|Home | About | Journals | Submit | Contact Us | Français|
Estrogen and progesterone are the defining hormones of normal female development, and both play critical roles in breast carcinogenesis. Cyclin D1 is a breast cancer oncogene whose amplification is linked to poor prognosis in estrogen and progesterone receptor-positive breast cancers. Here we report that cyclin D1 regulates progesterone receptor expression, consequently enhancing responses to estrogen and progesterone. Estrogen treatment of cyclin D1 transgenic mice increased progesterone receptor expression and induced mammary hyperplasias that were stimulated by progesterone and blocked by a progesterone antagonist. Progesterone receptor levels decreased in cyclin D1 knockout mice. Cyclin D1 regulated progesterone receptor expression through a novel estrogen- and cyclin D1-responsive enhancer in DNA encoding part of the 3′ untranslated region of the progesterone receptor gene. Small inhibitory RNAs for cyclin D1 decreased progesterone receptor expression and estrogen receptor binding to the 3′ enhancer region in human breast cancer cells. Since estrogen and progesterone regulate cyclin D1, our results suggest that cyclin D1's participation in a feed-forward loop could contribute to increased breast cancer risks associated with estrogen and progesterone combinations. Additionally, its regulation of the progesterone receptor identifies a novel role for cyclin D1 in ovarian hormone control of breast development and breast carcinogenesis.
We have long known that ovarian hormones control normal breast development (13). Likewise, the connection between ovarian hormones and breast cancer was first therapeutically exploited over 100 years ago (3). While estrogen and progesterone (Pgr) combine to regulate normal breast development during pregnancy, the Women's Health Initiative study of hormone replacement has clearly demonstrated an increased breast cancer risk when progestins are combined with estrogen (7). This result reinvigorated the view that combinations of progesterone with estrogen can be carcinogenic.
Cyclin D1 (CCND1) is a multifunctional G1-phase cyclin whose regulatory effects are particularly important in breast development and cancer (44). The cloning of cyclin D1 at 11q13 translocation breakpoints in parathyroid and mantle cell tumors was strong evidence of its oncogenic function, and evidence indicating that D1 was a G1 cyclin was a key finding that linked cancer to perturbations of the cell cycle. Extensive evidence links cyclin D1 to cyclin-dependent kinase regulation of the cell cycle (27), although a recent report demonstrates that it also acts in transcriptional control (4). 11q13 amplification has been repeatedly demonstrated in breast cancers (22) where CCND1 overexpression has been linked to estrogen and progesterone receptor (PR) status. This linkage was first attributed to cyclin D1's regulation by estrogen and progesterone (12, 37). In this model, CCND1 is proposed to contribute to poor treatment response of estrogen receptor (ER)-positive tumors by acting downstream to promote hormone agonist- and antagonist-independent proliferation (47).
The connection between poor patient outcome and cyclin D1 status in ER-positive breast cancers also led to explorations of a second potential explanation for cyclin D1-ER linkage. Identification of a potential ER coactivator interaction motif in the cyclin D1 protein suggested that CCND1 can interact with ER coactivators to activate estrogen receptor binding elements (ERE) in a ligand-independent manner (29, 52). Importantly, this direct activation of EREs by CCND1 was not inhibited by antiestrogens (52) and did not require an interaction between cyclin D1 and its associated cyclin-dependent kinases (CDK4/CDK6). This CDK4/CDK6 independence is potentially important, since CDK activity is not always increased in cyclin D1-associated breast cancers (45). Importantly, knock-in of CDK-independent cyclin D1 repairs a mammary developmental defect seen during pregnancy in cyclin D1 knockout mice, specifically confirming cyclin D1's potential CDK-independent functions (23).
A study employing cyclin D1 transgenic mice provided the first in vivo demonstration that cyclin D1 is an oncogene (46). These mice demonstrate extensive preneoplastic hyperplasias that precede tumor onset by months. Using laser-capture microdissection and microarray analysis, we identified genes that were upregulated in invasive tumors compared with normal or hyperplastic tissues (48). Remarkably, these genes were responsive to estrogen and responded to cyclin D1 in a CDK4/CDK6-independent manner, and the cyclin D1 and estrogen effects were additive. This result suggested that, as a third explanation for their association in breast cancer, cyclin D1 and the ER might share target genes. To investigate this possibility, we treated cyclin D1 transgenic mice with estrogen, causing hyperplasias that exhibited increased progesterone receptor expression. We now report that cyclin D1 and estrogen regulate progesterone receptor expression, which identifies a mechanism that may couple the effects of estrogen and progesterone in breast tissues and explain the tight association between cyclin D1 expression and hormone receptor status in breast cancer.
