Reversal of Endogenous Dopamine Receptor Silencing in Pituitary ...

CANCER-ONCOGENES

Reversal of Endogenous Dopamine Receptor Silencing in Pituitary Cells Augments Receptor-Mediated Apoptosis Haneen Al-Azzawi,* Kiren Yacqub-Usman,* Alan Richardson, Leo J. Hofland, Richard N. Clayton, and William E. Farrell Human Disease and Genomics Group, Institute of Science and Technology in Medicine (H.A.-A., K.Y.-U., A.R., R.N.C., W.E.F.), School of Medicine, Keele University, Stoke on Trent, Staffordshire ST4 7QB United Kingdom; and Department of Internal Medicine, Division of Endocrinology (L.J.H.), Erasmus Medical Center, 3015 Rotterdam, The Netherlands

Dopamine (DA)-agonist targeting of the DA D2 receptor (D2R) in prolactinomas is the first-line treatment choice for suppression of prolactin and induction of tumor shrinkage. Resistance to DA agonists seems to be related to receptor number. Using the MMQ and GH3 pituitary cell lines, that either do or do not express D2R, respectively, we explored the epigenetic profile associated with the presence or absence of D2R in these cells lines. These studies led us to explore pharmacological strategies designed to restore receptor expression and thereby potentially augment DA agonistmediated apoptosis. We show in GH3 cells that the D2R harbors increased CpG island-associated methylation and enrichment for histone H3K27me3. Conversely, MMQ cells and normal pituitaries show enrichment for H3K9Ac and barely detectable H3K27me3. Coculture of GH3 cells with the demethylating agent zebularine and the histone deacetylase inhibitor trichostatin A was responsible for a decrease in CpG island methylation and enrichment for the histone H3K9Ac mark. In addition, challenge of GH3 cells with zebularine alone or coculture with both agents led to expression of endogenous D2R in these cells. Induced expression D2R in GH3 cells was associated with a significant increase in apoptosis indices to challenge with either DA or bromocriptine. Specificity of a receptor-mediated response was established in coincubations with specific D2R antagonist and siRNA approaches in GH3 cell and D2R expressing MMQ cell lines. These studies point to the potential efficacy of combined treatment with epigenetic drugs and DA agonists for the medical management of different pituitary tumor subtypes, resistant to conventional therapies. (Endocrinology 152: 364 –373, 2011)

opamine (DA) agonists are considered a first-line treatment choice for patients with pituitary prolactinomas where they not only effectively suppress prolactin (PRL) but also reduce tumor size (1, 2). Despite their success, a small proportion of patients are intolerant to DA agonist therapy (3, 4), and in cases where resistance is apparent this seems to be related to the number of DA D2 receptors (D2R) expressed by the adenoma (5). Activation of the D2R by DA and DA agonists leads to reduced cAMP production through interaction with Gi/Go proteins (6).

D

The reduction in cAMP levels in normal and tumoral lactotrophs is considered integral to the inhibition of PRL synthesis and release (1, 6 –9). In contrast to our understanding of pathways regulating PRL release within the lactotroph our understanding of receptor(s) and their intracellular pathways responsible for tumor shrinkage and or apoptosis are less clear. Early studies suggested that resistance to DA but not bromocriptine (BC)-mediated apoptosis was a consequence of these cells not expressing a D2R (10). However, more recent

ISSN Print 0013-7227 ISSN Online 1945-7170 Printed in U.S.A. Copyright © 2011 by The Endocrine Society doi: 10.1210/en.2010-0886 Received August 2, 2010. Accepted November 11, 2010. First Published Online December 22, 2010 * H.A.-A. and K.Y.-U. contributed equally to this work.

Abbreviations: BC, Bromocriptine; ChIP, chromatin immunoprecipitation; CT, cycle thresholds; D2R, DA D2 receptor; DA, dopamine; DAT, DA transporter; DMSO, dimethylsulfoxide; HDAC, histone deacetylase; NRP, normal rat pituitary; PBGD, porphobilinogen deaminase; PRL, prolactin; siNT, nontarget siRNA; TSA, trichostatin A.

364

endo.endojournals.org

Endocrinology, February 2011, 152(2):364 –373


studies by Jaubert and colleagues (11) suggest that DAinitiated apoptotic responses are mediated through the DA transporter (DAT). Although this transporter represents an attractive alternate mechanism a recent study from our laboratory, using this same model system, namely, the rodent pituitary cell line GH3, reached a different conclusion. In this case, we showed that both DA and BC elicits an apoptotic response in these cells in the absence of a either a D2R or a DAT (12). Moreover, characterization of the intracellular apoptotic pathways engaged by these drugs show them to be distinct. We found that a c-Jun N-terminal kinases-mediated apoptotic pathway is activated by BC, but not by DA. However, the DA- and BC-mediated apoptotic pathways converge to activate the terminal caspase-caspase-3. Coincubation experiments in these cells reinforced and extended these findings where we observed a synergistic increase in apoptotic end points (12). Perhaps surprisingly, given its near ubiquitous but not invariant expression pattern in the different cell types of the normal pituitary gland and their cognate tumors (2), our understanding of mechanisms responsible for loss or reduced D2R expression are incomplete. In these adenomas, and in common with most other tumor types, genetic alterations act in concert with changes to the epigenome (13, 14). These epimutations are apparent as changes in both global and gene-specific DNA methylation and as histone modifications. In these cases, aberrations are frequently associated with or responsible for altered gene expression profiles that characterize the initiation, development, and progression of disease states (13, 15). In addition, changes to the epigenome may also impact upon treatment responses, and this is apparent in different tumor types. As example, in ovarian tumors silencing of genes involved in apoptotic pathway are a frequent finding (16) as are epigenetic aberrations in specific receptor pathways in primary breast tumors and their cell lines (17) and in these cases can lead to varying degrees of drug resistance. In this study, we investigated potential epigenetic mechanisms responsible for or associated with loss of D2R expression in pituitary cells and the potential for pharmacological strategies to “unmask” and thereby induced expression of silenced genes. Through use of pituitary tumor cell lines that either express or do not express endogenous D2R we were able to explore the potential to reinstate and/or augment receptor-mediated apoptotic responses.

