The focus of this study is on the expression and regulation of the estrogen-regulated breast cancer and salivary gland expression (BASE) gene that may function as a breast cancer marker. In MCF7 cells, BASE is repressed by estrogen in an estrogen receptor α (ERα)-dependent manner. Promoter analysis of the BASE gene led to the identification of a 2-kb upstream enhancer that harbors binding sites for ERα and FoxA1. The recruitment of both ERα and FoxA1 to this region was shown by chromatin immunoprecipitation analysis. Furthermore, mutation studies and knockdown experiments show a clear separation between gene expression mediated by FoxA1 and ERα-dependent gene regulation. Additionally, we provide information on BASE expression in human breast tumor samples. [Cancer Res 2008;68(1):106–14]
- estrogen receptor
- breast cancer
- gene regulation
In women, breast cancer is the most common cancer and accounts for most cancer deaths. The molecular mechanisms underlying this pathology are diverse and contribute to the complexity of the disease. Early diagnosis and detailed molecular characterization of tumors significantly improve patient prognosis. A limited number of breast cancer markers are already used for diagnosis, characterization, and determination of the most promising course of therapy ( 1).
A major criterion in breast cancer diagnosis is the presence of the estrogen receptor α (ERα), which is associated with better prognosis and often sensitivity to antiestrogen therapy. Current therapies focus particularly on blockage of estradiol availability and action. However, other markers are needed to further classify breast cancer patients.
A new putative breast cancer marker was recently identified by Egland et al. ( 2) in a screen for membrane and secreted proteins that could be used as targets for immunotherapies or diagnostic markers in prostate or breast cancers. The gene identified was termed breast cancer and salivary gland expression (BASE) due to its expression pattern. Several breast cancer cell lines and five of eight breast tumor samples were positive for BASE mRNA ( 2). With its restricted expression pattern and the possible secretion of the 19.5-kDa protein, BASE has the potential to be an easily accessible and potent diagnostic marker for breast cancer. However, its expression profile needs further extensive characterization, especially in terms of differences in expression levels between normal breast tissue and primary and metastatic tumors.
Neither the BASE gene nor the protein has been experimentally evaluated previously. The protein has no predicted domains, which could give insight into its function. It does share sequence similarity with latherin, a surfactant protein found in horse sweat ( 3), and has been assigned to the PLUNC gene family whose members are expressed in the upper airways where they may function in host defense ( 4).
Although ∼50% of E2-responsive genes are down-regulated in the presence of E2, only few studies focused on repression of transcription by ERα, whereas gene activation has been studied extensively.
The aim of this study was to investigate the mechanism underlying BASE gene expression and to further analyze the role of ERα in hormone-induced gene repression.
Materials and Methods
Cell lines, cell culture, and analysis. MCF7, T47D, SKBR3, MDA-MB231, HepG2, and HeLa were maintained in DMEM supplemented with 10% FCS, 2 mmol/L l-glutamine (Invitrogen), and penicillin/streptomycin (described as full medium; Invitrogen) at 37°C under 5% CO2. ZR75 cells were cultured in RPMI 1640 containing the same supplements. Cell lines stably expressing ERα [ERα46, ERα66, and ERα66mut (E203G/G204S, DNA binding mutant); ref. 5] derived from ERα-negative cell line MDA-MB231 were previously described ( 6) and maintained in DMEM under hygromycin selection (0.8 mg/mL). When hormone deprivation was required, cells were cultured in phenol red–free DMEM (Life Technologies) containing 2.5% charcoal/dextran-stripped serum and antibiotics (described as stripped medium) for 72 h before treatment.