Mouse mammary tumor virus (MMTV) cyclin D1 transgenic mice were maintained as previously described (46). Pellets containing 17β-estradiol (Innovative Research) (0.72 mg/pellet) (49), mifepristone (RU486) (Innovative Research) (35 mg/pellet) (34), and/or progesterone (Innovative Research) (10 mg/pellet) (40) were implanted into wild-type (WT) and cyclin D1 transgenic mice 60 days of age, and sham-treated animals were used as controls. This estradiol dose provides twice the physiologic levels found during estrus but 1/20th of the levels found during pregnancy (30). The progesterone dose is sufficient to regulate ductal development in ovariectomized mice (40). The mifepristone dose blocks tumor development in BRCA−/− mice (34). Whole mounts were prepared 2 months later by dissecting the mammary fat pad from the pelt, fixing in 10% buffered formalin, staining with carmine red, dehydrating in graded series of alcohol, and clearing with xylene. Eight animals each were used for the wild-type and cyclin D1 transgenic mouse groups or were subjected to sham or estradiol treatment. Morphometry was performed by counting structures in three photomicrographs of each of the 32 resulting whole-mount preparations. Numbers of branches (B), termini (T), and lobuloalveolar structures (LAS) were counted and totaled for each image. The means and standard errors for the numbers of branches, termini, and alveolar structures per image were then plotted. Four animals each were used for the experiments combining estradiol, mifepristone, and/or progesterone treatment. Mammary glands were fixed in 10% formaldehyde in phosphate-buffered saline (PBS) for routine hematoxylin and eosin staining histology. Additional mammary glands from the same mice were harvested for RNA and protein analysis. Cyclin D1 knockout mice (JAX 002537 mice; Jackson Laboratories) and matched controls were sacrificed at 8 weeks of age and after 12 days of pregnancy, and the mammary tissues were used for RNA and protein analysis.
mRNA levels in cellular RNA isolated from the indicated cells or mouse tissues by the use of Trizol (Invitrogen) were analyzed. mRNA was reverse transcribed into cDNA and converted to double-stranded DNA (dsDNA); quantitative real-time PCRs (qRT-PCR) were performed as described previously (48). Table Table11 presents primer and probe sequences.
For Western analyses, 50 μg of protein sample was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were incubated with rabbit polyclonal anti-cyclin D1 (48), anti-PR antibody (C-20; Santa Cruz, Santa Cruz, CA), anti-beta-catenin (BD Transduction Laboratories, San Jose, CA), anti-IER3 (Abcam Inc., Cambridge, MA), anti-etv5 (Abnova Corp., Taipei City, Taiwan), anti-mitogen-activated protein kinase (anti-MAPK) (K-23; Santa Cruz Biotechnology, Santa Cruz, CA), or antiactin (Chemicon International, Temecula, CA) as indicated. The secondary antibodies used were purchased from Santa Cruz or were those included in an enhanced chemiluminescence detection kit (Amersham).
Proliferating T47D cells were incubated in phenol red-free Dulbecco's modified Eagle's medium (DMEM) with 5% charcoal stripped serum for 3 days before transient transfections were performed. Cyclin D1 (Ambion) and scramble (Dharmacon D-1205-20) small interfering RNAs (siRNAs) were transfected using NeoFX (Applied Biosystems). To demonstrate the knockdown efficiency of the siRNAs, cells were harvested 48 h after transfection and subjected to Western blot analysis. We also stably reduced cyclin D1 expression in T47D cells by the use of a lentiviral construct expressing a short hairpin RNA (shRNA) targeted to CTCTAAGATGAAGGAGACCAT in exon 2 of cyclin D1 (RNAi consortium no. TRCN0000040040).
ZR-75-1, CAMA, HCC1428, BT483, and MCF 7 breast cancer cells were grown under the conditions indicated by the supplier (ATCC). They were also transiently transfected with scramble or small inhibitory RNA for cyclin D1 (siCCND1) siRNAs as described for the T47D cells. Western blot analysis and quantitative real-time PCRs (qRT-PCR) were performed as described above.
We performed chromatin immunoprecipitation (ChIP) as previously described (48). We evaluated estrogen receptor genomic binding in T47D cells grown in phenol red-free DMEM with 5% charcoal-stripped serum for 3 days and treated with vehicle or 100 nM 17β-estradiol for 4 h. We performed chromatin immunoprecipitation by a method that we previously described (48). Cross-linking was performed by adding 37% formaldehyde to reach a final concentration of 1.5% for 15 min at 25°C with gentle agitation. The reaction was stopped by adding 0.125 M glycine for 5 min. Cells were scraped and collected by centrifugation; the pellet was washed twice with ice-cold PBS (120 mM NaCl, 2.7 mM KCl, 10 mM phosphate buffer, pH 7.4) and twice with immunoprecipitation (IP) buffer (150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.5% NP-40, 50 mM Tris-HCl [pH 7.5], 0.5 mM dithiothreitol [DTT]) and resuspended with IP buffer supplemented with 0.5% SDS and protease inhibitor cocktail (Roche). Sonication was performed until the desired DNA fragment sizes were reached. The lysate was then subjected to centrifugation at 12,000 × g for 10 min. The supernatant was collected and diluted 5-fold with IP buffer. For immunoprecipitation, anti-ER antibody (HC-20) from Santa Cruz or an equal concentration of normal rabbit serum was added to the lysate and incubated at 4°C overnight with agitation. Protein A Sepharose beads (Amersham) were added to the mixture and incubated for 1 h. The beads were collected by centrifugation and washed twice with IP buffer, twice with high-salt IP buffer (500 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.5% NP-40, 50 mM Tris-HCl [pH 7.5], 0.5 mM DTT), twice with stringent wash buffer (10 mM Tris-HCl [pH 7.5], 0.25 M LiCl, 0.5% NP-40, 0.5% sodium deoxycholate, 1 mM EDTA), and twice with Tris-EDTA (TE) buffer (10 mM Tris-HCl, 0.1 mM EDTA). The bound fraction was eluted in elution buffer (TE buffer with 1% SDS) at 65°C for 10 min. Cross-links were reversed by incubating the eluate at 65°C for 6 h. Samples were treated with proteinase K at 45°C for 1 h and subjected to phenol-chloroform extraction and ethanol precipitation. The primer sets used in quantitative PCRs (qPCR) are listed in Table Table22.