Materials and Methods Cell cultures GH3 cells, a rat pituitary cell line (CCL-82.1; American Type Culture Collection) in the somatolactotroph lineage, were cultured as described previously (12) in Dulbecco’s modified Eagle’s


365

Medium (DMEM) (Biosera, Ringmer, UK), supplemented with 10% fetal bovine serum, 4 ␮g/ml gentamycin (Sigma-Aldrich, Dorset, UK), and 2 ␮g/ml ampicillin (Sigma-Aldrich) in a 5% CO2 atmosphere at 37 C. Cells used in the current study had undergone ⬍10 passages. MMQ cells, a rat pituitary cell line in the lactotroph lineage (CRL-10609; American Type Culture Collection), were grown as suspension cultures in Ham’s F-12 media (Sigma-Aldrich, Dorset, UK) supplemented with 2.5% fetal bovine serum, 10% horse serum, 4 ␮g/ml gentamycin, and 2 ␮g/ml ampicillin in a 5% CO2 atmosphere at 37 C. Cells used in the current study had undergone ⬍5 passages.

Primary tissue As control tissue, for some of the described experiments, we used DNA and RNA extracted from normal rat pituitaries. In these cases, the strain was Sprague-Dawley and these were obtained from a commercial source (Charles River, Kent, UK).

Chemicals The cytidine deaminase inhibitor and DNA demethylating agent, Zebularine [1-(␤-D-ribofuranosyl)-1,2-dihydropyrimidin-2-one] and histone deacetylase (HDAC) inhibitor, trichostatin A (TSA) were obtained from Sigma-Aldrich (Dorset, UK). Stock solutions were prepared in dimethylsulfoxide (DMSO) and ethanol respectively. BC, [2-Bromo-␣-ergocryptine methanesulfonate salt] and DA, (3-Hydroxytyramine) were also obtained from Sigma-Aldrich, (Dorset, UK) and stock solutions prepared in DMSO and PBS, respectively. DA D2R antagonists Eticlopride (18) and Haloperidol (19) were obtained from Calbiochem (Nottingham, UK). The drugs were dissolved in PBS and DMSO, respectively and stored in accordance with the manufacturer’s suggestions. The doses of antagonists were first optimized relative to the dose(s) of D2R agonists used (data not shown). Antagonists were used at a dose of 25 ␮M, and higher doses were found to be cytotoxic. The doses of agonists that were used in the current studies are shown in the figure legends. All experiments were repeated at least three times with triplicate determinations. The antagonists used in this study were used to determine the specificity of D2R-mediated response in GH3 and MMQ cells.

Treatments of pituitary cell lines with demethylating and histone-modifying reagents Because preliminary experiments showed in early passage GH3 cells that loss of endogenous D2R is associated with increased promoter CpG island methylation and histone tail modification (see Results) initial experiments were designed to induce reexpression of the D2R in these cells. To this end, GH3 cells were seeded into six-well plates and incubated overnight in growth media and challenged for 48 h with either Zebularine, or TSA alone, or in combination at the doses shown in the figures or their legends.

Response of D2 expressing GH3 cells to receptor agonist-induced apoptosis The D2R expressing GH3, generated as described above, were designated GH3-D2exp. These cells, and those treated with vehicle alone, GH3-D2veh, and not expressing D2R, were cultured overnight in fresh growth media in six- or 96-well plates before apoptotic drug challenges. The majority of experiments,

366

Al-Azzawi et al.

Restoration of D2 Receptor–Mediated Apoptosis


using these cells, were performed in six-well plates with the exception of the caspase assays (see below). In these experiments, to facilitate the caspase assay format, GH3-D2exp and GH3D2veh cells were cultured overnight in 96-well plates. In either case (six- or 96 well), semiconfluent cultures of GH3-D2exp and GH3-D2veh cells were challenged with either BC or DA for the time points and drug doses shown in the figures. In those cases where coincubation experiments with D2R receptor antagonists were performed, the cells were pretreated with the antagonists for 2 h before the drug was added. Similar studies to those described here were also performed on MMQ cells which constitutively express endogenous D2R. Importantly, GH3-D2exp and GH3-D2veh cells were only used in a single experiment and newly established cells, expressing D2R, and generated from the parental cell line, GH3, were used for each discrete experiment.

Expression analysis

D2R knockdown

Quantitative RT-PCR

To assess the specificity of the D2R-mediated response we first used an RNAi approach. To this end, Dharmacon SMARTpool siRNA to the rodent D2R (siD2R) (Thermofisher Scientific, Epsom UK) was used to knockdown induced or endogenous D2R transcript expression in the GH3 and MMQ cell lines, respectively. In addition, a nontarget siRNA (siNT) (Thermofisher Scientific) was used as a control to confirm that the effects of the siRNA treatment were specific for the D2R transcript. As described previously (12, 20), low passage GH3 or MMQ cells were seeded into six-well plates (Sarstedt, Leicestershire, UK) until they were between 30 –50% confluence. Cells were then treated with 20 nM of either siD2R or siNT using Lipofectamine 2000 (Invitrogen, Paisley, UK) according to the manufacturer’s instructions. The efficiency and specificity of D2R knockdown was first determined in time course experiments (data not shown). In these experiments, optimal knockdown, as determined by quantitative RT-PCR for the D2R, was apparent between 36 and 48 h posttransfection. Therefore, in all subsequent drug-challenge experiments of GH3 or MMQ cells, drug challenges were initiated at the 48-h time point. siRNA experiments were repeated three times independently.

The cDNA samples and primers described for the RT-PCR analysis were also used for real-time quantification of the DA D2R as previously described (12, 20). Briefly, reactions were prepared containing 1 ⫻ Brilliant SYBR Green QPCR master mix (Strategene, Santa Clara, CA), 400 nmol of forward and reverse primers. and 300 nmol of ROX passive reference dye (Strategene). All samples were analyzed in triplicate to account for technical variation. The target genes were normalized to the endogenous control and relative quantification was carried out using relative-standard curve and 2 ⫺⌬⌬ cycle thresholds (CT) methods, where, ⫺⌬⌬CT ⫽ CT(DRD2 cell line ⫺ PBGD cell line) ⫺ CT(DRD2 normal pituitary gland ⫺ PBGD normal pituitary gland) . In those cases where the calculated values for expression were 5% or less this was reported as loss.

Apoptosis assays After the drug challenges described above, cells were harvested as previously described (12). Cells were then stained using Hoechst 33342 (Molecular Probes, Paisley, UK) or analyzed for DNA fragmentation using a commercial terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL) labeling kit (Promega, Southampton UK) as described previously (12).