Transient transfections and luciferase assays. Transfections were performed in 24-well plates at 70% confluency with Fugene6 transfection reagent (Roche) for MDA-MB231 and with ExGene 500 (Fermentas) for all other cell lines used. Cells were transfected using 500 ng firefly luciferase reporter constructs and 100 ng Renilla luciferase reporter (phRL-TK vector, Promega). Cotransfections were performed with 200 ng of the expression constructs of hER46, hER66, or hER66mut. Cells were treated immediately after transfection with either ethanol, 10 nmol/L E2 (Sigma-Aldrich), or 1 mmol/L ICI182,780 (Tocris). After 24 h, cells were harvested and cellular extracts were analyzed for luciferase activity using the Dual-Luciferase Reporter System (Promega). Firefly luciferase reporter activities were normalized to Renilla luciferase activities and presented as relative light units (RLU). At least three independent experiments with triplicates were performed for each experiment.
Identification of putative transcription factor binding sites. Identification of transcription factor binding sites was performed with Dragon ERE Finder (version 2) and MatInspector v.2.2 ( 7). Both databases are available online. 3
Luciferase reporter constructs. The −2,418 to +23 bp region of the BASE promoter was amplified using FastStart Polymerase (Roche) and the following primers: gcgggtacctacatgactccaggctgtgg (2.4 kb forward) and gcgctcgagtgtgctgtcaagacactctgg (start reverse). The fragment was cloned into the pGL3-basic vector (Promega). The sequence was verified by DNA sequencing. All other constructs were derived from this construct by subcloning into the pGL3-promoter vector (P0) or the pGL3-enhancer vector (E0; Promega) or by digest and site-directed mutagenesis. The TFF1-luciferase reporter construct contains the −556 to +26 bp region of the TFF1 gene as Nhe1/BglII fragment. Supplementary Table S2 lists constructs and primers used for mutation studies.
Reverse transcription-PCR and quantitative real-time PCR. RNA was extracted from cells using the Trizol reagent (Invitrogen). cDNA reverse transcription using poly-dT oligos was performed on 3 μg of total RNA using Expand Reverse Transcriptase (Roche) according to the manufacturer's instructions.
Quantitative PCRs were performed on an ABI Prism 7500 (Applied Biosystems) using SYBR Green PCR Master Mix (Applied Biosystems) and gene-specific primers (Supplementary Table S3). Either β-actin or peptidyl-prolyl isomerase A (PPIA) was used as an internal reference gene.
Small interfering RNA experiments. MCF7 cells were transfected in six-well plates with small interfering RNA (siRNA) against ERα (Stealth siRNA, Invitrogen) or against FoxA1 (SMARTpool siRNA, Dharmacon) at a final concentration of 100 nmol/L using Lipofectamine 2000 (Invitrogen). The next day, cells were treated and, 24 h later, harvested for RNA isolation or analysis by Western blot. When hormone deprivation was required, cells were set E2-free 24 h before transfection.
Western blotting. Cells were washed twice with PBS and directly lysed in Laemmli buffer. Proteins were resolved on a 10% SDS-PAGE gel and transferred onto polyvinylidene difluoride membranes (Millipore Corp.) using wet transfer. Membranes were blocked for 1 h in 3% milk + 0.05% Tween 20 (Sigma-Aldrich), incubated 1 h with indicated primary antibodies, and then incubated for 1 h with horseradish peroxidase–conjugated secondary antibodies and developed using enhanced chemiluminescence Western blotting detection reagents (Perkin-Elmer) and Kodak BioMax MR films.
Electrophoretic mobility shift assay. Electrophoretic mobility shift assays (EMSA) were performed as described previously ( 8) with the following modifications. In vitro–translated proteins were synthesized using the TNT T7 Quick Coupled Reticulocyte Lysate System (Promega) and expression constructs for hERα66 (pcDNA3.1-hERα66) and hFoxA1 (pcDNA3.1-FoxA1). Oligos are listed in Supplementary Table S4.
Chromatin immunoprecipitation assays. Chromatin immunoprecipitation (ChIP) assays were performed as described previously ( 9) with unsynchronized MCF7 cells or synchronized by serum starvation followed by E2 treatment. Antibodies used are indicated in figure legends and primers are listed in Supplementary Table S5.