T47D cell nuclear extracts were obtained and analyzed using previously published methods (9). Gel shift activity for DNA binding proteins was determined using electrophoretic mobility shift assay (EMSA) binding buffer (20 mM HEPES [pH 7.5], 50 mM KCl, 7.5 mM MgCl2, 0.2 mM EDTA, 10% glycerol, 0.5 mM DTT) and 2 ng/μl poly(dI-dC) as a nonspecific competitor. Complexes formed in binding buffer were resolved on a 4% nondenaturing polyacrylamide gel containing 0.5× Tris-borate-EDTA (TBE) buffer at 4°C. Double-stranded oligonucleotides were purified by polyacrylamide gel electrophoresis and labeled using [γ-32P]ATP and polynucleotide kinase. Each binding reaction mixture contained between 0.1 and 0.5 ng of labeled oligonucleotides. Competition experiments were performed using a 200-fold molar excess of unlabeled oligonucleotides. Oligonucleotide sequences are listed in Table Table22.
Transient transfections were performed using T47D cells for the majority of the reporter assays (but see Fig. Fig.8,8, which shows the results of experiments in which we used MCF7 cells). The transient siCCND1 transfections were performed as described above. Cyclin D1 overexpression was accomplished using the plasmids pRC-CMV-cyclin D1 and pRC-CMV-mutant K112E as previously described (48). The plasmids hPRAB, hPRA, and hPRB were gifts from I. De Vivo (17). The 3× estrogen response element (ERE) luciferase plasmid was obtained from Addgene (14). The 1.3-kb nucleotide portion of the 3′ untranslated region (UTR) of the progesterone receptor gene (+94788 to +96092 relative to the B from the transcription start site [TSS]) was amplified by PCR using primers PR3F114788IF (GTCTGGATCCGTCGACGTCCTAAATTGCTTATCCTTACTTCACTAAG) and PR3R116092IF (AAGGGCATCGGTCGACTGGCAAAAAGCTTGAAACCACTGCTGTC) and cloned to SalI-digested pGL2 reporter vectors by the use of an In-Fusion cloning kit (Clontech). To insert the 3′ UTR in the opposite orientation, the same nucleotide portion was amplified using primers PR3F114788IV (AAGGGCATCGGTCGACGTCCTAAATTGCTTATCCTTACTTCACTAAG) and PR3R116092IV (GTCTGGATCCGTCGACTGGCAAAAAGCTTGAAACCACTGCTGTC) and similarly cloned into pGL2.
To clone hPR AB-CER, primers PR3F115435IF and PR3R115672IF were used to amplify T47D genomic DNA. To clone hPR AB-REC, primers PR3F115435IV and PR3R115672IV were used. PCR fragments were inserted into the SalI site of hPR AB plasmid by the use of an In-Fusion cloning kit (Clontech). To construct hPR AB-3XCER, a primer PR-P1F- and primer PR3R115672IF-amplified fragment was similarly cloned to the hPR AB SalI site. The second copy of the central element-containing region (CER) was inserted in the same site by a primer PR-P2F- and primer PR3R115672IF-amplified fragment. This step was then repeated to introduce the third copy of CER. To move the tandem CERs to the proximal position upstream of hPR promoter, the 3XCER fragment was amplified from plasmid hPR AB-3XCER by primers IF-3xCERXhoF and IF-3xCERXhoR and cloned to the XhoI-linearized hPR AB plasmid by the use of an In-Fusion kit. hPRAB-CERμ1 and hPRAB-CERμ2 were mutated from hPRAB-CER by the use of a Stratagene QuikChange Lightning site-directed mutagenesis kit. All constructs were validated by sequencing. The primer sequences used for cloning were as follows: for PR3F115435IF, GTCTGGATCCGTCGACATTCCAAGGCAGAGCTCAGG; for PR3R115672IF, AAGGGCATCGGTCGACCTTAAAAATAATGCCTCTCCC; for PR3F115435IV, AAGGGCATCGGTCGACATTCCAAGGCAGAGCTCAGG; for PR3R115672IV, GTCTGGATCCGTCGACCTTAAAAATAATGCCTCTCCC; for PR-P1F, GTCTGGATCCGTCGAGACGTCATTCCAAGGCAGAGCTCAGG; for PR-P2F, TATTTTTAAGGTCGATGCATATTCCAAGGCAGAGCTCAGG; for IF-3xCERXhoF, CATTCAGAATCTCGAGCTGGATCCGTCGAGACGTC; and for IF-3xCERXhoR, GGCATTTCCACTCGAGGGCATCGGTCGACCTTAA.