Caspase assays Semiconfluent cultures of GH3-D2exp and GH3-D2veh cells, in 96-well plates, were challenged with drugs for 6 h and activated caspases detected using a commercial kit that specifically detects the terminal caspases (Caspase 3 and 7) as previously described (12, 20). Assays were performed as suggested by the suppliers (Promega, Southampton UK) in triplicate determinations and experiments repeated at least three times. Data were analyzed after correcting for the cell numbers measured by crystal violet staining.

Reverse transcription-polymerase chain reaction Total RNA was extracted from normal rat pituitary (NRP) and drug-challenged cells lines as previously described (12, 20). cDNA was synthesized as previously described (12, 20) using 200U of M-MLV reverse transcriptase (Promega, Southampton, UK) according to the manufacturer’s instructions. The expression of the house keeping gene, porphobilinogen deaminase [PBGD, (21)] and the D2R were assessed using the following primers, PBGD, (forward 5⬘-ATGTCCGGTAACGGCGGC-3⬘ and reverse 5⬘-CAGCATCGCTACCACAGTGTC-3⬘, product size-126 bp) and the D2R receptor, (forward 5⬘- CCGTGGGTTGTCTACCTGGAG-3⬘ and reverse 5⬘-TGTACCTGTCAATGCTGATGG-3⬘, product size-136 bp) as previously described (12). The RT-PCR products were analyzed by agarose gel electrophoresis.

Bisulfite treatments and genomic methylation sequencing DNA was extracted from NRP and the pituitary cell lines pre- and postdrug manipulations using phenol-chloroform- proteinase K method as previously described (12, 20). Bisulfite treatment of DNA was performed using the EZ DNA Methylation-Gold Kit according to the manufacturer’s protocol (Zymo Research, Cambridge, UK). Briefly, 0.5 ␮g of DNA was treated with CT conversion reagent, followed by thermocycling for 3.25 h (98 C for 10 min followed by 64 C for 3 h). The converted DNA was desulfonated, washed, and purified by using Zymo-Spin columns. The converted DNA was eluted from the column and stored at ⫺20 C.

DA receptor CpG island The methylation status of the D2R CpG island (www.ebi.ac.uk; see Fig. 1) in NRP and pituitary cell lines was determined using two overlapping PCRs to amplify a 880-bp region. The primers used were D2RF1 5⬘- GTATAAGAGGGGATTAGTTT ⫺3⬘ and D2RR1 5⬘- ACCTTCTTTATCATTCCCATCTTAAATC-3⬘ (that amplified a 545-bp fragment from the antisense strand and includes the 20 CpG shown in Figs. 1 and 2) and D2RF2 5⬘GTTATTTAAAAAGTATAGGA-3⬘ and D2RR2 reverse 5⬘TTACCTTCAAACCATATAAC-3’ (that amplified a 581-bp fragment from the sense strand and included 60 CpGs). Amplicons, post T:A cloning (Promega, UK), were used for sequence analysis. In studies were cell lines were subject to drug challenges (Fig. 2)



FIG. 1. A, In silico analysis of the rodent D2R CpG island locus region. The CpG island extending from ⫺374bp to ⫹650bp, relative to the transcription start site (black bent arrow [⫹1]) was determined using CpG island searcher (http://www.cpgislands.com/) and the figure drawn using a web based program (www.ebi.ac.uk). The observed GC percentage is plotted across the region and is shown relative to the transcription factor start sites. Individual CpG dinucleotides are shown as vertical tick marks. The black bar below the tick marks (⫺538 to ⫹7bp) marks the amplicon that includes the 20 CpG reported for methylation status. B, Beads-on-a-string representation across the 20 CpG dinucleotides shown in A above. Individual clones from each of the cell lines and NRP are shown. The filled and unfilled circles (beads) represent methylated and unmethylated CpGs, respectively.

analysis was confined to the 20 CpGs within the amplified antisense amplicon. Sequence analysis was performed commercially (Bioscience Geneservice, Cambridge, UK) using Illumina Genome Analyser 2 and Applied Biosystems 3730 DNA analyzer platforms. The sequencing data were analyzed using the BIQAnalyzer web based program (Max Planck Institute of Informatics, Germany).

Chromatin immunoprecipitation (ChIP) analysis for histone modifications ChIP analysis was performed employing ChIP-IT Express Enzymatic Kit (Active Motif, Rixensart, Belgium) following manufacturer’s instructions. Briefly, DNA-protein cross linking of ⱖ107 cells was carried out in 1% formaldehyde (Sigma-Aldrich) followed by treatment with Glycine-Stop-Fix solution (Active Motif). The cells were washed in ice-cold PBS followed by lysis in ice-cold lysis buffer supplemented with protease inhibitor cocktail and phenylmethanesulfonyl fluoride. The lysate was homogenized in a (ice-cold) dounce homogenizer to aid the release of the nuclei. Pelleted nuclei were resuspended in digestion buffer supplemented with inhibitor cocktail and phenylmethanesulfonyl fluoride and the enzymatic shearing cocktail (Active Motif). The resultant supernatant, after centrifugation of the digestion mixture, comprised the sheared chromatin DNA for immunoprecipitation using antibodies to specific histone modifications. A fraction (10%) of the sonicated chromatin was set aside as “input” DNA before antibody affinity manipulations. The histone modifications used in this study were specific for active gene

367

expression, H3K9Ac (22) and for DNAmethylation independent gene silencing, H3K27me3 (23). The antibodies for enrichment of H3K9Ac and H3K27me3 were obtained from Abcam (Cambridge, UK). The sheared chromatin was antibody-enriched by overnight incubation on a rotary shaker at 4 C and precipitated using protein G–magnetic bead separation (Active Motif). The chromatin-antibody complexes were eluted followed by reverse-crosslinking and Proteinase-K treatment and DNA was purified using GenElute PCR Clean-Up Kit (SigmaAldrich). DNA was probed for enrichments of D2R using specific primers for quantitative PCR - forward 5⬘- ACAGTGCAGAGATAGTTCTG-3⬘ and reverse 5⬘- GGACAGCTCCGCGGAATCA-3⬘. In these cases, ChIP primer design and the specific region amplified was first determined using criteria described by Roh et al. (24) Percent input was calculated by 100 ⫻ 2(CT-adjusted Input ⫺ CT-Enriched). Input DNA CT was adjusted from 10% to100% equivalent by subtracting 3.322 CTs or Log210.

