–376.
Bundred NJ. 2001. Prognostic and predictive factors in breast cancer. Cancer Treatment Reviews 2:137–142.
Colnot S, Lambert M, Blin C, Thomasset M & Perret C. 1995. Identification of DNA sequences that bind retinoid X receptor-1,25(OH)2D3 receptor heterodimers with high affinity. Molecular and Cellular Endocrinology 113:89–89.[CrossRef][ISI][Medline]
Cordero JB, Cozzolino M, Lu Y, Vidal M, Slatopolsky E, Stahl PD, Barbieri MA & Dusso A. 2002. 1,25-dihydroxyvitamin D down-regulates cell membrane growth and nuclear growth promoting signals by the epidermal growth factor receptor. Journal of Biological Chemistry 277:38965–38971.
Dusso A, Cozzolino M, Lu Y, SatoT & Slatopolsky E. 2004. 1,25-dihydroxyvitamin D downregulation of TGFalpha/EGFR expression and growth signaling: a mechanism for the antiproliferative actions of the sterol in parathyroid hyperplasia of renal failure. Journal Steroid Biochemistry and Molecular Biology 89–90:507–511.
Freedman LP, Arce V & Perez Fernandez R. 1994. DNA sequences that act as high affinity targets for the vitamin D3 receptor in the absence of the retinoid X receptor. Molecular Endocrinology 8 :265–273.[Abstract]
Gonzales EA, Disthabanchong S, Kowalewski R & Martin KJ. 2002. Mechanisms of the regulation of EGF receptor gene expression by calcitriol and parathyroid hormone in UMR 106- 01 cells. Kidney International 61:1627–1634.[CrossRef][Medline]
Huang YC, Chen JY & Hung WC. 2004. Vitamin D3 receptor/Sp1 complex is required for the induction of p27 Kip1 expression by vitamin D3. Oncogene 23:4856–4861.[CrossRef][Medline]
Kimmel-Jehan C, Jehan F & DeLuca HF. 1997. Salt concentration determines 1,25-dihydroxyvitamin D3 dependency of vitamin D receptor-retinoid X receptor-vitamin D response element complex formation. Archives of Biochemistry and Biophysics 341:75–80.[CrossRef][Medline]
Kim RH, Li JJ, Ogata Y, Yamauchi M, Freedman LP & Sodek, J. 1996. Identification of a vitamin D3-response element that overlaps a unique inverted TATA box in the rat bone sialoprotein gene. Biochemical Journal 318:219–226.
Koszewski NJ, Alimov AP, Park-Sarge OK & Malluche HH. 2004. Suppression of the human parathyroid hormone promoter by vitamin D involves displacement of NF-Y binding to the vitamin D response element. Journal of Biological Chemistry 279:42431–42437.
Koszewski NJ, Ashok S & Russell, J. 1999. Turning a negative into a positive: Vitamin D receptor interactions with the avian parathyroid hormone response element. Molecular Endocrinology 13: 455–465.
Koszewski NJ, Malluche HH & Russell J. 2000. Vitamin D receptor interactions with positive and negative DNA response elements: an interference footprint comparison. Journal of Steroid Biochemistry and Molecular Biology 72:125–132.[CrossRef][ISI][Medline]
Kremer R, Sebag M, Champigny C, Meerovitch K, Hendy GN, White J & Goltzman D. 1996. Identification and characterization of 1,25-dihydroxyvitamin D3-responsive repressor sequences in the rat parathyroid hormone-related peptide gene. Journal of Biological Chemistry 271:16310–16316.
Li JJ, Kim RH, Zhang Q, Ogata Y & Sodek J. 1998. Characteristics of vitamin D3 receptor (VDR) binding to the vitamin D response element (VDRE) in rat bone sialoprotein gene promoter. European Journal of Oral Science 106:408–417.
Lichtner RB. 2003. Estrogen/EGF receptor interactions in breast cancer: rationale for new therapeutic combination strategies. Biomedicine & Pharmacotherapy 57:447–451.
Liu M & Freedman LP. 1994. Transcriptional synergism between the vitamin D3 receptor and other nonreceptor transcription factors. Molecular Endocrinology 8:1593–1604.[Abstract]
Liu SM, Koszewski N, Lupez M, Malluche HH, Olivera A & Russell, J. 1996. Characterization of a response element in the 5'-flanking region of the avian (chicken) PTH gene that mediates negative regulation of gene transcription by 1,25-dihydroxyvitamin D3 and binds the vitamin D receptor. Molecular Endocrinology 10 :206–215.[Abstract]
McGaffin KR, Atkinson LE & Chrysogelos SA. 2004. Regulation of growth and EGFR expression in MCF7, T47D, BT474, and BT549 breast cancer cells by retinoids and vitamin D. Breast Cancer Research and Treatment 86:55–73.[Medline]
Natha R, Hortobagya GN & Esteva FJ. 2003. Growth factor receptors in breast cancer: potential for therapeutic intervention. Oncologist 8:5–17.
Ozono K, Liao J, Kerner SA, Scott RA & Pike JW. 1990. The vitamin D-responsive element in the human osteocalcin gene. Journal of Biological Chemistry 265:21881–21888.
Pemrick SM, Abarzua P, Kratzeisen C, Marks MS, Medin JA, Ozato K & Grippo JF. 1998. Characterization of the chimeric retinoic acid receptor RARalpha/VDR. Leukemia 12:554–562.[Medline]
Roskoski R. 2004. The ERBB’HER receptor protein kinases and cancer. Biochemical andBiophysical Research Communications 319:1–11.
Ross TK, Darwish HM, Moss VE & DeLuca HF. 1993. Vitamin D-influenced gene expression via a ligand-independent, receptor-DNA complex intermediate. PNAS 90:9257–9260.
Stern DF. 2000. Tyrosine kinase signaling in breast cancer ErbB family receptor tyrosine kinases. Breast Cancer Research 2:176–183.[CrossRef][ISI][Medline]
Umesono K, Murakami KK, Thompson CC & Evans RM. 1991. Direct repeats as selective response elements for the thyroid hormone, retinoic acid, and vitamin D3 receptors. Cell 65:1255–1266.[CrossRef]