Cells were transfected in 96-well plates with 200 ng of luciferase reporter plasmids by the use of Lipofectamine 2000 (Invitrogen) with 0.5 ng of pCMV-renilla as a control. Cells were treated with 100 nM estradiol 24 h after transfection and assayed using a dual-luciferase reporter assay system (Promega) after 24 h of treatment.
Since cyclin D1 is particularly associated with hormone receptor-positive cancers, we sought to determine whether constitutive cyclin D1 expression would accelerate the response to estrogen in an animal model. We stimulated mammary proliferation using wild-type and CCND1 transgenic mice and continuous-release 17β-estradiol (E2) pellets (49). We initiated treatment at 2 months of age and evaluated the proliferative response of mammary glands 2 months later using whole-mount and histologic assessments. Estradiol and cyclin D1 and their combination all caused different changes in mammary glands (Fig. (Fig.1).1). Ducts in untreated wild-type mice had a simple branching structure (Fig. 1A and B) that terminated in single end-bud structures (see letter T in Fig. Fig.1A).1A). Estradiol treatment led to dilated ducts and aberrant buds at the termini of ducts in wild-type mice (Fig. 1B and F). Cyclin D1 overexpression increased levels of both side branching and terminal acinar structures in the transgenic mice (Fig. 1C and G). The combination of estradiol treatment with cyclin D1 overexpression caused more marked changes. First, ducts were dilated and showed multiple abnormally enlarged buds that were reminiscent of normal alveolar structures (see “LAS” in Fig. Fig.1D;1D; also see Fig. Fig.1H).1H). Second, multiples of such alveolar structures were found along the sides of ducts, whereas no such structures were found in similar locations in E2-treated wild-type mice, and those found in untreated CCND1 mice were small. These effects combined to produce extensive alveolar hyperplasia in E2-treated CCND1 transgenic mice. We analyzed whole mounts and histology using a total of eight mice from each of the four groups and performed morphometric analyses as described in Materials and Methods. Numbers of LAS in E2-CCND1 mice increased at least 10-fold over LAS in the wild-type mice and 5.8-fold over LAS in E2-treated mice (P < 0.001) (Fig. (Fig.1I1I).
We next assessed expression of cyclin D1, the ovarian hormone receptors, and several target genes to identify genes whose altered expression accompanied these histologic changes. Using species-specific primers, we found that levels of murine CCND1 mRNA increased in response to transgene treatment or to E2 treatment but that combining the two treatments did not cause additive increases (Fig. (Fig.22 A). Since the human cyclin D1 transgene was not present in the wild-type mice, we further evaluated the increase in its mRNA value with respect to the murine wild-type value and found that the transgenic CCND1 value equally increased in the absence and presence of the E2 stimulus, as expected, since the MMTV element does not respond to the presence of estrogen (32). Thus, estradiol was not acting simply to induce expression of the CCND1 transgene.
We then evaluated hormone receptor mRNAs and found that ER-alpha mRNA levels were only minimally affected by either estradiol treatment or transgene expression (Fig. (Fig.2B).2B). Instead, progesterone receptor mRNA levels increased in response to the presence of both estradiol and cyclin D1 and especially increased in the E2-treated CCND1 mice to a level that was 4-fold above the wild-type value (Fig. (Fig.2B2B [PR]; P = 0.001). The PR mRNA is transcribed as a full-length BA transcript and a 300-nucleotide-shorter A transcript. The B-specific extension can be identified using PCR primers specific for its unique 5′ region. As previously reported (1), PR expression in these nonpregnant mammary samples was nearly entirely the PR A transcript. Cyclin A2 (CCNA2) was evaluated as a potential CDK4/E2F target gene, and its expression increased 20-fold in level in response to CCND1 expression (Fig. (Fig.2C).2C). The combination of estradiol with CCND1 did not further increase CCNA2 numbers. We also evaluated expression of two candidate CCND1 targets that were found to be associated with tumor invasion in our previous study (48) (Fig. (Fig.2C).2C). Expression of both the immediate-early response 3 (IER3) and ets variant 5 (ETV5) mRNAs increased in response to the combination of cyclin D1 and estrogen, along with that of the increased progesterone receptor mRNA. Since IER3 is a known progesterone target (35), the ETV5 and IER3 increases may have resulted as a response to the cyclin D1-induced increase in PR rather than as a direct effect of CCND1 and E2 on regulation.