Results

Methylation status of the D2R in pituitary cell lines In silico analysis (25) identified a bona fide CpG island within the D2R gene (Fig. 1). A portion of this CpG island was analyzed by sodium bisulfite sequencing. In the NRP gland, across 20 CpG dinucleotides, 53% of CpGs, across the total clonal population, are not methylated. In contrast to these findings, in the GH3 cell line, the majority of CpGs are methylated and only 5% did not show this change. In the MMQ cell line, although this portion of the CpG island also shows a high frequency of methylation (⬃77%) relative to normal pituitary, ⬃23% of CpGs are not methylated. In agreement with previous reports, quantitative RT-PCR of D2R transcript revealed robust expression in NRP gland (12) and in MMQ cells (Supplemental Fig. 1B, published on The Endocrine Society’s Journals Online web site at http:// endo.endojournals.org/); however, early passage GH3 cells failed to express this transcript (Fig. 2). Drug-induced demethylation and reexpression of the D2R in GH3 cells To determine the contribution of CpG island methylation toward loss of D2R expression, GH3 cells were treated with either zebularine (26) alone or in combination with the histone-modifying agent TSA (27–29). Zebularine alone at the highest dose used (3 ␮M) induced D2R

368

Al-Azzawi et al.



methylation (H3K27Me3) (23) genes, in drug-challenged GH3 cells. Across the D2R promoter region, enrichment for marks associated with a silent gene (H3K27Me3) is apparent in WT GH3 cells (Fig. 3A). However, and as controls, in normal pituitary and in the MMQ cell line that express the D2R this mark is not enriched. In contrast to these findings, enrichment for a mark associated with active genes (H3K9Ac) was not apparent in wild-type GH3 cells, whereas enrichment is seen in normal pituitary and MMQ cells (Fig 3B). In GH3 cells, subject to drug challenges, ChIP analysis shows H3K9Ac enrichment and a decrease in immunoprecipitated H3K27Me3. Thus, these results are consistent with the known role of TSA as an inhibitor of histone tail deacetylation and promotion of an active chromatin structure (27–29).

Drug-induced histone modifications in GH3 cells ChIP assays were next used to investigate histone tail modification associated with either active, H3 lysine-9 acetylation (H3K9Ac) (22), or with silent H3 lysine-27 tri-

ChiP Enrichment for H3K27Me3

*

% INPUT

expression as determined by conventional RT-PCR, as well as by quantitative RT-PCR, that, in this case, was assessed relative to NRP. Higher doses of zebularine were cytotoxic (data not shown), however, in combined drug challenges, TSA augmented zebularine induced transcript expression but was ineffective as a single agent (Fig. 2). The potential for a mechanistic link between drug induced reexpression of the D2R and methylation of its cognate CpG island was further explored after each of the drug challenges shown in Fig. 2, A and B. Challenge of GH3 cells with zebularine or TSA alone induced or was associated with modest but reproducible demethylation (⬃10 –15%) relative to vehicle-treated cells. However, demethylation of this portion of the CpG island was more pronounced (⬃25%) at the highest combined drug doses used in these studies (Fig. 2C).

A

*

CELL LINE: DRUG

MMQ

NRP

-

-

B

GH3-WT -

GH3 +(ZEB)

GH3 +(ZEB+TSA)

CHiP Enrichment for H3K9Ac 100 75 % INPUT

FIG. 2. Expression and methylation status of the D2R in GH3 cells to challenges with zebularine (ZEB) or TSA alone or in combined incubations. RT-PCR (A) and quantitative RT-PCR (B) analysis of D2R expression after the challenges and drug doses shown in the figures. For quantitative RT-PCR, cell line expression is shown as percentage relative to NRP. C, Sodium bisulfite sequencing across the CpG described in Fig. 1. Drug challenge(s) are shown above each of the three clones analyzed after each of the treatments. The beads-on-astring representation is described for Fig. 1. *, P ⬍ 0.01 vs. vehicle alone. Data were analyzed for significance by one-way ANOVA with Dunnett’s multiple comparison post test.

Apoptotic end points in D2 receptor– expressing GH3 cells In coincubation experiments we found that 3 ␮M zebularine in combination with 30 nM TSA induced robust expression of D2R in GH3 cells (Fig. 2). Subsequent culture of these cells, designated GH3-D2exp (see Material and Methods), in the absence of these agents, showed that D2R expression is maintained for approximately 40 h in six-well plates (data no shown). Similarly, in those cases where cells were subcultured in 96-well plates before ap-

50 25

0 CELL LINE: DRUG

#

MMQ

NRP

-

-

GH3-WT -

GH3 +(ZEB)

GH3 +(ZEB+TSA)

FIG. 3. ChIP Analysis of the D2R promoter region. ChIP analyses are shown relative to input DNA for H3K27Me3 (A) or H3K9Ac (B). Enrichment in the pituitary cell lines (GH3 and MMQ) are shown after drug challenges shown in Fig. 2 and relative to enrichment observed in NRP. *, P ⬍ 0.05 and #, P ⬍ 0.01 vs. GH3-WT treated with vehicle alone. Data were analyzed for significance by one-way ANOVA with Dunnett’s multiple comparison post test.



369

ings were also apparent to DA challenge except that the caspase3 activation did not reach statistical significance. Specificity of D2R-mediated apoptosis in GH3 cells The specificity of our observation that augmented apoptosis is mediated through reexpression of endogenous D2R in GH3 cells was assessed using a pharmacological and genetic approach. In these studies, GH3-D2exp cells were preincubated with the D2R antagonists, eticlopride (18) or haloperidol (19) before challenge with either DA or BC. The antagonists, eticlopride and also haloperidol (Fig. 5A) effectively attenuated the augmented apoptotic response apparent in GH3-D2exp cells to either DA or BC challenge and relative to GH3-D2veh cells. Further support for a D2R mediated and specific effect was also

% Apoptotic Cells

A

100 ##

# 75

** 50

**

* *

25 0 BC DA ETIC HALO

B 100

% Apoptotic Cells

FIG. 4. Apoptotic response of GH3 cells to DA and BC challenges (A– C). GH3 cells induced to express D2R (GH3Dexp) and labeled D2R in the figure, relative to cells not expressing D2R (WT), were challenged with DA or BC at the drug concentrations shown in the figure. A, Apoptotic cell counts (30 h post drug challenges) as determined by Hoechst 33342 staining are presented and expressed as a percentage of total cell counts. B, Fluorescent microscopic images of terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL)positive cells are shown at 48 h post exposure to either BC or DA. C, Caspase 3/7 activation 6 h after each of the drug challenges. The results are the means of three independent experiments with triplicate determination in each case. *, P ⬍ 0.001 GH3-WT treated with BC. Data were analyzed for significance with a paired t test with Welch’s correction.