Expression analysis of BASE in human breast cancer tissues. Primary breast tumors were obtained from the Department of Surgery Biobank (Department of Surgery, University College Hospital, Galway, Ireland). Total RNA was isolated using Trizol reagent and cDNA was synthesized with SuperScript III reverse transcriptase (Invitrogen). BASE expression was analyzed in quantitative reverse transcription-PCR (RT-PCR) using Taqman Universal Master Mix (Applied Biosystems). BASE expression levels were normalized against the endogenous control PPIA gene expression levels by subtracting the average PPIA cycle threshold (CT) from the average BASE CT for each cDNA sample, yielding a level of mRNA expression for the target molecule relative to the endogenous RNA reference gene (ΔCT). The ΔCT for the calibrator sample, the breast cancer cell line BT474, was subtracted from the ΔCT values for all cDNA samples to yield mRNA expression relative to the calibrator sample (ΔΔCT). The relative quantity of gene expression for each sample was calculated using the formula 2−ΔΔCT ( 10).
BASE expression in normal and breast tumor tissue. To date, BASE expression has been examined using a commercially available tissue array and by RT-PCR for normal tissues and breast tumor samples ( 2). These showed expression of BASE in salivary gland as the only normal tissue and in ∼50% of the breast cancer samples. To further test BASE expression in breast tumor tissue, 50 tumor samples were tested by quantitative PCR and 26 were positive for BASE expression. Within this group, BASE expression was significantly higher for ERα-positive tumors (P < 0.05). Furthermore, four benign and four control breast samples were included in the study. The expression levels of BASE in control (two of four) and benign (one of four) tissue were 32-fold lower than in the tumor samples; thus, BASE expression differs significantly between malignant and control samples. The control samples in this study were control biopsies from patients diagnosed with either benign or malignant tumors. Recent studies using magnetic resonance imaging revealed a significant number of patients that had additional cancer foci in the ipsilateral and contralateral breast ( 11). Consequently, as these samples do not represent an adequate negative control, fibroblast cell lines derived from breast reduction tissue [kindly provided by Prof. B.F.C. Clark (University of Aarhus, Aarhus, Denmark)] were analyzed and found to be negative for BASE expression.
BASE is an estrogen-responsive gene. Comparative transcriptome analysis in the breast cancer cell line MCF7 showed a strong reduction of BASE transcripts in response to estradiol treatment. After 24 h of treatment with 10 nmol/L E2, BASE mRNA levels were decreased ∼6-fold compared with untreated control samples, whereas known E2-induced genes, such as CTSD and TFF1, were up-regulated 2.3- and 2.9-fold, respectively (data not shown). The microarray results were validated by RT-PCR.
To assess whether the effect of hormone treatment on BASE expression is mediated via ERα, siRNA was used to reduce receptor levels. Transcript levels of ERα and BASE were analyzed 24 h after transfection with ERα-siRNA in the presence of E2. ERα mRNA levels dropped to 30%, whereas BASE mRNA was increased to ∼2.6-fold ( Fig. 1A ). Under estrogen-deprived conditions, reduction of ERα levels has no effect on BASE expression, but when E2 was added, repression was only observed when ERα was present ( Fig. 1B). These results indicate a role for ERα in repression of BASE expression.
Prolonged exposure to hormone could allow regulation through secondary, ERα-independent effects. To address this point, the kinetics of BASE regulation in response to E2 and pure antiestrogen ICI182,780 were analyzed ( Fig. 1C). Time course experiments were performed in MCF7 cells in which BASE transcript levels were determined at different time points up to 24 h after treatment with E2. Repression of the BASE gene was found to be very rapid, showing an ∼30% decrease in transcript levels after 1 h, which continued to decrease until the 24-h time point. This indicates that BASE down-regulation after hormone treatment is a direct rather than a secondary effect. However, a mRNA degradation/destabilization-dependent mechanism cannot be excluded.