To confirm the mRNA results, we performed immunoblot analyses for the progesterone receptor as well as for several progesterone receptor and cyclin D1 target genes (19, 36) (Fig. (Fig.2D).2D). Progesterone receptor protein levels generally paralleled mRNA changes, and only the 94-kDa PR A isoform was seen, as previously described (1). PR levels clearly increased in the cyclin D1-expressing transgenic tissues. β-Catenin, a known target of PR function, increased in level only in the E2-CCND1 mice. This increase in β-catenin levels, together with the increased PR levels, suggested that the E2-CCND1 combination enhanced both PR expression and PR function (16). Indeed, the histology of the E2-CCND1 mice resembles that seen in PR overexpression (41) and in increased β-catenin signaling (26). Both the IER3 and ETV5 protein levels increased in response to the CCND1 transgene (Fig. (Fig.2D),2D), although increases in IER3 and ETV5 protein numbers did not fully match the mRNA increases, suggesting that additional posttranscriptional mechanisms may limit additional estrogen-induced expression. Taken together, our results suggest that combining E2 with constitutive CCND1 expression causes a hyperplastic abnormality in virgin mice and that that abnormality is accompanied by increased PR and PR signaling in those mice.
Loss of cyclin D1 in knockout mice causes mammary growth failure during pregnancy, when progesterone receptor signaling is most critical for normal reproductive development (43). To determine whether PR expression is decreased in CCND1 knockout mammary glands, we measured PR mRNA and protein levels in glands harvested from virgin wild-type and CCND1−/− mice at 8 weeks of age and in glands harvested from wild-type and CCND1−/− mice at 12 days of pregnancy (Fig. 2E and F). Loss of cyclin D1 led to decreased PR mRNA levels in both stages. PR protein changes paralleled the changes in mRNA, and, as expected, we observed a PR B isoform band at 114 kDa in pregnant tissues. The results seen with a keratin 19 blot experiment suggested that the observed changes were not merely the result of changes in luminal cell composition in the knockout tissues. PR expression was increased in tumors in the cyclin D1 transgenic mice (Fig. 2G and H).
Given this evidence for increased PR signaling and the phenotypic changes that were consistent with a shift toward enhanced PR function in estrogen-treated cyclin D1 transgenic mice, we next sought to determine whether progesterone (Pgr) agonists or antagonists would enhance or block effects of the E2-CCND1 combination in virgin mice (Fig. (Fig.3).3). First, mifepristone (RU486) blocked much of the effect of E2 on CCND1 mice (Fig. 3F and H versus Fig. 3B and D). Importantly, the addition of progesterone pellets to E2 and CCND1 markedly stimulated the lobuloalveolar development beyond that achieved by E2 treatment and clearly increased hyperplasia compared to the results seen with E2-Pgr-treated wild-type mice (compare Fig. 3J and L to Fig. 3I and K). In contrast, treatment of wild-type and CCND1 mice with progesterone alone revealed few differences (Fig. 3M and O and Fig. 3N and P, respectively). We repeated these experiments using mice subjected to ovariectomy at 7 weeks of age. The histologic changes that accompanied estrogen and progesterone treatment of wild-type and cyclin D1 transgenic mice were not affected by ovariectomy, suggesting that these phenotypic changes were not dependent on endogenous ovarian hormone production (not shown).
To determine whether the regulatory effects of CCND1 and estrogen on PR expression could also be found in human cells, we tested the effects of a small inhibitory RNA for cyclin D1 (siCCND1) on PR expression in T47D breast cells. Progesterone receptor mRNA levels decreased 60% ± 1% in response to siCCND1 compared to those seen with scramble siRNA-transfected control cells. This decrease was seen primarily in the shorter PR A isoform analyzed using isoform-specific primers. Western blot experiments confirmed that siCCND1 transfections decreased both cyclin D1 and PR levels (Fig. (Fig.44 A), and image analysis confirmed a 2-fold decrease in cyclin D1 protein levels. We then evaluated the effect of this PR change on IER3 expression and found that it also decreased in response to decreased CCND1.
Although previous investigations failed to discover PR promoter sequences that contained the canonical estrogen receptor binding sequence (agGtCAnnnTGaCct; uppercase indicates conserved sites) (6), several investigators identified alternative promoter elements that responded to estrogen, including Sp1 (39), AP1 at +90 from the PR B transcription start site (TSS) (33), a G/A GATA5 breast cancer risk polymorphism (17), an Sp1-ERE half-site (33), a second AP1 at +745 (38), an RNA polymerase II binding site (6), and a rat promoter site that responded to E2 (20). We used ChIP-PCR to directly evaluate effects of estradiol and CCND1 on estrogen receptor binding to these candidates (Fig. (Fig.4B).4B). According to the results of our chromatin immunoprecipitation studies, none of these previously proposed alternative promoter elements showed increased estrogen receptor binding with estrogen stimulation. Two of the eight sites showed a significant decrease in estrogen receptor binding after cyclin D1 knockdown, and they were located at the core progesterone receptor promoter sites for the B isoform (centered at −60) and the A isoform (centered at +658).