siRNA

+

+ -

+ + -

+ +

Veh. BC DA

75

**

**

**

**

+ -

**

+ + -

25

+ +

**

50

0

optotic drug challenges (caspase-3 activity, Fig. 4C), D2R expression was apparent before either BC or DA challenges and at each of the time point used to determine apoptotic end points (see Supplemental Fig. 1A). As control, GH3 cells, which were not subject to the drug challenges designed to induce reexpression of endogenous D2R (GH3-D2veh), were included in these studies. In these cases and at each of the time points investigated we did not detect expression of D2R. After the Zebularine and TSA coincubations experiments, the cells and their vehicle-treated counterparts were allowed to recover for 24 h and then subject to challenge with drugs shown previously to induce apoptosis in wild-type GH3 cells (10 –12). D2Rpositive cells (GH3-D2exp) challenged with BC (Fig. 4, A–C) show significant augmented apoptotic responses relative to their counterparts, GH3-D2veh cells. Similar find-

+ -

# # * *

WT -

D2R -

D2R -

D2R si-NT

D2R si-D2R

FIG. 5. Specificity of Apoptotic response of GH3 cells to DA and BC challenges. A, Wild-type GH3 (unfilled bars) and GH3 cells expressing D2R (filled bars) were challenged with DA (25 ␮M) or BC (20 ␮M) as described in Fig. 4 and preincubated with the D2R antagonists, eticlopride (ETIC) or haloperidol (HALO) agonists, and/or antagonist used are shown in the figure. #, P ⬍ 0.001 vs. GH3-WT treated with BC. ##, P ⬍ 0.01 vs. GH3-WT cells treated with DA. *, P ⬍ 0.005 vs. GH3–D2R cells treated with BC in the absence of either antagonist; **, P ⬍ 0.005 vs. GH3–D2R cells treated with DA in the absence of either antagonist. B, Cells treated with the specific agonist challenges shown in panel A and transfected with a D2R specific siRNA (siD2R) or siNT. As further control, in D2R cells in the absence of either siRNA (⫺) cells were cultured either in the absence or presence of the transfection reagent (Lipofectamine) respectively. *, P ⬍ 0.01; **, P ⬍ 0.001 vs. GH3-WT vehicle alone in the absence of a specific siRNA to the D2R. #, P ⬍ 0.001 vs. GH3–D2R expressing cells, transfected with siNT and treated with either BC or DA. In both panels the proportion (as a percentage) of apoptotic cells are shown on the y axis. Data were analyzed for significance by 2-way ANOVA with Bonferroni post test.

370

Al-Azzawi et al.


apparent in the genetic knockdown approach. In these experiments, D2R expression in GH3-D2exp cells was subject to siRNA mediated knock-down. Relative to cells transfected with a siNT the augmented apoptotic response to either BC or DA challenge in D2R-knockdown cells is attenuated (Fig. 5B). Specificity of D2R-mediated apoptosis in MMQ cells The MMQ cells line expresses endogenous D2R and provided a means of more rigorously assessing potential confounding secondary effects of drugs used for epigenetic “unmasking.” We therefore assessed the ability of DA and/or BC to induce apoptosis in vehicle-treated MMQ cells and those subject to the pretreatment regimen used to generate GH3-D2exp cells. These experiments (Fig. 6A) do not show any discernible differences in apoptotic responses between vehicle treated or pretreated cells to either DA or BC challenges. In these cells, and in agreement with our observations in D2R-expressing GH3 cells (GH3-D2exp), the D2R antagonists attenuate, with ap-

A 60

WT Cells Zeb +TSA Treated

% Apoptotic Cells

50

# #

40 # #

30 20

* *

* *

* *

+ + -

+ +

+ + -

*

*

10

0 BC DA ETIC HALO

B

+ -

% Apoptotic Cells

60 50

+

+ -

+ -

+ +

si-NT si-D2R

40 30

*

*

20 10 0

Vehicle

BC (20uM)

DA (50uM)

FIG. 6. Specificity of apoptotic response of MMQ cells to DA and BC challenges. The experiments on MMQ cells (that express endogenous D2R) were performed essentially as described and reported in Fig. 5, A and B. A, #, P ⬍ 0.001, vs. MMQ cells treated with either BC (20 ␮M) or DA (25 ␮M) in the presence or absence of Zeb and TSA .*, P ⬍ 0.0001 relative to MMQ cells challenged cells with either BC or DA in the absence of either antagonist. B, *, P ⬍ 0.001, vs. MMQ cells treated with either BC (20 ␮M) or DA(50 ␮M) and siNT. In both panels the proportion (as a percentage) of apoptotic cells are shown on the y axis. Data were analyzed for significance by two-way ANOVA with Bonferroni post test.


proximate equal efficiency, the apoptotic response mediated by either BC or DA. Finally, and similar to our observations in GH3-D2exp cells, knock-down of endogenous D2R transcript in MMQ cells attenuated DA and BC mediated apoptosis whereas siNT was without effect (Fig. 6B). Collectively these experiments provide convincing evidence that DA- and BCmediated apoptosis is achieved principally through the D2R. Furthermore, pretreatment strategies with drugs designed to uncover/unmask silenced gene (Zebularine and TSA) do not lead to nonspecific sensitization of these cells to apoptosis inducing agents or augmentation of D2R expression in these cells (data not shown).