In E2-deprived medium, where ERα is inactive, no regulation of BASE by ICI182,780 was observed (data not shown). Therefore, the kinetic studies for ICI182,780 were performed in normal medium in which ERα is activated by E2. As expected for ERα-dependent repression, BASE mRNA levels increased after treatment with ICI182,780 ( Fig. 1C). However, the induction was delayed for about 6 to 12 h compared with the repression in response to E2. The delay suggests that degradation of ERα or of a repressor is required for BASE derepression. Opposite trends were observed for the well-characterized E2-inducible TFF1 gene, which was included as positive control (data not shown). In conclusion, these results indicate that activation and degradation of ERα are the main cause for BASE regulation in response to E2 and ICI182,780.
Analysis of the BASE promoter. Bioinformatic analysis of the first 2.4 kb of the BASE promoter using MatInspector ( 7) predicted four ERα binding sites (−2,377, −1,926, −560, and +7). To further investigate expression and regulation of BASE, a reporter construct was generated containing 2.4 kb of the BASE promoter upstream of the luciferase gene. This reporter construct resembles the regulation of the endogenous gene in transient transfection assays in MCF7 cells with a strong repression by estradiol and slight induction by ICI182,780 ( Fig. 2A ). As expected, the opposite trend was observed for a TFF1 reporter construct.
To identify crucial promoter regions for BASE expression and regulation, a series of deletion mutants omitting different segments of the promoter were created. In transient transfection assays in MCF7 cells ( Fig. 2B), deletion of the most upstream region I (−2,419/−2,353, ΔI) had no effect on response to E2 or ICI182,780 compared with the full-length reporter construct. In contrast, deletion of the region between −2,352 and −1,689 (segment II, ΔII) reduced gene expression to almost undetectable levels. Absence of segment IV (−916/−179, ΔIV) also reduced expression strongly, whereas deletion of segment III (−1,688/−917, ΔIII) resulted in slightly increased expression, indicating that this region could harbor repressive features. All constructs that were expressed also showed repression by E2 and induction by ICI182,780 (data not shown).
In conclusion, promoter segment IV (−916/−179) contributes to BASE gene expression but is not required for repression by E2. Segment II (600 bp) is required for full expression of the BASE gene and can be considered to be an enhancer. Whether it also plays a role in the repression by E2 could not be assessed with this approach due to the very low expression levels.
BASE expression and regulation can be separated. To better define the regions in the enhancer that are essential for regulation, reporter constructs containing deletions within region II (each omitting only ∼150 bp) were generated and tested for their activation potential and hormone sensitivity ( Fig. 2C). Deletion of the region between −1,990 and −1,835 (ΔC) reduced expression levels to the same extent as deletion of the whole enhancer, indicating important binding motifs in this region. The absence of segments A, B, and D had less effect on expression.
Segment C was further separated into three subparts, each 50 bp long. Corresponding reporter constructs were generated and tested in MCF7 cells ( Fig. 2D). The strongest reduction in transcriptional activity was observed when region C2 (ΔC2) was omitted, whereas again regulation was not affected. Varying reduction levels were observed for ΔC3.
FoxA1 is essential for BASE expression. Sequence analysis of the C2 region using MatInspector v.2.2 ( 7) indicated a forkhead factor binding site cluster in the enhancer region, including two binding sites for FoxA1 ( Fig. 3A ). Disruption of these predicted sites strongly reduced the expression of the luciferase reporter construct in transient transfection assays in MCF7 cells ( Fig. 3B).
To confirm the ability of FoxA1 to bind to the BASE enhancer regions, EMSAs were performed using radiolabeled oligos (50 bp) containing either segments C1 or C2. In vitro–translated FoxA1 was able to bind to the C2 oligo (containing the BASE FoxA1 binding sites) and could be supershifted with the corresponding antibody ( Fig. 3C). ChIP experiments indicate the association of FoxA1 with the BASE promoter in vivo ( Fig. 5A). These findings confirm the bioinformatic prediction of a FoxA1 binding sites within the C2 region.