The Genome Browser of the University of California at Santa Cruz (UCSC) identifies three highly conserved, canonical estrogen receptor binding sites ([agGtCAnnnTGaCct]) (ERE 1-ERE 3) in the progesterone receptor gene that are located at 51,200, 79,800, and 95,585 nucleotides downstream of the transcription start site for the B isoform of the progesterone receptor (Fig. (Fig.4C)4C) (21). A recent genome-wide screening showed that only the site at +95585 binds the estrogen receptor in response to estrogen treatment (6). We used ChIP together with quantitative PCR (ChIP-PCR) to directly evaluate effects of estradiol and CCND1 on these candidate sites (Fig. (Fig.4D).4D). We first confirmed that estradiol induced anti-ER-bound DNA only for the ERE 3 site, which also showed the highest levels of αER-specific binding. The siCCND1 small inhibitory RNA decreased ER binding at all three sites, but only the binding to the ERE 3 site was both stimulated by estrogen and blocked by siCCND1.
To confirm the ChIP experiment results, we stably expressed an shRNA that targets exon 2 of cyclin D1 in T47D cells. This shRNA knocked cyclin D1 mRNA levels down by 50%, which resulted in a decrease of PR mRNA to 10% of its levels in control cells (Fig. (Fig.4E).4E). We then evaluated DNA-protein interactions across the 240 nucleotides spanning the ERE 3 region by the use of electrophoretic mobility shift assays (EMSA). We first compared the DNA binding activity of lysates harvested from the vector control to that of CCND1 shRNA cells for 13 overlapping oligonucleotides that spanned the entire region. Cyclin D1 knockdown significantly altered binding at the canonical nucleotide 95585 ERE site (Fig. (Fig.4F)4F) without altering binding at other adjacent sites (not shown). Using estrogen stimulation and cold-competitor oligonucleotides, we confirmed the estrogen response and sequence specificity of this ERE site.
We cloned a 1.3-kb genomic segment containing ERE 3 into standard luciferase reporter gene vectors to determine whether it could enhance transcription. Using PCR cloning, we placed it downstream of luciferase in reporter vectors lacking a heterologous promoter (pGL2-Basic [pGL2N]) and containing a heterologous promoter (pGL2 promoter [pGL2P]). We transfected reporter constructs together with either a scramble control siRNA (scramble) or siCCND1 (Fig. (Fig.5A5A and andB)B) into T47D cells. In addition, we cotransfected the reporter constructs with either a control expression vector or a CCND1-expressing vector (pCCND1) (Fig. 5C and D). We performed these experiments using cells treated with vehicle alone (Fig. 5A and C) and cells treated with estradiol (Fig. 5B and D). The PR 3′ UTR sequences did not significantly increase luciferase expression in the absence of promoter sequences (Fig. (Fig.5A5A [pGL2N-UTR versus pGL2N]). However, they caused a 6-fold increase in expression when combined with a heterologous promoter (Fig. (Fig.5A5A [pGL2P-UTR versus pGL2P]).
Estradiol stimulated the PR UTR when combined with a heterologous basal promoter, taking estradiol's inhibition of the isolated promoter into consideration (compare Fig. 5A and B). This estradiol responsiveness of the PR UTR was best seen in the experiments using scramble control oligonucleotides and siCCND1 to determine whether decreased cyclin D1 could block the 3′ PR UTR's enhancing function (Fig. 5A and B). Importantly, siCCND1 decreased PR 3′ UTR-driven reporter activity to 78% of control levels in vehicle-treated control cells and to 42% of controls in estradiol-treated cells. Transfection of a cyclin D1 expression vector increased the reporter activity of the PR 3′ UTR (Fig. 5C and D). It doubled pGL2P-UTR activity in vehicle-treated cells and increased pGL2P-UTR reporter activity nearly 3-fold in the presence of estradiol treatment.
To determine whether these effects were related to cyclin D1-CDK interactions, we compared the effect of a wild-type cyclin D1 to that of a cyclin D1 K112E mutant that cannot interact with CDK4/CDK6 (15) (Fig. (Fig.5E).5E). The KE mutant had effects equivalent to those of the wild-type cyclin D1 on the pGL2P-UTR reporter vector containing the 3′ UTR of the progesterone receptor in vehicle-treated cells. Interestingly, the KE mutation abrogated wild-type cyclin D1's 2-fold increase in activation of the pGL2P-UTR in estradiol-treated cells. Finally, a reporter construct containing a 3-fold repeat of an isolated estrogen receptor binding sequence (ERE) (14) did not respond in the same manner to the CCND1 expression plasmid or to siCCND1 (Fig. (Fig.5F).5F). Taken together, these assays identify the 3′ UTR region of the PR gene as a unique cyclin D1- and estrogen-responsive element.