Discussion Recent reports, using the pituitary tumor cell line GH3, have explored the apoptotic pathways activated after challenge with either DA (11), or relative to, or in combination with BC, a D2R agonist (12). Despite the absence of D2R in these cells, and in contradistinction to earlier reports that DA-mediated apoptosis is dependent upon expression of a functional receptor (10) these studies describe engagement and activation of bona fide apoptotic programs. Although Jaubert and colleagues concluded that DA mediates its apoptotic effects through a DAT (11), our own studies of this cell line were not able to confirm this finding. Our conclusion, therefore, from that study, in the absence of a detectable D2R or DAT were that the mechanism by which DA induces apoptosis had not been definitively identified (12). However, in these studies (11, 12) we had also noted that high doses of DA or BC were required to induce an apoptotic response, and we proposed several possible explanation to account for these observations (see Ref. 12). To extend these previous studies and observations, and to specifically address the contribution of the D2R to agonist induced apoptosis, we explored potential mechanisms responsible for loss of D2R in these cells, strategies to induce reexpression of the endogenous receptor, and finally, the effects of receptor occupancy on apoptotic end points after dopaminergic challenges. With limited but important exceptions, mutations leading to activation or loss of oncogenes or tumor suppressor genes are an infrequent finding in pituitary derived primary tumors (14, 30). However, and in common with many other tumor types, epigenetic aberrations are a frequent finding. Indeed, multiple reports described methylation-mediated gene-silencing of tumor suppressor genes, those involved in cell cycle control and in apoptotic pathways (14, 30, 31). No less important, but subject to more limited investigation, are reports of histone modifica-


tions that lead to either inappropriate gene activation or silencing (31). Because epigenetic gene silencing is potentially reversible it provides an appealing approach to reestablish cellular functions that are silenced through this change (32). We now show, for the first time, hypermethylation of a portion of the D2R CpG island in GH3 cells relative to that apparent in either NRP gland and or to that seen in MMQ cells that express the D2R. To investigate a potential causal relationship between epigenetic change and gene silencing, GH3 cells were subject to pharmacological “unmasking” with drugs designed to inhibit DNA methylation and/or histone deacetylation patterns: in these studies, we used either zebularine (26) alone or in combination with the HDAC inhibitor TSA that, in some cases, have been shown to be synergistic toward reexpression of silenced genes when used in combination with a demethylating agent (27, 28) and in other instances is sufficient as a single agent to “unmask” genes silenced through histone modifications alone (29, 33, 34). Although incubation of GH3 cells with either agent alone was responsible for a marginal, but reproducible, decrease in methylation across the CpG island, reexpression of D2R transcript was only apparent to zebularine challenge at the highest dose used (3 ␮M). However, coincubation with both agents (zebularine and TSA) was responsible for a synergistic increase in D2R expression and a more pronounced decrease in CpG island methylation. Throughout this report, we have interpreted D2R expression of 5% or less as loss, however we recognize that this can also be regarded as low or barely detectable expression by some investigators. However, we do not believe that these low/barely detectable levels of transcript would impact upon our conclusion. Previous studies in multiple other tumor types and their cell lines have shown that zebularine, principally through DNA demethylation, reactivates gene expression, and in many of these studies the cognate CpG islands show only partial drug-induced demethylation (32, 35, 36). Histone modifications, either alone or in concert with changes in DNA methylation patterns, can lead to either gene activation or repression that is dependent not only on which residue are modified but on the type of modification present (13). For example, acetylation of histone 3 lysine 9 (H3K9Ac) correlates with transcriptional activation (22), whereas trimethylation of histone 3 lysine 27 (H3K27me3) is presentatpromotersthataretranscriptionallyrepressed(23).In GH3 cells that fail to express D2R, and relative to normal pituitary and MMQ cells that express this receptor, histone modifications, as determined by ChIP analysis, are consistent with repressed and active genes respectively. We found, in normal pituitary and MMQ cells, enrichment for H3K9Ac whereas,


371

enrichment for modification associated with repressed genes, H3K27me3, was present only in GH3 cells. However, combined challenges of GH3 cells with zebularine and TSA resulted in enrichment of H3K9Ac and decrease in H3K27me3. These findings are consistent with these agents inducing histone modifications, and in combination with zebularine-mediated CpG demethylation, lead to reexpression of D2R in these cells. To our knowledge only a single investigation has induced reexpression of endogenous D2R in GH3 cells. These studies, from the Spadagroup(37),used9-cisretinoicacidtoinducereexpression of this receptor in primary human pituitary tumors and in GH3 cells. A recent report, in this case in promyelocytic leukemia cells, has shown that pharmacological doses of all trans-retinoic acid induce changes in H3 acetylation and gene expression (38). Epigenetic change after retinoic acid challenge of GH3 cells has not thus far been explored but remains a testable possibility. However, it is also possible that occupancy of retinoic acid response element in the D2R promoter might override the epigenetic aberrations apparent in these cells. The reexpression of endogenous D2R in GH3 cells that is consequent to zebularine and TSA-mediated inhibition of DNA methylation and histone deacetylation provided an opportunity to examine apoptotic end points in these cells and relative to those seen in cells that did not express endogenous receptor. These studies, in cells that differed only in D2R status, showed an augment and presumed D2R mediated response to drug challenges with either DA or BC. In these cases these agents had been shown previously to induce apoptosis in these cells (11, 12). A caveat associated with the findings thus far described is that even in the same cell line nucleoside analogs do not necessarily have the same effect on the transcriptome (reviewed in Ref. 39). For example, Flotho and colleagues (40) showed that after individual treatment of the same cell line with the nucleoside analogs, decitabine, azacitidine, or zebularine, the identified (“unmasked”) genes showed remarkably little overlap, indeed a considerable number of genes were down-regulated. Similar conclusions were reached when comparing the effects of decitabine, a DNMT knockout model and the HDAC inhibitor, TSA (41). In this case, the effects on gene expression did not seem to be dependent on dosage and duration, which would be expected if the drugs exert their effects solely by incorporation into DNA during the replicative, S-phase, of the cell cycle. Thus, these reports and others prompted us to examine the specificity of our observation in greater detail. The specificity of a D2R-mediated augmented apoptotic response to DA and/or the DA agonist, BC, was addressed through pharmacological (receptor antagonists) and genetic (RNA interference) approaches. In these studies, inhibition or interference of dopaminergic signaling

372

Al-Azzawi et al.