To analyze the contribution of FoxA1 to BASE basal and hormone-dependent expression, siRNA experiments were conducted. Knockdown using siRNA against FoxA1 dramatically reduced BASE expression ( Fig. 3D). FoxA1 siRNA reduced transcript levels of FoxA1 to ∼40%. In these samples, BASE mRNA levels decreased dramatically (∼90%). These results strongly support a role of FoxA1 in BASE gene expression. Hoxwever, repression of BASE expression by estrogen is not diminished in the absence of FoxA1, as a further reduction of BASE transcript levels is observed in the presence of E2 ( Fig. 3D). This result is consistent with the luciferase reporter assay ( Fig. 3B).
The enhancer region is required for hormone response. Deletion of region II reduces expression of the reporter construct to such a low level that a potential hormonal response cannot be detected ( Fig. 2B). Further dissection of this region also did not indicate the mode of repression because regulation was not affected. To evaluate the role of region II in BASE regulation, promoter segments of increasing length were subcloned into the pGL3-enhancer vector (E0), which harbors a SV40 enhancer. These constructs possess increased basal expression and therefore allow the regulation of constructs omitting the BASE enhancer region to be assessed.
MCF7 cells were transiently transfected with sequential 5′-deletion constructs and constructs either omitting or exclusively containing the enhancer region ( Fig. 4A ). The region from −917 to +23 bp (segments IV to V) was identified as the core promoter (E2 and E7). It imparts basal activity and has been shown to contribute to full promoter activity in previous experiments ( Fig. 2B) but not to the response to E2 or ICI182,780 (data not shown). Regulation and further increased expression was observed only when segment II, the enhancer, was included (E4 and E9). In summary, the identified minimal hormone-responsive region consists of the 600-bp enhancer and the transcription start site (TSS; E9).
The enhancer is essential for expression and regulation, but the importance of the TSS in BASE regulation remained to be elucidated. To assess this question, the TSS of the BASE promoter (−178 to +23) was replaced with the SV40 promoter that does not confer response to E2 (P0). The −2,419/−179 promoter region (segments I to IV) and sequential 3′-deletions were subcloned upstream of the SV40 promoter and analyzed for their hormone response in the absence of the BASE TSS in transient transfection assays ( Fig. 4B). Consistent with previous results, all constructs containing the 600-bp enhancer showed hormone responsiveness (P1, P2, and P3), indicating that the TSS is not essential for BASE gene regulation by E2. However, the enhancer alone was not sufficient to mediate the hormone response (P6). Only inclusion of segment I (P3) immediately upstream of the enhancer provided hormone sensitivity. Taken together with the results from Fig. 4A, the minimal sequence required for hormone response consists of segment II and either segment V (including the TSS) or segment I (66 bp directly upstream of the enhancer). Thus, the results point toward a two-component regulation, involving an interaction between the enhancer (II) and either the TSS (V) or a small region immediately upstream of the enhancer (I).
ERα is present at the BASE promoter but direct binding is not required for repression. ERα is essential for BASE repression by estrogen and all three regions of the BASE promoter involved in mediating hormone response contain bioinformatically predicted estrogen-responsive elements (ERE; Fig. 4, E9 and P3).
To investigate whether ERα can bind to the BASE promoter, ChIP analysis was carried out using MCF7 cells cultured in complete medium ( Fig. 5A ). ERα and FoxA1 were found to be present at the enhancer and the TSS. The absence of polymerase II from the TSS is consistent with BASE repression in the presence of E2.