We further analyzed the 1.3-kb PR 3′ UTR region to determine whether it functions as an enhancer (Fig. (Fig.6).6). We first combined the 3′ UTR region with the progesterone receptor's own promoter region (17) and found that it activated reporter expression from the PR promoter 13-fold more potently than it activated the heterologous basal promoter in the pGL2 vector. We next found that its core 238 nucleotides that are conserved from lizards to humans are nearly as functional as the full 1.3-kb region when combined with the progesterone receptor's own promoter. The functions of both the full 1.3-kb UTR region and its core 238 nucleotides were independent of orientation, and increasing the core sequence copy number increased its function. The core 238-nucleotide element responded to estrogen in a position-independent manner. Finally, mutations of its ERE binding sequence (ATGTCATGCTGACCT) to AgGgCATGCTGACCT or ATGTCATGC-----T reduced its enhancement of promoter function and its estrogen response.
To evaluate cyclin D1's effects on PR isoform-specific promoters, we cloned the PR 3′ UTR enhancer into reporter constructs containing isolated PR A and B promoters (17) (Fig. (Fig.7).7). We repeated cotransfections with cyclin D1 and siCCND1. The PR A promoter is somewhat responsive to changes in CCND1 expression on its own, and its function and response to CCND1 are markedly increased by the addition of the PR 3′ UTR enhancer sequences (Fig. 7A and B). Using isoform-specific qRT-PCR, we also found that PR A mRNA decreases accounted for most of the response to siCCND1 (55% of levels in scramble-transfected controls). A strong reaction of PR A to cyclin D1 changes was seen in response to both siCCND1 and CCND1 overexpression. Interestingly, the PR B promoter was not responsive to CCND1 on its own, but siCCND1 blocked estrogen-induced increases when the PR B promoter was combined with the PR 3′ UTR enhancer (Fig. 7C and D). While the reporter gene result differs somewhat from the behavior of endogenous PR B (Fig. (Fig.4A),4A), the added estrogen in these experiments likely accounts for the difference. The PR gene 3′ UTR enhancer therefore activates both PR isoform promoters, with the A isoform showing greater interaction with cyclin D1 and the B isoform being more responsive to estrogen.
Having identified a direct link between CCND1 levels and PR expression, we sought to determine whether additional breast cancer cell lines exhibited this CCND1 control of PR (Fig. (Fig.8).8). siCCND1 transfection effectively reduced CCND1 protein levels in five additional cell lines (Fig. (Fig.8A),8A), and PR was consequently reduced in two of them, namely, the ZR75-1 and MCF7 cell lines. We further found that IER3 and ETV5 levels also decreased in response to CCND1 and PR decreases in the same two cell lines. We used isoform-specific qRT-PCR to analyze PR isoform transcripts (Fig. (Fig.8B).8B). We found that levels of the combined PR A-and-B transcript, predominately, those of PR A-specific mRNA, decreased in response to the siCCND1 knockdown in ZR75-1 and MCF7 cells. Finally, we evaluated the response of the 3′ UTR enhancer to cyclin D1 expression in MCF 7 cells (Fig. (Fig.8C).8C). As described for the T47D cells, decreased cyclin D1 levels decreased the PR UTR enhancer activity, increased cyclin D1 levels increased the PR UTR enhancer activity, and the cyclin D1 K112E mutant remained active in these cells.
While studies of cyclin D1 have long established its primary role in cell cycle regulation, additional studies examining its effects on gene expression have suggested that it might have additional CDK4/CDK6-independent effects (29, 52). Since those previous studies initially relied on reporter assays, endogenous targets were not identified (51). By identifying genes associated with cyclin D1's effects on tumor invasion, we recently identified genes that responded to the combination of estrogen and cyclin D1 (48). Given that these genes responded to this combination, we treated cyclin D1 transgenic mice with estrogen. We now report that the combination of estrogen and constitutive cyclin D1 expression increased progesterone receptor expression, increased PR function, increased PR-target gene expression, and increased PR effects (Fig. (Fig.11 to to33).
The phenotypic changes accompanying cyclin D1 overexpression shown here, together with the previously reported phenotype of cyclin D1 knockout mice, support the idea that the PR is a cyclin D1 target. The histologic appearance of cyclin D1 transgenic mammary glands treated with estrogen (Fig. (Fig.1)1) strongly resembles glands in mice constitutively expressing additional PR (41, 42). The failure of mammary development during pregnancy in cyclin D1 null mice (43) is nearly identical to the loss of mammary development in PR null mice (28). In contrast, mammary glands in mice lacking the ER do not resemble glands in mice lacking cyclin D1 (11). Our data suggest that cyclin D1 may act both as a cell cycle mediator of ovarian hormone activity and as a developmental switch that increases in mammary glands to augment their effects on mammary development. This model implies an additional potential explanation for the decreased tumor formation seen in erbB2 and ras transgenic mice when they are cyclin D1 null (50), since the decreased tumorigenesis could result from loss of PR function and/or loss of cell cycle control. Notably, a progesterone receptor antagonist blocks tumorigenesis in BRCA mutant mice (34).