pathway(s) was initially investigated in GH3 cells, where endogenous D2R expression had been “unmasked,” and then in the MMQ cell line that expresses endogenous functional D2R (42). In D2R-expressing GH3 cells the augmented apoptotic response to either DA or BC challenges was attenuated through preincubation with the D2R antagonists, haloperidol, or eticlopride. These findings were reinforced using siRNA-mediated knock-down of the D2R in these cells, whereas a siNT was without effect. In MMQ cells, receptor antagonist experiments also resulted in attenuated apoptotic responses to DA and BC. In these cells, prechallenge with zebularine in combination with TSA did not augment or attenuate apoptosis, however this end point was effectively attenuated by receptor antagonists. Finally and similar to the findings apparent in D2R expressing GH3, siRNA-mediated knock-down of endogenous D2R in MMQ cells attenuated DA- and BC-induced apoptosis. Although these studies support the conclusion that reexpression of the D2R in GH3 cells is indeed specific for the observed augmented apoptotic response, there may be other yet to be identified aberrations in this apoptotic pathway. In this context the limitations of extrapolating from in vitro findings should also be born in mind. This study, in a model system, may however have important consequence for the clinical management of primary pituitary adenomas, that irrespective of subtype express, to varying degrees, DA receptor subtypes (2–5, 7, 37, 43, 44). In this context, several reports in different tumor types now describe strategies, where epigenetic therapy is combined with chemotherapy or radiotherapy with improved treatment outcomes (reviewed in Ref. 39). In addition, this approach, where DNMT and HDAC inhibitors are used in combination with, for example, antiestrogens to restore estrogen receptor ␣ responsiveness to previous ER- and antiestrogen-resistant tumor cells (33, 34, 45). Thus these types of approaches might prove useful in the provision of more efficacious treatment strategies of primary pituitary tumors. Indeed, promising results in primary tumors and particularly in the GH3 cell line were described in the report by Bondioni and coworkers (37) where retinoic acid induced expression of D2R and the generation of proapoptotic signals in this cell line to DA challenge. In conclusion, our study uncovers the epigenetic aberrations that are responsible for D2R receptor silencing in the pituitary cell line GH3. Restoration of functional receptor through an epigenetic therapy strategy reestablished a functional dopaminergic pathway that is sensitive to DA and DA agonist challenges. These studies will prove useful and encourage investigation of D2R receptor status in primary pituitary adenomas and may uncover com-


bined treatment approaches for the medical management of different pituitary tumor subtypes.

Acknowledgments We thank Farjana Rowther for the technical contribution. Address all correspondence and requests for reprints to: W.E. Farrell, Human Disease and Genomics Group, Institute of Science and Technology in Medicine, School of Medicine, Keele University, Stoke on Trent, Staffordshire ST4 7QB United Kingdom. E-mail: [email protected]. Disclosure Summary: The authors have nothing to declare.

References 1. Ferone D, Pivonello R, Resmini E, Boschetti M, Rebora A, Albertelli M, Albanese V, Colao A, Culler MD, Minuto F 2007 Preclinical and clinical experiences with the role of dopamine receptors in the treatment of pituitary adenomas. Eur J Endocrinol 56(Suppl 1):S37–S43 2. Hofland LJ, Feelders RA, de Herder WW, Lamberts SW. Pituitary tumours 2010 The sst/D(2) receptors as molecular targets.Mol Cell Endocrinol 326:89 –98 3. Colao A, di Sarno A, Pivonello R, di Somma C, Lombardi G 2002 Dopamine receptor agonists for treating prolactinomas. Expert Opin Investig Drugs 11:787– 800 4. Colao A, Pivonello R, Di Somma C, Savastano S, Grasso LF, Lombardi G 2009. Medical therapy of pituitary adenomas: effects on tumor shrinkage Rev Endocr Metab Disord 10:111–123 5. Molitch ME 2005. Pharmacologic resistance in prolactinoma patients. Pituitary 8:43–52 6. Missale C, Nash SR, Robinson SW, Jaber M, Caron MG 1998. Dopamine receptors: from structure to function. Physiol Rev 78: 189 –225 7. Spada A, Nicosia S, Cortelazzi L, Pezzo G, Bassetti M, Sartorio A, Giannattasio G 1983 In vitro studies on prolactin release and adenylate cyclase activity in human prolactin-secreting pituitary adenomas. Different sensitivity of macro- and microadenomas to dopamine and vasoactive intestinal polypeptide. J Clin Endocrinol Metab 56:1–10 8. Ben-Jonathan N 1985. Dopamine: a prolactin-inhibiting hormone. Endocr Rev 6:564 –589 9. Picetti R, Saiardi A, Abdel Samad T, Bozzi Y, Baik JH, Borrelli E 1997. Dopamine D2 receptors in signal transduction and behavior. Crit Rev Neurobiol 11:121–142 10. An JJ, Cho SR, Jeong DW, Park KW, Ahn YS, Baik JH 2003 Antiproliferative effects and cell death mediated by two isoforms of dopamine D2 receptors in pituitary tumor cells. Mol Cell Endocrinol 206:49 – 62 11. Jaubert A, Ichas F, Bresson-Bepoldin L 2006 Signaling pathway involved in the pro-apoptotic effect of dopamine in the GH3 pituitary cell line. Neuroendocrinology 83:77– 88 12. Rowther FB, Richardson A, Clayton RN, Farrell WE 2010 Bromocriptine and dopamine mediate independent and synergistic apoptotic pathways in pituitary cells. Neuroendocrinology 91:256 – 267 13. Sharma S, Kelly TK, Jones PA 2010. Epigenetics in cancer. Carcinogenesis 31:27–36 14. Farrell WE 2006 Pituitary tumours: findings from whole genome analyses. Endocr Relat Cancer 13:707–716 15. Egger G, Liang G, Aparicio A, Jones PA 2004 Epigenetics in human disease and prospects for epigenetic therapy. Nature 429:457– 463