Furthermore, ChIP experiments were performed in full medium in the absence or presence of ICI182,780 and in stripped medium in the absence and presence of E2 (Supplementary Fig. S1). As expected, treatment with ICI182,780 resulted in increased association of PPol with the enhancer and at the TSS. No change was observed for ERα. In stripped medium, treatment with E2 for 4 h resulted in slight decrease of PPol. However, when sequential ChIPs were performed in 10-min intervals after treatment with E2, both ERα and PPol rapidly disappeared from the TSS ( Fig. 5A). In contrast, cyclical association of ERα and PPol was observed at the enhancer. Interestingly, the cycling time of ∼40 min has been reported before for the estrogen-induced gene pS2 ( 9). Furthermore, it is of note that the levels of ERα and PPol observed at the TSS were generally lower than at the enhancer.
To investigate whether ERα can bind directly to the BASE promoter, EMSAs were performed using in vitro–translated ERα and radiolabeled oligos (50 bp) containing either region C1, region C2, or TFF1 promoter sequence ( Fig. 5B). Binding of ERα to the TFF1 promoter containing the ERE, which functioned as positive control, was observed. In the BASE C1 region, which contains a predicted ERE, only weak ERα binding could be detected. To confirm the observed weak binding of ERα to the C1 region, a competition assay was carried out. Oligos containing a perfect ERE were 32P labeled and incubated with in vitro–translated ERα, whereas different cold oligos were used for competition ( Fig. 5B). Only the perfect ERE and the oligos covering region C1 + C2 competed for ERα binding, therefore confirming the ability of ERα to bind to the BASE promoter in the enhancer region.
In conclusion, ERα can bind to the BASE promoter in the enhancer region in vitro and in vivo. However, whether ERα binds only directly to the predicted EREs or whether it is also recruited through protein-protein interaction remains to be determined, as indirect binding of ERα cannot be excluded by these experiments.
The complexity of promoters allows the integration of different signal pathways, which makes assessment of single promoter features rather challenging. Consequently, in addition to the full promoter context, the mutational analysis was also performed on the enhancer region (II) and the TSS (V) where the intervening promoter sequences were not present ( Fig. 5C). This combination is still able to confer the hormone response [although to a lower extent ( Fig. 4A)], and association of ERα was observed for both sites in ChIP experiments ( Fig. 5A).
The most proximal ERE is located in segment V, directly flanking the TSS. By binding to this site, ERα could interfere with the assembly of an activation complex and thereby represses BASE expression. To test this hypothesis, the ERE was disrupted, whereas the ATG of the TSS was preserved. In both contexts (full-length and short), these mutations reduced expression to ∼50% compared with wild-type levels while not altering estrogen- and ICI182,780-dependent regulation ( Fig. 5C). Similar results were obtained when the ERE in the enhancer region was mutated. In conclusion, direct binding to the predicted EREs is not required for repression by E2 and derepression by ICI182,780.
BASE expression in different cell lines. In the single literature report, BASE expression is reported to be limited to salivary glands ( 2) as the only nonpathologic tissue, whereas high expression was detected in different breast cancer cell lines and breast tumor samples.
To evaluate expression of BASE in various cell lines with regard to their ERα and FoxA1 status, cDNA was generated and analyzed for the presence of BASE transcripts using semiquantitative (end point) PCR. To verify the quality of the cDNA templates, separate control PCRs were performed for PPIA. Different breast cancer cell lines (ERα positive: MCF7, ZR75, and T47D; ERα negative: MDA-MB231 and SKBR3), a human cervical adenocarcinoma cell line (HeLa, ERα negative), and a hepatocellular carcinoma cell line (HepG2, ERα negative) were tested. In agreement with the previous study, BASE was undetectable in HeLa and HepG2 cells but detectable in all breast cancer cell lines except for MDA-MB231 ( Fig. 6A ). BASE expression did not correlate with ERα status. FoxA1 mRNA was detected in all cell lines but MDA-MB231. The presence of the FoxA1 protein was confirmed for the ERα-positive cell lines as well as for HeLa and HepG2 cells. For SKBR3, detection of FoxA1 protein was dependent on the antibody used (data not shown). In summary, the FoxA1 expression pattern supports the importance of FoxA1 as a key factor for BASE expression because FoxA1 protein was present in all breast cancer cell lines that express BASE. However, FoxA1 itself is not sufficient, as HepG2 and HeLa do not express BASE. It implies the requirement of additional factors that are cell specific.