After years of studies of estrogen regulation of the PR gene, a clear estrogen-responsive enhancer has not yet been identified in its sequences (31). Various studies of PR promoter proximal elements particularly failed to identify canonical estrogen receptor binding elements. Here we show that a 238-nucleotide sequence in the 3′ UTR of the progesterone gene is a classic estrogen-responsive enhancer. It contains a canonical estrogen receptor binding element whose sequence is highly conserved. The homology plot in Fig. Fig.4C4C shows that this element is the most highly conserved region of the progesterone receptor gene, excluding its coding exons. The estrogen receptor binds this sequence in an estrogen-dependent manner. In reporter experiments, it activates estrogen-regulated expression of both the PR promoter and a heterologous promoter. Its activation is independent of position and orientation and increases with increased copy numbers (Fig. (Fig.66).
Given the importance of progesterone receptor expression, exploring unique features of its enhancer is important to understand how ovarian hormones regulate breast development and carcinogenesis. We have shown using ChIP, siRNA knockdowns, and cyclin D1 overexpression studies that this enhancer is uniquely responsive to cyclin D1. Its response to cyclin D1 is independent of the presence of CDK4/CDK6 in the absence of estrogen, although its response is dependent on the presence of CDK4/CDK6 in the presence of estrogen (Fig. (Fig.5E).5E). Estrogen and cyclin D1 additively stimulate its function, potentially explaining their combined in vivo effects on mammary development. In contrast, a reporter construct containing a 3-fold repeat of the canonical estrogen response element did not respond to cyclin D1 decreases or increases (Fig. (Fig.5F).5F). Importantly, the estrogen receptor binding element was required for the PR enhancer's function both for promoter activation and for estrogen response (Fig. (Fig.66).
Our results provide the first in vivo identification of a gene containing an estrogen receptor binding element (ERE) that responds to cyclin D1 in the manner previously proposed (29, 52). On the other hand, we did not confirm that cyclin D1 could activate an isolated ERE (Fig. (Fig.5F),5F), demonstrating that other elements in the PR enhancer are required for its cyclin D1 responsiveness. Our results suggest that cyclin D1 has at least two mechanisms responsible for activation of the PR enhancer. Further, we could not demonstrate direct binding of cyclin D1 to the PR enhancer in ChIP experiments, largely because available cyclin D1 antibodies immunoprecipitate several nonspecific bands. Given our results, the PR enhancer obviously contains additional DNA binding elements likely to contribute to its unique response to cyclin D1. Brown and colleagues have found that FOXA1 binding in combination with the ER defines enhancer-driven lineage-specific transcription for a number of ER-targeted enhancers (5). The PR 3′ UTR enhancer region lacks a FOXA1 binding site, although the results of another genome-wide ChIP scan previously suggested that FOXA1 can bind in this region (25). Additionally, there are no E2F sites in the PR enhancer, and PR is not known to be regulated through the cell cycle. Rather, a transcription regulatory element search (TRES) of the core region of this PR enhancer identified conserved sites for AP4, E47, Sox-5, and HOXA9 (18). Since the PR enhancer is now a well-characterized endogenous target sequence that is regulated by both increased and decreased cyclin D1 levels, identification of combinations of transcription factors involved in its regulation would be important to gain a better understanding of mechanisms explaining cyclin D1's role in gene expression control.
The role of progesterone in breast cancer pathogenesis has long been controversial (24). Progesterone's contribution to rapid signaling through the activity of mitogen-activated protein kinases is important in the deregulation of breast cancer cell proliferation (8). Hormone replacement therapies that combine estrogen with progesterone clearly demonstrate that these hormones together can promote breast carcinogenesis (7). We now show that cyclin D1's tight association with the ER and PR status of breast cancers may result from the combination of ovarian hormone control of cyclin D1's cell cycle effects together with a feed-forward regulation of PR expression by cyclin D1 and estrogen. This mechanism would provide a strong selective pressure for increasing cyclin D1 expression in hormone receptor-positive breast cancers. We further show that this regulation is encoded in a novel 238-nucleotide enhancer in the progesterone receptor gene. Identification of additional regulators of this enhancer should offer important insights into cyclin D1's CDK4/CDK6-independent transcriptional effects and could potentially identify therapeutic targets whose manipulation could inhibit the combined effects of cyclin D1 and estrogen in cases of breast cancer.
For this work, C. Yang, L. Chen, and E. V. Schmidt were supported by a grant from the National Cancer Institute of the National Institutes of Health (grant RO1 CA69069). C. Yang received additional support from grant R01 CA112021, and both E. V. Schmidt and L. Chen received support from grant U01 AI07033. C. Li was supported by the Harvard Breast Cancer SPORE grant P50 CA89393.
We gratefully acknowledge review of the histology results determined for our mice by Robert Cardiff of the University of California at Davis.
Published ahead of print on 19 April 2010.