16. Balch C, Montgomery JS, Paik HI, Kim S, Kim S, Huang TH, Nephew KP 2005 New anti-cancer strategies: epigenetic therapies and biomarkers. Front Biosci 10:1897–1931 17. Pathiraja TN, Stearns V, Oesterreich S 2010 Epigenetic regulation in estrogen receptor positive breast cancer–role in treatment response. J Mammary Gland Biol Neoplasia 15:35– 47 18. Martelle JL, Nader MA 2008 A review of the discovery, pharmacological characterization, and behavioral effects of the dopamine D2-like receptor antagonist eticlopride. CNS Neurosci Ther 14: 248 –262 19. Seeman P, Ulpian C 1988 Dopamine D1 and D2 receptor selectivities of agonists and antagonists. Adv Exp Med Biol 235:55– 63 20. Bahar A, Simpson DJ, Cutty SJ, Bicknell JE, Hoban PR, Holley S, Mourtada-Maarabouni M, Williams GT, Clayton RN, Farrell WE 2004 Isolation and characterization of a novel pituitary tumor apoptosis gene. Mol Endocrinol 18:1827–1839 21. Fink L, Stahl U, Ermert L, Kummer W, Seeger W, Bohle RM 1999 Rat porphobilinogen deaminase gene: a pseudogene-free internal standard for laser-assisted cell picking. Biotechniques 26:510 –516 22. Jenuwein T, Allis CD. 2001 Translating the histone code. Science 293:1074 –1080 23. Cao R, Wang L, Wang H, Xia L, Erdjument-Bromage H, Tempst P, Jones RS, Zhang Y 2002 Role of histone H3 lysine 27 methylation in Polycomb-group silencing. Science 298:1039 –1043 24. Roh TY, Cuddapah S, Cui K, Zhao K 2006 The genomic landscape of histone modifications in human T cells. Proc Natl Acad Sci USA 103:15782–15787 25. Takai D, Jones PA 2003 The CpG island searcher: a new WWW resource. In Silico Biol 3:235–240 26. Cheng JC, Matsen CB, Gonzales FA, Ye W, Greer S, Marquez VE, Jones PA, Selker EU 2003 Inhibition of DNA methylation and reactivation of silenced genes by zebularine. J Natl Cancer Inst 95: 399 – 409 27. Garcia-Manero G, Issa JP 2005 Histone deacetylase inhibitors: a review of their clinical status as antineoplastic agents. Cancer Invest 23:635– 642 28. Cameron EE, Bachman KE, Myo¨ha¨nen S, Herman JG, Baylin SB 1999 Synergy of demethylation and histone deacetylase inhibition in the re-expression of genes silenced in cancer. Nat Genet 21:103–107 29. Carew JS, Giles FJ, Nawrocki ST 2008 Histone deacetylase inhibitors: mechanisms of cell death and promise in combination cancer therapy. Cancer Lett 269:7–17 30. Dudley KJ, Revill K, Clayton RN, Farrell WE 2009 Pituitary tumours: all silent on the epigenetics front. J Mol Endocrinol 42:461– 468 31. Ezzat S 2008 Epigenetic control in pituitary tumors. Endocr J 55: 951–957 32. Cheng JC, Yoo CB, Weisenberger DJ, Chuang J, Wozniak C, Liang G, Marquez VE, Greer S, Orntoft TF, Thykjaer T, Jones PA 2004 Preferential response of cancer cells to zebularine. Cancer Cell 6:151–158 33. Yang X, Ferguson AT, Nass SJ, Phillips DL, Butash KA, Wang SM, Herman JG, Davidson NE 2000 Transcriptional activation of estrogen receptor alpha in human breast cancer cells by histone deacetylase inhibition. Cancer Res 2000 60:6890 – 6894


373

34. Yang X, Phillips DL, Ferguson AT, Nelson WG, Herman JG, Davidson NE 2001 Synergistic activation of functional estrogen receptor (ER)-alpha by DNA methyltransferase and histone deacetylase inhibition in human ER-alpha-negative breast cancer cells. Cancer Res 61:7025–7029 35. Cheng JC, Weisenberger DJ, Gonzales FA, Liang G, Xu GL, Hu YG, Marquez VE, Jones PA 2004 Continuous zebularine treatment effectively sustains demethylation in human bladder cancer cells. Mol Cell Biol 24:1270 –1278 36. Scott SA, Lakshimikuttysamma A, Sheridan DP, Sanche SE, Geyer CR, DeCoteau JF 2007. Zebularine inhibits human acute myeloid leukemia cell growth in vitro in association with p15INK4B demethylation and reexpression. Exp Hematol 35:263–273 37. Bondioni S, Angioni AR, Corbetta S, Locatelli M, Ferrero S, Ferrante E, Mantovani G, Olgiati L, Beck-Peccoz P, Spada A, Lania AG 2008 Effect of 9-cis retinoic acid on dopamine D2 receptor expression in pituitary adenoma cells. Exp Biol Med (Maywood) 233: 439 – 446 38. Martens JH, Brinkman AB, Simmer F, Francoijs KJ, Nebbioso A, Ferrara F, Altucci L, Stunnenberg HG 2010 PML-RARalpha/RXR alters the epigenetic landscape in acute promyelocytic leukemia. Cancer Cell 17:173–185 39. Kristensen LS, Nielsen HM, Hansen LL 2009 Epigenetics and cancer treatment. Eur J Pharmacol 625:131–142 40. Flotho C, Claus R, Batz C, Schneider M, Sandrock I, Ihde S, Plass C, Niemeyer CM, Lu¨bbert M 2009 The DNA methyltransferase inhibitors azacitidine, decitabine and zebularine exert differential effects on cancer gene expression in acute myeloid leukemia cells. Leukemia 23:1019 –1028 41. Gius D, Cui H, Bradbury CM, Cook J, Smart DK, Zhao S, Young L, Brandenburg SA, Hu Y, Bisht KS, Ho AS, Mattson D, Sun L, Munson PJ, Chuang EY, Mitchell JB, Feinberg AP 2004 Distinct effects on gene expression of chemical and genetic manipulation of the cancer epigenome revealed by a multimodality approach. Cancer Cell 6:361–371 42. Gruszka A, Ren SG, Dong J, Culler MD, Melmed S 2007 Regulation of growth hormone and prolactin gene expression and secretion by chimeric somatostatin-dopamine molecules. Endocrinology 148:6107– 6114 43. Tateno T, Kato M, Tani Y, Oyama K, Yamada S, Hirata Y 2009 Differential expression of somatostatin and dopamine receptor subtype genes in adrenocorticotropin (ACTH)-secreting pituitary tumors and silent corticotroph adenomas. Endocr J 56:579 –584 44. Neto LV, Machado Ede O, Luque RM, Taboada GF, Marcondes JB, Chimelli LM, Quintella LP, Niemeyer Jr P, de Carvalho DP, Kineman RD, Gadelha MR 2009 Expression analysis of dopamine receptor subtypes in normal human pituitaries, nonfunctioning pituitary adenomas and somatotropinomas, and the association between dopamine and somatostatin receptors with clinical response to octreotide-LAR in acromegaly. J Clin Endocrinol Metab 94:1931–1937 45. Ferguson AT, Lapidus RG, Baylin SB, Davidson NE 1995 Demethylation of the estrogen receptor gene in estrogen receptor-negative breast cancer cells can reactivate estrogen receptor gene expression. Cancer Res 55:2279 –2283