After investigation of the endogenous gene expression levels of BASE, the luciferase reporter construct containing the 2.4-kb promoter fragment of BASE was assessed for its ability to drive expression in cell lines other than MCF7. This construct was sufficient for expression in the cell lines where endogenous BASE transcripts were detected ( Fig. 6B). No luciferase activity was detected in HeLa, HepG2, and MDA-MB231 cells stably expressing either full-length ERα66 (MDA66), the short isoform ERα46 (MDA46), or a DNA binding mutant ERα66mut (MDA66mut). Cotransfection of both full-length ERα and a DNA binding mutant into SKBR3 induced hormone response, again indicating that presence of ERα but not direct DNA binding is required for BASE repression by E2.
Breast tumor detection and characterization at an early stage increases the 5-year survival rate dramatically ( 12). Therefore, much attention has been devoted to the identification of markers and expression signatures indicating malignant phenotype, metastatic potential, tumor growth, and therapy prognosis ( 1). The recently identified BASE gene, which encodes a putative secreted protein, shows an expression pattern restricted to breast cancer cells and salivary gland. It therefore has the potential to serve as a breast cancer marker ( 2). This study aimed to understand the mechanisms underlying BASE expression and regulation in breast cancer cells.
This article identifies BASE as a highly estrogen-repressed gene whose expression and regulation can be separated: with gene expression being dependent on FoxA1 and estrogen-mediated repression requiring ERα.
Exclusion of FoxA1 from the BASE promoter either by siRNA against FoxA1 or by mutation of the FoxA1 binding site greatly decreased transcription of the gene. FoxA1 is a pioneer factor that can bind compact chromatin and initiate chromatin opening events ( 13), which has been shown to support ERα-mediated transcription. Thus, FoxA1 might allow BASE expression by increasing the accessibility of the chromatin and thereby facilitating binding of other transcription factors, including a breast cancer–specific and salivary gland–specific factor that is essential for BASE expression. Alternatively, FoxA1 might displace a repressor.
Repression of BASE in response to E2 requires ERα. Absence of ERα, either due to gene silencing, treatment with the antiestrogen ICI182,780, or siRNA directed against ERα, abrogates repression in the presence of hormone. However, as revealed by mutation studies, direct interaction with DNA is not crucial. Nevertheless, ERα can bind to the enhancer region in vitro and is present in the same region in vivo. It is conceivable that ERα is recruited via protein-protein interactions, which are then stabilized by a direct binding to the ERE. The very rapid repression of BASE in the presence of estrogen argues against the requirement of de novo protein synthesis. However, activation or inactivation through posttranslational modifications cannot be excluded.
The clear separation of gene expression and hormone-dependent regulation was a major finding of this study. Although also expressed in a subset of ERα-negative tumors ( 14) and cell lines, FoxA1 expression is often associated with expression of ERα ( 15, 16). Furthermore, ∼50% of ERα-binding sites are accompanied by forkhead factor binding sites in close vicinity ( 17, 18), and this applies for estrogen-stimulated and estrogen-repressed ERα target genes. In this subset of genes, FoxA1 was required for recruitment of ERα and gene expression in MCF7 cells ( 19).
Although essential for BASE expression, the presence of FoxA1 in the BASE-negative cell lines HeLa and HepG2 points toward the requirement of breast cancer–specific factors for BASE expression. Contradictory reports also exist about FoxA1 regulation by estrogen. Whereas some groups report estrogen-induced down-regulation of FoxA1 ( 20), Laganiere et al. ( 19) observed up-regulation of FoxA1 within 4 h after estrogen treatment. In our experiments, FoxA1 mRNA levels remained unchanged in response to estrogen exposure (data not shown). Thus, reduced FoxA1 levels are not the main reason for down-regulation of BASE expression in the presence of estrogen.
ERα is likely to cooperate with other factors or might even regulate BASE indirectly. Three regions in the promoter were important for hormone response: the essential enhancer (region II) and either a small region immediately upstream of the enhancer (region I) or an ∼200-bp segment including the TSS (region V). Bioinformatic analysis predicted binding sites for INSM1, also called zinc finger protein insulinoma-associated 1, in all three regions. INSM1 protein was detected in all cell lines tested except for SKBR3 (Supplementary Fig. S2) where BASE is expressed but not regulated by estrogen. When BASE is expressed, INSM1 could contribute to estrogen-mediated repression. It is of note that the predicted INSM1 binding site slightly overlaps with the confirmed ERE in the enhancer region of the BASE promoter but all highly conserved bases were maintained when the ERE was mutated. This could explain the retention of regulation by E2 even if the ERE was disrupted.
INSM1 has been shown to repress the neuroD/β2 gene in conjunction with cyclin D1, a well-known ERα target, by recruiting histone deacetylases ( 21). Preliminary ChIP experiments show the presence of INSM1 at the BASE promoter in MCF7 cells (Supplementary Fig. S2). There are further indications from the literature that might argue in favor of this model. Cyclin D1 is a well-known estrogen-induced gene that is overexpressed in ∼50% of breast cancer tumors ( 22). Interestingly, BASE expression is also detected in ∼50% of the tumors tested. It remains to be determined if a correlation (positive or negative) exists between expression of cyclin D1 and BASE. An altered expression in some types of cancer has also been reported for INSM1. For example, it is reexpressed in neuroendocrine tumors ( 23) and its expression is increased in small cell lung cancer ( 24). However, no link between INSM1 expression and breast cancer tumors has been reported thus far.
BASE is a predicted member of the PLUNC gene family, which is expressed in the upper airways ( 4). However, in a commercially available tissue array, no BASE expression was observed in lung or any other normal tissue except for salivary gland ( 2). A single study on BASE expression in breast cancer samples ( 2) and the results obtained in this study indicated that ∼50% of breast tumors are positive for BASE, whereas normal breast samples are negative. This expression rate is higher than for the accepted breast cancer marker HER2, which is overexpressed in only 30% of breast and ovarian cancers ( 25). Based on the results obtained in this study, one can speculated that BASE expression may correlate with FoxA1 status. In addition, BASE was found to be more frequently expressed in ERα-positive tumors. This agrees with the positive correlation between ERα and FoxA1 expression ( 26). Nevertheless, the changes resulting in the transition from a normal to a malignant cell subsequently inducing BASE expression remain to be determined. A link between salivary gland, or more precisely salivary gland tumors, and breast cancer has been reported. Female patients with a salivary gland tumor have a 2.5 times increased risk of developing breast cancer ( 27). However, reports in the literature are contradictory, ranging from no increased risk to up to 8-fold increased risk ( 28, 29). The expression of BASE in salivary gland tissue does not undermine its potential use as a breast cancer marker because salivary gland tumors can be easily distinguished.
The potential diagnostic relevance of BASE is supported by the predicted secretion of the protein, as a correlation between the levels of the breast cancer marker HER2/neu (c-ERBB-2) in saliva and nipple aspirates and breast cancer in women has been shown already ( 29, 30). Detection of breast cancer markers in saliva or nipple aspirates opens the possibility of a noninvasive and inexpensive diagnostic tool for early detection of breast tumors and improved treatment response.
Grant support: 6th European Community Framework Programme grant CRESCENDO (FP6-018652), European Molecular Biology Laboratory, and European Molecular Biology Organization.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Brenda Stride for critical review of this manuscript, Prof. B.F.C. Clark for providing fibroblast cell lines, and David Coyle for his technical contribution to the human breast cancer study.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received September 26, 2007.
- Revision received October 29, 2007.
- Accepted November 2, 2007.
- ©2008 American Association for Cancer Research.